SAP Aktie

Great. Good morning, everyone. Thank you for joining us. We are delighted to have the CEO of SAP, Christian Klein, join us today at the conference. Christian, thank you for joining us.

Yes. Thanks for having me, Mo.

That's great. Before we start, I just need to read a safe harbor statement, so if you could bear with us. SAP will make forward-looking statements, which are predictions, projections or other statements about future events. These statements are based on current expectations, forecasts and assumptions that are subject to risks and uncertainties that could cause actual results and outcomes to materially differ. Additional information regarding these risks and uncertainties may be found in SAP's filings with the Securities and Exchange Commission, including, but not limited to, the Risk Factors section of SAP's 2025 annual report or Form 20-F.

So with those sort of niceties out of the way, Christian, maybe to start with, can you give us a sense of the conversations you're having with your customers? You come off the back of what were very strong Q2 results, at least certainly, given where fears were in the marketplace. What are the kind of key priorities in the context of both the geopolitical uncertainty, but all the structural disruption around AI that you're seeing? And maybe your sort of perspective on kind of the discussions of AI deployment in the boardroom and particularly how do they think about before they commit to spend those returns?

Yes. That's a lot in this question. So let's start with the geopolitical situation. I mean, throughout the whole year, we were always mentioning the geopolitical conflict, especially in the Middle East. I mean, now we are so much into the year that I can say, at least for SAP, we don't see now an impact at a global scale. Obviously, the Middle East is not a small business for us. So it's a very well-run business for SAP, but definitely, it will not have an impact now on the cloud revenue anymore. Now on CCB for Q4, obviously, it's still a sizable business, but that's why it could have a certain swing, but it's not something what we are heavily concerned about anymore.

Now the Board level, the C-level conversations, and I can also shed some light on the conversations we're having with our ecosystem and partners where you also hear a lot about what they are hearing from the market and the customers. I mean, when we are going to send our customers to the cloud, I mean, 12 months ago, 24 months ago, it was more about, okay, Christian, let's get it done, automate my business processes, let's get rid of the custom code and show me the outcome of the RISE journey. So how you're going to make it work together with us.

Now the C-level conversation is, Christian, I can't wait another 12 months for these business processes to be automated or infused with intelligence via AI. So AI is now in every conversation, yes. Even if the customers in the C-level doesn't have really a concrete idea, they just want to see AI connected to it. And that's why when we are now doing this, it's very important for us to really look into what -- first of all, what business use cases have the best ROI. Customers also are asking more and more for, okay, the TCO token -- because they see the token spend running away. And then the last piece, which I feel is a very good sign for all the announcements we did at Sapphire is that customers who even build some agents with open-weight models on third-party platforms, they are coming saying, "Christian, I'm honestly completely overwhelmed my IT organization by making sure that these agents deliver the right outcome." Accuracy is a big problem oftentimes and the governance of these agents. I mean, you have so many compliance requirements, auditability of a financial close, payroll needs to run 100% perfect. So a lot of customers are saying, "Hey, SAP, can you not do this for us before my IT teams are dying a lot of complexity on running and managing those agents."

So obviously, we're going to delve into AI because there's a lot to talk about with kind of upcoming releases. But maybe just touch on the kind of Wise journey and the kind of S/4HANA migration. Can you kind of tell us kind of where you are in that? Because I think CCB is still growing quite healthily. And how much of that journey is there? And there were some fears, I think, earlier in the year around potential slowing of that migration, but Q2 sort of told us that perhaps that's in decent shape. So I would love to get your perspective on that.

Yes. Being here in San Francisco, it's also -- memories are coming back at the time when I was the CFO of SuccessFactors. And right after the acquisition, we didn't have so many good times also with some other acquisitions, not integrated, sitting on a stack which was not really scalable, not performing. I mean that has changed a lot. Looking at the CCB growth of 26% in Q2, it's not only S/4. All LOBs are winning market share. And I'm so happy that when we did this cloud move, a lot of customers were really quite concerned, "Oh, how do we do this? Are the products competitive, et cetera." In the meantime, we are winning market share in all LOBs, which was definitely not the case in 2 or 3 years.

And also the channel is performing even better than our direct sales, which I always find a good indicator that we are not only converting installed base to the cloud, but that also we are winning a lot of new family members with GROW with SAP. Also here in the Valley, a lot of start-ups who want to scale their business, say, okay, SAP has localization for over 100 countries. The platform works, public cloud end-to-end, why not SAP. And also some private investors are saying, "Okay, let's really close a framework contract for SAP because our start-ups should just use the best platform to scale their business." So that's good.

Now going forward, we talk a lot about AI. But in my financial plan for the next years, I mean, still there's 60% of our installed base is not yet in the cloud. So there's stuff to come. Work to be done. Just this morning, I talked to Julie Sweet from Accenture. And she asked me, Christian, I have so many SAP consultants. How long will this party continue? And I said this party will continue, but in a different way and form also for the existing -- for the customers who are now moving to the cloud because they want to be moved to the cloud with AI tooling, with the AI use cases, which are then showing them the value and not just do a technical migration. So there is enough potential also on the cloud side.

The maintenance is coming to an end, standard maintenance. 2030 is then the end of the extended maintenance. And again, 3 years sometimes for such an ERP modernization is not a long time. But again, it's not even the dominating factor for the next years, why to move to the cloud. Even some customers who moved before my tenure as a CEO to S/4 on-prem, they see now the value with AI and to make the move to the cloud. So yes, there is enough business left to be converted. And yes, I'm not seeing that our subscription business will decline. It's just the opposite. I still see very healthy growth.

Maybe one last piece, which is maybe also important for the audience to understand where now Lockheed Martin, BMW and others are coming and say, Christian, we need more features in our S/4 system, I said features. I would love to talk about agents. So how can we convert that? And even there, we are now seeing that -- we see a lot of growth also in the subscription business by adding features, but we are not only adding features for spare parts management, we are telling the customer, okay, we can add a feature, but only if there is some data missing to the system of record. And then we want to build an Agentic AI layer right away on top. And so you see both the subscription business and as well as, of course, the consumption-related AI business. We'll see growth in the future, I'm sure about that.

Got it. So maybe let's turn to kind of AI. You laid the road map out at Sapphire earlier this year in May. And obviously, you've got the autonomous suite, the SAP Business AI platform. Maybe for the benefit of the audience, kind of help us kind of understand that sort of vision, what you sort of see in terms of the role SAP plays, particularly as we move in the LOB solutions to more kind of Agentic workflows.

Yes, happy to do so. And I saw also some comments about is SAP maybe a little bit late on shipping all of these assets now. I don't believe that we are late. We want to get it right. And as I mentioned also before at earnings, there was a learning curve. It's not like that we understood everything perfect right away from the beginning when we started to build the first AI use cases 2 years ago.

Now what is different with the announcements we did this year? First, I told my product managers, which are all gathered together in our all-in-one AI workshop here in Palo Alto. I told them, "Hey, when you believe we are developing an agent for document scanning and we are going to win with this agent, you are completely wrong." Our ERP wants the most complex business processes in the world, and we need to solve the most complex challenges of our customers with our agents. So all the agents we are shipping, 400, 200 now in September, will solve complex business challenges, not the low-hanging fruit, which you can easily do with any LLM you pick from the market, first.

Second, lessons learned was accuracy. Data, ontology, BDC is super important. I just saw Sridhar here on the floors, Snowflake, Ali, Databricks. We don't want to give away our crown jewels, our semantics, but we want to make sure that our agents also understand non-SAP data. That the ontology layer, the semantic layer, you can build it with Palantir, but with SAP, you get a lot of that out of the box. We have 7 million triples in the ERP system. And this 7 million triples need to be understand. I mean the data correlations need to be understand via knowledge graph, via the semantic layer, which sits on our new AI platform. So that is the first element.

And then the second element, what we really saw in our learning curve is what I mentioned before. A lot of customers came to us and said, Christian, even for this simple AI use case, identity management, authorization, has the agent now read access, change access or write access. Compliance requirements, FedRAMP here in the U.S. In Germany, we are not short of bureaucracy as well. So there are a lot of requirements as well, which is understood by our ERP, but not yet understood if a customer tries to run those agents on their own.

So with the new platform, we are going to manage every agent which is built on this platform, either if it's built by SAP, built by our partners or built by a customer, which is a huge differentiator next to the contact player because we are saying no matter where this agent needs to operate, in which industry, in which country, we ensure that it's governed and compliant 100%. And then last but not least, some of our peers announced great partnerships with one LLM provider. We believe that with Joule Studio on the new platform, we want to be model agnostic because we don't see that one model always delivers the best price outcome. So we are checking our agents with at least 6, 7 different models before we go GA. And we don't see that only one model is winning. So this being agnostic, switching the models, delivering the best price outcome is also very key for our customers. So it's the context, it's the governance which comes via SAP. It's the extensibility. You can build your own agents, partner can build agents, but it's all provided the context of SAP and it's governed by SAP.

Got it. So I think you said you've got a kind of important launch coming up in the fall this year, which [indiscernible]. And I think there's a kind of wave of additional kind of use cases coming. Maybe talk us about the early feedback you received from some of the pilot customers around these products and your expectation of product rollout.

Yes. I mean, first, just on the commentary on why is SAP may be a little bit late with the new platform. We could have shipped the platform already at the beginning of the year, no problem. But then again, every SAP customer would need to manage the agents by their own, which is a lot of feedback again coming where say we are dying in this complexity, we can't do it. Please do it. And now no matter if it's a Concur and S/4, logistics, supply chain, CX, sales, all A2A scenarios are managed by SAP. And I told my product managers, we are not going GA as long as we cannot prove that we can manage the whole life cycle.

Now what is coming in fall is the new platform. With that comes Joule Work. [indiscernible] announced their own engagement layer. I don't want to be headless as SAP. So we have Joule Work. You look at Joule Work as you are used to Fury or the good old SAP GUI. First of all, Joule Work is beautiful. We have 150 customers. They love the auditability, the traceability. They love the contact Joule Work has. And they are even saying, why should I use even for the non-SAP users? And I really want to emphasize that. ALDI, a German retailer is saying, why would I not use Joule Work for all of my people at ALDI because you can automate a power -- you can create a PowerPoint, you can automate an Excel report, but you're always getting the SAP context, the SAP data.

And wait a second, you are checking authorizations, data access, you are managing these A2A scenarios underneath Joule Work. Why is Joule Work not becoming my one engagement layer for the company? In SAP, we have 100,000 people, a bit more. 80,000 are using Joule Work. Feedback is great. And so we are confident in what is coming in fall. So it's the platform, it's Joule Work. It's 400 agents. And again, this morning, I said to Julie, move all your SAP consultants a bit away from typical ERP migrations, use the tooling for migration, AI tooling, but also let them build agents on top of the new platform because extensibility will be key in such complex scenarios, what we are going to deliver. We have seen it in the ERP. We have a lot of customizations and also the agents. They need partners who are going to extend it, and we need partners building also their own agents on top of our platform.

Yes. Okay. That's great. So I wanted to come back to the sort of discussion on the moat. And obviously, one of the kind of underappreciated aspects is the contextual layer, the knowledge graph and hence, the decision to sort of use a third-party LLM system versus we're going with SAP. Can you expand upon that kind of richness of that sort of data moat? Because you also across multiple domains, the ability to bring in now non-SAP data. And so ultimately, in that conversation with the customer of why kind of Joule Work works over kind of building your own DIY type of solution.

I mean, obviously, when the concerns were coming up, so visible in our share price now, we are doing better and confident that this will continue. But obviously, I asked my portfolio and my strategy team immediately said, okay, which app is really up for disruption. Which app can be easily replaced by any LLM. The answer was we have 2 solutions in our portfolio where I would say the moat is not really great. Concur bookings, which is, God, thanks, a very tiny business of our overall Concur business. Concur is really used often for travel expense, recording expenses. So a lot of regulations, et cetera, not easily replaceable and not easy to disrupt. But the bookings part, yes, this is where we now build a partnership with Amex and others.

And then the learning business in the success factor because to create a personalized learning, [indiscernible] can do that. Now even for those 2 apps, which is only a minor part of our portfolio, for learning, I said, hey, the SuccessFactors team, we need to provide an A2A scenario where we connect learning to the total workforce management AI scenario we are building. What skills do you have today? What skills do you want to need to have tomorrow? How is the recruiting agent connecting to that, the learning agent and so on to safeguard this business because stand-alone, no chance. And then obviously, the moat for all the other solutions. I mean, when we build the integration between S/4 and all of our acquired entities, I mean, I learned myself how deep the data model is, how deep also the process complexity is between a logistical solution and finance solution. And so that is -- for us, we need to -- the teams who are building that this ontology layer on our new platform are very experienced domain experts who once build our payroll solution, logistical solution, coupled with data architects, data engineers who really build this together. And for us, this is doable. But I don't believe that this is something what you can do with people who are checking today LLMs and is an LLM accurate, yes or no. For me, these are 2 different type of people, type of profiles, type of skills. So I can sleep well enough.

Again, we need to get our acts together with the new platform. We need to prove that our agents can work more accurate. The Joule Work is way more accurate. We need to prove that the governance is managed by SAP. And if we can prove that, I mean, we will use all of these LLMs, not one, so they will be partners, but I see not a scenario where I would now say, "Oh my God, this could disrupt my ERP business or the overall portfolio of SAP." It's 1 plus 1 is 3 on our platform and not 1 plus 1 is 2.

And I just want to touch one other product area. Business Data Cloud has been out for kind of 18 months. It's somewhat been sort of forgotten a little bit. But as we now roll out kind of the Agentic workflows and the kind of data consumption drives, where are we in terms of kind of the adoption, the kind of customer feedback? You've obviously signed up now a very comprehensive list of partners as well.

And also, I can answer this question also giving a little bit of context to the acquisitions we did in the data space. BDC, obviously, is fundamental for our new AI platform. Why? We see even that for our billing assistance, sometimes you need to check a ticket in a ServiceNow system to see, okay, why is the customer not paying, there is an escalation, et cetera, et cetera. So extensibility to non-SAP data sources is key. And if a customer now has Databricks or Snowflake, et cetera, they love that they have 0 copy. They don't need to change now the data layer. They can just use BDC, and we build a semantic layer as part of the BDC data platform.

With Reltio, then we can also govern the data, now also non-SAP data because data quality is key so that the agents really understand, okay, the whole context of the business and the task they need to fulfill. And so these partnerships are great. Now with Dremio, we did it because take, for example, H&M, we built a shopping assistant. They want to have social media data to not only do the same non-intelligent recommendations on certain products they have in stock. They want to check the sentiment on social media. They want to check the sentiment coming from other stores, et cetera, et cetera. So they said, no, SAP, I have this data nowhere now. I want to store it, but please give me this AI scenario end-to-end SAP. So we infuse Dremio if a customer really says, okay, I want to be agnostic. I don't want to use one of the other data lake providers. So we can do that. And again, with Reltio, I already explained the data governance part.

So yes, BDC is actually a core part of the new platform of the contact layer. And in every AI scenario, we are now going to launch and which we are already implementing, we see, of course, there is a great pull-through now also to the data platform because you need sometimes more data products, sometimes you need a better governance of the data. Sometimes you're missing certain parts of non-SAP data, okay, give me Dremio. It has great TCO. It's built on Apache. So it's all good. And that's why BDC is for us connected to our AI platform and we'll see, of course, a huge upside potential now just by building more and more agents on the new platform.

Got it. So you're obviously on track now on the product rollout. You've got this target of sort of 30% consumption revenue by the end of the decade. Can you maybe give us a sense of now is we're clearly going to be in an execution phase? What gives you that confidence around kind of hitting that target? And do you need to do anything in your go-to-market in terms of sort of getting there?

Yes, we are on track. I mean I don't want to downplay this. After the cloud transformation, this is a transformation which is equally big for our workforce. I mean in product management, I told the people, I said, look at the standard AI use cases, what we shipped at the very beginning, everyone can do it. Go out, find the domain experts on the customer side and design agents in a week, prototype them with AI and come back and build real AI agents who are solving real problems.

And sometimes you are the victim of your own success. When you have a backlog on features, which is enough for the next 3 years, why would you talk anymore to the domain experts? So some of our product managers needed to learn how to go back to the customer and really design our agents, which are really solving complex business problems. And again, the 400 we are going to ship will make a big difference in that. Accuracy, ontology building, accuracy, not shipping features anymore, but really making sure that the agents we are going to ship have the right accuracy, big change. I mean data scientists play suddenly a much bigger role than in the past and even bigger role than in the past.

And now on the go-to-market side, think about all of the customers who already moved to the cloud. I mean these account teams did sell us will wake up next year and going to see, oh my God, the bookings, the app bookings are gone. I see a huge quota now on data and AI consumption, consumed ACV. We are moving them to an even more consumption-related business model. And so for them, it will be also a change where we are now obviously preparing them also already to do it. The product, the platform is there, we see now it's working, and now we are going to do the shift on the go-to-market side as well.

And when you have seen our service revenue, maybe a slight decline, I mean the reason why is I don't want to tell our consultants anymore as to do ERP migrations, let's give this to our partners. I want to use our consultants to, first of all, making sure we are extending our agents so to really solve even bigger problems first. And second, to use our data architects, our data engineers to drive the adoption on the platform much better than to support any ERP migration project out there because, again, this is something what the partners can do. So also the consultants next to our customer success managers, they will be pulled much more into this AI project to really make sure that the adoption is there, the value is there and that we can monetize, obviously, all the stuff what we are going to sell in our AI portfolio.

So I wanted to touch briefly on gross margins. There's sort of the fear out there that even if sort of consumption revenues grow, this will be negative for gross margin. But it's interesting that you're taking this very agnostic approach, particularly with regards to some of the LLM providers. But you also emphasize the point about the more complex business cases. So give us a sense of kind of how you think that gross margin evolves, both short and sort of medium term?

Yes. I mean I'm -- sometimes it's good to have an own opinion, an own vision, an own strategy and not just follow what everyone tells you. There were people in my Supervisory Board who told me 3 years ago, let's bet the whole company on OpenAI. Then everyone told me let's bet the whole company on Anthropic. And now we are so happy that we have a model switching where we can switch the model within a day. And we have agents who have seen, in the meantime, 5 different models because we see that not one model always delivers the best outcome. And it can change with the next generation of models, maybe not the one model is leading, the other model is leading. So switching models is very, very important also for the gross margin because we have seen, of course, at the beginning when we delivered the first agents, the gross margin didn't really look good. But then, of course, we have proven the value now on the new platform. The customers are really excited about what they see. The billing assistant works super smart. Cash flow, we have the first supply chain agents now for the demand and the supply chain optimization, the inventory optimization. So that works.

And now, of course, what we are seeing is that over time, we can heavily improve the TCO to also shift some of these agents to some open-weight models because, again, they deliver sometimes as good outcomes as some of the frontier modules. And then last but not least, if we come to our new engagement layer, Joule Work, I mean, sometimes customers are also telling us from a [indiscernible] perspective, let's use Mistral in Europe. I want to use -- and again, the accuracy is good enough to also do an HR report, create an HR report with SuccessFactors data with Mistral. So absolutely, let's shift to Mistral. We can cut the cost by 5 compared to some other frontier modules, and we deliver still a good enough outcome for the customer. So yes, so that's why all of these concerns will have this shift a negative impact on the gross margin. I don't see it. I mean, obviously, when you land an agent, you should now not be overly aggressive on price. You should first prove the value. But then with the model switching we have in place and that we see the new platform is really providing the value, providing strong contact, we manage the customers -- we manage the agents for the customers. I mean, we definitely -- I have no concerns that the gross margin will see now not a healthy development in the future.

And maybe just touching upon how you are using AI efficiencies internally. You've outlined the sort of EUR 2 billion of efficiencies over the medium term. Maybe talk us through how you've been managing internally, not just cost of sales, but how much potential and runway there is in the company to kind of extract more efficiencies.

Maybe we have also the -- we have done a little bit of different approach than some of our peers in the industry. You have now not seen a big restructuring from SAP in the last 12 months because I told the team, I mean, first of all, we are not running out of development ask from our customers, co-innovation ask. So despite we have seen a huge increase in development productivity, of course, there's a lot of stuff to deliver, first and foremost, all the agents what we promised.

Now still, we also needed to make sure that we could accelerate the delivery of code a lot, 30%, 35% increase. Now still, you need to check the code. You need to check is all data privacy requirements fulfilled, cyber, et cetera. And we need to make sure that we cannot only generate much more code, but that we can also ship it to the customer way more faster. And that's a good example where I said, let's get first the homework done, and we are on a very good track to really get the productivity not only on coding, but really getting the code out to the customer faster, which we are now doing. And then also roll out Joule Work inside SAP. We are seeing now productivity gains in finance, in HR and other functions of up to 20%, which is great.

And now we are also, of course, having the discussion on, okay, what do we do with this higher productivity? We need to transform our workforce. And with that, you can also expect that by the end of next year, you're not going to see the same kind of SAP workforce like what we have today just based on the productivity gains, what we are seeing.

Got it. So maybe in closing, with every sort of technology cycle, there may be a commoditization of certain processes, but it opens up new frontiers in terms of -- from sitting in your seat where your software kind of powers a lot of the enterprises of the world. What are some of those sort of use cases or workflows that you see that kind of excite you and when you speak to your customers that kind of drive that leg of growth?

I mean just look at myself when I prepared for this conference here, I asked Joule Work, okay, what are the analyst estimates? What are the concerns? What is the sentiment in our investor base and connect it to my forecast for half year 2. And I, by myself, created all of this stuff in the past. And now I have in 1 minute, a rock solid analysis about what my forecast says, what my pipeline says, what the expectations are and how confident, how bullish can I be for this conference. And then -- and that is, of course, something where you needed an amount of people in Alexandra's organization in controlling and FP&A to do that. And that excites me. I mean the hours I spent in my life to pull this stuff out of SAP is enormous.

Or when an Infineon CEO says to me, Christian, this order management agent is -- it's so great because it has so much intelligence on, I have not a demand problem at Infineon, I have more a supply problem. And now this agent makes sure that in thousands of orders coming in every day, how to route it to which customer who gives me the best margin. And it's unbelievable the kind of intelligence what we're going to see. And then the last example, even if the German user group, imagine that they are so critical and Germans tend to be more critical. But the user group, they are testing now Joule Work and they are saying, Christian, this is unbelievable. The user experience, the results, what we are getting out of that. We have -- they even did -- one user group member did a post on LinkedIn and they say the results are incredible. And you see the potential. And now with the platform, with the agents we are going to ship, we are at the beginning. But I see there is a lot of potential for our customers. And obviously, when you then multiply that with the installed base, what we are having, it's, of course, also for us an exciting time.

Yes. That's great. I think we're up on time. Christian, thank you so much.

Thanks for having me.

Thank you everyone, for joining us.

Thank you.

Ladies and gentlemen, thank you for standing by. Welcome, and thank you for joining the SAP Q2 and Half Year 2026 Financial Results Conference Call. [Operator Instructions]

I would now like to turn the conference over to Alexandra Steiger, Global Head of Investor Relations. Please go ahead.

Good evening, everyone, and welcome. Thank you for joining us. With me today are CEO, Christian Klein; and CFO, Dominik Asam. On this call, we will discuss SAP's second quarter 2026 results. You can find the deck supplementing this call as well as our quarterly statement on our Investor Relations website.

During this call, we will make forward-looking statements, which are predictions, projections or other statements about future events. These statements are based on current expectations and assumptions that are subject to risks and uncertainties that could cause actual results and outcomes to differ materially. Additional information regarding these risks and uncertainties may be found in our filings with the SEC, including, but not limited to, the Risk Factors section of our annual report on Form 20-F for 2025. Unless otherwise stated, all numbers on this call are non-IFRS and growth rates and percentage point changes are non-IFRS year-on-year in constant currencies. The non-IFRS financial measures we provide should not be considered as a substitute for or superior to the measures of financial performance prepared in accordance with IFRS.

And with that, over to you, Christian.

Yes. Thank you, Alexandra, and a warm welcome to everyone joining this call. Q2 was an outstanding quarter, highlighted by our flagship customer conference, Sapphire. The event was a huge success. We saw record attendance generated significant additional pipeline and received very, very positive feedback on our Autonomous Enterprise launch. In parallel, our AI transformation is progressing well, both among our customers and our employees reflected in growing AI consumption and tangible outcomes. This momentum contributed to our strong top line performance in Q2.

Let's look at this in detail. Current cloud backlog grew 26%, an acceleration compared to Q1. And after two quarters where our CCB was lagging behind cloud revenue growth, it was a welcome trend reversal in this important forward-looking indicator, a great result, especially given the volatile environment.

AI and SAP Business Data Cloud were embedded as key pillars in more than 90% of our 50 largest deals, giving us strong confidence for the second half of the year. Cloud revenue grew 24% to EUR 6.3 billion, backed by a solid execution of on-premise to cloud ERP migrations in our installed base.

Total revenue was up 11% to EUR 9.9 billion. This comes despite a slight decline in services revenue as we reallocate consultants to build agents and foster AI adoption. Our top line performance translated into an operating profit of EUR 2.7 billion, an increase of 9%. Our indirect channel continues to be a strong growth pillar. Q2 growth again significantly outpaced direct channel cloud revenue, reflecting our successful go-to-market transformation over the last two years.

As mentioned, a key highlight in Q2 was, of course, SAP Sapphire and the launch of the Autonomous Enterprise. It resonated strongly because it gets to the root of why enterprise AI is so hard and what that means for our customers. The reality for many enterprises today is that LLMs don't understand business data, processes and governance. AI token spend doesn't mirror outcomes. Log-in to single frontier vendors is a growing concern and AI sovereignty is becoming more important. At Sapphire, we explained how the combination of the Autonomous Suite and Business AI platform will solve those challenges.

The new business AI platform is being delivered this quarter. The platform covers three key pillars. First, the build pillar, where we offer the best experience for pro code and citizen developers to create and extend agents. With Joule Studio, you can choose from a range of leading LLMs, including Anthropic, Cohere, Google, Mistral AI, OpenAI and other open weight models.

Joule Studio is integrated with the second pillar, the context and reason pillar. This pillar provides the agents with the context and the knowledge they need to run business processes autonomously. It starts with our data foundation. SAP Business Data Cloud provides broad data access to the agents. And with our latest acquisition of Dremio and its Apache Iceberg native technology, we are bringing mission-critical SAP and non-SAP data together to become a true enterprise lighthouse, meaning SAP and non-SAP data can be analyzed together in real time without moving or copying it first.

On top of the data foundation, we are able to build one semantic data layer for SAP and non-SAP data. By joining data products, we are creating semantic data models centered around the customer, supplier, material and other master data objects of a company. Our acquisition Reltio will govern these master data models end-to-end to ensure high data quality. And then these semantic models are connected with our ontology layer and knowledge graph, which we offer for every LOB and industry domain. They are the plane infused into the agents developed with Joule Studio on the new platform. Most importantly, all three layers will be extensible by partners and customers to cater for their specific business needs. Our acquisition of Prior Labs will enable agents to generate accurate tabular predictions out of the box.

After providing the agents with the context, they also need to run the business with trust and high scalability, which brings me to the third pillar, one and govern. This pillar addresses another key challenge of AI adoption. Here, we take the complexity of our customers' shoulders by managing and governing the agents embedded in the Autonomous suite and beyond.

SAP manages the agents across the complete agent life cycle, including SAP, partner and customer build agents. Specifically in the governance layer, we will ensure the agents meet your compliance frameworks and data privacy requirements from over 130 countries, checking all the identity and authorization rules to ensure the response is not only accurate, but also compliant. Furthermore, we are able to switch between different models safely and dynamically inside customers' SAP landscapes. This means we will run the agents without any lock-in, adhere to local sovereignty requirements and ensure the best price-to-outcome ratio. Powered by the new platform, our Autonomous suite will consist of SAP partner and customer agents, all managed by SAP.

The AI Agent Hub is our command center to discover, manage and govern SAP and non-SAP agents, MCP servers and more. The AI Agent Hub gives customers transparency across a universe of agents for every LOB and every industry. And finally, this quarter, we will also launch our new end-to-end user experience, Joule Work. It's a single entry point and interface across all our portfolio solutions for all tasks where users can collaborate with our AI agents. Connected to our business AI platform, Joule Work dramatically accelerates outcomes for our 350 million end users. For example, a salesperson can create a complete data rich customer pitch in just a few minutes. A finance business partner can pull together a financial analysis, including all structured and nonstructured data from his or her company.

After Sapphire, the beta programs for our new platform, Suite and Joule Work were immediately oversubscribed and initial customer feedback has been excellent. This makes us very confident about the successful launch in Q3. In addition, we will release close to 50 assistants by the end of Q3, underpinned by more than 400 Autonomous suite agents by the end of the year. To accelerate our customers' journey to the autonomous enterprise, we are also releasing three additional ERP migration assistants with 10 underlying agents later this quarter.

Let me now share some tangible outcomes from our customers. On the Autonomous Suite side, SAP and Amadeus, a platform for global travel, developed an AI agent that autonomously reconciles unstructured payment data already clearing around 40,000 incorrect transactions. One example from our Business AI platform. To prepare for business AI, Norsk Hydro transitioned from a legacy BW to an end-to-end data platform with BTP. This delivered significant agility, cutting BI solution build time by around 75% and accelerating report creation time by 50%.

Moving on to industry AI. With NTT DATA, Denmark's largest wholesaler for steel and technical equipment, Lemvigh-Muller deployed custom AI agents to verify purchasing orders. The solution achieved over 90% touchless processing and 98% matching accuracy. For AI agents to deliver the accurate outcomes at scale that all of these companies need a harmonized data foundation and simplified process layer is essential. That's why the modernization of legacy system landscapes is still very important.

To support our customers in this transformation, we launched our new RISE and GROW with SAP offering, which has already been very well received in Q2. As part of this new offering, we are seeing a strong uptake of our AI ERP migration tool chain. Customers are achieving faster time to value and up to 30% lower ERP migration costs. A great example of this in action is Dexco, who used their RISE with SAP migration to eliminate 97% of legacy customizations, driving a 75% faster accounting growth. In parallel, our new offering also includes a firm commitment to our customers to activate and adopt AI assistants and agents within the first year of their journey.

We also saw many Wise deal highlights in Q2. They include Shell, Morgan Stanley, Samsonite Group, Vonovia, Eli Lilly, retailer Shoprite Group and Electrolux. We also see great momentum around SAP GROW with companies such as Parloa, Gooroo Credito, Modular Data Centers and Techem Energy Services.

Turning to the AI deals. Key wins included PwC, one of the world's largest professional services firms. They selected our AI to transform a complex billing process, cutting a 35-minute task to just 5 minutes while improving accuracy and end user satisfaction. Travel platforms, Booking.com and GOL as well as Oki Electric Industry selected many of our LOB and industry AI offerings in addition to PC. Our Software and Cloud offerings also gained significant momentum with key wins, including companies like Airbus. Successful go-lives included Fonterra, Dohler and Natura Cosmeticos.

To deliver our AI vision, we also continue our own transformation. We are moving with full speed to turn SAP into an autonomous enterprise. In engineering and technology, we are transforming our operating model from software development to building AI at scale. We are doubling down on our oncology development with our best domain experts working on knowledge first for every industry and LOB. And we are accelerating innovation, targeting complete agent delivery in under three weeks. By applying tools like low-code, we're also increasing overall developer productivity by up to 30%. In go-to-market, we are also evolving our operating model. The consulting AI factory is a prime example. Over 3,000 SAP consultants are driving AI adoption directly with more than 2,000 customers. Over the next few weeks, we will roll out Joule Work desktop internally to drive additional productivity across all functions.

While we are driving significant efficiency gains with AI, we are making investments in our workforce, both by investing in world-class AI talent as well as up and reskilling at full speed. We are rolling out a range of code camps and in-person training offerings across our key locations with the target of reaching more than 90% of our employees over the next few months. In addition to upskilling our people, we are focusing our hiring efforts to bring in the industry's best data scientists and AI experts. Their leading skills will complement our deep business process and domain know-how.

Let me summarize. In Q2, we delivered a strong quarter with strong momentum in our business. In the age of agentic AI, SAP is leading the way. The autonomous enterprise is anchored in AI agents that can run end-to-end business processes accurately, compliantly and cost effectively and always with the human in the loop. SAP successfully completed our transformation to the cloud, and we will once again successfully transform in the AI era to deliver accelerated growth and profitability.

And with that, I'll hand over to Dominik.

Thank you very much, Christian, and thank you all for joining us this evening. To build on what Christian shared in his opening remarks, Q2 was a strong quarter for SAP, supported by sustained current backlog growth and further improving free cash flow generation. These results were delivered against the backdrop of a complex and uncertain operating environment with the ongoing conflict in the Middle East continuing to weigh on customer sentiment and decision-making. Despite these headwinds, we remain focused on executing our strategy and continue to see progress across our strategic priorities.

At Sapphire in May, we outlined our vision for the autonomous enterprise and the expanded role of AI across our portfolio. While these innovations are still at an early stage, we believe they can create new commercial opportunities over time and support durable growth beyond the current cloud transition. Importantly, we continue to invest in these areas while maintaining our commitment to the operating leverage framework we've laid out. These results reflect the resilience of our business model and give us confidence in the path ahead.

Now let me provide more details on our financial highlights. Current cloud backlog reached almost EUR 23 billion, up 26%, benefiting from the first-time inclusion of Reltio, which only contributed less than 1 percentage point to the constant currency growth rate. While CCB growth sequentially accelerated, we continue to expect a slight deceleration exiting the year.

As you are all aware, the situation in the Middle East remains fluid and the longer it persists, the more it weighs on customer decision-making, particularly in directly affected industries and supply chains. That said, the breadth of our pipeline, the mission-critical nature of our solutions and the fact that the second half of the year typically accounts for the lion's share of our bookings give us confidence in our ability to execute against these opportunities in front of us. As expected, the year-over-year cloud revenue growth rate declined sequentially to 24%, reflecting several quarter-specific effects that particularly benefited the 2026 year-on-year comparison in the preceding first quarter.

You might recall the comments we made in that regard in our last quarterly earnings call. Also recall that in Q2 2025, we had roughly 2 percentage points higher cloud revenue growth than in Q1. So there is a strong basis effect here. SaaS and PaaS combined continue to perform strongly with growth again far above the overall market. Cloud ERP suite revenue increased by 27% in Q2, now accounting for 88% of total cloud revenue. Software licenses revenue decreased by 32%. Finally, total revenue in the second quarter was EUR 9.9 billion, up 11%.

Now a brief look at our regional performance. In the second quarter, SAP's cloud revenue performance was particularly strong in APJ and EMEA and solid in the Americas region. Brazil, France, Germany, Italy, India, South Korea and Spain had outstanding performance, while Australia, Singapore and the U.S. were particularly strong.

Now moving down the income statement. Our IFRS cloud gross margin in Q2 was 74.3% and non-IFRS was 74.6%, down 0.7 percentage points year-over-year at constant currencies. IFRS operating profit increased by 8% to EUR 2.6 billion. Non-IFRS operating profit was up by 9% to EUR 2.7 billion. The quarter-on-quarter deceleration in IFRS and non-IFRS operating profit growth is mainly caused by lower cloud and total revenue growth in Q2 as compared to Q1 2026, an unusually low stock-based compensation expense in the first quarter of 2026, accelerated investments into research and development, higher marketing expenses in the quarter of the launch of the Autonomous Enterprise and finally, the slightly dilutive impact of the Reltio acquisition, which just closed on May 7. Additionally, we are investing to accelerate our own AI transformation. This includes rapid internal adoption of our AI solutions as well as targeted hiring in the most critical roles. These investments enable us to further enhance our products and drive efficiencies. At the same time, we're applying disciplined governance to manage our cost base and improve spend predictability.

The IFRS effective tax rate was 26.5% and the non-IFRS effective tax rate was 30.8%. The IFRS effective tax rate is lower than the non-IFRS effective tax rate due to tax benefits from tax-exempt income. Free cash flow in Q2 was robust at EUR 3 billion. Finally, IFRS earnings per share increased by 30% to EUR 1.89 and non-IFRS earnings per share increased by 6% to EUR 1.59.

Now on to the outlook. As you will have seen in the quarterly statement issued earlier today, we are maintaining our financial outlook for all top line parameters and free cash flow. We are adjusting our operating profit outlook by EUR 0.1 billion and now expect EUR 11.8 billion to EUR 11.2 billion to reflect the dilutive impact of the recent Dremio and Prior Labs acquisitions. As announced last quarter, we are planning to fully offset the slightly dilutive effect of the Reltio acquisition on our non-IFRS operating income. As you will see in the half year report, Reltio has generated an IFRS loss after tax in closing since closing on May 7 to the tune of EUR 10 million in Q2. Adjusting to non-IFRS operating profit by taking taxes and items such as amortization of acquisition-related intangibles out, we are down to high single-digit million euro amount. However, Dremio and Prior Labs in combination will weigh on H2 2026 with a very low triple-digit million euro amount.

We feel that the current environment and the investment needs in our own AI transformation do not support what will effectively be a noticeable upgrade of the underlying organic non-IFRS operating profit outlook we gave at the beginning of the year. To cut it short, we are fully on track on our initial non-IFRS operating profit outlook, which obviously did not include any M&A effects despite all macro headwinds so far this year.

We continue to target an 80% to 90% expense to revenue ratio despite the J curves of the recent M&A investment coming. We have great confidence in the impact of AI-driven productivity measures as they are going to pay off over the coming years.

With respect to current cloud backlog, our expectation of a slight deceleration over the course of the year remains unchanged. The second half typically accounts for the lion's share of our annual bookings, and we remain focused on converting the pipeline we have built. As the situation in the Middle East remains in flux, the range of possible outcomes for the metric continues to be wider than we would like.

To close, while there has been no shortage of volatility in the macro environment and massive noise around the alleged SaaS apocalypse over the last quarters, the underlying trajectory of our business remains fully intact as evidenced by solid and sustained current cloud backlog growth. By virtue of starting to harvest the fruit of our successful cloud transformation, we're well positioned to honor our commitments to capital markets, while at the same time, heavily investing in our own transformation towards an Autonomous Enterprise to ensure the sustainability of our strong growth trajectories for years to come.

The recent debate about exploding token costs at most enterprises supports our strategy of leveraging a unique combination of both deterministic, highly scalable and low-cost mission-critical enterprise applications on the one hand and probabilistic agentic AI-powered solutions on the other. We're highly assured deterministic solutions are not yet attainable and heavy human intervention is the baseline, AI can very effectively compete with labor. This ambidexterity at unrivaled levels of functional breadth, reliability, semantical richness, industry-specific process know-how, cost competitiveness and last but not least, enterprise-grade governance makes us the partner of choice for those enterprises who do not see AI as a destination, but a means to reach better efficiency.

All this without enterprise grade assurance requirements at risk. We are more convinced than ever that our strategy not to be locked into any generic large language frontier model, but to flexibly benefit from the vibrant competition amongst them in terms of both performance but also cost is the right one. And in times of high geopolitical uncertainties, customers do value the resilience of this model delivered by a provider actually headquartered in Germany. Sovereign requirements are taking center stage for more and more customers. We will continue to work very hard every day to reearn the trust they put into us. Our priorities for the second half of the year are clear: sustain the momentum in our cloud business, delivering on operating leverage we have committed to and close the year with strength.

Thank you, and we're happy to take your questions now.

Alright. We will now take your question. I would like to remind you to only ask one question when prompted. Operator, please open the line for the first question.

Ladies and gentlemen at this time we'll begin the question-and-answer session. [Operator Instructions]

We'll take our first question from Adam Wood with Morgan Stanley.

It's probably one for you, Dominik. I think investors have got used to the kind of beat and raise cycle from SAP on the operating income line and just maybe a little bit of a surprise to see the weaker second quarter. I appreciate there are good reasons behind that. But maybe also the cut for the full year suggests there was a little bit less room for maneuver than people thought. I know you've given that 80% to 90% expense to revenue growth guide, continue to guide for that for '27.

But could you just talk a little bit about going into a bit more depth there around how you think about how much room there is for margin expansion in the business over the next 12 to 24 months? Has there been a shift in focus in terms of what you need to invest in the business and what you need to invest to drive top line growth versus driving the margins higher? And maybe specifically on the AI investments you're making internally in the business, any more detail you can give us on the scale and time frame of payoffs of those?

Well, that's a lot of questions, but all centering around, I think, the operating profit for Q2. I mean, first of all, let me reiterate that it's, from my perspective, not really conducive to look at one single quarter, but it's really over several quarters. For instance, if you look at the first half, you see that our operating leverage envelope has been well respected. And we will also see -- we also see that for the full year '26, we'll be well in that kind of operating leverage formula. And that's even including the acquisitions, all the acquisitions we have made. And so of course, it pushes the kind of point a little bit more to the downside of the range. But we think the real absolute highest priority is to drive the AI transformation forcefully and to protect the top line in the current environment, while still preserving discipline on the growth.

In many areas like development, of course, we are in a stage where we are ramping capacities. We mentioned the hirings we did for selective -- very selective but very high-performing resources, but also the widespreading talking where basically we are also now improving the efficiency and improving the bang for the buck, introducing model routing technologies and so forth to optimize the bang for the buck. So this is why I mentioned in my introductory remarks, from my perspective, there is no point in extrapolating just one single quarter where we had a concentration of all the factors I've already mentioned, which I can also reiterate if you're interested in. It was a little bit of a special quarter, so don't read too much into that.

Now -- it is true though that the M&A acquisitions we did, they will have actually an impact which is in excess of EUR 100 million. And we, of course, had a debate now should we basically upgrade the underlying organic guidance and say we absorb that on top or should we keep the bigger room to drive aggressive growth and we decided to go for the latter. So that's the backdrop, I think, on that.

And maybe, Adam, just to build on what Dominik said, I mean, as we are also now building the plan for the next 12 months, and I always draw a little bit of comparison to our cloud transformation. I mean, the good piece this time is when you think about our gross margin, I mean there is nothing which now stops us with the ability to switch models to always choose the best model for an ideal price outcome. And we see already that for many agents we are now developing, especially for Joule Work, we don't always need to use the expensive frontier models. So that will always help us to manage the gross margin at a very healthy level.

Second, when you think about the productivity overall of the company, when I look at development, I mean, right now, we are very busy to shift the backlog from SaaS features, UI enhancements to AI development. But what you need is what we already have in-house. We have a lot of domain know-how. We have a lot of domain know-how in data and business processes. Now we need to hire a few great data scientists to build the ontology layer. That's what we need. But there's nothing like now in DevOps where we need to build up a massive operations like in the cloud transformation.

The good piece is on the go-to-market side, we need our consultants now to drive massive agent extensions and also adoption. And there, we are now reshuffling parts of our consulting to really work with the customers' hands-on in the hybrid landscapes to really drive that. And so we have a lot of the capabilities already in-house. In the second half now, it's all about also reskilling, enabling our workforce to work with AI. But when I also look at the productivity levels, what I just mentioned in development, depending on the area, we see on an average a 30% productivity increase. And I'm convinced, especially when it comes to now adding more agents, adding more data layers to our applications, I actually see that this is not even the end. We can easily reach a higher productivity level, but we need to give ourselves a little bit of time in the second half.

But again, the outlook for the next 12 months, I don't see anything now similar to the cloud transformation where we had to invest massively into the buildup of our operations.

We'll move to our next question from Mohammed Moawalla with Goldman Sachs.

My question was more focused on the top line. Given you haven't really seen any deceleration sort of to date in the CCB, can you talk about the kind of now the visibility you have for the rest of the year on the cloud revenue.

And then just sort of to extend that one step forward, Christian, can you talk a bit about how the pipeline has evolved kind of coming out of Sapphire with obviously the launch of the road map and some of the agents. So when do you sort of expect to sort of drive some of the -- both the adoption but monetization of some of your kind of AI solutions? Could you see that effect potentially towards the end of the year? Or is it still kind of more into next year and beyond?

I can get started and Dominik, please add. I mean being seven months now in the year, I mean, of course, the predictability on cloud revenue is becoming better and better. So we are very confident to hit our guidance for the year despite the volatility we still see and we had in -- especially in Q1.

Now the pipeline after Sapphire is for half year 2 better than expected, and it's a better coverage than the last year, especially where customers saw the new platform, and we have a lot of customers, hundreds of customers now in the beta testing, the feedback is extraordinary good. The good piece is also that a lot of our customers, obviously, truth to be told, build custom agents. And -- but they all somehow came back and said, hey, we are really missing a good price outcome ratio. We are missing the efficiency gains, the value we expected. And also the IT teams are oftentimes really completely overwhelmed by managing the agents in the hundreds of countries. So also the governance is not easy. And there, the customers really saw now SAP with the new platform, with the announcement we did. That definitely also reconfirmed the belief that SAP will deliver the leading AI platform.

So net-net, despite the macro volatility, we see a very positive pipeline for the half year, too. And obviously, now when the customers also see now by building the agents in their hybrid landscapes, they also see continuously the need to modernize the landscapes. I mean that's also a very important pillar now in Q2. You saw the big RISE deals we closed. And all of these customers said, hey, with the current data quality, with the current complexity in my ERP landscapes, AI is going nowhere. So we need to modernize while we need to implement and drive adoption of AI. And both they are getting with SAP. And that actually also now was really reassuring in Q2 that a lot of customers are realizing, hey, we need to do both. We can't stop on ERP migrations while, of course, we want to lever the power of AI.

Quantitatively, I think just comparing the evolution of '26 to what happened in '25, in '25, we had a quite a higher-than-expected deceleration in CCB growth. We guided slightly and then it turned out to be more than slightly. Now we had a much better start actually in '26. We guided slightly, but you've seen not much attrition even if you adjust for M&A, it's actually a very stable CCB development. Now the reason why we still stick with slightly is very much the macro uncertainty for the second half of the year, where the outcomes can be more nuanced, different depending on what type of escalation you might see.

But I think operationally kind of depolluted CCB growth development is much more stable this year. And also in relation to the current -- sorry, the cloud revenue growth, we now see that while last year, there was a big gap of cloud revenues were below CCB growth that now has flipped. So that's really positive from our perspective.

We'll move to our next question from Ben Castillo with BNP Paribas.

Just really on that cloud revenue outlook, please. H1 is running slightly ahead of your guidance. So we're looking at quite a material deceleration in cloud revenue growth in the second half to get to your guidance midpoint. Again, this despite adding more M&A despite CCB growth reaccelerating in Q2 and also growing ahead of cloud revenues, which is usually positive directionally for trailing cloud revenue growth. So I guess how much of this is just prudence given what's going on in the world, but how much is actually what you really expect? And how can we get comfortable with that deceleration that's now implied in the second half?

Just a quick follow-up would be just on the macro side. Did you actually see any sort of impact on sales cycles and pipeline conversion in Q2? Could the backlog growth to be better in Q2 without that?

Yes. I mean because of the macro now in Q2, definitely, I mean, look at the CCB, super strong. We don't see any major backlog because of macro. I mean, obviously, in the Middle East, a few deals here and there got delayed, but definitely not at the broader scale. And let's hope that it continues like that in the second half.

Now obviously, with regard to the revenue in the second half, very important for me to mention is, I mean, all three acquisitions now we did was not to acquire growth. I mean, first of all, all three acquisitions have only a very, very minor impact on CCB and revenue. The latest two actually have no impact on cloud revenue. And so for us, this was very important to just strengthen our data and ontology layer, what I just outlined in my intro.

And then last but not least, obviously, now with being seven months in the year, as I just said, of course, the predictability becomes better and better. So we are more and more confident that we can also hit the guidance what we outlined at the beginning of the year despite all of the volatility we see out there in the market.

We'll take our next question from Kirk Materne with Evercore ISI.

Christian, can you just follow up on that last point on the most two recent acquisitions you've done. I'd be curious sort of if you could just go over again the thought process behind specifically Prior Labs and the idea of bringing on technologists, obviously, around tabular models. And then sort of what is the strategy for taking that IP and then monetizing it across the customer base? Can you just give us an idea of how we should think about sort of the return on that acquisition in particular? Obviously, I realize Dremio plays into it a little bit as well.

Yes. I mean, first, I mean, I said it last time, we went definitely over the last two years through a learning curve in development when it comes to building accurate and reliable AI. Now what we are now doing with the new platform, and again, the test, the accuracy tests we are doing very, very promising.

First, the agents, they have access to a lot of mission-critical data in the ERP, the most mission-critical data of a company, which is great. Now when you're doing replenishment, when you're doing financial forecasting, when you're doing workforce planning, obviously, oftentimes, you not only need SAP data, you need also non-SAP data. That's why we acquired Dremio, and that gives us access to this data without copying it, we have real-time access. That is very important.

Now with Reltio, obviously, the second layer is then now you have a lot of data. I mean you can have that also with Databricks with Snowflake and others. But now you need high-quality data. We have a very strong master data governance solution, but we were missing a master data governance solution for non-SAP data because we want to build one semantical data layer. So, in development, we are shifting now our developers more and more to really not only expose data products on our new platform, but also joining data products. I mean a customer model for customer churn, for example, includes over 100 data objects across the ERP, plus oftentimes over 400 data objects in the non-SAP world. So, for example, in the Salesforce system, also sometimes even on social media on the web. And we are joining the data products to build the semantical modules.

And then Prior Labs comes in, and we want to keep this open source, but we want to monetize it via our agents. I mean, we are pricing our agent based on value. And Prior Labs will give our agents the ability to predict more accurately than any other agent in the industry because we will use the tabular AI modules to really -- where we can source SAP and non-SAP data and they can run predictions without curating data, without managing a data pipeline and they drive predictions up to a level of 99% accuracy where you need it before a bunch of data scientists to get this done.

So we can deliver these predictions out of the box. And when you look at predictions, you need it in finance, you need it in sales, you need it in a lot of industry AI agents. And that's why we want to use the Prior Labs tabular AI module to include that into our agentic AI layer and monetize it via the agents. We don't want to monetize the model on its own. We want to monetize it via the value of our agents.

Maybe one -- maybe at the risk of stating obvious, what is also interesting about Prior Labs is that it's trained on tabular databases. I mean, no secret, SAP has probably the biggest reservoir of data of tabular databases. And that is not public data, it's proprietary data. So, that is, I think, a very different ball game from the large language models where a lot of public data is great. So the combination of that technology with the unique treasure of data that we have in that format, I think, positions us extremely well to run this frontier model here.

We'll take our next question from Michael Briest with UBS.

Great. A question on the R&D side of things. So headcount looks to be up 3% year-on-year and the costs 14%. In Q1, the numbers were 2% and 2%. And I know Gina was talking about some sort of compensation structure change. Can you elaborate on whether there's been some either targeted or sort of significant increase in rates for the R&D staff? And also where your token costs would go? Would those go into cost of sale or if they're related to developing products, would they go into R&D? And is that part of this increase?

Good question, Michael. I mean, first of all, in the last 12 months, indeed, we still invested into new job profiles in R&D, data scientists, data engineers, we invested into full stack developers for industry AI. But we will now continuously heavily slow down the hiring for the other profiles because now that the AI productivity is kicking in and indeed, the highest token consumption is in R&D, we see productivity gains of an average of 30%. So there is no need anymore now to hire additional people.

You also see that the costs are more up than the headcount, that is actually the token effect, yes, because we charge the tokens, of course, to the functions who are using it. And then second, obviously, we also -- for very few top caliber people, we hired a few top caliber people, which came in with a higher personnel expense per FTE than the average what we are having. And that's what also what Gina was alluding to. But you can expect now in the next 12 months, not a further increase of headcount. It's just about getting a few experts in and then, of course, driving the productivity of R&D up in line with the token consumption.

Maybe one addition. I guess what is also very important, Michael, what we are doing, and that also depends -- it also relates to our customers. We are now also managing with our customers, the development backlog from SaaS features to AI development. And obviously, with this change in the backlog, I mean, we don't want to incrementally just add to the development backlog. We want to now change the backlog from feature development into agentic AI development. And obviously, that is also a transition period, not to forget. We cannot just say from one day to another, we stop feature development of our SaaS solution. But this will be a transition, which is, by the way, already on its way. The customers, of course, are heavily now interested in shifting their feature requirements into agent requirements, and that process is also ongoing to really, at the end, also see the wide R&D productivity in the next 12 months.

We'll take our next question from Charlie Brennan with Jefferies.

Can you just say something on the recent European ruling on maintenance? There's some suggestion, I think, that giving customers more maintenance options potentially opens the door to them staying on ECC for longer. Do you think this ruling actually changes any behavior on the ground? And then maybe as a small modeling follow-up, Dominik, you've given us the loss run rate for Dremio and Prior Labs. But can you say anything about the revenue and the CCB contribution from them?

Let's start with the last question. It's negligible. So contrary to Reltio, which had less than 1 percentage point incremental CCB, not much different on cloud revenue growth. it's basically negligible on these two acquisitions. This is why we didn't comment on it much.

Now you said EU ruling, this is an agreement between the EU and SAP to commit to certain mitigants in terms of flexibility on maintenance. I mean, first of all, I want to stress that the maintenance is extremely highly valued by the lion's share of our customers. They clearly see the value of being current on cyber patches on compliance patches, legal patches on some functional improvements we bring to the table. And we have actually also in the past, granted flexibility and opportunities for these customers to adjust their software spend to the needs. And that has been, to some degree, now formalized in this agreement. And in some degrees, some more flexibility has been granted in limited scenarios for the customer. And for some customers that are actually prioritizing spend, lower spend over the advantages of maintenance and support I described, there might be an impact, but we think we can manage it. And don't forget that one is also phasing out more and more as we convert the customers onto cloud and RISE. And that has nothing to do actually with ECC versus S/4 transition because at some point in time, that maintenance will anyhow expire. In 2030, ECC maintenance is basically zero anyhow.

So, and the last point I want to mention, we do actually see quite a nice pickup in returns from third-party maintenance. that gives us also confidence when people try that for a while, they tend to be more nervous over time about incidents happening and then come back. And actually, some of the discussions in that context are about how do we deal with the customers who are knocking our door and say, we want to come back and how much back maintenance do they need to pay and stuff.

Our next question comes from Frederic Boulan with Bank of America.

Frederic at Bank of America. If I can come back on the cost side, if you can comment on the decline we've seen in cloud margins this quarter and more broadly on the R&D side, so a bump as well in terms of percentage of sales. Is it something structural in terms of rebalancing from sales and marketing into R&D? Or as you were saying, it's more about an initial investment that should then normalize over time?

Absolutely the latter one. I mean, give us some time, please. I mean, look, I mean, this is the second transformation we are now in. And as I mentioned before, first, you need to reshuffle the backlog. I mean our backlog was full of feature requests from our customers, and we need to have the time to reshuffle this now to AI -- and we are on a good path there. We already see that the share of agentic AI development in the backlog has substantially increased. And the second part of it, now we invested into getting the right experts into SAP. We invested into the AI tools. But at the same time, we are already seeing the productivity gains. And now in the next 12 months after I have done that, you will see very healthy R&D ratios going forward. So there is no structural shift. And the same is true for the gross margin.

Now in Q2, we had some onetime investments into the test environments for our new platform into Joule Work. So we had to do that. And we have a cost increase for the sovereignty environments we are delivering end-to-end. But still for the bulk of our cloud operations delivered by the hyperscalers, we see no cost increase. So, also there, there were some minor onetime impacts now in Q2, but nothing where we believe will continue now in the second half of the year on a 12-month outlook.

And recall, we always gave that kind of formula to say that total expenses will grow at 90% of the revenue growth. And we always said we want to have some wiggling room within any specific line item. I think that served us well. That gives us flexibility to optimize the business while sticking to that envelope. And we have no reason whatsoever to change that. So I think the longer we wait, the more it turns out to be a very solid corridor with regards to how we can leverage the revenue growth down to the bottom line.

Yes. And maybe one last point. When our AI business now is really starting to scale, I mean, think about the price levels, what we can also then achieve in the market. I mean, for 50 years, this company has sold system of works, ERP, first on-prem, then cloud. But the customers were used to a certain discount level.

Now we always, in the last years, very successfully maintained very healthy price levels. You have seen this in the gross margin development of SAP. We have thought that we're going to achieve such a gross margin, by the way, five years ago. Now with AI, I mean, now we can completely reset the price level. I mean now you're going to a customer and said, "Hey, it's not your end user. It's not your financial accounting team who does the financial close. It's the agents doing this autonomously. -- and here is outcome-based pricing.

So there is nothing that you can relate to when you come from the system of record pricing. And that is, in my eyes, a unique chance also for SAP to do a reset and really price outcome value-based. And that is our clear task also to our salespeople to make sure, hey, don't even go to the SaaS world, don't even go to the price levels you have given. That is a new way of selling, and it's a new way of actually pricing our solutions. And for me, this is a unique chance, which, first of all, will hopefully end up in the next 12 months in an acceleration of our AI cloud revenue, but second and also in very healthy gross margins going forward.

We'll take our next question from Toby Ogg with JPMorgan.

Perhaps just on the new EBIT guidance, Dominik, so EBIT slowed to 9% in Q2 for the reasons that you laid out, but you had a strong Q1. So overall H1 EBIT growth constant currency was 16%. And the guidance midpoint implies sort of mid-teens or so EBIT growth in the second half, which would imply a reacceleration relative to the Q2 EBIT growth. Can you help us understand what would drive that reacceleration from the Q2 EBIT growth rate? And then what gives you the confidence in that as we think about the second half?

I did mention in my introductory comments that Q2 was a little bit of an abnormal situation because also from the cloud revenue growth acceleration, we had a strong contribution. We are frankly optimizing massively now on how we spend tokens by virtue of a very tight controlling. We can now, on a very granular basis, see who's using what tool and what is output driven here, and we will funnel the tokens in a way that gives us a better bang for the buck. And there are also measures on cost containment to really make sure that we focus our resources where it really matters. So -- but you nicely summarized the thinking we have in terms of H1, H2. This is exactly what we're going to do.

We'll take our next question from Michael Turrin with Wells Fargo.

And I just want to ask a little bit of a different flavor on the questions around bookings and margins. So CCB growth improved this quarter. I think that's a surprise to many given the backdrop. But the margins coming in a touch, Dominik. Maybe walk us through both what drove the Q2 growth improvement. And then given that, why that doesn't flow through to operating income? Is there anything outside of M&A impacting that relationship? Are you seeing on the organic side, things are generally improving and it's all tied to just some of the dilution impacts.

Okay. So let me do it step by step. So if you look at the deceleration, so to speak, in the growth of the non-IFRS operating profit in Q1, which was extremely high, 24% versus Q2 -- and first of all, we have to understand there was a very big strong positive from lower stock-based compensation because the EUR 60 roundabout share price drop in Q1 that didn't reoccur in Q2. So that is sequentially going down.

We mentioned a heavy R&D investment that was discussed at length on this call already. We had a slower growth in cloud revenue growth. And this was predominantly driven by the comparison in the prior year, as I tried to explain. So we had basically in the prior year, an increase of 2 percentage points in the cloud revenue growth from Q1 to Q2. And that just mathematically means that the comparables on cloud revenue growth are less favorable. So we had much harder comps in Q2. And by the way, that will, of course, flip around also in the next quarter.

We did have some pointed marketing investment around the launch of the Autonomous Enterprise. And then there was also a minor effect, but still negative on first-time inclusion from Reltio in early May. So that is giving you basically the bridge why Q1 had a higher growth rate on non-IFRS shopping profit than Q2. And again, if you put it together, you see a mid-teens increase in operating profit for the two quarters taken together, and we are expecting something similar with all the measures I have been highlighting, we are going to take in the second half of the year. And of course, this is actually so to speak, a little bit more comfortable to achieve than top line topics because it's entirely under our control. We can do these things.

So the CCB is the more important measure from our perspective in terms of bringing that home because that is also requiring customer to sign a contract, whereas our spend is something we have entirely under our control.

Yes. And maybe just to close it out on the bottom line. Just last week also, we reviewed our hiring plans for the next 12 months and for 2027. And seeing the increase of the AI, seeing the increase of the productivity, now not only development with Joule Work, we're going to see it cross company. We have now 4,000 users inside SAP using Joule Work with tremendously good feedback that will now hit the market in Q3 as well. I mean we also adjusted our hiring plans. For this year, we will nowhere near hire the number of people we plan to hire at the beginning of the year. And also for next year, we see that we are balancing AI to consumption and own headcount in the right way so that we see the productivity increase to hit the 80% to 90% revenue cost revenue ratio we committed to you.

Ladies and gentlemen, the conference has now concluded, and you may disconnect your telephone. Thank you for joining, and have a pleasant day. Goodbye.

For those who don't know me, Fred Boulan, I lead software research here at Bank of America. So thank you all very much for coming.

We're delighted to be hosting SAP. We have Philipp Herzig, CTO, who's probably one of the leading AI experts and architects in SAP. So it's extremely topical.

Before we go into kind of the Q&A side of things, I've got a quick disclaimer to read. And then -- so we'll first start with the conversation on some of the key topics, and then we'll open up for Q&A. All the mics are open. So just during the session, just bear that in mind. So if you want to make sure, you keep that in mind.

So quickly on safe harbor. During this fireside, SAP will make forward-looking statements, which are predictions, projections and other statements about future events. These statements are based on current expectations and assumptions that are subject to risks and uncertainties that could cause actual results and outcomes to differ materially.

Additional information regarding these risks and uncertainties may be found in SAP's filings with the SEC, including, but not limited to the Risk Factors section of SAP's annual report on Form 20-F for 2025.

So with that, thank you very much, Philipp. Thank you, Alexandra as well.

And maybe starting high level, it would be great to hear from you key points around SAP's AI road map. You introduced a kind of autonomous enterprise I think vision at Sapphire, what does it bring to your current capabilities?

Yes. First of all, thanks a lot for taking the time here in this big round. I really appreciate that.

Look, I mean, the vision in my mind, has just now emerged right to the overall company strategy, right? The AI strategy so far was, of course, it was very consistent, but I think now we have really taken it to the next level. And if you look at this overall AI strategy from SAP, what we are bringing to the customer is that we really strive with the whole strategy to directly bring business outcome to the end customers.

And that we do on 4 levels. With Joule work, that's the kind of new Joule 2.0 version. We have -- and we can talk more about that in detail. That's our new version that both from a technology perspective, but also from a user experience perspective is our next version of how users in the business will interact with SAP software and also with non-SAP software if they choose to.

Then underneath, we have all the Joule assistants and Joule agents. We announced that we will bring -- so there's this taxonomy that we are using. First, we start with the autonomous domains, where we have the autonomous domains for finance and for HR and for spend and for customer experience and supply chain. And underneath each of these autonomous domains in the suite, we have assistants. And assistants are representative of today's personas in the enterprise.

So think of a treasurer, for example, in finance or an HR manager or an HR business partner in the HR organization or just a demand planner, for example, in the supply chain and the COO area, for example. And we're aiming to release by Q3, 50 of such assistants across these autonomous domains, where then underneath each of such assistants, these assistants are kind of orchestrators of a bunch of agents that are necessary in order to fulfill the assistance capabilities in order to make that persona more efficient or to help them to achieve more, for example, on the sales side with less resources.

So these are kind of -- and we have underneath these 50 assistants, 200 agents that we are delivering all coming out of the box with the underlying SaaS applications that we are shipping, because what we are striving for is a fully SAP managed Software-as-a-Service offering where such agents don't need to be built, right? But you can basically right away start consuming them out of the applications and the landscape that you have today.

All of that, and that is where then the third element is coming in is, of course, doesn't work without a comprehensive platform. And I mean, you see this everywhere, right, that the big problem is people spend a lot of tokens, building agents, don't even know what to build for, what's the return on investment, what's the outcome that they are building for.

And then, of course, also how do you then -- once you figured out a few agents that you may be ship to production, how do you then also consistently govern them, improve them and also manage that across agents from SAP, but also non-SAP agents. And this is where we announced the Business AI platform, which brings together, of course, our existing business technology platform as well as the business data cloud as kind of the foundational element.

But we put on top a new version of Joule Studio as well that will, together also with Signavio, allow customers to, first of all, study as part of their existing business processes, where is there an opportunity to become better with AI and then directly from this analysis from SAP Signavio, finding maybe inefficiencies in the process or finding opportunities in the business to, for example, sell more with improving their existing products, then they can take this over directly as a product requirements definition into Joule Studio to build the right agents with the right context with much less tokens compared to what they can maybe achieve with other platforms.

And once they are ready with these agents, then they can bring this into our AI agent hub and from there observe and continuously improve the agents in that platform. And I think with all of these 3 main elements, I think we have a very -- not only compelling vision, but also a very differentiated offer where you get the best of both worlds out-of-the-box agents that directly make your business better, while simultaneously building custom and bespoke solutions that are deeply grounded in the business process and the business data that many of our customers obviously have already, which they can use to not only build better AI experiences, but also far more efficient AI experiences in the next few years.

Great. So before we move into the whole kind of monetization and pricing side of things, I think on Joule, you were actually -- SAP fairly kind of self-critical at Sapphire about some of the limitations of the initial products you ship. It would be good to understand a little bit with the kind of new version that's coming up soon, what that can unlock for your customers.

Sure. I mean, first of all, what were the limitations? I mean, in that sense, look, I mean, look left or right, everybody is using more or less the same technologies. And of course, Joule suffered from the same limitations any other chatbot in the enterprise space presented, right, in terms of limitations.

Too few integrations, right? Sometimes too much -- it was not getting then sometimes the right information or maybe the coverage was not broad enough, right, depending on if you're looking specifically in the ERP space, right, where there's a lot of scope, right, that needs to be covered, where simply more often than not, Joule came back and said like, I cannot provide the answer because maybe the integration is still missing and so on and so forth.

Now of course, specifically in half year 2 last year and also beginning of this year, there was a tremendous progress on the technology side of the house. I mean, you have studied that all intensively. The models got better, right, starting with the cloud models, [indiscernible] right? Tremendous progress on that side. And obviously, there was a lot happening on the harness side with -- I mean, most prominently, obviously, open claw, right?

I mean, forget about the enterprise problems, right, with security and integration and so on, but it showed us how a better orchestrator looks like beyond the pure RAG-based system, for example, skills like -- there are many puzzle pieces on the technology side that came together that allows us to build just simply a better orchestrator and build better integration, right, with the SAP system.

But at the same time, what we also clearly did is -- and we have invested in that for quite a while now is, of course, also the thing that you don't get just by off-the-shelf technology components out there, such as, for example, with our Knowledge Graph. Because the Knowledge Graph is very key to provide actually the right metadata, right, the right structures to the AI models, so they need to guess less. So they don't need to guess through MCP, for example, all the various APIs all the time, but they directly -- you get at the first pass already the right API, just call it, right, which saves you tokens and gives you a far more accurate answer.

And that combined with the latest harness methodologies, with the latest models that gives us the big boost from an accuracy and a performance perspective. While simultaneously, we also have realized some of our research work from last year like our generative UI that -- where we are now able to actually -- it's not just a conversational experience, but also things that, for example, also OpenAI did recently with sites and so on that we can generate just the applications basically on the fly.

If you have an analytical question, you don't need to model anymore the chart or the KPI, right? We just get a visual representation that gets generated on the fly. We call this concept now Spaces in Joule, but that's basically our generative UI approach to move slowly, but steadily away from classical UIs that were designed for humans to an AI-led approach to generate UIs, for example, on the fly.

And we are also now shipping as part of that, by the way, that has been generated, 95% of that code has been generated with AI is we're also not just delivering it for the web and mobile. We are also going to release a new version for the desktop as well. So it runs on Windows as well as also on Mac. So you can even contextualize further with what people have on the computer, invoices, other business documents that need to be processed in context with the rest of the business. So these are some of the innovations that we're bringing forward.

Great. Can you talk about in order to address the kind of Agentic orchestration layer. So we have OpenAI yesterday. That's one of their play. Microsoft has a play around that with the Frontier AI platform solution. To what degree is there an incentive for enterprises to deploy agents that can then manage multiple applications? And to what degree you think you can play in that orchestration layer and directly help your customers?

Look, I think there will be -- there are 2 dimensions here, right? One is the -- what's the UI that is being used, right? And then also, of course, what is the underlying orchestrator, so to speak.

Because I don't believe in a world where there's just essentially just one UI and one orchestrator, right? And of course, depending on where a user types it's prompt, right, or where maybe an e-mail gets sent to if it's a system triggered agent asynchronously. And of course, it's then the orchestrator, first of all, that's being addressed, right, by that event. But that orchestrator needs to be able to talk to other orchestrators because there will be not this single -- I think this is a very naive idea that there will be this one Uber orchestrator that just orchestrates everything on top. So there is opportunity or there is other orchestrators that are required to do the heavy lifting, because there is like all these orchestrator to me, they form like a hierarchy, right?

You go from the highest level orchestrator and a branch out to a bunch of other ones, which then they get delegated to other orchestrators and so on. I showed this all the time in my keynotes. So there's kind of a hierarchy of nested loops of orchestrators, right? Because the problem is still with AI, you want to keep on each level the things as narrow as possible for better accuracy and performance essentially.

And so that is why we are saying, look, you can consume also the Joule orchestrator, which does the heavy lifting, let's say, for the SAP part, right? And it also is possible to branch out if customers choose, we believe in choice of customers choose to also use that orchestrator to orchestrate non-SAP agents, they can do this as well. But you can consume Joule orchestration itself through A2A, through the A2A protocol and tap into any of the SAP provided agents.

And then that's, of course, associated -- we talk about commercials, I think, in a second. Of course, then the meter spins in terms of our consumptive model in such a headless experience basically of Joule, which then gets also orchestrated into another orchestrator. We heavily did this already. I mean, if I talk to customers, what they -- what most customers demand is that at least SAP, Microsoft Teams/copilot and/or Gemini do work. This is what we are seeing. And of course, we build these integrations already out, right?

So that, with the more casual use from an SAP perspective, the more casual users probably would rather go through Gemini or would go through Copilot/Microsoft Teams and the more power users, they usually start in Joule, right, and the SAP screens. And then, of course, they want to complement that also with non-SAP data from any of these other orchestrators, so to speak. So this is what we are predominantly seeing, but then customers there will be customers that consume other experiences also from OpenAI, from Claude and so on, and we are open in this regard.

Great. So yes, let's move to your kind of monetization strategy. If you can share a little bit your approach, your philosophy, what's the kind of pricing model? How has it changed? How do you see it evolving?

Yes. Now, look, I mean, there's always -- I feel -- let me start from the basics, right, because I felt and also in the previous conversations, there are some misconceptions overall as it relates to our overall commercial approach with AI. So clearly, the way how we commercialize -- and first of all, we divide between base and premium AI. I mean, that's probably known and base AI comes just like, for example, expensive with Concur, right? I mean, you have the Concur app, right, and you upload a receipt or your taxi receipt or whatsoever, right? I mean, that's part of the base subscription basically, right, at no extra charge. So there are a bunch of these capabilities that are just table stakes from our perspective that customers come to expect as part of the product.

And then, of course, there are premium capabilities where there's a willingness to pay by customers because there is great value that this provides to the customer in addition to what the underlying base software or SaaS software brings to them. And all these premium capabilities that we are shipping, roughly 200 as of today that are there, you find 400 -- it's roughly 50-50, I would say, maybe 60% to 40% I need to do the accounting on that. But that you find also on this AI discovery, if you go to Google and say SAP AI features, you probably land on this catalog, you see roughly 400 AI capabilities and then you see there what is premium, what is space.

So if you want to get entitled to use any of these premium AI capabilities, you have to purchase this concept SKU called AI units, right, that you probably have hear of. And with that AI unit, you basically get entitled for all premium capabilities across the entire portfolio of SAP. So no matter whether it's a capability in supply chain and finance and for the IT function, that's true for consultants, true for developers, all of them roll up into this AI unit concept. The AI unit concept by nature is a consumptive model with consumptive revenue recognition, right?

So also to make that clear, it's by nature, it has been from the get-go designed as a consumptive model because it was very clear already when we released this in '23, it was very clear to me, yes, there will be some headwinds because customers don't like consumptive model, but the AI world will -- AI value will be consumed in a consumptive way. So I think we were pretty fairly -- fairly early in that assumption.

Now of course, what we do underneath now how do we price. So what we don't do is we don't directly -- this consumptive, but we are not just passing on tokens, right? Our customers really don't like tokens. They like business outcomes. This is what SAP stands for. So what we basically do is when we commercialize something, like document AI or let's take Joule for Consultants, which, by the way, is still per user per month seat-based today, but under AI units. So it's kind of seat-based charge consumptively under AI units, if you understand what I mean.

So what we are doing when you talk about value, first of all, you need to have a hypothesis what is the value. And just let me make the example of Joule for Consultant, which has tremendous value, because it directly translates into reduced billable hours that customers spend with their SIs on an SAP implementation, right? So they can say, okay, if I use Joule for Consultant, I pay SAP so much more money for my IT staff, so many hundred, 200,000 users. And that translates directly into 20%, 30% reduced cost in terms of billable as measured by billable hours towards the SI. Very simple deal, win-win situation, so to speak, from a business model perspective.

So whatever the price now is, whatever the value is, could be $1, could be $1 million, it could be $10 million. We say 100% of that value, 80% is for the -- 70% to 80% is for the customer. So we apply a take rate of 20% to 30% that gets charged to the customer, value-based, right? And that's, again, numbers of days sales outstanding reduced, numbers of days in consulting and billable hours reduced. And so on something the business can measure where you can go to a CIO, to a CFO, to a CHRO and say, okay, we're going to deliver that value against you.

Now what we need to make sure, and that's exactly the beauty of this commercial model. Now in this take rate of 20% to 30%, right, and we need to charge. We need to, of course, make sure that the tokens in the cost of goods sold structure that's required to produce that, of course, the costs need to be in the margin profile structure that we need, right, so that within the price, boundaries of this 20% to 30% take rate, we can, of course, produce the outcome with the respective margin profile with, of course, the respective cost of goods sold profile, right? So that's kind of the overall principle on how we approach commercialization for SAP, value-based and then working backwards.

Of course, now the question comes, I get this a few times earlier today is, okay, but what does this mean now for margin, right? What does this mean? Can you actually uphold an 80% cloud gross margin, for example. And so, then the question is -- but usual that's very usual, right, in software development throughout the years is, of course, in an early product, that is a different statement than in a very mature product, right?

And take again, Joule for Consultant, very mature product with meanwhile more than 1,000 or a couple of thousand customers on that product with customers who are using it every day, right? And we -- I can -- I'm not sure if I'm allowed to say this, but just anecdotally, take it as, that is a very mature product. We have optimized the hell for the token optimization, and we are run with margins beyond 95%, 95% on a very differentiated product because we sell on value and we optimize the tokens that are required to produce that value.

Now if you turn towards Joule Studio and any of the new releases that we are doing, of course, we communicated at Sapphire that it will be free until the end of the year, okay? And I can tell you what the margin is, right? Will it always stay this way? Absolutely not, right? Of course, we favor adoption in the first place because only an adopted product will then lead, of course, to the realization, oh, it actually provides value, then people are willing to pay money for that, so we can recognize revenue.

And then in the next step, I can optimize the margin because the beauty is once adoption sets in, we can collect data, much, much data about how the users are using it, which scenarios are being used. We get data that we can use to fine-tune, post-train models and so on. So we can then get actually the same experience that maybe requires to start with an Opus 4.8 model that burns through a hell lot of tokens at a very high price.

We can actually start to reduce and optimize because under the covers, we are switching that. To give you just again the Joule for Consultant example, of course, we started 2 years ago on the latest Opus model, right? And the margin profile wasn't there. Today, Joule for Consultant runs on 5 different models and maybe tomorrow on 7 different models. And they're all very small, and they're all doing different tasks in order to do that.

And with that, you can start optimizing the overall system, which is -- and that becomes an architectural problem. This is not an AI model problem. It becomes an architectural problem. And this is how we approach this in general, but also in order to make sure that from a cost perspective, over time for mature products, we are then also converging towards the margin profile that we -- that is our ambition.

Great. I think one question that I think we get a lot on SAP is, what is the kind of killer app on the AI side. So I think it would be interesting to understand what you see your most advanced users doing, people that really embrace that at scale. So probably some of your most advanced cloud users, what areas you see the fastest deployment of AI at scale?

Yes. Look, I mean, I talked already -- I mean, coding, obviously, that's also true for us, that's also was true for developers because, I mean, there are a bunch of reasons why development is kind of natural good candidate for an AI killer use case in that sense. The same, of course, for Joule for Consultants, lots of unstructured data. Usually, I would say the killer cases are in the world of unstructured data.

Document AI, we have huge -- it's one of -- I mean, it's fairly simply explained, right? It's a service that basically instead of people keying in information into an SAP system, right, coming from an invoice, coming from a purchase order, coming from a bill of lading, a delivery note, right, if a truck comes right with all the goods into a plant and an automotive company, right, still people need to enter stuff manually into the system, unbelievable, right? That's actually it was even mind-boggling for me why that is, but it is a very simple reason.

And now we are basically processing this all with Document AI. We have processed alone in '25, 750 million documents. I think the run rate is roughly 70 million to 80 million now on a month that we are processing. If you would just stack this paper all up, you have the height of Mount Everest by just processing through Document AI every single month in an automatic fashion with AI.

Again, unstructured, right? And of course, high value with respect to the amount and time that you save, right, by just entering stuff into the system or reconciling business processes that are disconnected from a variety of sources.

Then clearly, HR, customer experience, Sales Cloud, Service Cloud, enterprise service management. This is, of course, where we see the biggest adoption. It's a bit harder still. We are getting there, but still a bit harder on the, let's say, more structural things like think finance, for example, or think supply chain where there's a lot of optimization also involved, where there's a lot of number crunching involved and so on and so forth. But this is exactly where our investments with also with our own Tabular Foundation Models, also the recent acquisition that we announced with Prior Labs are coming in to also solve that in a more systematic fashion.

Great. So maybe, I mean, one topic in the, I think, debate is all the kind of token maxing and cost of really AI being a problem now, how do you help enterprises frame their return on investment on the products that you deploy? How do you give them visibility on those kind of AI credits and avoid them kind of fearing a bit consumption side?

Yes. Well, first of all, the model I talked -- the problem is with tokens themselves, tokens is not yet the outcome, right? You want to -- and that's exactly -- again, I talked already as part of the commercial model about this. That's the beauty of in my mind of our commercial model because I always say, hey, look, if you have runaway costs and if you have runaway tokens, you never know really. Token is like measuring the performance of the company based on how much electricity they are consuming, right? That's not the performance indicator.

You can consume a lot of electricity and still -- I mean, maybe just because you keep the lights on the entire night, nobody is being here. Right? So that's not a good measurement, right? It maybe serves as a proxy variable, but it's not a good measurement.

In our commercial model, it's beautiful because you see now, aha, you have saved so many hours, your days outstanding down by 1, 2 days, right? And you can directly translate that into your business metric, what that means. So if you have runaway costs, it's almost certain that you also have runaway value, right? So that's the first thing. But of course, still customers, specifically at the beginning of the journey when they don't trust yet that return on investment, of course, they need visibility, they need experimentation. That is exactly why we're saying with a new product where customers still first need to gather trust, we give it away for free. We get used to it. They see how much cost it is in terms of AI units.

So they see this and we have this thing called SAP for Me. Not sure if you're aware of this, like every customer can log in and see across the entire portfolio, what they spend on AI and see their finances, see the T&Cs, see which agreements they have with us, like all the contractual and financial things that customers have with us. And there, they, of course, see the AI unit consumption. They can also budget. We don't have this yet, but they will also be able to budget end of this year then the consumption in the various buckets as well.

And of course, then if they see, the consumption is going up. So obviously, people are using it. And then they also see the value that is associated with it, right? That, of course, is the trust building exercise you need. So then also basically the upsell and the renewal becomes a no event, right, because they see the value. And of course, then they purchase more and expand more as a result of that motion.

Great. And anything you can share around the kind of upsell or spend increment you've seen with some of those kind of early users? I mean, to give us a bit of an order of magnitude of what that can represent? I mean, I think as a firm, you shared some ambition in terms of AI revenue, but it would be good to understand a little bit how meaningful that is for some of your customers.

I'm not sure if I fully understand the question.

How much people are -- I mean, in terms of people that are actually are paying and using AI credits, I mean, to what degree that's a meaningful change in terms of their overall spend with SAP?

It's hard to say. I think this is difficult to overall put at this point in time into context, right, because there are so many other factors, right, that are also playing into that. I mean, we are overall very happy with general uptake, right, in terms of our customers are not only purchasing these AI units, but then, of course, also the uptake, right? And then we measure very consistently also the consumed ACV, so what is actually being then consumed by the customers. And that's a very, very clear hockey stick that we are seeing without disclosing our specific numbers. And I think it will only grow from there.

Anything you can say around your industry AI road map as well? I mean, you presented that at Sapphire as well. I think you have 10 -- I mean, you have a number of industry that you've identified. So you were talking about the kind of 10x growth in that piece of business as well. So I mean, it would be good to understand a bit the fork.

Yes. No, look, I mean, I think we -- in all honesty, I think we neglected industries for a bit too long in that sense as a company because that was always a big strength also of SAP to be very present in certain industries. Obviously, in some of the core industries that we are supporting, oil and gas, retail, public sector, right, professional services and so on. Of course, there's financial services. So there's still -- with Pioneer, there's a big focus. But I think from an AI perspective, we haven't done enough justice to industries.

Now what we do, of course, with industries really is also to go in these high -- because obviously, in the industry, when you really go into the core process and the core value creation of that industries, there usually also is the highest -- we see a lot of the high-value scenarios, right? Like take, for example, asset management, right, in oil and gas, for example, there are very high-value scenarios there.

And so the challenge on the other side is with AI, you really -- and I don't want to use now this word forward deployed engineer term because it is so both ill-defined and conflated. But what you really want to do is to sit with the customer, right, and design with AI this value creation together in these industries. And this is exactly what we are aiming for there.

So we are creating a -- or we have created a dedicated team that really is -- yes, works in this kind of forward deployed engineering fashion closely in those industries with the customers to build high-value scenarios, which then again will -- once they work for the first 5 customers, we're bringing them back into the standard into this commercial framework that I outlined, right? So it then from there on scales to more customers via the platform and in the commercial boundaries because then we have better proof point about how differentiated it is from a value perspective and as well as also what is the -- we already have some optimization applied already, so we can really then scale this more consistently for other customers in such industries as well.

Great. Last one for me, and then we'll open up for Q&A. So one concern out there is that some of your customers will try to extract value from your data and develop agents using other -- I mean, either internally or using other providers. BDC, for instance, is one tool that you're actually offering to help customers extract data and manage in a Databricks environment. So to what degree is the kind of market missing some of the issues around that ability for an external provider to help leverage SAP data, drive insights. I mean, is this a realistic threat? Or do you think this is completely a misconception?

It works for some use cases, but it certainly doesn't work for all. And I mean, unfortunately, almost no topic in IT or in software is a binary 0, 1 thing, even though we program in 0s and 1s, are not anymore. We have floating point numbers and GPUs anyways. The problem really is, how do you still connect this with the transactional system. Because like I've seen the craziest ideas then all of a sudden that emerge out of this thought that, hey, you put everything into one central data lake and have all the data there. That is great for read, right, if you can accept that it's maybe not real time. That's great.

But the question is, it gets -- people then say like, oh, yes, but then I can even have the discussions in BDC. Oh, then I can write back to BDC. So the agent, the result that the agent has can actually write back to the data lake and BDC. No, no, no, how should that work, right? Because it still needs to go and check all your transactional rules and check the validity and check the referential integrity with the rest of the business if this is actually legitimate need to go to the -- like all these things. It needs to run through the transactional system. It cannot just be stored and the like, right?

So -- and then you need to close all the time the loop, you need to remodel authorizations, right? And it becomes this, this very complicated thing that you are creating with a lot of effort. So I'm not advocating against doing that, right? But obviously, what we are trying to do is, to get -- to minimize and make it much, much easier for the customer, like, for example, with these SAP data products, right? So that you don't have to build your ETL, right, you extract, transform and load pipeline and remodel all the authorizations on top, but you basically connect BDC to as 4 SuccessFactors, RBA, Concur, and boom, you directly have access in the lake, right, with full referential integrity and real-time updates as well, right? So that's the first angle to really make it simple for the customer from a manageability perspective. And I think this is why BDC works pretty well.

On the other side, that's exactly where, again, the Tabular Foundation Models are coming in. I think we have an amazing opportunity here to disrupt also in that space in general. Because if you think about -- forget LLMs for a second, GenAI, agentic stuff based on Claude and OpenAI, forget it for a second. Nobody really has cracked the nut on structured data yet, i.e., on tables on all these 100,000 tables that are on SAP systems and Oracle systems, Salesforce systems and your dusty old SQL server that is standing somewhere in the corner and use that, right?

Because there's a lot of business data and a lot of data that's required for decision intelligence in the enterprise. And so what happens today, you still need to resort for such things to classical machine learning, right? Because classic LLMs will not help you in this, if you want to do a good demand forecast, for example, or a good cash flow forecast or predictive maintenance and asset management. And what we have built and also with -- we are doubling down now on this with the acquisition of Prior Labs is, I think we can do the same for predictive and structured data, what large language models did for the unstructured world.

Because if you look at the results, they're actually pretty phenomenal. I mean, in classical machine learning, right, the predominant algorithms are XGBoost, Random Forest, if you're a little bit into that, and it's AutoGluon, right?

And the recent models with our own RPT 1.5 as well as Prior Labs, we beat XGBoost with a small amount of data already in 100% of the cases and AutoGluon that runs for 5 hours in 80% of the cases. And I believe by end of this year, we will beat them in 100% of the cases as well.

While simultaneously, these Tabular Foundation Model, actually, you don't need to be a machine learning expert at all anymore. You just provide your table into, then boom. You can start making predictions, asking questions on top of the table and reason over the structured data.

Because what happens today still is -- and that's why we are doing this is if you have such a question, a CFO asks for a cash flow forecast or whatever or the COO ask for demand forecast in your stores, then of course, you go to the data science or the machine learning COE and they build their stuff in the Jupiter notebook or whatsoever because this is where they live and breathe, right? And of course, as a result of that, that is why the gravity is like in order to train such models, you need to get the data. Some of SAP data, some of non-SAP data and so on. Otherwise, you cannot train those machine learning models. Now with the Tabular Foundation Models, the plan is to flip this around because the Tabular Foundation Models, everybody can do this. You don't need to be a machine learning guy anymore.

You put this on top of BDC where you can call out into any data lake that is out there is exactly what I said at Sapphire. So you get tables on the fly for BDC and then predictions on the fly. And I think this will be a pretty powerful technology going forward to build something that is technologically very differentiated and provides a lot of value to businesses. That is today for me, the other big missing part of the coin that nobody is really talking about yet.

Great. Do we have any questions?

I have a lot of questions. I understand and please correct me if I'm wrong, that one of the customer frustrations with Joule has been that it's not been possible to access the data that sits in SAP on-premise systems recently. Is that correct? I guess, I understand and please correct me if I'm wrong, that one of the customer frustrations with Joule has been that it's not been possible to access the data that sits in SAP on-premise systems recently. Is that correct? I guess, there's a lot of value, I guess, for customers over the years in terms of the data that's there.

Yes and no. So what we always said is -- so first of all, the frustration more was coming from that we haven't yet able specifically in S/4 to cover the broad range of cases that customers would come to expect. It worked quite well, for example, in sales and distribution. But for example, we had flaws and classic good old direct procurement material management and so on and so forth, right?

And this is where I think -- and then back to what I described earlier with the new -- with the advances in Knowledge Graph and with the new harness and skills and so on and so forth, we see a lot of improvements that are coming there.

And the on-premise frustration, well, first of all, we always said and we are staying true to this, that out-of-the-box capabilities only come with the cloud. right? Because only there, it's not a technical argument. It's a life cycle management and speed of innovation argument because only in the cloud, we can ensure that the customers are getting the latest and greatest basically overnight, right?

Case in point, I tell this to every customer, right, who doesn't believe that. I call a good friend, the CIO of a small medium-sized company. He is always on the latest and greatest of SAP. And when we showed Joule work, the mobile app -- he has everything on mobile. He doesn't even have Joule in S/4 and so on and then SuccessFactors, everything just in the central mobile app with so far it was Mobile Start and then have this little Joule icon in SAP Mobile Start where you could open Joule and then talk to your SuccessFactors, Concur, ERP and behind that.

And he did that work very well, he's big fan. And he's on the latest and greatest. And he just pulled out his phone in Sapphire, clicked on Mobile Start and it opened and said, Hey, Mobile Start is now Joule work. There's a new icon you want to have the new experience, click Joule, and he got the new experience.

That is not usually the experience customers have with SAP, specifically as they have this good old on-prem ECC or even S/4, as safe. And that's, of course, where we want the customers to get into, right, in order to participate with the speed. However, this is where it's still solvable because this is why we also said value while modernizing. That was -- we are just now messaging this a bit maybe clearer. It was always there because obviously, you can use the platform and our extensibility mechanisms to always connect and build custom things against your ECC, because it's a platform because there's no technical boundary condition.

But of course, that's then work the customer needs to do. And we have many customers who have used the former Joule Studio, for example, or the Joule Command Line Interface to build custom skills to do whatever, to integrate an ECC system, to integrate with ServiceNow, to get a bunch of stuff out of Salesforce and integrate that into Joule. I mean, technically, that's all possible, right? But of course, it's then require -- it requires work by the customer, and it's not this out-of-the-box capability that we are shipping what we are calling SAP managed as opposed to customer managed. So the customer has to do the work. I hope that makes sense.

Thank you. We're going to leave it there. We're on time. Thank you very much, Philipp. Thank you, Alexandra for coming. And thank you, everyone.

Good morning. At this time, we would kindly ask that you please take your seats and silence all devices as our program is about to begin. Thank you. Please welcome to the stage SAP Global Head of Investor Relations Alexandra Steiger.

That was too fast. Good morning, everyone, and thank you for joining us at our annual financial conference. We hope you're enjoying Sapphire so far and had a chance to walk the floor and explore all the exciting innovation here on display in Orlando. A warm welcome as well to those joining us virtually from around the world. Today's agenda offers a great chance to hear directly from our executive team, take a closer look at some of the key developments across our product portfolio and see how our technology and strategy are coming together here at SAP. As AI continues to reshape the technology landscape, and the way our customers operate their business, this year's conference is an opportunity to reflect on that change and also the progress we are making against our own vision. Last year, we shared how we laid the groundwork in enabling AI and data capabilities across our existing product portfolio. This year, we're building on that momentum by going all in on AI. We have a strong agenda for you today with our executive sharing updates on our strategy, product road map, execution, our workforce transformation and our financials.

So let's get started. First of all, welcome our CEO, Christian Klein, on to the stage to share his perspective on our vision and how our strategy differentiates us in the age of AI. Next, Muhammad Alam, who leads SAP's product and engineering of the board area, will share an update on our technology portfolio and how we're continuing to innovate across the business. Following this, Thomas Saueressig, our Chief Product Customer Officer, will share his view on the progress we are making on our cloud migration journey while helping customers realize value in the business. Afterwards, well, welcome Gina Vargiu-Breuer on stage, Chief People Officer and Labor Director, to discuss our people strategy and how we're transforming our workforce to support SAP's next phase of growth. We'll then hear from Sebastian Steinhaeuser, our Chief Operating Officer; and how we're accelerating strategy execution and simplifying operations internally. Last but not least, our CFO, Dominik Asam will provide an update on our financials and the key drivers of our top and bottom line growth.

Together, these sessions will hopefully offer you a broad view of SAP's strategy and execution across our business. After Dominik's update, we'll take a short lunch break and then afterwards the entire SAP executive team will come back on stage for an interactive Q&A session.

Before we begin, it's always a great pleasure to reach you a disclaimer. So let's get out of the way. In this presentation, we will make forward-looking statements, which are predictions, projections or other statements about future events. These statements are based on current expectations and assumptions that are subject to risks and uncertainties that could cause actual results and outcomes to differ materially. Additional information regarding this risk and uncertainties may be found in our filings with the SEC, including, but not limited to, the Risk Factors section of our annual report on Form 20-F for 2025.

Unless otherwise stated, all numbers in this presentation are non-IFRS and growth rates and percentage point changes are non-IFRS year-over-year at constant currencies. The non-IFRS financial measures we provide should not be considered a substitute for or superior to the measures of financial performance prepared in accordance with IFRS. That out of the way, I'd like to ask Christian on to this stage.

Hello, everyone. Welcome to all of you here joining us on site in sunny, sunny Orlando and of course, also welcome to all of you joining us virtually. What a difference 12 months can make? I was here on stage where we talked about the acceleration of our cloud transformation. I guess we can all agree, a very successful cloud transformation. We talked about Joule, and now we are talking about the future of the software industry in the age of AI. And after yesterday's keynote, I would like to share a little bit more about what is actually the right to win for SAP in AI. Why will be -- why SAP will be a winner.

And then second, of course, what does it take to become a winner in AI. And to start, maybe a quick look back into our cloud transformation because some of the things we did, which you see behind those financial numbers are now very, very important also for the success in AI.

Now first, when you remember, we acquired a bunch of cloud companies starting 2012. And then when we took over we said, hey, we want to harmonize our portfolio. And harmonization means we want to actually build a best of suite. We want to harmonize data. We want to harmonize processes, and I will come later to that, why this is so important that customers and our stack are not sitting on a bunch of data silos and not on a bunch of pro business processes, but that we are now really building AI on a harmonized foundation. And second, I mean, what was equally big, I would say, is actually the transformation we did on the go-to-market side. I mean in 2019, there were not many people in SAP know how to manage subscription consumption, not how to adopt actually drive adoption of consumption-related business models. There was a lot of change happening inside SAP to not only equip our sales force for that, but then, of course, also to change the whole operating model.

And then third, no transformation actually just will happen bottom up. I mean it needs leadership, it needs strong people, it needs strong expertise. It needs a culture when you turn around a company which is big 110,000 employees, I mean it means something. And that needs to happen at fast speed at fast pace. And we also changed our culture so that we are really also having our employees understanding that this change is actually needed and that there is a clear plan in place, which we all have to execute now.

And then you, of course, see it in the numbers. I mean, Q earnings in my eyes was really great. I mean, we are outperforming all of our competitors. And of course, even in the cloud world, the ERPs, the SaaS apps, they will not go away. I mean there's a lot of potential still for -- to grow our business in the cloud, but of course, now much more with the value equation happening on the edge and AI.

So we transformed SAP once. And yes, I can tell you, we do it a second time. Now what does it take from a CEO perspective to transform and make a transformation successful. In my [indiscernible], there are 4 pillars, 4 levers to make this a success. First product who [ I met ] Philip and I showed yesterday in the keynote, what does it really mean? What kind of value will the autonomous enterprise deliver to our customers today. We want to share a bit more insight, what is really differentiating SAP from the West.

And also then, of course, how do we going to make it happen? How do we develop it? And then second, of course, go to market. The way how we sell, the way how we deploy it, the way how we drive consumption going forward, again, we'll also need some transformation on the go-to-market side. Thomas will share more details about that. But obviously, there was a certain reason why we also did certain reorganizations. Why we did also why we are now doing also certain re-skilling on how to vibe code in presales and show the value and then later on, of course, drive the consumption and drive the adoption of our agents.

And third, I mean, super important is, of course, the people there. I mean, going back into the cloud transformation, I can't remember where we were. I looked around in SAP and did we have enough people to know how to actually operate the cloud at scale. Do we actually have enough people who develop a multi-tenancy enabled architecture for our core products? No, we brought new people in, but we also did a massive re-skilling inside SAP and the same needs to happen now all over again. And then, of course, with Sebastian on the operations side, it's also now very important that we become on our own and autonomous enterprise. First, I mean, you are looking at us saying, okay, first, how will SAP deliver AI at scale, at speed. And then, of course, also, how can we also deliver the efficiencies you're going to -- you need to see coming out of, of course, deploying our AI inside SAP. For us, I also want to be very transparent on that. There will be also, of course, major investments. You have seen some of our latest acquisitions. But of course, we need to bring top talent in. And obviously, this will also then require certain investments, targeted investments in certain areas to make sure we have the best workforce to win in AI.

Now coming to our why to win. We won ServiceNow. We did won a few best-of-breed applications in the past because some of our acquisitions made the mistake to not choose SAP, don't ask me why. But every time when we then migrate it away from these best of breed or no, you're wanting a ticketing system, I mean there's not a lot of domain know-how in the data feels, I mean, you can actually now with AI, you can migrate those over. The switching costs are getting lower and lower. Talking about SAP, it's a little bit different. I can remember well here with Thomas when we started 5 years ago to harmonize our data model. I was actually overwhelmed by -- to understand, oh, there are 1 million of correlations between logistics and finance. And there are another 2 million between the payroll and the commission system and this -- and then you're adding this all up, and you're counting over 7.5 million data fields. And now think about it, at that point, it was about harmonization.

Now the agents need a graph where they can correlate this data. So it's actually 7.5 million multiplied by a factor, which is not imaginable where we now need to make sure that we are also bringing this semantically rich data contact sitting in our ERP into the new platform I will talk about in a second. And then obviously, we are running thousands of business processes in these companies. And I talked last earnings about the learning curve we had. I mean clearly, 1 learning curve with tool was also that these agents need process context. You can talk about agendic AI use cases all day long, but if they don't understand your process logic and you cannot also flexibly extend that because every customer has a different way of doing sourcing has a slightly different way of running a payroll. Obviously, it's not going to work.

But all of that sits in the brain of a company, and this brain is actually the ERP. So I'm very confident that our ERP, our apps, they will stay. And of course, we now need to build the autonomous enterprise and actually enrich it with world-class AI. How do we do that? Yesterday in the keynote for purpose, we didn't start with the business side. We started with the platform. Why? Talking about lessons learned, we know that Joule is not perfect. And when you are talking to some of our partners and customers, probably also here, the results are not really accurate. Is it really compliant? Is it really governed in the right way? Or does it really require a lot of handholding. And I can tell you, why do we create examples? We can deliver world-class AI, but also, I have to admit, there's probably a lot of handholding behind to let the agents connect to the wide data fields and making sure they understand the process context. And then Muhammad, Philip and I were sitting together and said, "Hey, this can't be", because our Asian buildup was on BDP. The data layer was on BDC. The process context sits in [indiscernible] and in Erland in some other places. So then we said, hey, it doesn't make any sense because when we are the plane of every company and you are developing agents, I mean, obviously, the context needs to come right away with that. And I hope you have seen with the new better version of the platform, and in one month, this platform is GA, we are going to deliver now an agent builder where you can bring your own tools.

And on the building side, you can use Anthropic, you can use OpenAI. Next week, we talk about Mistral, because of sovereignty in Europe. The differentiation on coding and agent is 0. That's why you don't have to come to us. But then when you start developing it in Joule Studio 2.0, you're going to see that half, I'm developing now a pricing agent. I need access to the material data in S4. My pricing data, it's sits either in Salesforce or in SAP. I come in the second to non-SAP data. Okay, this is how you quote. This is how you price. This is -- okay, I need to understand how to follow this process steps. I understand the approval steps. And then finally, I come to the governance part in a second. So the community will -- we'll see in a month from now, together with our own developers already building agents with that now, they're going to see Aha! now I have the context. Now I have access to the point of every company. But here, we don't stop. I know there are certain concerns around BDC, what data do we share? What data we don't share. We actually share more data. We do -- we do [indiscernible] copy share and also that people don't need to move the data around, that's good for TCO, that's good for cyber reasons. So we do that.

But then actually, what happens on BDC and only on BDC, we joined data. We -- a data product alone, it's a raw data. But then when you join data, when you build a semantical module, suddenly, you can say, Aha! my pricing agent actually needs data from SAP and maybe from Salesforce or maybe from another CPQ solution out there in the market. And then you can do these data joins and the way how Muhammad and Philip now engineer that is you can actually say, okay, I joined data, customer data between SAP and Salesforce, and now I link it into the [indiscernible] because then suddenly, the agents also realize Aha! this is how I should understand the semantics of my customer data in the company.

The same we do for material, et cetera, et cetera. We build this really, really which semantical data layer between SAP and non-SAP data. You have seen our acquisition of [indiscernible] so that we also then can take care about the master data quality, again, because agents need high-quality data. They should not compensate for a potent data module. And then last but not least, of course, with [indiscernible], we have a Lake House to really also develop own agentic AI scenarios with our own lake house built on Apache Iceberg. So that is actually the context layer. And that is the heart and then we are also now launching SAP domain modules. So we are training these models with our code to even better understand the specifics of our customers' business on how they want certain business processes, what approval steps do they have, et cetera, et cetera.

And then last but not least, comes the governance layer. And that is actually really everything what I showed yesterday which sits under the iceberg. You can build, of course, a lot of fancy agents with LLMs, they will miss the context, but especially, you're going to deal with a ton of complexity in the governance part. I actually had in the commerce area, a customer of us telling me yesterday, Christian, we built over 50 agents. And now I have a problem. These agents actually don't understand really the pricing logic. They actually share pricing data with consumers, they should never ever share. They are, for sure, also not actually adhering to certain regulations with our data privacy standards in Korea or actually with the data privacy standards in Germany.

I mean this is completely out of control. I can double my IT organization to really make sure that the governance layer is actually working. I said, yes, but hey, SAP is investing over EUR 600 million each and every year for the localization of our software for the certification. I mean here is the logic. It sits in our software. So that's why we said, "Hey, when we are talking about the business [indiscernible], here's the build layer, here's the context there, and here is the governance layer, very, very important. And then other taxes are debating or who has now the orchestration layer and who owns the agent layer. I actually -- we give this away for free. Why? Because we want to monetize the success, the value of our agents. We want to have the best HR agent. We want to have the best source to pay assistant. We want to have the best record to report the system. We want to have the best planning assistant. But the orchestration, if customers want to really also block in third-party agents, of course, this is for free.

And I know there was some talk about this API policy. The only thing that we actually manage now with the API policy so that we manage actually the access to our APIs, to our MCP servers so that we really can guarantee high governance standards. And second, obviously, that we also can still guarantee with millions of API calls coming in that we can manage the system performance. But everything that you're going to build on the SAP business AI platform. You can use any model you want, no differentiation, but then comes to context and then comes to governance. And that will be world-class delivered by SAP and some of that is actually already there.

Now the autonomous suite, when we are now having the platform, I mean, yesterday, we presented at, okay, what is actually then our vision of wanting our autonomous enterprise will one. So first, you're going to have your air systems, design to the personas in your company. The agents will actually take over a lot of task, will enable you to do more with less or actually enable you to do things which were not even possible before, especially when it comes to prediction, you saw RPT 1.5, you saw prior labs. I mean there will be a lot of business scenarios we are now going to enable, which were just not possible before. Everything on on our AI platform. And then next to whole center, obviously, it's very important that it's outcome-based. So I just had a U.S. public sector customer, 4 rooms away. Actually, they ask me, Christian. I now build in custom agent and okay, understands he -- the agent understands, listen, of course, but actually now to change something in the system of record and to do -- and to wide back into the system of [indiscernible] it's not possible.

And actually, that is not good because I want to want this autonomous. This was about a payroll. I said, he see. This is actually the power of the SAP platform together with our autonomous domains because you can actually tell us, does this agent need to read, change or wide back even into an SAP system. So we are going not only from listening, understanding, we are getting all the way down to execution. And there's always a human in the loop, but the customer can decide, okay, how far do I want to go with regard to the level of autonomy, I want to give these agents, very important, audit-ready. [indiscernible] was smart enough, super smart as always to say, okay, hey, we have so many certifications for our software. I mean, at the end, the same what we need to have for our software now needs to happen for our agents. So SOC certifications. I mean we are really going down now, not only giving you the full traceability of the actions what every agent is performing now. We're going even 1 step further. We are actually telling you when you won your financial close with our assistant, "hey, this is actually SOC-certified." This has all the certifications you were used of running your financial close with S/4HANA, with our software, extensible by design.

When I talk about learning of tools, my god, maybe we should have thought about that before. But again, everyone is still learner these days. extensibility. I mean, when I look at our policies inside SAP, still Sebastian is fighting it a lot to make it simpler, simpler, simpler, but we have different policies. And sometimes, we have new ideas about new policies, new pricing policies, et cetera. So extensibility is super important. There will be not this 1 standard agent for pricing, not this 1 standard agent for sourcing and customers will also have different policies. So that's why extensibility is key. But what is also very key, and I hope you saw that yesterday with Joule Studio 2.0 that you can extend also your autonomous sweet layer now with new agents and with new skills on the fly. We just had -- actually, we showed the new platform to a customer right now, and they are building actually a personalized AI agent for commerce. And actually, we said, "Hey, okay, you actually have this one more step." You want to have this agent to do. You go in and say, okay, as a business user, you can say when you have the authorization, while you can just say, okay, with [indiscernible]. Here is 1 more step the personalized AI agent. And you should consider when you are making this recommendation to commit consumer. Think about that. Today, we are thinking about in ERP on-prem, we sold about 8-year long upgrade cycles. In the cloud, we moved to quarterly, then to monthly and now you can add agents and skills where we are, and we deploy it and we govern it. So the speed, the agility and the extensibility is very, very important. And think about that. Now you could build all of this with Anthropic. Let's assume that. Your business is changing so fast every day, not only the tech world is changing.

I mean how do you want to actually make sure that these agents are always extensible everything when everything is custom, have fun with your IT organization to make sure they understand all the business model changes, all the business process changes is the business wants tomorrow and not in 3 months.

So industry AI. Yes, some customers told me a question yesterday in the keynote, so, so happy that we heard this term industry, again, from SAP. I can tell you that when we sometimes say SAP is so sticky because we are the point of every company. But actually, I take this as a huge credit credit to SAP is that we delivered incredible value. I mean why would someone run all of these industry applications with SAP because it's sticky now, because it has value. And especially in these domains here, what's the best [indiscernible] and the team picked, I mean look at that, when you do asset management, when you do commodity management, yesterday, we showed unified commerce. There is no scenario, not 1 single 1 who doesn't need a tight integration into the ERP of SAP, not single -- not 1 single one.

And in many, many cases, again, you need to even have a wide bag because otherwise, forget about autonomous, and forget about running autonomous asset management, when you cannot maintain or change certain assets or trigger actions in the SAP system to actually trigger a predictive maintenance or get one worker from place A to place B, in order to fix the asset. How is this possible? You can always say it's autonomous. But what you're missing is the last mile with the deep integration into the ERP. And that's what we are going to deliver. And then yes, Palantir has the copy wide for the FTE. But [indiscernible] actually told me just 4 weeks ago, I said Christian, you know what, with this autonomous suite, it's great. I like the agents. Over 200 agents are already here now, but you need to completely reinvent, tell your product management, forget everything about what they learned in the past and just think greenfield and think about how do we deliver autonomous asset management in the future.

When we talk to Coca-Cola now and Kona about last-mile delivery, we talk to the truck driver, talk to the people in the warehouse to understand not about how to incrementally apply AI on top of our transactional application. No, no, no. Just to completely reinvent this process because now truck optimization happens with AI agents without anyone sitting in front of Tableau or Power BI and trying to somehow manage it on top of a transactional system. No. That is now done via autonomous industry AI use cases. We're pulling now together our IBU, our industry experts with our industry developers. You're going to see in that space that we will also build here a huge ontology layer for each industry. We are now started with -- you saw H&M yesterday on commerce, on retail, oil and gas, everything around related manufacturing because this is high value. This is -- these are complex scenarios, but they are high, high value for all of our customers. And actually, by the end of the year, we want to actually scale this business to over 200 projects. More to come. I can tell you after I told everyone yesterday call, Sebastian,, if you want to do something. His inbox is exploding and the pipeline looks actually anyway really, really good because again, what is the why to win? The why to win is it is so natural that you need to not only collaborate with the brain, you need to also wide it back and the agents need to understand the full logic of our suite.

Now when we are now building the autonomous enterprise, when the platform gives everyone a great reason why to build agents with SAP. Why to an autonomous with SAP. Now 1 lesson I learned during my whole career is technology means nothing if you don't bring it to adoption. And every great technology needs redesign, needs business process change. you can plug in the best agent into your sourcing. But if your procurement department actually is not really leveraging it, not really thinking how can I shortcut the process now? How can I actually use this agent now to also do more intelligence sourcing, it's useless. And that's why we said, "Hey, we really want to also deliver at Sapphire a new [indiscernible] with SAP offering." And what does this mean? Two things. When we decided to give Thomas sales and consulting, it was not that I want to have more time for my family, maybe that as well. But at the end of the day, we also said, hey, in order to do that, we are not now inventing new roles. Now what we need to do is actually right from the beginning is to start prototyping is actually start to show value. And we can't have our customers waiting for 3 years until they finish the modernization of the system to actually use AI.

And to underpin that, we said, hey, we are not only saying this in some nice words. I say, we contractually commit to that, and we come on site. And we're going to take all of these great AI assistance and agents coming from for Muhammad and the team now and bring it to adoption, not only activate it, bring it to adoption. And now, of course, I heard, "Oh, Christian, you changed your strategy because now you're also allowing it for on-premise systems." No, no, no, it's also not a defensive move. What we are now doing is after a ton of feedback, I mean think about that, there are customers here who are actually shifting now and harmonizing over 100 ERP systems over the next years. I come to the acceleration for that. But now, of course, these customers are saying, Christian, my CEO, my CFO is telling me, "Hey, I need to deliver AI tomorrow. I can't wait here until I have finished all the modernization". And even for some of the on-premise systems, I mean, they are mission-critical systems in the company, I see great value AI use cases. If you -- if I'm not doing it with SAP, I do it with someone else, custom, but I would love to do it with SAP.

And that's why we build this on-premise connector to really now also then connect our assistance and our agents to these on-premise systems. But of course, what we would like to see is for us the AI foundation, the AI platform needs to be there, and there must be a commitment there to also then go with us on the modernization journey because needless to say, when I talk about extensibility, if you have a heavily customized landscape, obviously, there is more handholding, more extensibility required than when we are talking about the [indiscernible] a very clean system with a lot of standardized business processes and a standardized data module underneath. Then ERP migration, acceleration. I can still remember when Alex Karp called me and said Christian, I would love to switch to SAP, but my god, my CIO tells me just takes that long. That needs to be done faster. And I said, Alex, we are on it. We have tool for consultants, for developer. But please feel free to join us. I mean I'm welcome every partner to shrink this time line for getting our customers from place A to place B. And when I'm now talking to Accenture to Deloitte to PwC, everyone. Everyone is investing into that.

Somehow we are disrupting someone's business model here -- but that is overdue. And so that's why we are now also today, in Thomas keynote, we launched our AI-led ERP migration platform, where we are SAP on its own is shipping new assistants for data migration, for business reconfiguration. And we are very confident by the end of the year when we are testing this now with our customers, that we can actually cut the time and the effort and the expenses by sometimes up to 50% when we are deploying all of these assistants now available for the ERP migration.

And then we are pulling in twice enters for test automation, AI-based. We are pulling in Palantir. We are pulling in Accenture. And there are many, many more now to join. So that journey will be accelerated without any doubt. And then the customers have both. They have actually a very harmonized data module. They have a very clean system of record where you then can plug in the AI agents. You don't have to wait for that anymore 3 years, even if you have an on-premise system, we can start tomorrow while we are modernizing your landscape.

And that actually resonates extremely well with our customers, and I feel we can combine here really the 2 primary objectives of our customers, meaning modernization and of course, AI value. Now talking about the go-to-market model. Thomas will deep dive on that, so I want to keep that short.

Look, at the end of the day, what we are going to do is you can show a lot to customers on PowerPoint. What is more convincing is that 4,000 of our presales experts can already wipe code prototypes. That is what we need to show. In industry really also even with more on-site presence to really build that while we are still defining the value, showing the customer the value and then come to sign a value-based transformation outcome based on industry AI, where we then, okay, say, okay, let's go and really bring those agents now fully into production. And here, I mentioned already before, we didn't want to add new roles to our go-to-market model, we just simplify that. But then with the consultants we have, the data consultants, the LOB consultants, the industry consultants, I mean, they can hear help. So that we don't need to onboard new people. We don't need to build an AI deployment engine in the SAP. We have it.

We need to do some reskilling, but we have it. And when I see the amount of consultants who can today code already with our new platform with Joule Studio 2.0 or also understand how to optimize an LLM module and how to actually fine-tune it. I mean it's remarkable and we will double down on that and come in a second tool at the people section.

And of course, that we need to use our own agent led tool chain is, of course, a mandatory task to all of our people in the post sales area. And then yes, when you actually then see when they have adopted a suite when they have adopted BDC. And now with the data catalog and you can then build these data [indiscernible] and you can add it into the KG, we, of course, also see that the AI consumption is already going up because now finally, we see, okay, in ERP, a best of suite, best of suite will win, not best of fleet. And then finally, actually, we also then see much higher consumption and much higher penetration not only of our LOB solutions but also with AI in our installed base. The people side quickly. First, again, we are investing. We are investing into top, top experts, but do we always need to deliver certain AI agent or a feature, less and less features, more and more agents. Do we still always need a team of 10? No. Of course, you're going to see productivity gains. But again, I also want to highlight to get the best ontologists, the best data scientists to get the best full stack developers.

I mean, these are all the roles at AI architects. These are all the roles where we also then hire from the top, top universities in the world and bring those in, but even more important is, of course, the rescaling and here, you see, I mean, we call it mandatory, I actually call it, it's actually an offer to our employees to have a quite career within SAP to enjoy all of these learnings. And the managers are always in the lead. It's not about that we have -- here is IT and here's your AI tool. No, no. We really want to have cross SAPs, the managers and some frontrunners on AI showing about what is possible with AI in the one or the other job and we are doing this [indiscernible] SAP. Obviously, with more intensity in engineering. And as I mentioned, the vibe coding and also then the LLM trainings we are doing with our consultants. Sebastian is responsible not only to simplify SAP and make SAP AI-ready to really deliver a consumption business model from the beginning to the end at scale, but of course, also to make SAP more efficient. Here, you see by function, again, with the business and the lead how we are now applying AI internally.

We already have realized efficiency gains triple digit. I'm not allowed to share the exact number, but it's actually quite substantial. And of course, we see a huge potential to make this even a bigger number than 2 billion. And again, don't only take this as efficiency. I mean we will take some of that money, of course, also to reinvest. But again, it's not about the quantity of people in development. It's about having the best people in certain areas. The market, I mean, believe it or not, I'm 100% confident that our SaaS past business will not go away. AI agents don't work without a brain. The brain is SAP and having a best of suite versus the best of breed, I see definitely scenarios where customers are replacing SaaS apps, but the domain expertise is not so deep where the data models are not so mission-critical. But this is why I actually see a unique chance now with the suite. We always said the suite will win. I guess, in the age of AI, there is one more reason to believe that. And then, of course, when we are growing our market share, and there is no doubt that we are going to grow our market share even further, then of course, we rather see this $5 trillion market here as an opportunity because if our platform now delivers what we here presented and it will, and when we then deliver these high value-add use cases for the different domains plus industry, there is no doubt that we can accelerate now also our growth towards 2030.

Here, 1 last point on the commercial modules. I said this as earnings. So over 2/3 of our cloud revenue is already today see [indiscernible] meaning. It's nonuser based. So we have all kinds of value metrics. So no one needs to be scared that, oh, efficiencies, users, not anymore needed. And suddenly, the revenue goes down. No, no, no. We price a lot of our apps already today value-based. And then, of course, you see now with AI with the platform, with our migration tools with the autonomous domains that, of course, we see that at least we're going to see a 30% consumption-related revenue share in our cloud revenue in 2030.

Again, it's there's some time to go, but we see that we are on the right path. We have the right strategy and the execution is there. Now look, that's how I see our growth journey. There was, first, an enabled AI. Maybe we can also a learning AI journey, where we delivered SAP Business Data Cloud. Let's not forget. This offer is only one year old. We actually fixed a lot in our AI foundation over the last 2 years. We delivered the first assistance and tool agents. And then, of course, we worked further on the harmonization of our suite. Now we are definitely in the scaling phase. Now I say, hey, with this platform, there is no reason why in the SAP context, you are not building agents with SAP. The autonomous wheat layer. By the way, not only that we are delivering 200 agents for 60-[indiscernible] systems this year. This morning, the customers who are the partners who are better testing the platform already have developed over 600 agents. And that will be over 1,000 attacked, and there is more to come. So when the platform now starts to scale and the autonomous suite will become bigger and more mature and with higher value, I mean, that will, of course, help the adoption of AI massively. And then last but not least, I mean we also then have our migration tools, and we definitely want to leverage them, and we will monetize them because again, this is a big, big market, customer spending big, big money on that, and we will hit it right there where we say, no, you don't have to spend that money, spend what are your money on our AI migration tool chain. And then, of course, for me, it's actually now very important to build the community on the platform, creating the belief in the ecosystem with the new platform being in better and 1 month productive. And then actually, we build this. And then we build the graphs for the different industries. We build the graph and we will get more and more mature for the horizontal process layer in the industry. The domain modules will get better and better.

[indiscernible], you can feed in a lot of more additional process logic. We have [indiscernible] with our AI government hub, where you can free of charge actually also then see the transparency and govern your AI layer. And that's actually what I believe will happen within a year from now. And I guess this year was a very, very important sapphire for us to show that we are now really accelerating both on product, go-to-market, people, operations. And when this all comes together, there is no doubt that this will be a second successful transformation for SAP. Many thanks.

Good morning. It's kind of hard to follow that act. Thank you. I think Christian already covered most of the product strategy. I think what I'd like to do here in a few minutes is maybe go through 3 things: one in not as much detail. But thing number one is I want to at least share with you that as we think about our product strategy, what were the shifts that we saw both in the tech and in the market that guided the strategy that you saw yesterday and what Christian just walked you through. And how internally are we really mobilizing that strategy to build a product that we believe our customers need. So that's one. The shifts that we see and hopefully, in an effort to see if those are the shifts that you see in the industry as well. Thh second is the product strategy.

A lot you saw yesterday, Christian covered a lot in detail as well. But related to the product strategy, what I also want to attempt to do is share a little bit more of the conviction that we have as to why this particular strategy is both differentiated and unique then the strategy that you're hearing from some of our other tech peers as well. Because at the category level, things are probably starting to sound very similar, right? Everybody has a set of agents, everybody has an agentic platform.

They think that it's the world's best and people want to control the orchestration layer as well as the engagement layer as well. But I want to sort of build a little bit more conviction in you at least share the conviction we have as to why what we're doing and what Christian just walked you through is different than the rest of the tech space.

So let's start with the shift. The basic shift that we see is happening in the industry is the tech stack that hasn't changed in a few decades is going through a fundamental shift. It has a new layer that's getting added on top of it. And that layer, the top most layer, historically has what's commanded the most value, both in terms of what it creates for the customer and then the value it generates for the provider. And that has been the SaaS applications now for the recent past, if you will. But what's happening now is AI instead of being in the SaaS applications, you can put on top of the SaaS application.

So there's a new layer on top, and that's what everybody is trying to buy for. So what that means, though, is two things if you think about. One, as SAP, we clearly need to make sure that this is a new layer that's coming on top of it of the agentic experiences, we can take the experiences and the value and the products that we have to shine in it, hence, the autonomous suite and the industry AI, but the second thing is the layers underneath become more commoditized or the better way to look at them is to become more of a platform layer.

Now what that means for SaaS applications is particularly in the category of SaaS applications that we play in is that they're not going away, they effectively become a platform layer that provides and feeds the agentic layer on top of it. Now the thesis is not that all SaaS applications become platform, there is for sure, a class of SaaS applications that are going to be disrupted, that will go away that can probably entirely be subsumed and consumed by the agentic layer. But as you think about finance, as you think about GL, as you think about supply chain, as you think about core employee management, that layer isn't going away because it has a level of specialty compliance laws, regulations, capabilities, logic that needs to be executed.

Now can it be executed better with intelligence and generative AI on top of it and predictive models? Absolutely. Can it be executed in a decoupled manner with the apps? Absolutely. And that, hence, comes the new layer on top of it. But what that means is the more this new layer takes foothold in our customers' landscape for us, two things would happen: One, if we can provide this new layer of compelling agentic experiences for the customers to consume out of the box as much as possible with the right extensibility, we will get a share of the value that we're creating for our customers. Hence, the autonomous suite.

Two, even if the customers then say, choose another agentic platform to build that agentic layer on top for the class of business process domains that we live in, the platform underneath will still continue to grow as well, the more you interact with it. Hence, our openness that from an a to a perspective, from an orchestration layer perspective, can we be the top Layer? Absolutely. Do are we buying to be the top layer? Absolutely. Do we need to only be for a customer landscape, do we have to be the only top player? The answer is no.

They can choose what they need to and what they've already bet on, but we're still going to win because the SaaS layer that we are underneath that needs to exist for it to consume.

So there's going to be a level of growth that we're going to see in both, if you will. But this, to me, as you think about the shift in the tech landscape, right, and the stack that hasn't happened in multiple decades, there's -- we sort of whittle that down to 4 different shifts: One, the fact that AI on apps era is gone, it's now AI on apps. And that now allows everybody to be a player in this new agentic layer. Now some SaaS apps may go away. But as we talked about, the class of apps we belong to will now become a platform layer by definition because it sits underneath that highest layer up down. And there are going to be a pull-through on that. But of course, we want to also play in that new agent layer up top that by definition, is loosely decoupled to the layers underneath.

The second thing is, is now how we think about our product strategy and what you just saw Christian talk about is this app boundary versus landscape boundary. Now as we think about, for instance, source to pay, historically, what we've shipped our product in Ariba, right? There's sourcing, there's contracting, there's buying and there's invoicing. And any features we ship sort of stayed within the boundary lines of whatever the scope that we addressed in those applications, that could be S4, success factors and anything. The mindset shift now is this agentic layer at the customer level that they're going to go build is going to be one agentic layer for a source to pay, if they're using us for Ariba and they've got 2 other tools completing that process, we, by definition, whatever a genetic layer we provide has to factor in the additional tools that they have because the customer will only have 1 layer.

If we only provide agents that work on our app, they're going to have to find another layer on top to rule at all that can connect to those other applications because the customer is not going to stitch together that new layer up top by multiple agentic platform.

So that's why, as you now think about the core properties of our agentic platform, the new business AI platform extensibility and our partnerships with N8N is so core because there's hundreds of connectors that N8N brings that's embedded in our business AI platform, the extensibility that says, hey, because we generally are the core of that process, right, as for finance or even a rebind others and the other tools are sort of secondary and subsidiary to it, that we have more of a right to be that agentic layer.

And with the open extensibility we have connect and really create that landscape view. So again, if you think about and step back and believe in that thesis, ones that have a smaller share of that app landscape will have a harder time becoming the agentic layer. But because in all of these core processes record to report, hire to retire, make to deliver we are the core application. We have a lot more of a right to say, listen, majority of this is what comes out of the box with SAP and will extend with N8N or the vibe coding experience that you see with business AI platform and Versal and others to expand.

So we do believe that in that SaaS landscape, we do belong generally in a category of 1 as we believe it in our customer base. I think generally, the category is a category of 2 because there's another player that has the breadth that we do, but we don't really compete head-to-head with them from that perspective. But in the SAP customer base that are running these processes, chances are that if you get 60%, 80% up to 90% of the process that you run on the core application with that new agentic layer out of the box with our autonomous suite, you'll go cover the rest of it with us as well. And as you do that, the stickiness of let me go do a few more things would come into play as well. That's the app boundary versus landscape boundary that's so core to the product strategy that you saw.

The third one, and we've been -- that's been received very positively is, listen, historically, we've said -- you won't get the value that you're going to get from our innovation is when you get to our latest products. But that's shifting now, right? Because this layer up top is loosely coupled with the platform layer. So we have to shift our product strategy as well to say, listen, we're going to build this agentic layer, the autonomous suite that is loosely decoupled with the application here because it has its own ship cycle. I'm not shipping agents on S/4 every 6 months like I'm shipping features on S/4 every 6 months. Those agents are coming monthly and soon in a weekly basis, working with S/4 as well because we're building them in a loosely decoupled manner as well. And hence, it allows us to now make the commitment to the market that says, hey, as long as you're in the modernization journey, which is valuable in and of itself because there's components that are going out of support, you need the compliance, the regulations and things like that, that the modern solutions have. We will, as SAP, allow you to take that agentic layer and connect to your ECC and S/4 environment. So your value can start on day 1, not on day 700 and however many years it takes you to sort of get to the modern solutions, and that's been received very positively as well.

And the last 1 is, of course, because modernization is still important and so core to our sort of business model as well as the compliance and regulations for the customers and the value, we want to make the cost of getting to the modern solutions much faster as well. And that's what Christian talked about with our modernization assistant as well. But these 4 shifts, I believe in the class of applications we live in are fundamentally sort of changing how customers think and are reshaping our portfolio. So as you think about SAP now in this autonomous suite, it's not what we've been shipping for the last 4 decades is fundamentally a new product line that we're going in with that a much bigger TAM, not just the new cloud solutions, but as S/4 on-prem and ECC. And not just that, it has also those additional tools and maybe non-SAP applications that sit adjacent to us because we have the right being one of the largest tech partners for most customers, right? It's either us or a hyperscaler. Those are the top 2 if you ask some of our larger customers to who your largest 2 tech partners are. Now how this shapes and leads to a strategy is the picture that you've seen.

And this largely has 3 layers, right? And that's what Christian walked you through. The layers on the right, unfortunately, don't map the layers on the left. But if you would allow me to the bottom most as a business, AI platform, built on that is this decoupled genetic experiences layer that can work on any application landscape and you can connect it to non-SAP landscape, too.

Now that's the agentic layer, right? Agents can run headless, but agents need a place for the user to interact with them when there's an exception and hence, comes due. So those are the 3 core things. Now you can argue, listen, you can go to any of the leading tech conferences and you'll probably find these 3 layers there, at least the bottom 1 from a platform and the top 1 from an orchestration layer perspective. Nobody else largely plays in saying, listen, we're just going to build an end-to-end autonomous suite for you. Nobody else can do that besides just anyway because they don't have the right and the applications and the need to run it.

Now there could be small start-ups just saying, oh, I've got the finance agents that can really run close for you, but they're not doing supply chain, right? And they're not doing HCM. And a customer doesn't want 22 different agents running in them, right, turning to different platforms, agents on different platforms because that proliferation, the security nightmare is going to be pretty massive.

So we're one of the only ones, again, in our installed base that has the breadth of that autonomous suite, and that is what we talked about yesterday to say, from this process to the end process and not just that, with industry AI, we're going to go to your core processes last mile distribution, distributed energy unified commerce adaptive manufacturing going to go build those as well to complete your story on it. So that's what these 3 layers are. But I'm not going to spend -- because I think Christian did an amazing job, makes me proud sitting there. on what these platforms are and our differentiation. But what I do want to spend a few minutes in each 1 of those is talking about what are our differentiators and uniqueness is, right? And starting with the autonomous suite, right, the one in the middle.

We're, of course, building 200 agents that we talked about. It's going to be more than double by the end of the year. There's a set of characteristics that these agents have that are unique. One, because we run the underlying application, we can be far more committed in terms of the outcomes we think that these agents can generate.

And not just in aspiration, but in reality because we know what that data model is we can track those outcomes, and we can prove that in the AI agent that, that's there because that's the biggest issue that some of the customers have. Second, they are extensible. And that needs to be there, like I said, because it needs to be that landscape view. Third, the audit-ability part of it and the traceability is unique in what we can offer because we have one of the world's leading process mining platform in Signavio, and we're using the same tech to have agent mining logs for customers to see how these processes work.

And second, from an audit-ability perspective, we already made a significant portion of our portfolio through audit validation, audit readiness for those audit relevant places, and that's what we're working with external partners on. That's just a natural thing for us that we do, and that's what we provide to agents, that's what people need. Works where you are, we've talked about in the openness from an [indiscernible] perspective. These assistants, of course, come together in each 1 of those domains. And hence, there's only 1 organization in our customer base that can canvas out of the box set of patients that allow you to get to an autonomous enterprise stage, and that's us.

And we've talked about the numbers. the stat on the right, from an engineering perspective, just to give you some appreciation for it, again, is the uniqueness of the platform layer that we're going to get into it, right? The ability to take an idea on what agent you need to go build and take it to GA at an enterprise level is what this platform enables us to be able to do by the end of Q2 in under 10 days, and we're going to get to under 5 days. And GA is a high bar, right? You can go use Perplexity or any tool to find us listen, and what does it take to really GA an agent? And what you're going to find this timelines that's from 3 to 6 months or even further because there's a lot of complexity at an enterprise level, you have to account for the -- under the waterline iceberg that Christian talked about, that we have platform made as a platform in our agent governance and business AI platform.

So this ability to innovate at a pace that's going to be very hard for anybody else to be able to go do is what's the differentiator for our platform.

Moving on, we've talked about this. I'm going to skip this. The fact that we've really expanded the TAM without risking our modernization story to say, hey, the AI consumption can now expand massively to the customers' ECC estate and the SFC on prem state alongside the cloud state is a big, big plus for us with not just, hey, you can go Custom Connect to ECC, but there is a connector that we've built that is part of our agent gateway that truly understands the ECC data model and the patterns because we're, of course, the writers and the publishers of that software as well.

Moving on to the platform layer. Again, I'm not going to spend too much time here, but I want to sort of give you the highest level of the strategy here, right? We're partnering with things that we believe are going to be commoditized in the AI tech landscape, too, meaning the public LLMs, we believe is a commodity. Of course, that's where a lot of the early money and revenue is going, but the race between an Anthropic and Gemini and OpenAI and who else to come with public models is going to be a race that's going to see new winners and new laggards probably every quarter, if not every month, soon enough, right?

So we want to make sure that we bring the best of that to our Agentic platform. Second, the design experience of the agent is also going to be a commodity. Like do we want to go build a design experience that's better than any -- and of course, we want it, but is that sort of really what our secret sauce is not. It's not. So we want to go partner with an [indiscernible] to say, "Hey, if customers are choosing an experience to go build something, we just want to make that part of the business AI platform a way that we can add another N9 or N10 and whenever it comes up in the future to make sure the platform itself is complete.

Same thing we do on the hyperscaler side, on the connector side. The thing that differentiates us is this, and I want to spend 30 seconds that I don't have on this slide still, which is listen, the public LLM part is what people can generally go build an agent on, and they're getting smarter every day, right? The OPUS 4.6 and the new models are going to continue to get very smart. But the things that we do that again, as you sit in another tech conference that you should think about, like, does this really exist when they say they're going to be the best agentic platform is a knowledge graph of our canonical model, right? That's already over 7 million fields and the relationships that we need to go maintain because that's very hard to go do.

The process graph because we cover end-to-end process, these 120 top level but thousands below as to where things fit. We also have domain models, over 2 billion lines of code that is not in the public domain that we train, pretrain and then make it available with the public LLM to say, if you're building an extension or an app or an agent and SuccessFactors, ours is always going to come out to be closer to the pin on the first iteration than just using the public model. I mean, that's just common sense, but we'll prove that with e-valve as well.

And the fourth one is predictive models. This is where we shine because we don't just have the shape of the data as our IP, which is the knowledge graph. We have the data itself with tens of thousands of consent an aggregated anonymized manner, that we're training this model on to say, "hey, we can give you predictions on the fly and really make machine learning science a thing that could be disrupted next with a model business, right?" These are the tabular model. So we've done that for SAP data, which is Rapid One. But then we said it's not going to be enough just for SAP data because customers always live in a world where they have a lot of non-SAP data and hence, prior labs because we want to bring the best of both to be able to bring into our context.

These 4 things, I would ask, and you can humble me if you have an answer that I don't have, like who else can give on top of the best public models because it's not an or, it's an in. And then finally, that's the SAP context, it's the customer context that we already know that we can embed in the platform, which is every customer has extensions. So the knowledge graph is not just what we canonically ship, but the thousands of customizations and extensions that customers have, we know that. We bring that in, in a dynamic way. Company memory, which is the customer context that's process mining, process insights, process models, e-mails, chats and others, we announced that yesterday as well as part of our Agentic platform.

And finally, the agentic run time which has the governance built in, which has the sovereignty built in because it's natively on BTP. It can run anywhere you want. Like you add all those 3 things, and it gives our Agentic platform a differentiation that's going to be very hard for a hyperscaler to go do because that's a bit hard from a sovereignty perspective. They don't have access to the shape of the data or the data at the scale or the predictive models and so forth. But anyhow, this is what leads to -- you can see the evals here, right?

There's about 5 agents that we tested multiple times with just a different kind of public model and then a model that has that goodness in it. And you can see we beat each 1 of them every time. And that's only going to get better. Now the public models are going to get better. But of course, the context that we have and the company context that we build every time a customer runs on it, is going to get even better and better. So both will continue to show some difference in movement, if you will.

Now finally, I'll close with because we've talked about the data. We've got the [indiscernible] on the [indiscernible]. Again, on the [indiscernible] part, you should think of that we have MDD, but we knew because we have to go from an apps app boundary to landscape boundary that we have to provide a solution for non-SAP master data management, hence, Reltio that comes together with our MDG. Because we have to, again, go to the landscape view, we need to be able to be able to access that data that's not SAP on the fly as well and hence [indiscernible] on top of BDC to be able to go to. So you can see that story is now building up from a clear product strategy perspective.

And finally, I'll just stop here. governance, we've talked. Actually, the reason why -- because you also hear a lot of control towers, you'll hear a lot of agent 365. Like our right again to win on this is the platform that this cycle is built on in a seamless way is a set of platforms that we're largely leading in already. So LeanIX is one of the leading enterprise architecture platforms, which already understands your landscape beyond SAP as well as the best place to understand where all your agents are. And that's why we already have 55,000 agents registered on them, and they're not all SAP agents, most of them actually are not SAP agents, but [indiscernible] already know them because they are the enterprise architecture solution in our customers' landscape.

Signavio already does mining. So adding agent mining was just in the world's best mining platform, another capability to add. Cloud ALM was already a leading platform for observability, and we continue to active. It's our right to win in agent governance. And then finally, sort of tackling it all the way to the end to success factors to say, listen, the total workforce discussion for our customers is going to be a pretty important one as it is for us, as Sebastian is going to talk about that, your total workforce is not humans, if not FTEs and contingent, but FTE's contingents and agents and how we brought that together with SuccessFactors is pretty unique.

But finally, I'll stop at this and then hand over to the next presenter here is the Joule, the engagement layer. And what's our right to be in this. Now there's 2 points here that I want to talk about, right? One is there's a lot of feedback out there in the public domain on Joule. And I think some of it is actually fair. Now there are some good successes as well. But we know, as Christian talked about earlier that we had a lot of learnings in Joule. So if you look at what Joule is today and what we're now launching and already we have with some early customers is a cloud-based harness that completely reimagines Joule, both from a deployment experience perspective as well as the value perspective.

Internally, honestly, we call it V10 because we thought the difference between what we have today in the market, B1 this was so massive and so redone that it's not even V2 or V3 or V4. Externally, we didn't brand it like that. Our product marketing folks prevailed on us, but this is a massive uplift. So you will hear feedback if you call a customer and say, "Hey, what do you think about Joule?" And I think some of it is actually fair. There's obviously success stories like Ericsson out there, too, but this is what we're completely reimagining.

And then finally, we announced something called Joule work. This is that engagement layer, right? And the reason why we believe this is something we again have a right to have a significant market share in as we have over 300 million end users, right, that engage with our applications, so they're just in the cloud space. There's another few hundred depending on how you estimate in our on-premises estates in ECC. So for us to provide an interaction layer that understands those applications to say, hey, now we're actually solving our evils of the past because we have a lot of bad rep on user experience as well. You can argue in our earlier core ERP apps, is something that we believe that will resonate and is resonating with a lot of customers as well.

Now I'll stop with this and hand over to, I think, Thomas, because I was having this conversation earlier. Listen, if you go out to the market now, and you pick any random pie customers to say, "Hey, I want to get your feedback or maybe any random [indiscernible] partners, right?" I think what you're going to find is a couple of things: One, the strategy that we outlined yesterday is a strategy that is not a vision. It's not Figma, it's not vaporware, right? We have a lot of early customers that through FTEs we've been working with the logos that you see here, the examples that you saw all day yesterday and what you're going to see today are people -- customers that we've been working with through this FTE program that we've had now for about a year to be able to go work with them. They have the experience on the new platform. The others, and there's a lot of them are still working on our previous platform because we're now just publicly announcing our new stock and mid-June is when we're going to scale out the rollout of it to any customer that wants it is part of our early adopters.

So you're going to find 2 classes of feedback. And you need to be able to ask the question, "Hey, now, do you have the experience and the new one that was just announced at Sapphire? Or is it the old one?" And you need to be able to discriminate in your assessment to say, "Hey, what strategy are you outlining here?" And as as Christian pointed out, I think the industry AI part -- I didn't touch on it. Christian did, but it's a big, big focus for us. And this is where we're setting up an organization that have now thousands of FTEs from SAP with already deep understanding of industry and our products that are going to engage with customers at scale a majority of them in engagements to activate these domains and this industry AI as well.

So now is when we're ramping up, hey, we've got the platform that we feel is now in some cases, better than best-in-class because it's using the best-in-class and adding our context that slide, if you remember on top of it, and you want to go unlock value if you hopefully, this makes sense. Now I'll hand it over to Thomas. Thank you.

Hello and also welcome from my side. And Dominic challenged me a little bit. If we can accelerate migrations by up to 50%, if I can accelerate my presentation up to 50% as well in light of the time. So let's give it a try. No, but I'm really looking forward to talk with you about how we evolve our go-to-market model in the company. And in order to do so, I think it's always good to basically remind ourselves which world and environment our customers are living because that's basically the expectations we want to not only meet but actually exceed. And when you look at the world, I think we all see the volatility, we all see uncertainties. We see that the world is changing faster than ever before. Geopolitical fragmentation, economic uncertainties, supply chain disruptions, and this is just compounding. And for sure, this is a pressure on all of our customers. In parallel, we see this little disruption called AI, but we also talk today a lot about that. So on the one hand side, there's cost pressure. On the other hand side, is how to get value of AI. And that's coming together. So the customer expectations are raising up. And for sure, they want to have faster outcomes because the pressure is high.

The world is changing more quickly. But also here, it's clear that resiliency is becoming more important for them as well. Resiliency is one of the highest value driver they see. And then for sure, a huge pressure in the market about return on AI because everybody is piloting and POC-ing some of the AI capabilities, but also if you see the various research in the enterprise context, for sure, it's clear that it's a little more complex. And that's what we discussed the last 2 days here at Sapphire how we, as a company, want to help our customers to overcome it.

But it also means for us, we need to adjust ourselves that we need to serve our customers in that world even better. And that's what we do and what we already prepared. So for sure, we focus on adoption services. So our services transformation really driving to get into an AI engine, a deployment engine for our customers, how we evolve our business model with consumption-based, outcome-based pricing in that sense. Also thinking about the consumption of value realization, which we see with all of the agents, all of the AI capabilities what we drive and also our responsibility to help our customer in doing so. And that's why we basically also elevated our services and support portfolio to the next level and included the AI capabilities and deployments as part of our success plans with our customers.

That's also the reason why we brought together all the respective customer functions from presales, sales, post sales, service and support and cloud operations to really have a delivery engine for our customers to make that happen. And we said also, we simplify the entire customer engagement throughout the customer value journey. And this is, for sure, something where the customer experience is improving dramatically. It gives us also the opportunity to inject the relevant resources in the mix, and I will come to that as well as Christian [indiscernible] to it, how we leverage our technical skills on the services side, also in presales and sales to really drive these AI conversations with real prototypes, with POCs on the spot when we are at the customer because that's one of what [indiscernible]. But we should also ground ourselves where we are. I mean we talk about 440,000 customers.

If we zoom in now to the Global Fortune 500, 91% of them are SAP customers. And 68% of them are already using SAP Business AI. That's the scale what we have. And scale for us is a critical component, which we for sure want to enable with our workforce. And if you think about this customer value group, which we brought together, touching the entire customer life cycle, as I mentioned, the seamless journey is something what we want to enable, but also think differently how we work. So also embracing AI and the best team we talk about how we internally embrace AI to serve our customers. For sure, we changed the entire motion about consumption-led models. And that's also, by the way, part of the bonus plans of all of the people in this organization. So harmonized KPI, which is fully consumed ACV on our customer side. also leveraging the deeper expertise and especially with AI, it's clear that we need to think about how can we embrace AI in a complex enterprise landscape.

And that means we use our architects to make that happen. So for the deployment of AI, we leverage our colleagues as part of the success plans to really activate all of these cases for our customers. And that's, I think, a critical aspect of what we want to see.

And then we talked for sure about also how we leverage AI itself to proactively guide our customers to proactively deflect tickets and support and the like to really put that on steroids. Now if you think about the growth lever, I think there's a common misunderstanding and believe that we only grow by basically migrating our on-premise customers to the cloud. This is wrong. And I will show you a little bit where we are here because actually, our growth is extremely diversified from new customers where we land and expand our existing Forshaw customers with our proud installed base, which we have, but also thinking about the cross and upsell based on the new portfolio based on the new innovation is what we see with data, but also the huge opportunity we have in sovereign cloud.

And I also want to touch on that one a little bit in a second. Now if you think about purely the cloud growth from 2020 to 2025 and think about the new customers, the new logos, which we acquired, daily up to close to EUR 3 billion in cloud revenue, which but you also clearly see already here the indication how much cross and upsell is happening there. So our new cohorts, which are coming since 2020 are significantly accelerate with cross and upsell the revenue as well.

What we also see, because our mid-market engine is really now at steroids where we also use the indirect channel partner territories. And here, you see some of the numbers, I mean, 1,000 partner territories more than 3,000 sales partners, which we have. And they also -- when you see the cloud revenue growth comparison of the indirect and the direction are really at speed. And that's what we see on aggregate, which means our grow with SAP offering for small and medium-sized enterprises, 4 start-ups, 4 scale-ups is extremely important for us.

And also here, I mean, if you think about companies like Snowflake, Databricks, I mean for sure, they want to grow infinitively and which is the single only 1 ERP system on this planet, which you cannot outgrow, which scales infinite, it's SAP. Once they sign up with growth SAP, they never need to worry about any local market, 160 local markets, all legal regulations, all taxation system. You don't just do what you just can grow. When you even come to customers like NVIDIA, which shows how amazingly these systems can scale when Jensen said that he wants to do $1 trillion in revenue next year. This is 1 ERP system. This is just unlimited scale.

And we saw the other customers on stage today, which clearly shows that this is a big advantage. So basically, you already see the contribution of 20% of our cloud revenue growth in the last 5 years coming from new customers. And we see with the SME share that there's even a huge potential for us based on our new portfolio to really accelerate the growth also of our net new customers.

Now if you think about our installed base, if you think about a like-for-like comparison from the support to the cloud, then you see it is 2x multiplier. But actually, with rise with SAP, we're in the moment of the landing of the offering already do up and cross-sell in that motion. That leads up to the 3x potential what we see. But also here, you see -- if you think about the numbers, again, how much cross and upsell we do in our entire installed base. So it's a nice mix of new customers about customers migrating to the cloud. But even more aggressively the cross and upsell what we see across our installed base, which I believe is a huge opportunity.

And you see some of the facts here on the right side as well with our customers use our cloud ERP and our BDC as a beta platform, then you see that 9 out of these 10 customers have more than 4 cloud solution in place. So you clearly see the drive, but also are starting this flywheel from AI data and applications. And with that also to cross-sell across the portfolio because to Muhammad's point, the more you in aggregate in this agentic AI world bring together, the more value it adds. So 1 plus 1 does not equal 2, but actually 3 in the world of AI and what we do here. Now how does it look like in our usual customer cases?

For sure, we start with cloud ERP. We start with [indiscernible] with SAP conversation. And for sure, business transformation management components are already vehicle because for sure, to support the transformation and accelerate the migration. They need these tools like Signavio, Linux, which we directly also package together for our customers. But also in a rise migration, what is important for you to know, it's not just that we do on steel, and that's basically the flat line. Actually, we have customers like Bosch in Germany, they have more than 350 productive ERP systems. Now we can multiply that by 5 or 6 from a system landscape perspective. So we see thousands of ERP systems which we migrate into the cloud, which means you see the ramp and the journey over time, which means that the revenue is increasing over time even further.

In the meantime, to give you some context, we operate more than 190,000 system [indiscernible] for [indiscernible] with SAP customer at scale. So it's the largest operating scale in the market and growing. And here, for sure, on top, we use already the business technology platform for all the customer extension, for the customer bill, the differentiation, the integration capabilities, which is part of that and further compounding where we then also see the business data cloud as the flywheel to really add more and more of our cloud LoB solutions on top of that. For sure, in order to help our customers, we started already our services transformation 2 years ago to really shift away from a traditional services business, but really going to an adoption engine to basically really focus on adoption and consumption for our customers. And that led also to the establishment of the new success plans, which we have in place. And here, we've embedded the activations. We've embedded it. We help our customers building up AI COEs. So it's all natural part. And also, we evolved our engagement model that with rises SAP, with our enterprise architects, we actually help our customers to get to that point. And for sure, with our agent-led migration and tool chain, we not only support the migration in modernization, but also the continuous innovation, the continuous transformation our customers do, and that's a huge advantage.

Now when we talk about a reduction of the migration time and effort for 35% to 50%, that also means that we can also shift this capacity in order to support our go-to-market capabilities exactly as Christian mentioned, from the presales capabilities to the adoption capabilities, and we can handle that within this workforce transformation in itself. What is for sure, super important for us as well to handle that volume is that already 85% basically of tickets from our customers are reflected by self-service, by AI capabilities that the tickets are even not getting created.

And that is with the scale of the growth we have with new logos is super critical for us. For the remaining tickets, which actually get opened, already 20% get actually fully solved by agentic AI. And actually, customer efforts so the customer satisfaction is go for our customers is even better as it would be a human so we see even a further customer experience improvement with AI.

So AI is not only for the productivity, but also for the customer experience in itself. Now we talked about these migrations. And I think I don't want to go into details in light of the time, but think about this market of EUR 100 billion plus for SAP migrations. And for sure, with our [indiscernible] migration, which we just launched today and showed in the keynote how that looks like with the reduction potential which we have. We not only on the 1 hand side, get the benefit of additional revenue streams based on that, but also for sure, the customers are freeing up their budget.

And that means they can also invest more into innovation results, which is a good thing. And another good thing is there was a lot of worry in the market about, oh, what is happening with the maintenance and yes, ECC maintenance is 2027. The extended maintenance is 2030. Now with AI, we can help our customers to accelerate these transformations into the new world. And that's also a benefit for our customers, which we leverage here big time.

Another growth lever is Sorin cloud. And here, just to remind ourselves, we talk about 60,000 critical infrastructure customers. 23 NATO armies are running fully on SAP, just to give you some context where we are also in the defense industry. You've seen customers like Lockheed Martin on stage RTX, [indiscernible] across the world are betting on SAP, 9 out of 10 defense manufacturers here in the U.S. rely fully 100% on SAP. That's where we are.

And because of our architecture because how we engineer the serenity, we can serve those markets. And I think this is a huge opportunity what we have. What is important is that if you think about serenity, what does it really mean? What's the definition of serenity. And what we offer, which is unique to us is on the one hand side, the data serenity, an operational sovereignty, a legal sovereignty, especially if you think about also Europe and Germany and the likes, and also technical sovereignty which is absolutely essential. We already delivered it in various forms of shapes.

In the U.S., we have NS2. In Germany, we have a vehicle called Delos Cloud, but also we have capabilities like sovereign cloud on-site options, where even for really critical customers in highly defense activities, bring our Sorin cloud on-site on the customer's place with the physical security of this defense organization in itself.

And this flexibility, what we have is based on our platform, which is on the one hand side, agnostic from an infrastructure, so can run on our own SAP infrastructure can run on all the hyperscalers, can run on partners in Germany like [indiscernible] or [indiscernible] Systems because this platform is super flexible and agnostic.

On the other hand side, in this platform, based on our partnerships, we natively embed all the AI capabilities from here, Mr. [indiscernible] OpenAI in this platform to be able to only deliver our entire cloud portfolio, but also the agentic AI world, which our customers need. And that's differentiating us in the sovereign cloud. And that's for sure, a huge growth driver, which we see, especially based on the geopolitical evolution, what we see in this world.

At the end of the day, I think the key aspect is always when we talk about customer and go-to-market that we actually talk with customers and get some insights from our customers first time about how they work with SAP, what they do with us in the cloud with AI specifically and here, it's actually my pleasure to welcome on stage the CIO of Ericsson, Marlin [indiscernible].

Good to have you here with us. And perhaps we start with the -- for the audience a little bit to set us a little bit up with kind of what triggered and started the Ericsson transformation and the challenges we were addressing on the outset results.

I think that your starting slide was actually spot on, right? We are running a quite traditional company. and our core business being hardware being network technology, fueling the global connectivity is quite cumbersome. There's very few customers, and they are struggling over their margins. So in order for us to be relevant, we needed to transform. So we started that kind of journey. And the same thing goes when you're having that kind of landscape of customers that you really need to accommodate, right, it also drives complexity internally. So the way we're operating was really hampering us from both safeguarding bottom lines, but of course, also be relevant towards our customer base.

Yes. No, absolutely. -- nothing about where you are today because I already mentioned earlier about what we achieved together. Can you describe a little bit the path how we got there. And also even more important, the status today, the outcomes you have and some of the learnings based on the journey which we have together.

No, great. As we started out, of course, it was more about the clean core and what can be utilized on that actually be a catalyst for necessary simplification and transformation. But as we have progressed of course, it's evident the trajectory of the development of AI is just massive. And of course, when we're talking about Gen AI and Agentic, there's a lot of things that need to be done. And I think that we have truly showcased how we are parking up on that. And I think the communication you did on the BDC last year, that was a missing piece of the puzzle because that was really the foundational element that kind of allowed us to fix the fragmentation that we had.

So when we're looking into where we're at at the time being, starting point, I would say. There's still a lot of things that we are exploring and looking into, and we have high ambitions for this year together. But what we have done and what is in live practice is for instance, some really good agents within the HR domain. They are highly appreciated by all our employees and me being a manager also highly appreciated [indiscernible] it says a huge amount of time then there are more perhaps complex ones being implemented in the supply chain, where, of course, we can get substantial tangible values.

You mentioned already the AI and how you use it in supply chain in the employ space. Can you give us also a little bit more data about kind of what kind of numbers do we talk about here with the adoption of AI in your business. And with that, for sure, the employ experience and how the people in the daily work fundamentally start changing to work based on AI.

Of course, it's starting to change. And I think we are beyond that phase when it's more exploration. So all of our employees actually have access through SuccessFactors and other thing, which of course, is simplifying data. So personal productivity is obvious, right? That's already there. And then we have the agent build of different sorts, which is getting much more in the hands of the democratization of quite a few uses.

No, absolutely. So we've talked about more than 80,000 users who use it actually on a daily basis, and that's real today. And that's what you heard at appreciation, how people will start working differently. That's exactly what we want to do. And that's exactly where with our customers, we bring that not only to our portfolio, but as Muhammad and Christian shared, really also with these new capabilities now on steroids, which is certainly amazing.

Now if you think about a little bit your churn, also a little bit ahead, what do you see in the future? Where do you see us playing also? You mentioned already data as well. So where do you see us in our joint journey also for the next couple of quarters to com.

I think, to be honest with you, we actually changed our data strategy. That was a necessary thing when we're looking into agent trade flows, right? But when you communicated the BDC last year, we actually fit things around. So we are depending heavily on your trajectory and road maps within the full domain of the autonomous enterprise, right? We are moving quickly. We have already quite a few things in line running in first quarter, but the ambition levels for '26 is that we're going to have more than 100 in live running. p

Well you heard that 100, but also, I think it's a good example because you mentioned, I mean, RISE with SAP, BDC, the flywheel with AI and how that's coming together in real life on a customer. And I think that is for me the perfect example how to drive that change. And just perhaps as a last question because I truly believe in the world we are living is also a little bit about -- it's about leadership, and we talked also about that.

What do you think in such transformation now embracing AI in the enterprise and some customers for sure, struggle and you fundamentally changed this upside down in Ericsson to really be AI first and do all of that. So give us some context about also from a leadership perspective, what you see in the age of AI?

I think, firstly, it needs to be a commitment top down. So when you're talking about AI first, that can easily be something that you see as the [indiscernible] for it and then you're facing master resistant entire organization. So it's everything about how you're changing your metrics, your incentive models to actually drive a new set of behaviors and also allowing for a safe zone for people to explore and learn because if you're just saying that is for productivity to reduce workforce, of course, you're going to get that resistance. So I think that leadership in this never been more important. I think that is 1 of the key things that we need to add.

No, absolutely. No, first of all, thanks so much for sharing on stage here with us.

Thank you so much.

So I think what you have seen is here, AI is there is here now. driving value for our customers already now. And I think Ericsson is a perfect example. More than 80,000 people experience SAP's business AI every single day. I think it's a great testimony of the strategy, which also is again proving basically the topic about these various components, which we now bring together enable us to take that to the next level, which for sure from a go-to-market perspective is exciting because the new platform, the new autonomous suite will help us to even more simply scale through the customer base with the capabilities which we have. It's certainly an exciting time. And I'm certainly looking forward to work with all of our customers to bring our autonomous enterprise to life with them. We are ready to do that. We have the right organization in place. So I'm super excited about what's coming. And with that, thank you so much for your attention. I want to hand it over to Gina. Thank you.

Thank you, Thomas, and also good afternoon, I can say already also from my side. So actually, you have heard Christian presenting the strategies. So he was setting the strategy for the autonomous enterprise and my job is now to focus on what we are caring about next. And this is actually how to do a repeatable execution because we need to have the talent. We need to have the skills. But we also need to have the operating model in order to enable delivery at scale. So it all goes hand in hand because AI won't be a growth driver because it's powerful. AI will be a growth driver because we have to enable organizations, only then we are able to produce value. And autonomous enterprise means actually that we have no agents in place that do the work proactively end to end.

And then you have humans in the loop that is making this the decisions and also the judgment because it's important that agents are operating in boundaries with clear accountabilities and also a clear governance around that. And this is actually why values created only when technology operating model, but also the workforce capabilities move in lockstep. And we have looked at that very holistically and I will also go into the pillars in a second with very concrete examples as well.

Because to translate also AI investments into consistent results, we understand that we have to fundamentally rewire also how we as SAP run. It's an operating model shift. And it's not just putting AI as a layer -- on top as a layer. Christian already shared our growth formula when we talked about product times go-to-market times people times operations, and this is leading to AI led growth. And the CHRO, of course, I have to underscore the importance of our people because this is actually how we can then also unlock capacity speed but also the operating leverage.

In order to achieve that, we have now built the transformation backbone, how we call it with 4 integrated pillars and this is all supported with our skilled foundation. Because with that, we can drive actually scalable and repeatable execution, but we can also expand execution efficiency, which is also extremely important. So let me quickly walk you through the 4 pillars.

So the first pillar is very much about enablement and adoption, extremely important 1 because we are investing heavily in our employees through let's work the Away -- we are combining structure change management, together with hands-on enablement, which is important, that is not theoretical and abstract and also real tool usage in daily workflows because our employees have to test that have to experiment that have to use it in order to get adjusted and learn about that.

And this is, of course, also reinforced and this targeted app and reskilling. We have put out a role-based and skill-based learning journeys and a lot of learning recommendations. And I will also speak about that in a second to make it even more concrete.

The second pillar is extremely important as well. So it's all about structure and process redesign because organizational development is so important but when we are moving towards an autonomous enterprise because as humans and also agents work together hand in hand, we have to redesign the roads because now you have an overlap of tasks, agents will take over task. So we have to make sure that we know, okay, what is in the role and who is doing what, that you have a clear task it. But we also have to change organizational run models and also processes. So -- and when you are also redesigning everything that also requires that we have to look and reduce hierarchical complexities. This could also mean that we have -- it also includes actually that we have smaller teams, more agile teams with less decision makers. And this is also helping us to drive faster results and also have go-to-market speed.

The third pillar, which is also important, is now strategic workforce planning. As I just said before, because now we have tartan accountabilities together between split between agents and also humans. And that's why we also have to make delivered by [indiscernible], build and also automate decisions. And then we have to say, okay, where do we invest now also in strategic AI skills. And [indiscernible] also includes simulations, actually, of workforce compositions, the capacity but we also have to look where are skill shifts happening and what is the right role and also location mix.

So those are the questions we are also answering with the execution backbone. And this translates, of course, them into the right decision to say how do we do also talent shifts because we can't be combining here very clearly skills led hiring, but also we are buying skills with selective M&A activities. And we are also investing, as I said before, in target up and reskilling for our 110,000 employees that we are securing also critical AI talent and critical AI skills. So this goes hand-in-hand. It's all 3 that are important. And because skills are changing fast, we are always talking about a so-called skill flux.

So because the duration of the shelf level of skills is only 6 to 18 months. We always need to have a system in place that gives us a current view of our capabilities. We were building now a 700 skill taxonomy. This skilled taxonomy, we have translated now into an updated drop in kill profile is extremely important. And we will also have that up and running in our SuccessFactors growth portfolio by middle of the year because we have to measure proficiency and we also have to see, okay, what is our scale inventory. You need the transparency. Otherwise, it's we are unable to drive the workforce transformation. In this -- and this data is also informing of the game, okay, where do we have to hire and where do we have to also improve our learning portfolio. So bottom line is we are reviving the organization to enable AI, but to also -- and building also the execution engine that makes AI outcomes also repeatable, extremely important.

But now let's go and deep dive a little bit also into the pillars. So the first one is, as I said, AI transformation really fails on technology. It normally fails on the lack of adoption. And that's why we also drive enablement and adoption through our unified approach, how we call let's work the way campaign. So this transformation program is actually combining different formats of learning, different communication formats but also different change management formats. Because you have to drive the transformation at scale. And we don't -- we cannot stay actually with only localized or very fragmented usage of AI.

So we have to make sure that we scale across the organization. And for that, everything starts also with communications. We have designed a global narrative that is also saying, okay, why is it so important for everyone at SAP -- for every employee SAP to use right, and to make sure that we are freeing up time also of our employees for higher value tasks.

And we are also reinforcing it with our growth culture. Also, Christian talked about that, and I think we have proven we have a very strong culture that has proven that we were able to adjust for over 52 years meanwhile. But it's important to keep the culture strong and to always make sure that we are putting it into new context for our employees. And we are also doing that to apply performance management. So in 2025, and I think I spoke about it also last year on stage, we introduced the growth culture. And meanwhile, we have driven transformation journeys for more than 2,000 leaders, and we have also reached more than 15,000 employees in growth summits around the world in order to strengthen the innovation, the customer focus and also the impact.

And one of the flagship promotes is also what we call grow with AI sessions alone this year, we have already reached more than 13,000 employees with these kind of formats. Those are formats where people come together and where we can also share experiences with AI usage with tools, but we are also using very focused change agents to speak about, okay, what are the barriers today in order to use AI? And what are the misperceptions also? Because it's a holistic approach.

It's not just a tool usage and to understand, okay, what is possible. I think it's also important to engage the employees in these kind of conversations. And then from there, we say, no, we are scaling actually also the enablement with 3 levers. So the first lever is actually that we say is skilled learning. Skilled learning is, I said it before, that we have now AI-based and role-based skills and learning journeys. So we're investing heavily also in trainings. We have one example I would like to share this is the -- our AI developer program with 2 learning tracks. So the first 1 is tool developer. It's for developers who are then building and building extensions and integrations with Joule Studio. And then the second track is actually intelligent agents. This is for developers who are building autonomous agents or multi-agent systems. But we are also complementing formal learning with practical learning out of the projects we are running with our customers.

And we are bringing that back together with our centralized AI customer research team in order to have practical learning and then bring it back into the project. Additionally, we also have launched just in January this year, our so-called Learning Navigator. This is a single point of entry for our AI training. We have meanwhile more than 49 learning journeys for the selected profiles, and we have more than skills who are directly linked to trainings. And we keep that also current actually this quarterly learning priorities. We also set out a target last year to say 15% of the learning of the working time should be also invested for learning.

And we are complementing that on the one hand side, with quarterly learning priorities. We have also more than 3,000 AI courses internally and externally. And we will also make that adoption sustainable even further also after Sapphire because we are putting out 2 hours of protected learning time per week for our employees. Because this is the feedback we're receiving also in the employee services that give us more time to learn.

So this is what we're doing because AI capability is a must. It's extremely important for everyone to learn and embrace how to use AI. The second pillar or the second lever, how we are also scaling is actually experimentation. So we are running code camps across SAP, codes camps enhanced on build sprints, where cross-functional teams are working actually on real business AI problems for our customers. And this is also a very practical way how we can transfer learning directly into impact for our customers. In addition, and I also said it before, people have to use it. So we have also released more than 200 tools for our employees. It's clearly linked to roles, but also to the workflows because it has to be relevant also for our employees.

And those are tools like dual work or also Claude Code. The third lever is how we scale is actually through our multipliers. So this is also when it comes to change mine trend, when it also comes to large-scale adoption, it's a peer-to-peer influence. We have meanwhile an ambassador network with more than 9,000 ambassadors. We have, since this year, beginning of this year, increased that network by 60% alone. And this is also how we try to remove the adoption bottleneck.

Let's come to the second pillar. And this is something where we really have to look on how can we wire SAP. We are experimenting, and I also say that very clearly. So this is something where we have pilots, and I will also speak about it in a second, but it's important that we have a very clear [indiscernible] star of how does an agentic company looks like. And how can we also design the operating model that is actually integrating humans at agents in the operating model. Because it's very clear that humans have to stay at the top and always have to stay in control.

I think this is extremely clear. But when you look at the current models we are having, we have verticals, we have old boxes. And now we have to shift into layers because you also see -- saw before on the slides, how it works. So we have, on the one hand side, you have the so-called strategic decision layer. This is where the human is always in control and is also making the decisions and also decides, okay, what are the problems, what do the agents have to do and then we also have the second layer. And the second layer is how we call it the orchestration layer. So this is where Joule sits, and this is where the Joule is orchestrating the agents, and then you have the execution layer underneath.

And the questions we are asking at the moment is, okay, how do we bring these 3 layers in swing. What do we have to change, especially on the strategic decision layer because this is where our team sits. And then we also have to ensure how do we integrate actually agents together with our teams. So the separation is, of course, deliberate because humans have to decide actually what and why we are doing things, Joule orchestrates and the how and then the agents actually execute at scale. But now the question is how do we actually transfer that now into our own organization. We have started with a pilot in Muhammad's organization. It's the pilot we are driving because in the P&E area, you can see that this is where work is changing extremely fast at the moment.

And this is where we can also see where execution discipline is also most visible to our customers. So we have introduced last year, the so-called AI native harmonized product operating model in [indiscernible], which brings products, engineering and UX under 1 shared accountability. The point is here that we have fewer handoffs that we have faster decisions and also faster time to market. And that was the foundation actually to say, okay, how can we now evolve the system end to end? And how can we now integrate all the questions I have asked before. So how do all look like? How do skills look like? What do we have to infuse. How do we have to make sure that our people learn how do we put change formats in place and how do we also hire.

And we have tightly connected everything in order to start shifting also the operating model in P&E. So first of all, of course, we had to look at the role of responsibility. So we have a deep pay design of the most critical engineering roles, engineering and technology roles with very explicit AI responsibilities and oversight and also here, we are already considering actually the impact of the tool usage but also the integration of agents. And then when you look at skills, we have now infused more than 30 AI skills into our skilled taxonomy because it is cleared taxonomy, what I mentioned before, the 700 is not static. So this is changing all the time. We are adding every quarter, we are adding skills, we are removing skills again because of the relevance. So just recently, we added 30 new AI skills also into the taxonomy. Those are skills like contact engineering, rapid field prototyping or also AIS is development. And everything is directly embedded into the role architecture.

Year-to-date, when you look at learning and the P&E organization, we have more than 7,000 AI learning completions already recorded. We have run more than 65 AI experiments with real AI business problems. And we also have tested more than 73 tools that informs also the SAP wide adoption. And in June, we will also start rolling out a code camp's in Muhammad's organization in more than 26 locations in order to make sure that we're bringing our teams together and that we have hands on learning again.

And then we also shifted the hiring, I'll also say a few words about that in a second. So we hire for AI critical skills through very targeted and very active, proactive sourcing AI hubs around the world.

So we are hiring in Munich, Berlin, Singapore, Palo Alto, but also in Bangalore. And then change management. Also here, we have used existing formats already because it's important that the contact is relevant for the employees that we are not putting artificial change formats on top. So we have used existing formats in order to drive change. So actually, P&E is, for us, the foundation. It's true point for us. We are now seeing, okay, how is this evolving? Also with the Norstar I have explained before in mind. And then we will also roll that out across all other board areas, of course, with adjustments because for corporate functions, it looks differently than for the P&E organization, but there are some organizational design principles in common, and this is what we do.

So let's look next also how does it look now for the overall workforce investment and mix. So strategic workforce planning is extremely important because we are now continuously analyzing actually how workforce implications look like. And we also have to make sure that we have the right role and skill mix in place because our internal agentic AI road map is now in place, and now we have to translate that how does it look like? How does it impact the workforce composition? How does that impact also the role compositions, which task is automated, which tasks are stay and how do we then have to redesign the roads. So first of all, I also want to share, okay, where are we now leaning in because we have to make strategic investments in certain profiles and just as an example, so we're investing here very clearly in machine learning, engineering. We're investing in data science, enterprise architects, but also business in data platform consultants.

And then you also have roles that needs to be reshaped very clearly. So we have 1 example because autonomy increases, we have to see, okay, how is that role impacted and where do we have to invest to make sure that we are that we are redesigning the role, but that we also have the right learnings and the right training in place for our employees. So roads such as quality management and user assistance developers, supporting roads and shared service roads, those are roads that need to be reshaped when you also have the vision at the new North Star in mind.

And then we also try to make that, of course, actionable. So how do we make that actionable. So we have defined now, on the one hand side, core AI skills that go across all roads, context engineering, AI assistive prototyping, iTrust and verification. You can see that on the right side of the slide. But we also have come up, as I said, with the AI skills that are now embedded in the role architecture of selected profiles. So here you see machine learning engineer, the skills we have embedded are, for example, genetic engineering, cemented with people or evaluation and benchmarking engineering. So this is important because we say, okay, how again, we translate that into learning, but we also translate that very clearly into hiring plans depending on how do we have to infuse the skills.

And depending on the strategic workforce planning, do we have to also buy or build the capability of the market. So this is actually how how we create structural operating leverage. So -- and we are allocating capability to way AI I can scale execution.

So now let me also share how are we now investing also in these profiles and how do we also approach the hiring. So we understand, of course, that we -- that AI execution requires also different talent in very specific roles. And this is just an example from the roads I showed before on the previous slide. And here, we have a twofold approach. We have 85% at the moment where we say we are doing up and reskilling. And 15% is where we invest in hiring of senior experts in the market in an expert talent I have spoken about up and reskilling, but I would love to speak about now how hiring has changed using the 4 profiles as an example.

So first, what's very different in our hiring approach, and you saw it already on Christian's slide because we say we have a 3:1 ratio. That means that we are instead of hiring the developers or standard developers with standard skills, we are now investing also in top caliber AI talent because, on the 1 hand side, the productivity impact is meaningful and especially in development where we see roughly a 30% uplift already. But that doesn't mean that the cost base is going down because we have to invest in these care talent. And [indiscernible] talent is quite in high demand in the market. So we also have to make sure that we have premium salary packages in place. And that's why we are also redesigning also the com package at the moment in order to be competitive out there. But we also have shifted now the way we are recruiting our talent because we are coming from optimized cost location mix, and we are shifting towards a top AI talent market to secure these expert profiles. And we are also scaling seen an expert hiring where most directly drives build velocity, but also customer value organization. And you can see it also in the numbers already, how we are shifting and how we are changing our hiring.

So you can see the numbers that our AI relevant senior hires have nearly tripled from '23 to 2026. And since 2025, we have made already 6,000-plus relevant hires in engineering and technology alone. And in data science and also machine learning engineering, we have seen a 200% increase in hiring just about the last 6 months. But we are also bringing in exceptional key AI executives because from the market because the caliber of talent is necessary to accelerate also our AI ambitions. And what is also important is that we are also building our own internal pipeline because even though we have -- we are more selective also in the volumes we are hiring, we absolutely firmly commit that we are also getting AI native talent into SAP through our academies and our vocational trainings.

We have meanwhile in every vocational training academy program, AI installed also as a curricular as a strong pillar. And 80% of occasional trainings are already in IBD programs, especially for machine learning, engineering or data engineering. And what is also maintaining strong that we are also -- also this year saw in Christian slides that we are also collaborating with top-tier universities like Stanford, [indiscernible] or UC Irvine. And altogether, this is actually how we are also removing the talent constraint and to make sure that we're able to execute our autonomous enterprise. So with that, I would love to hand over to Sebastian. Thank you very much.

Thank you, Gina. By the way, I love what Gina showed not just because it's super important, but also because all of that is powered by our autonomous HCM solutions, and that's just 1 domain. But now, look, my job as a COO typically is to keep us on time and in budget. I have some good news to share on the letter, and I will make up some minutes on the former. So hello, everyone, right to be back. Now my real job is actually to turn SAP into an autonomous enterprise. And that means both in terms of making our vision for SAP and at SAP. So both in terms of what we commercialize for our customers, as well as how we run SAP itself. So let's first dive into the commercial side. You already heard from Christian about the gradual mix shift, incremental consumption growth on top of a very resilient subscription model.

Let's click a bit deeper into these different models, starting with the subscription side. Now here, we see significant resilience already today. But you can see the majority of our subscription base already, so 55% is non-seat-based that's even excluding the consumption share. And there are some great examples here, like BRIM, billing and revenue innovation management measured on revenue spend. By the way, a great reminder that we are the company bank rolling the software industries, commercial infrastructure.

So if anyone can be flexible for commercial metrics, that's definitely us. There are many more examples here. What I like to call out is success factors. Now you might say, oh, total employees, that's a disruption risk. I would say no, think of -- I mean, Starbucks presented down there with hundreds of thousands of employees at the front line. So even there, I would say, this is super resilient. And then on the cloud ERP side, by the way, we have a battle proven model that has been driving automation for decades and decades and decades, but the dominant share is also coming from nonseat-based components already. So let's move on to the consumption side. Christian and Thomas already talked about accelerating consumption growth. You will see that both on the autonomous suite layer that Muhammad's described, so through consumption of Joule's systems and agents as well as on the business AI platform layer where we already see significant growth in things like AI and BC. Now how do we do that?

That's not just a product strategy thing. It's actually we are turning the entire company already since years frankly, in our field incentives and our ecosystem investment in simplified commercial models and so many, many more levers to ensure we see this accelerating consumption growth, and we are on a great track here already. Now third, that's an area I'm personally very passionate about there's a lot of talk about outcome or value-based pricing. And that's where we will see where we will actually monetize our most premium offerings. And to give 2 examples here, Thomas talked about our migration assistance. This is where we are pricing based on the proven outcome that we show our customers how we can actually reduce the migration cost. And think about the trillions in services TAM we address with that.

Second example is industry AI. Here's where we deliver autonomous outcomes. Think of an autonomous batch release for a pharma company. We talked about yesterday, an autonomous maintenance ticket closed for an oil and gas company. So in a nutshell, here, we will be monetizing on a value basis. We need the outcomes and savings that we prove to deliver to our customers. So that's it on the commercial model. Now let's turn into how we actually turn SAP itself into an autonomous enterprise. And let me start with a very clear statement here.

You heard us talk about investments but our commitment remains absolutely clear. We are continuing to drive increased operating leverage with expenses growing 80% to 90% as a percentage of revenue growth with continued improvements across all key KPIs, decoupling the expense growth from the revenue growth, and that's despite significant AI investment. Now how are we going to do this? First of basically by demonstrating that our AI is how we scale SAP profitably by actually also simplifying SAP tech, for example, the operations function since I took over, we drove for more than 30% productivity in that function, but really then by turning SAP into an autonomous enterprise itself.

My team is fully focused on making this vision real, transforming our internal processes and driving AI adoption, acting as our own customer 0. Our ambition here is clearly that a lot of simple tasks and executional tasks are going to be executed autonomously for SAP so that our employees can focus on the highest value work.

We already see a triple high triple-digit amount of value realized in budgets today, and we are committed to deliver over 2 billion in productivity by 2028. And that's a value that has already measurably increased since we first talked about it. I think it was in Q4 2025 based on basically what we showed you yesterday, the significant jump we've made in the AI, we are shipping to our customers as well. And that means delivering productivity gains in every single function, north of and sometimes significantly north of 20%.

Now let's take a closer look at some of these domains starting, well, how could it be different for SAP with finance and spend where we see a clear head start, I guess, that's no surprise. You know Dominik. From planning to risk and compliance to invoicing and reporting. Our AI assistance and agents are already taking over strictly governed process execution. And the operational impact we do already see is material. We see 45% productivity gain in Contract Management, over 40% time reduction from offer to collected cash. And not to mention there, the significant jump in intelligence, we can arm our people with when planning by making every finance professional, a business data cloud user.

Let's look into 1 example, the financial planning assistant, which supports budgeting and forecasting, detect margin compression risks early identifies actions to protect profitability or increase top line, collaborating agent surface here, actionable financial insights, uncover cost and margin drivers and enable data-driven recovery scenarios driving 35% or more productivity gain in planning, which leads to more planning cycles, which leads to much better decisions and much more accurate and real-time planning.

Now let's look into the development side of the house. So here, of course, we are deploying AI tooling jewel agents and our own business AI platform as well as best-in-class third-party tools. And Muhammad talked about it in a nutshell, what we see is a compression from starting to build an agent to GA from month and month to a few days. How does that work? Well, our own team builds our agents with Joule Studio already with the agent buildup for low code and pro code development.

Every developer SAP has access, for example, to Claude Code. Claude code via our own agent hub on top of -- sorry, AI Hub on top of BTB, using our Joule SDK for Pro code development. And then, of course, our other developers have tool for developers as well. Then we contextualize these agents with easy access to business data Claude more than 300 data products available to all of our developers and our knowledge graph.

So it's a super fast process and honestly pretty full proof. So it really accelerates how we build and ship our own agents like it is accelerating how our customers and partners are going to extend and ship new agents. And then on the governance side, that's basically typically in the development process, the hardest work, and Muhammad already talked about it, it's basically taken care of in the majority for agent development for our teams based on the AI agent hub, which takes over the majority of life cycle tasks, so you don't have to build it. So the intent is clear. It's basically not to reduce our R&D ambitions or R&D investments but to unlock more innovation for every euro we invest into the R&D function.

Now Gina talked about the workforce side of the house. We are already operating here a highly, highly efficient shared service center set up, by the way, saying true for finance and many other areas. Now in autonomous HCM, the objective is really to elevate the employee experience as while taking away burdening administrative tasks. Our AI assistant support the full employee life cycle. And the impact is adding up quickly. So around 15 hours saved for simple onboarding per year, 25% productivity gains for performance preparation tasks, 80,000 hours saved an early rollout across as many, many operational HR workflows already within SAP. And with that, we are really turning Genus function into a truly experienced lab high-productivity HR function.

A good example here is our career and talent developments. By the way, all of that is either live or in rollout within SAP. It supports creation of development plans. It automates talent discovery. You now talked about how critical that is for us as it is critical for many others and proactively build succession plans and the impact is tangible. We already saved more than 45,000 hours per year across SAP, and that's in the initial rollout phase. So we don't even yet see the full productivity impact. That results not only in productivity, but in a much more personalized employee experience that gives additional productivity gains and enables us to do better talent planning.

Now last but not least, that's my favorite area of productivity and AI, autonomous customer experience. So -- our focus here is really on optimizing the end-to-end customer journey from demand creation to adoption and expansion. The journey Tom has talked about. Juul becomes really our single engagement layer already for our customer-facing colleagues save more than 15% for their preparation and coordination of customer meetings.

Our consultants save 1.5 hours per day roughly with AI guidance on implementations and consulting activities, account executives gain more than 20% productivity in lead qualification. Ultimately, really, what we do is we free up time of our customer-facing teams to do what they should do, spending time customer. But way it's also a great example here how this AI layer and the underlying architecture layer, fit together because we are continuously also modernizing our internal stack here. We just went live with a new version of CPQ, our own product for quoting, which actually led to reduction in quoting time and time to produce a quote within now AI agents coming on top to further reduce the time spent on quoting, which is a TDS activity, as you can imagine, especially for large and complex customer.

Take one example. Thomas already touched on it, something I'm very proud of, which my team in SAP's IT build on top of SAP AI platform with really the golden part we provide for Pro code-based development through the Joule SDK in this case. And basically, what you see here is how we build a fully autonomous software support multi-agent system. It has, I think, 26 agents knowledge sources, all of that connected through Business Data Cloud, all of that built on BTP. It has human in the loop capabilities.

And what we achieved with that basically of the 2 million cases that we don't reflect, we get more than 10 million cases a year. And you can imagine that stack was already highly, highly optimized. By now 100% of these remaining 2 million tickets get a very strong recommendation from Joule already for our technicians. And actually, 20% of these hardest to solve tickets are now solved fully autonomously. And that's early rollout. So you can imagine the productivity gains as well as and Thomas mentioned that the improvements actually we are seeing even in customer experience goes on support at the same time. So one thing to close off is also very clear for us, you cannot have autonomy without governance. That's why we have established a strengthen and holistic AI life cycle and value management within SAP with our business transformation portfolio.

We manage more than 300 AI use cases cutting across more than 2,000 processes within SAP with an average time to value 3 months. And that comes with rigorous value management, so I can commit more than $2 billion to Dominik in productivity, but also with rigorous change management in partnership with Gina in how we roll out these capabilities, using all of our own tools to do this. And these productivity gains with that are not just claimed but realized measured and continuously improved across hundreds of use cases. So I hope you take away from this short presentation, we are firmly on the journey of running SAP itself as an autonomous enterprise. The very same vision we provide to our customers is what we are driving internally at SAP as well. Thank you. And with that, over to Dominik.

So also a warm welcome from my side. It's great to have so many of you here. It has been a pretty wild ride over last year. So let me just kicking off now with some numbers. And then I go to a more conceptual part about the autonomous enterprise for finance, my board area. And last but not least, and I want to also go back to the numbers part. So the first point I want to make is, while there has been a roller coaster on some metrics like our share price, there has been actually a great stability in what at least you, the sell side sees as the free cash flow estimate for next year. It's actually slightly up from last year here and at a weaker U.S. dollar exchange rate. And there is one key number that I just calculated this morning, our free cash flow yield on 2027 free cash flow estimates is at about 7% now. That's up from 3% 1 year ago. And that's interesting because when I then -- Chris, the Investor Relations department, what is actually our weighted average cost of capital because I want to back solve what's the growth assumption SAP. I hear it's between 7%, 11% and the median and the mean is pretty exactly 9%. So if I deduct that 7% free cash flow yield, from the 9% weighted cost of capital, I think go back to business school, Corporate Finance 101. I see the implied pet growth rate is 2% nominal and that's a negative absolute growth rate in the perpetuity. So basically, we get a very clear message from the market. You're going to be kind of disappearing in the world economy, while the economy is growing at 2.5% -- real, your real 0 or shrinking.

And now what I want to do is share with you why I'm not convinced about this hypothesis from my point of view. So I want to sneak first in the shoes of somebody who has been a CFO role for a large group for now in my 16th year and not only at SAP but also before and share with you how we view the world and what the component can do to us.

First of all, I think we need to conceptualize 2 hemispheres in what is a finance department. And I want to characterize 1 as deterministic. And what is that kind of deterministic world. It is a world of very clear rules, standards, transparency, compliance, auditability, internal controls IT general controls, ensuring that you have a high degree of repeatability that when there is a certain input exactly the same output will come. And you would not be surprised that this deterministic world is where CFOs like myself, really feel at ease where audit committees feel at ease because this gives you the assurance that critical processes are run properly.

Think about producing the financial statements, you are all looking to the tax returns. And by the way, in the Executive Board in Germany, all of us signed with our blood that these documents are correct. And in U.S., the CEO and the CFO are signing that. So there is just no tolerance for risk in these processes. And this is really what is the absolute key priority of any finance department. I always say various kind of jokingly staying out of jail.

We want to just make sure that the assurance levels are really met. And by the way, sometimes we have some errors. These errors tend to be always related to the more probabilistic world. I come to later, maybe humans doing stuff. And I can tell you when the small errors occur, luckily, so far, nothing material. The Audit Committee is reminding us in a very stringent fashion about the necessity to have proper internal controls in place throughout this enterprise. And I think it's not only the audit committees that should care about it, but investors should care too because after [indiscernible], I think it showed that these internal controls have a certain meaning.

Now comes the fun part for CFO is the probabilistic world. There's no fixed rules. This is kind of the exception handling thing. And there's a lot of creativity in here, tribal knowledge, test acknowledge you require subtle reasoning. Inputs are unstructured. So what do I mean unstructured inputs. Think about us quoting to a customer. When we do that, then suddenly exceptions are popping up. So the nice kind of regular process that has been described before is busted because some deviations occur, what could be these deviations.

Let me start with one extremely unstructured example. It could be as little as a rumor from somebody that this is what Christian client has promised to some customer at some point in time or even a Board member who is not even there anymore or this is the pricing we had whispered into the years, 5 years ago to somebody in the supply chain or some professional that was in another company leading to another company and now that kind of knowledge about pricing has migrated to another company.

So I think you can all agree that it's very hard to put that kind of fluffy stuff into a deterministic process. And this is why we have a significant amount of work. I would actually venture the lion's share of the work in a well-organized finance department like an SAP on these exceptions. These are what I call the exotic animals that are put in our Zoo which are really hard to deal with. So -- and by the way, this is a compliance risk. It's extremely difficult to put controls around it, proper authorizations. And yes, that's where AI plays such a huge role suddenly, there is a recipe how we can even deal with these much more complicated topics in finance, but it requires a certain framework of guard rates.

The key point I want to make here is that it's not about either classical software deterministic stuff or AI. It's really the fusion of the two, which I would call is really the solution to squaring the circle of a flood of assurance requirements, new regulations, the scrutiny we are under and the efficiency requirements that we are all subjected to because of course, productivity is absolutely paramount. So we have to combine that kind of neural part of the world which the more algorithmic deterministic symbolic part of the world. And in order to come to that autonomous enterprise, we have to really make sure that the software is not a tool you work with anymore, but it's really becoming a tool that works for you.

So the systems that will achieve that are the ones which connect that neural intelligence I just described in the kind of probabilistic world with the hard reality of what we require in finance in the deterministic world. And I can assure you I'd say, if I think about my mission as a CFO, it's kind of chasing everything that's fluffy and probabilistic and moving it to the deterministic world to have no issues to make it cheaper and all these type of things.

So let me share some deep convictions we have on that front. The A models are not substitute for traditional deterministic systems, and I give you 3 reasons for that. First of all, I mentioned all the shortcomings of probabilistic in what trouble they kind of into. Secondly, simply economics. Once you've cracked the nut of putting something into a workflow in a deterministic way, the marginal cost of getting the job done is fat. In contrast to a kind of prompting process, which is regarding a lot of GPUs, a lot of kind of compute power and who knows where the cost of the token will need to go at some point in time to amortize the kind of trillions of investment that is currently bold into that machinery.

Third point, very trivial, but important. This stuff preexists. Our customers are really busy. They want to prioritize on what's adding value on what's differentiating the enterprise. They don't want to reinvent the wheel on something that's already working quite properly at almost no cost. So it's also that this kind of preexisting knowledge what has embarked in our systems is not publicly available.

So yes, of course, you can run pretraining on data. But in this part of the world, it's kind of hidden in every single customer, and we can then aggregate, of course, the findings of this data in these systems. So that's, again, the reason why the deterministic world will not disappear and will continue to use the software.

Next point is the quality of the agents really depends on the quality of the underlying systems. And I would venture to say that because of the data gravity in our systems and because of the fact that we are triggering so many hard monetary or other transactions in the company. The

SAP system is like the Alpha and Omega of many of the transactions of the lion's share of transactions that enterprises are actually going through. So not surprisingly, we have already the data gravity now we extend it with BDC. On the other hand, we have the transaction point on many of these transactions, and now it's about inserting that probabilistic model in that flow to also create that very autonomous user experience that has been highlighted by the colleagues here in such a powerful way.

So it's all about having the quality of the data, context which having a golden record and [indiscernible] also, of course, is exactly in that spirit, it was mentioned before. It's also about the governance of the system access. I mean who can see what, who is governing the whole discussion and the guardrails we need, and that's a very important point that is often overlooked.

If you look at the lion's share of [indiscernible] at [indiscernible] today, these are often 1 shot type of transaction. So it's a customer calling a call center and saying, what do I need? And then it's over. If you make a mistake here, you've spoiled that 1 transaction. In these processes, we run in finance, we have often multistep transactions. So there's many things from a quote all the way to the cash in all the way to the financial disclosure where when you only make 1 mistake in 1 step of this huge chain, it's over, you have the wrong number in the disclosure. So what that all means is that the probability of failure in 1 single step is compounding through the processes like in automotive manufacturing, when you have 1 part that is wrong, the whole car is broken.

And that's very different from this 1 shop processes, vibe coding is a one-shot process. You vibe code, you get a product. It's not a black box. You can test it. That's very different in the type of world in this kind of transactional long chain world that we are exposed to in financing. Third one is on the probabilistic side, yes, it's absolutely super useful to use these large language models. There are good ones out there, we can use. But we also have now this Rapid 1 model on large tabular data, which is, again, proprietary SAP data in SAP systems, and now we extend that with the third-party data by virtue of the acquisition we've recently done. I don't need to go in the details. I think Muhammad has really explained that extremely well. And last but not least, and maybe most importantly, I think it's a huge mistake to think that wipe coding is automatically the same thing as creating enterprise-grade ERP system engineering. We benefit a lot, as has been shown before. We gained huge productivity with that. But the new long pole in the tent is really to making it enterprise great to put what I call a trust trip around it to satisfy all the club compliance requirements to govern it properly. It's almost like an insurance policy for the customer.

And by the way, we see very little customers, especially in this highly sensitive area where they come to us and say, "Look, I would love to do all that stuff myself to the contrary." They want to put everything they can under that SAP trust wrapper to say, this is the thing, this is my insurance policy. These processes are all auditable, that they can be traced that we have documentation around them that when law is changing in a certain country, we know that it has been updated accordingly. And also, it's a floor to think that the cost of development is purely the functional code development.

It's then that whole hardening in an enterprise context, plus the maintenance is a moving target. A lot of the parameters that make it hard are changing over time, like legislation, cyber attacks and so forth and the higher share of the total cost of ownership of the software is actually in that maintenance cycle.

So here, I summarize some topics where you can see how we think about it in finance. I don't want to go through this because we are already far beyond time, but I hope I made the point clear that actually what we have to achieve in I also wanted to be modern and use an AI agent to coin something is that kind of symbiosis that neuro-symbolic system of action to become an autonomous enterprise. Now back to numbers.

We keep that very short because sorry to say, not much new stuff. It's pretty much the same thing we have told you last year. Here, just a summary, 1 year added, the CAGR on cloud revenues over that time frame has been 23%. And if you focus on what is really the center of gravity of our activity, which is the fast-growing SaaS pass layer. You know that Infrastructure as a Service is phasing out. The only specialty, so to speak, we're going to push is the infrastructure for sovereign applications. And also now looking a little bit in the forward indicators that give us a glimpse into the future I mean the CCB is growing healthy, 24% and the TCB is growing even faster at 34%. That means the ratio between TCB and CCB is actually expanding and this is not only driven by longer deal duration. It's also driven by steeper ramps. I think you heard us talk a lot about that in the last quarterly calls.

So we think that actually the numbers that we can produce are giving a lot of substance to the growth story, the TCB growth inflection, so to speak, is good for a couple of percentage points acceleration and helping us, so to speak, to offset that kind of expanding base. We have to compare ourselves against them. It's, of course, more difficult to generate that ultra-high growth when the cloud revenue base is growing massively.

Let me quickly double-click on our support revenues -- these support revenues have come down by around about $1 billion over the year 2025. And you see that we continue to believe that over the time frame through 2030, we should see that maintenance base be cut in half, and this is predominantly related to the fact that, of course, as mentioned before, we continue to see the end of maintenance for ECC and older versions of SAP, the mainstream maintenance, '27 and then the extended maintenance 2030.

There's 1 thing I want to highlight on top, which is that actually the ratio of conversion today is about 2/3 coming from ECC and 1/3 coming from S/4. So we're not only converting legacy directly to rise, but also the S/4 is going more and more to rise and when customers start the journey, they're not taking the detour through S/4 anymore. They're really going straight to the rise offering on S/4. In most cases, why? Because of what had been depicted before the processes themselves are massively reengineered with AI.

So if you do the blueprinting of an S/4 transformation and you go on-prem, you're basically barking up a little bit the wrong tree because you have to design on processes that one of the ideal processes once you're in cloud, and it's not very efficient to make that big detour and kind of try to cross the Canyon into steps. So that is why you see also an acceleration there, which I think bodes well for the sustainability of our conversion story because it's actually also tracking well on S/4.

Margin-wise, I think the most important message here is that this 80% to 90% operating leverage, as we call it, I see, i.e., the total expense growth versus the revenue growth is still intact. And Sebastian mentioned the more than 2 billion AI efficiencies will embark over the next years through 2028. And if you think about that and you then correlate it with that ratio you can roughly estimate that there is about a kind of mid-single-digit billion increment on OpEx.

If you assume on top of the gross margin is more or less stable, if you say that's the operating leverage will not come much from the gross margin but comes more from operating expenses, but that's a fair assumption to take. So we have actually a lot of incremental firepower to drive our cloud to AI transformation in a very rapid way. And also to do tuck-ins and you would not be surprised, and you will see the numbers in the not-too-distant future, that if you -- the last 3 tuck-ins, they were kind of depending on which 1 you look at more or less early stage. We are talking about transactions that are dilutive that have a J-curve, but this will not bring us to de-commit from these numbers, but we create the efficiencies to fund that aggressive acceleration of our road map on cloud.

So last but not least, cash conversion, always near to my heart. As you know, we don't see any deviations in the model. Yes, there is a little bit of investment from the sovereign cloud topics we talked about, yes, there is some headwind from higher hardware prices. But given the size of this operation, again, we will offset that in other parts of the business. We are still working on cash conversion as we have done in the past, and we will continue to grind that to make sure that we can adhere to these rules. If you depollute for the restructuring we had in 2024, there is like a 80%-ish cash conversion, which is the free cash flow divided by the non-IFRS operating profit.

And the way you have to think about it is that we basically slam the tax rate on our EBIT and then add back the delta between stock-based comp and what we accrued in the P&L for stock-based compensation. cash out versus P&L, and this gets you to these numbers. And this is also the reason why we think actually we are going to grind up further on the famous Rule of 40 target that we have set a while ago. Capital allocation, no change on the policy, a very prudent allocation in terms of M&A. And I say we are not buying growth per se in M&A. That's not the idea.

It's really about buying technologies, complementary capabilities that are really accelerating the time to market. You can often then just debate, is it a kind of make or buy an opportunity here, but time is really of the essence. So I hope you also understand that this is extremely useful to complement our capabilities to really become credible and execute on the autonomous enterprise. No change on our policy in terms of rating. We like that super conservative balance sheet we have. especially with some big risks looming. I mean, who knows how long this [indiscernible] will still be shut and what's happening if it's shut for too long. It's good to have a strong balance sheet because really, we want to execute on our strategy and don't want to be bogged down by leverage topics or anything of that nature in case something really bad happens.

Let's all hope it doesn't. But SAP is fully prepared for that, too. Share buyback, we have announced EUR 1 billion after the smaller program. We had in the past about EUR 5 billion. We have stepped it up to EUR 10 billion. We've done 1/4 of that already. And now the remain to do, which is about 3/4 of the EUR 10 billion will become quite linear. We don't speculate on the share price here. We just do that in a very disciplined, rigorous fashion over the next 1.5 years. So to conclude and bring it all back together, you see a CFO here who is a little bit desperate to see that market is telling me this company is going to decline in real terms because it's not at all what we plan. I have been always pushing back on giving long-term guidance on growth.

Some people said, "Dominik, why can't you say it's kind of mid-teens is the new norm?" now we are in a completely different environment where everybody says, well, will this business still be there? I hope the entire Executive Board has been able to give you some feeling why we have a deep conviction that we have the ability to grow significantly in real terms for long.

This is a sustainable enterprise. And now I don't want to be any longer between you and your lunch. We will do the lunch very quickly because we have been spending more time with all the passion we have about the topic here. And then we will open up for Q&A after you grab your lunch boxes and maybe we can reconvene here to then continue the session. Thank you so much.

[Break]

Please welcome back to the stage the SAP Executive Board.

Welcome back, everyone. I hope you had a good quick lunch. Also, thank you for the Executive Board to be with me on stage for an interactive Q&A session. Muhammad is actually joining us in 1 second. So we're now going to open it up to questions from you in the audience. So if you would like -- you would like to ask a question, please raise your hand. And we have mic runners here to bring you the mic. Okay, let's get started with Adam, please.

It's Adam Wood from Morgan Stanley. Maybe if I could dig in a little bit around how SAP is using its own technology. And I wanted to start the big discussion we have with investors is we're seeing the Frontier labs massively monetize what they're doing, seeing kind of unprecedented acceleration. And we're not seeing that happen within a lot of the enterprise software companies at least today. Could you talk about your use of your own technology? What's your experience been in terms of frontier lab usage, your own software.

If you were paying for your own software, how far away would you be from accelerating your spend with SAP? And then could you talk a little bit around how you think about cost and return on investment, both from your own software internally, but also what tokens are costing and how you monitor and manage that whole usage of technology internally and whether you feel you're getting a return right? It feels like you're being judged a little bit differently in terms of purchasing decisions than maybe some of the other players in the market.

Maybe I start. Thanks, Adam. So first of all, I mean, I showed some of the cases we have. We have over 300 cases. We see significant productivity and that productivity of north of EUR 2 billion we committed to you and is committed to our budget. And we see that actually across every function, I mean, we see a significant -- the north of 20% productivity lift we expect from go-to-market to corporate functions to development. And then if you compare usual productivity tools then the cloud code of this world and what we see, actually, it's a good healthy mix development.

I think all of our people have C code. I know all of our people have C code available through BTP, by the way, which takes care of a lot of the governance, IP concerns and so on that we have. So that's a very good model. But the real productivity but what we found is actually when our developers worked on enterprise-grade code, this is now by coding, that actually the productivity gain was still there, but the real uplift came when we now infuse things like the domain models Muhammad talked about, made our KG available, the agent life cycle management. Actually, in many cases, now the productivity bottleneck becomes not so much the coding itself. I would argue in most cases, that has never really been the bottleneck.

It's actually all the approval processes that come after data protection, security and so on. But with what we are shipping in the -- with through Studio 2.0 and the SDK that goes with it, take care of a lot of that for our internal development teams. And with that, what we've seen is actually a significant uplift on top of frontier led AI model usage. So based on measured productivity. And we see right now, I think a good measure, I can say, on data pull request. I'm not saying that's the natural yardstick. We see a significantly north of 30% acceleration in terms of development velocity that we are seeing. But then I always like to say the amount of unwritten software in the world is infinite. So I think this unlocks great opportunities for us to provide more solutions like in the industry AI space. And then we have a full road map. So every function of SAP is co-innovating with Muhammad's team, Dominik's team, Gina's team to make sure the jewel assistance and agents we are shipping actually solve real problems for us and deliver real and measurable productivity.

I just want to add, let's not forget that the autonomous enterprise is, of course, a very, very ambitious goal because we have these complex processes I described where the compliance and assurance and quality requirements are just excruciating and you don't have the same easy ability for giving ability to mark up code that you have generated by one of these coding tools because that's a human being can easily do that.

So I think it's a little bit like the autonomous car where once you get to these high stakes, of course, it takes longer to activate it. And I would venture to say that today's Frontier labs, none of them has any meaningful big revenues on these more complex food chains we discussed here. But the ones we also massively use like coding, like writing text. So the things that human beings can easily control as opposed to a big well-oiled machine where you really have to make sure that it's kind of end-to-end, high-quality, high fidelity, and that's the focus. And we're going for that more ambitious goal, and that's why it takes a little bit.

So I mean, we looked at that. I cannot disclose the numbers, but we are a right customer. We, of course, use our full suite. And if you go across AI and BDC, it would be a measurable, measurable uplift we would see on top of that. If we were a paying customer luckily, I don't have to pay usually for our own software. But yes, absolutely. And I think that's also one of the sessions where we had a session on SAP runs SAP. I'm not sure if you joined. Those were some of the best booked sessions where we actually show what we are using within SAP and what's working for us and how do we make it work. So we would pay a significant uplift on our [indiscernible] and AI assistance already today.

Okay. Let's go to Toby, please.

It's Toby from JPMorgan. Maybe just on the cloud revenue evolution chart. I think you showed that the consumption mix is about 10% today, and you sort of expect that to ramp up to over 30% by 2030. I think if we look at that on absolute numbers, that could imply quite a significant scaling in terms of absolute revenue. So could you perhaps talk through specifically what are you -- what gives you the confidence sort of in that consumption ramp? And any more detail you can share around the big kind of product components that would drive that sort of ramp in consumption revenue?

Yes. First, a small comment on the financial part. The key message we want to get across on the financial part is that this is not a massive impact like we had on the cloud transition where suddenly a lot of revenues are cannibalized and then we shift all the revenues to the right because we move from license. This is a very gradual movement where we have the growth in the market and on top of that, a replacement of seat-based revenues or other types of revenues by more consumption because indeed, it doesn't make sense to have a seat-based model when you are reducing the number of seats by virtue of the tools.

I honestly think that the concern is a little bit overblown because what really matters is can you deliver some differentiated capabilities to the customer that create value at the customer. And if you have that leverage with the customer because they want to have access to that value, then you can discuss a new monetization model with the customer, and you will be able to get your fair pound of flesh. If you don't have that product, it's a little bit theoretical to think about seat-based, consumption-based, whatever. But if you have that leverage and have a differentiated product, and if the customer cannot do it more cheaply himself or herself, that's the floor, so to speak, then I think we have definitely the ability to reconfigure our monetization model. Now on the growth trajectory, maybe Thomas, do you want to?

I mean, absolutely, what you see, what I've showed in the slide. First and foremost, we see with the business AI platform for sure, the consumption-based scale in that sense. So the cross and upsell within. And that's something if you think about the more data, which is processed, the more agents which are running, this is a compounding effect. And as you see the cross and upsell potential in this business AI platform capabilities, this is for sure a huge opportunity for us for our customers. So basically get the majority of the consumption-based revenue.

For sure, on top of that, we also talk about some of the large-scale RISE transformation, which is also in a consumption-based model as well, which for sure will continue to continue. And that gives us, quite frankly, a lot of confidence that this is exactly in that direction. And we see, I mean also talking here on the floor, I mean, there's huge excitement about the potential by bringing all of these things together in this flywheel, what we described earlier.

Yes. And maybe one last word, what was actually, for me, a little bit of a reminder about our cloud transformation 5 years ago. I can still remember when we announced RISE with SAP, many partners came and customers and said, "Oh, now it's for the first time, also understand how can I leverage better my contracts, my consumption commits with the hyperscalers." I met here a lot of partners who said, oh, I signed this consumption commit with [indiscernible] with Entropic. And now I see, oh, I can use this on your platform. I can actually drive consumption on your platform. Plus, of course, now I see how I get the context and the governance part into the agents. And so let's not forget that also to Adam's question that there is always what we are selling is the end-to-end. We sell the LLM plus the governance and the context part.

Jackson Ader at KeyBanc Capital Markets. The one that I had was, if I do some envelope math on the cloud starting point for revenue in 2020, take out the migrations, new customers, whatever, to the EUR 21 billion that you ended for 2025, it's, call it, like a doubling, right, of that like EUR 7.6 billion. So if we think -- I'm not asking like -- I'm not going to like keep you to it. But if we think about like the new base, right, '21 in 2025, is there -- understanding there's law of large numbers, like should AI be an accelerant? Should we expect a shorter -- like a compressed time to doubling that cloud revenue because these AI products in a consumption model? Just curious how we're thinking about like this new base in the next 5 years.

You know my notorious reluctance to do a guidance on revenues because if you think back, we committed to say we see some acceleration in '26, '27, then we had 3 impacts. We had the trade dispute, then we had the Iran situation. And we also had this shift of saying we are going to, in '26, reduce the hours we bill in services and use more hours to adopt. So that teaches you that you have to be always very careful to say that this is what it is. The very strong point we can make and it's trivial is that we can't see a scenario where this company would shrink in real terms.

But to the contrary, it's more a question about how do we modulate the growth rates we are currently seeing to the upside how much to the upside? Are there some risks of macro economy? Is there some headwind indeed from some [indiscernible] stories? So that's the modulation. But in general, we feel that the underlying growth trajectory is -- the puts and takes are very favorable in total for SAP. So that's why we said it more an opportunity than a threat from our perspective with all the reasons we try to give in this session.

Maybe my answer to you would be -- I mean I talked last earnings about the learning curve. And I feel this also speaks for the honesty to say, hey, I don't want to sell you here now a shiny world on AI while we are still learning. And we learned about the context. We learned about the governance. I mean you do your channel checks, we, of course, got a lot of feedback. And I guess -- but now when I look at the feedback what we are getting on the better version, which now will become GA on the platform, this time, I can say we are coming out of this learning curve. And also with RISE, I mean, when we launched this offering, there's so much learning which we then said, Signavio and then LeanIX, and we need to govern and support the customers even more with the architects. And the same is true here.

But I see also now a phase coming where we say, wow, now there is this point where we see, oh, now we can start scaling because now we see the accuracy. Now I see in the beta test things, oh, the accuracy, the compliance, the output is getting better. We measure that. And so that's why I have the confidence to say we are over the learning phase and now we are in the scaling phase. And then to Thomas' point, what is now very important that also our ecosystem is embracing the platform because similar to the ERP world, now we need partners, ecosystem customers building on the platform, building the expansions, building new industry AI use cases. And when that is happening, then you really then go into the accelerate phase, what we also outlined on the slide. But of course, Dominik is right. You never know what happens next in the world. But just looking at our AI journey, we are definitely now entering the next phase.

Let's do Charlie, and then you can do Ben afterwards.

Yes, Muhammad, I like some of your slides, particularly the one where you showed the fourth Agentic layer and the current battle to see who wins that opportunity. How important is it for SAP to be the dominant player in that part of the market? And if customers decide to build their AI away from SAP, are you still comfortable that you'll monetize that through API access that in either scenario, it creates an attractive financial outcome for SAP? And then if it is important for you to be a dominant player there, did you think about being a little bit bolder in some of your investments? I think we heard EUR 100 million co-investment with your partners to drive AI adoption. Did you consider bigger numbers to make sure that you're the dominant player?

Yes, I can start, and then I think we can also maybe reflect a little bit on the investments. To me, I think it's absolutely important for us to be able to sort of go win in that new layer of top, if you will. And I do believe -- I think the thesis around our growth, going back to a couple of questions as well. there's a belief in the thesis that, hey, the token consumption in the world is going to grow, and it's going to grow exponentially in some ways, and that's sort of what we've been seeing. Then you can sort of separate that growth into 2 buckets.

Certainly, there's going to be one on the consumer side and then there's one going to be on the enterprise and the commercial side, right? On the commercial side, there's going to be some maybe on the SMB side and the other is going to be on the enterprise side. On the enterprise side, further, there's unstructured world, like the office, the productivity stuff and then there's enterprise applications. On the enterprise application part of that token growth, the thesis, the proposition that we have is whomever in that value chain of ultimately Agentic experiences creating value for those enterprises, there's going to be a few things, right? And we think uniquely to whatever is -- whoever is the best LLM out there on top of what we can add creates the most value.

And then I think you have to sort of think through, we believe we're sitting on a pretty unique position that we can actually, for a good set of our customers, be that layer up top because to the best LLM, and that might change quarterly, right? That might change monthly, that might change every half year because the ability to shift from one LLM to the next LLM while the stuff on top continues to run is very simple. It's going to get even simpler. And the science around what LLM should you use for what workload is also going to get more sophisticated because you don't have to use the most expensive one for the simplest tax, right?

So to me, where the value ultimately our customers are seeing and will see is the thing on top that can do the smart determination of which LLM to pick to drive that token consumption to create them and that's where we shine, right? So to me, I feel like we've got the unique value proposition to be the layer of top to benefit from the thesis of token consumption explosion by providing the value to the customer, and that same. Now let's say that doesn't happen for a percentage of our customers because it might not, right? I think there's different reasons why customers might select others.

For that purposes, the way I look at it, our frame B is, again, if you go back to that stack, our application layer becomes, by definition, a platform layer, right? Because you're not rewriting GL, you're not rewriting supply chain. So there's going to be, even if it's somebody else's agentic layer, the consumption will pass through us ideally through our orchestration layer through A2A or other means, if you will. So we will obviously monetize that as well, not as much as the scenario A, but we're going to be in that cycle one way or the other, if you will. So to me, However, way you guys want to sort of lay out the thesis for token consumption, there is a material portion that we add value would come through us.

And the LLMs, I feel like would be the layer that there should be a lot more skepticism on that is the durability of one LLM provider in the fullness of time really that strong because that's the interchangeable layer. Not the stuff that we add on top of it, that's SAP context, as the customer context, the company memory, but the ability to switch out that is super simple. So anyhow, it's a bit longer and complicated answer, but I hope that makes sense. Now in terms of investments for us to be able to go win big in that agentic layer, one of the things Christian announced, 2 things Christian announced yesterday, if you remember, is not just this $100 million for our partners to rewrite and replatform on this new platform that we have, but the fact that we are giving design time of this agentic layer free of charge to the customers, design time, if you think about this is where Anthropic is making all their money right now, right?

It's when you go engage an LLM to say, hey, build this or build that because that's still design time. It's not run time, right? Run time happens later. That's where N80 also makes a lot of money. And we're saying to our customers that, listen, we're going to give you the design time that's not just the best of what's out in the public, but with our context layer. And not just that for a period of time, we're going to give you run time free as well. So to us, that's significant investment to be able to say, hey, let's get the foothold, the stickiness with the higher value stuff with our customers on scenario A, which is we really want to be that agentic layer for you. So hopefully, that makes sense. But Dominik and Christian, you want to add something?

That's from a go-to-market perspective. I think the partner incentive is one of the aspects. But also for partners, we have many means of various fundings and investments what we have. But also we are more prescriptive how do they need to work. in order to reach the quality levels, but also the acceleration in all of what we want to achieve. To give you an example, I mean, all of our RISE with SAP validated program partners commit on the methodology and the tool chain. So they all use Signavio, they all use [indiscernible] and tool for consultants and tool for developers, which is super critical to reduce the cost for our customers. And that's also something where we basically ensure that also here, we see the AI acceleration through partners.

If you talk to KPMG, PwC, Accenture, the like, they talk about thousands of users for tool for consultant and tool for developers, which they now include into the project work, and we will for sure. We also have, to your point on incentives for this world, also customer incentives, which we also associate with some of the RISE with SAP transformation where we for sure on purpose, want to also invest into the AI adoption as we discussed. So basically, partner and customer incentives in that sense.

I would like to add one very common sense, maybe trivial observation on this question. First of all, what is important is really when does the customer need to get a license for our dual capabilities, whether he put something in between or not. I mean that is really triggering a lot of value creation for us. And then dominant is a super strong word. We are not and we have not been dominant. We have been always in fierce competition. So I don't want to hear that word from an antitrust point of view anyhow. So we have been kind of attacked by SaaS companies left right and center.

So I don't see that it's now so completely different with the Agentic thing. And then last point I want to make is I think it very much depends on the persona, how big our market share will be. The shared service center guys who are working exclusively on SAP system, why the hell should they use anything but [indiscernible], obviously, at some point in time. But if you are kind of higher-level executives in a media company, maybe you have different needs and then you kind of make that jewel capability connected with Copilot, a kind of invisible slave below, so to speak. But as long as we get the license for that customer, too, commercially, it also gives us a lot of value.

And maybe just on the internal side because the question. I mean, I would say actually that engagement players is our to win and not to lose. What I've seen with the initial employees that have access to the new Jewel experience is exploding. And I think there's this misconception I hope you took away. I mean a lot of our application base is already in a way, lights out, and I couldn't tell us if it's a user or an agent that's triggering a an invoice that's triggering a purchasing order. Our commercial model is immune to that, and I'm seeing that within SAP, too. But actually on the engagement layer, I mean, let's face it. I mean, in many cases, we've been abstracted for a decade plus away from the end users who ultimately produce a report. And what I'm seeing now is actually for the first time that our own team as customer, as user is going absolutely crazy about the experience that we are providing, and that's something that makes me incredibly proud of what this team has achieved over the last couple of months.

It's Ben Castillo from BNP Paribas. Maybe one for Sebastian and Dominik. It's reassuring to hear your level of comfort in that cost to revenue growth ratio. But as you alluded to, that's a net number. And Sebastian, you're extremely focused on delivering that at least EUR 2 billion of AI efficiency. So when we're looking at that gross investment wallet over the next 3 or 4 years, mid-single-digit billion euros, it seems a very large number, particularly if I compare it to the last invest cycle that you made when you did this cloud transformation wave.

So I guess my question is that 80% to 90% feels quite conservative. Why could that not be maybe slightly more aggressive and maybe be below 80%? And then the follow-up question might be, well, if it is in that range, could you just help us understand where that sort of quantum of spend is going, headcount, talent acquisition product, just to help us get a sense of why it's needed to be that high?

Maybe I'll start with some ideas on that. So it's true. It's a big amount of money and a lot of firepower. And we don't want to change that target. We don't think it's the right time because it's all about speed. It's executing our plan as quickly as possible. And I did mention that behind some tuck-ins we did recently was also very much the idea of time to market. So the make option might have been available, but it might take too much time for us to then benefit from that capability in our offering.

So we would err right now also because we see that, frankly, from a valuation point of view, the growth is more favored by investors, we would err on really keeping an aggressive investment. And if there were opportunities to get even more productivity, then we would redeploy that into growth initiatives. And we've done that in the past. I mean we didn't even think about BDC 3 years ago, and then we had the J-curve from BDC. We absorbed that in our model because we created the room by better than targeted productivity. honestly, these acquisitions we do, there are actually, to some degree, dilutive in the coming years, but we don't have to change the model because we are also seeing great positive surprises on the productivity gains, leveraging AI internally.

So we feel we have actually reached a pretty natural balance between how we can aggressively grow and sustain the top line and how we can see a grinding up on the margin. And I don't see any reason why we should deviate from that model for the time being at least.

And maybe to add, first of all, you will see Dominik and me being relentless in ensuring these productivity gains are met, the 80% to 90%. Because for me, it's also showing that this autonomous enterprise vision is real. It creates the space we need to invest in areas like the EUR 100 million fund for AI adoption that we can commit to you without changing -- touching that guidance. But what's even more, I would say, encouraging to me is we announced EUR 2 billion, approximately EUR 2 billion in Q4.

We've actually now worked over the last month incredibly close our internal functions with our development team. And yesterday, I saw the Head of our finance shared service, and you can trust this is a highly optimized operation that we are running, decades of automation that went into that, proudly presenting how she co-innovated on the financial close assistant where she expects significant productivity increases that gives me a lot of confidence not only in that productivity number, it gives me a lot of confidence in what we are shipping now to our customers in terms of the productivity and with that also the growth that can unlock for us.

Johannes Schaller from Deutsche Bank. A lot of your customers are in kind of different stages in their AI innovation journey. And I think for the ones in the early innings, it's very easy to see how your offering is very compelling. Also, Christian, you mentioned some that have built agents and they're probably not working that well, also very easy to see where the value add is. But then there is this kind of small set that are very advanced in the AI journey.

Some of them have even been a bit vocal kind of how this has enabled them to cut back maybe even an SAP spend to a certain extent. Can you maybe zoom in a bit on the pitch specifically to this customer group and how you can win them back? And then just a quick question for Gina. You talked about the evolution kind of the skills mix investing versus reshaping. Maybe can you give us a bit of a time line for that and also what you think that will do to the overall headcount of the company?

Let me start and I give my pitch I just given to a customer from Indonesia here, a large conglomerate. And then Thomas, you can give your pitch. Look at this customer in Indonesia, they are running around about 40 ERP system, even 4 or 5 non-SAP acquisitions. And now they asked me that question said, "Oh, Christian, you said in the keynote, we can use AI also for S/4 on-prem or even ECC, what should we now do? Should we just build now the agentic AI layer? Or should we start modernizing?" And I said, hey, there is AI tooling now to speed up the migration because when I look now in your architecture, and there was actually with LeanIX an architecture which we built for the customer and said, hey, the handholding to get all of these policies now what you have in the different systems, which are absolutely not harmonized.

This will create a lot of work to get the agents now up and running because there's still a lot of customization in which we now need to reflect also in the agenda. We can do that, Muhammad build the connector. So we can do that. And you should pick these scenarios where you see the highest ROI, but let us also start working now with our architects on the modernization because then you will see that we are reaching a completely different scale. And also the TCO then of running those agents will also come and I guess this is not -- and also for many customers, it's not really an either/or. It's really about both. And this on-premise option now is by no means a defensive move. It's just about giving our customers this option because, of course, we have seen there's a lot of pressure on them as well, similar to Dominik and Sebastian here to deliver these efficiencies, what we are all shooting for.

And I think what is super important actually also for these critical customers what you mentioned, also they, for sure, explore what is the right way to go. What they now see is actually exactly the benefit with this business context, business data, business processes and governance and the governance as should not be underestimated how much simplification that brings because, again, in the end of the day, systems need to be audible, auditors need to testify, Otherwise, people go to jail. And that is certainly an aspect where nobody is joking around.

I'll give you first one very concrete customer in that sense, always good to talk about concrete customers. They are in Germany, one a very advanced technology-wise customer, which we have. And basically, with the new platform went to Bayer and said, look, let's go with an FDA approach. Now basically, actually this weekend, which is coming, they will have with 8 agents from SAP seeing the new power, and they basically leverage the platform. They also fully embraced tool for consultant as one of the early customers.

In the beginning, for sure, also here, feedback was, yes, it's nice, but we have so much amazing SAP consultants in Bayer. So it's not yet that we would roll it out to many people. Fast forward 2 months ago with all the adjustments and evolutions because what Muhammad said, innovation speed is unparalleled actually unbelievable how quickly we add more content, more context into that. Now Bayer is rolling out tool for consultants for all SAP professionals in Bayer. And actually, even more so, they now ask all the system implementers to do the same because they expect the productivity gains from tool for consultants now for the implementation projects.

These are the examples where I clearly see the proof that the business context, what we deliver as part of our offering is the differentiation, what we have. And I think it's a great example where the new world, what we've shown today, which again, the customers already can see and use is a totally different world. And I'm absolutely convinced, quite frankly, also after the [indiscernible], this will be a big boost in AI adoption, by the way, around the world. And I hope that when we then see the statistics from some of the research institutions end of the year that the statistics for enterprise AI adoption will be totally different end of the year than today.

I think the other thing I would add, just specifically on the -- taking it from the perspective of they bet on an Agentic AI platform, and they don't want to go change that because they've already deployed a bunch of agents on top of it. There's a few customers in that category, right? Or either they've done that or they built their own because they believe nobody else, they can't use anybody -- anything that's off the shelf with you. I think that I haven't run into a customer that has done that, but still doesn't see value from the autonomous suite. And this is why I think the characteristic of our strategy that says, hey, it's open.

What the question they ask is not that, hey, I want to now get rid of my Agentic platform. The question they ask is, hey, can I call your autonomous suite from my orchestration layer and get the value of what you're doing without having to replace. So this and part really excites them to say, "Hey, for my corporate functions and all the things I'm doing, I don't want to go rebuild that because I bought the application from you, and I'd rather just get the agent, but don't make me change my orchestration where I can just..." and the A2A allows us to have an and story as opposed to always [indiscernible] or story if you already bet on it. But now as you do the end, this is where the beauty of the context comes is that now as I continue to add SAP agents, I can build on this platform, but have orchestration on whatever I built internally. So this and to me is the answer to those class of customers, if you will. It's not an [indiscernible] discussion or are we going to try to go compete and say, get rid of that completely now. We're in the game and you should entirely use us.

So you asked on the time line. I think it's difficult to put a time line out because this is an ongoing effort. So we are trying to build a depth of organization that we can -- that we are able to say, okay, how do we up and reskill. So this is the first lever. As I also said, 85% at the moment in the most important profiles or to build up the AI skills is that we are saying we do up and reskilling and 15% for the roles we are getting from the outside market. But this is ongoing.

This is the cycle of strategic workforce management. You always have to ask yourself, okay, what do you buy, what do you borrow, what do you build and what do you automate. This is a constant cycle. And that's why I think we cannot say what is the time line. We have to make sure that we have the right skills on board in order to deliver on our customers and to our customers and to our products. So that's why we have an ongoing effort in order to do that. And from a headcount point of view, it's very much that we are keeping that flattish.

Okay. Let's go to Mohammed and Michael.

Mohammed Moawalla from Goldman Sachs. There's obviously been various discussions from -- in the marketplace around the momentum around the migration of the transformation cycle. In your opinion, to what extent is the sort of perhaps the lack of a road map in the past or the high cost of kind of implementation being a kind of factor because still a small percentage of the base has moved to the cloud.

And now post the announcement of the road map, stuff around dual consulting, how big of a step change independent of the macro can this drive in your opinion? Because you've obviously talked about in the next couple of months, some of these products coming. So when you measure this versus, say, the time of kind of cloud, how big of a step change in terms of that adoption do you see going forward? And what's been the feedback also on the kind of the road map from customers here?

I mean I can get started, and please team share your feedback. I mean, especially after that week, I mean, of course, we have customers who didn't yet move with us to the RISE journey. This was less of, okay, we don't see the value. I mean many customers in this camp also have seen their own transformation. They divested, they invested, they changed their portfolio. They said, "Oh, the organization will look differently in 1 or 2 years." And these customers said, SAP, we are going to make a move, but please give us the time, give us a year until we have figured out how we want to run this company. And there were quite a few.

The way how I now see it going forward is, I mean, first, these AI migration tools will help a lot because the much they like that, oh, after that, there is no ERP upgrade anymore. Of course, it costs a hell lot of money to migrate an ERP system. And so the AI migration tools get incredibly good feedback, and they will definitely help. The second part, what will also help is the on-premise connector because now we are saying, okay, go with us on the journey, and we actually help you to drive AI adoption from day 1 on while we are together modernizing your landscape. So all 2 announcements, I would say, make a ton of sense and will help us also to get now the remaining customers over the line. By the way, don't -- because you're looking at the maintenance and you say, oh, there are many, many customers still left.

I mean, of course, there are customers left. But oftentimes, it's also a customer has already started the journey, but then there is still started with 10 ERPs, but they are 50 out, gives these customers time, we are touching the most mission-critical system in the company. That's why you see also in our total cloud backlog always this WA because exactly that customers want to time that also in the right way.

Yes. I think that's an important perspective. If you look about the support revenue base, and already a portion of the support revenue base is already a customer who signed up to RISE. And based on, as I mentioned earlier, customers with 350 productive ERP systems need 1 or 2 years to move all of them to RISE. And then for sure, you see the shift in the support revenue. So that certainly is something we clearly see an acceleration as an expectation on acceleration based on the migration and the agent-led migration, which we've put out. And I think that's certainly something what is helping our customers also in light now what we discussed for the maintenance and which we see with ECC. So we are quite positive.

And let's also think about the commercial decision you have to make when you're sitting on ECC and think about what am I going to do? And like you do, they do a risk return analysis. They just say, okay, how much would it cost me if they even think about that, to rip out my entire system, to pay all the maintenance for that versus -- and then the risk associated with that not delivering to the excruciating governance and compliance standards that you need in enterprises.

By the way, I have been just checking in most companies, the recurring kind of SAP fees are lower than what they spend for important insurances. We talk about from middle of revenue type of tickets. So why would a CFO say, I take that risk, that my job to do all that myself. And it might not even be cheaper because they will also crunch a lot of tokens. The need to maintain that and they can get all that hassle that and sometimes it's a necessary evil for them. They can get that from SAP with a super high degree assurance level. We see exactly the contrary that they think how could I sneak more of the kind of critical things under that kind of trust SAP umbrella to make sure that I can use AI but can use it safely.

And I would say, I mean, I meet many customers on a peer-to-peer basis. Most -- actually almost all customers, I mean, they know that clustering AI on top of something that's broken 3 levels down. The only thing it gets you to is with Lightspeed into an RPA 2.0 disaster and the only thing that will go through the roof is token consumption. So actually, I don't meet customers really that really question the modernization itself. Often, it's a question of timing, when to do it.

Sometimes there are constraints around data centers. They still have to retire I can emphasize with that with running our own IT operations. So I don't sense there is hesitancy always stay on-premise. It's timing. And then, of course, they love everything we can do to reduce the cost and time to get there.

Go to Michael, please.

Michael Briest, UBS. So a question for, I guess, Christian and Thomas. Thomas, I think you talked about leaving no customer behind. Christian, you've been very passionate about public cloud ERP. And my takeaway from the last couple of days is it's a hell of a lot easier to run AI on public cloud. Where are you on -- or what are you doing to drive more customers down that path? Dominik, I think at our conference 18 months ago, you said maybe it was an order of magnitude smaller than private cloud. Could you maybe sort of give an update on that? And then just a very quick one for you, Dominik, just a yes, no answer, but do you think you can get to the rule of 40 before the end of the decade?

Should I start with public cloud. I mean, for sure, I mean, with public cloud and SaaS for sure, by definition, all the integrations with AI go out of the box. So basically, what you will see is that our customer can simply go to SAP for me, press the button, agents will be connected because that's the beauty of a SaaS public cloud ERP. We see a significant actually what I shared, acceleration increase of the public cloud share what we have. It's the mainstream model for all net new names, actually, which go to market.

We are very prescriptive on that one. So basically, all the net new names and logos you will see going on will be, by definition, grow with SAP customers. And for sure, we leverage significantly here on the one hand side, the scale with our indirect channel, what we have in the partner ecosystem for the mid-market, but also we leveraged opportunity with private equity companies with all the portfolio companies that for sure also they have an interest to use and we have a dedicated program called Grow Fast, actually with a dedicated fixed price associated frame contract for them that their portfolio companies quickly can jump to modern cloud ERP with AI embedded. And with that, again, they can grow infinitively. So for me, this is absolutely clear that this is the predominant net new name engine for the company.

We talked last year about partner-driven territories. We see phenomenal growth in that area.

The way I would react to your question on Rule of 40 is the following. We have a deep conviction that no matter whether it's normal, but of course, also in real terms, we will show significant growth going forward. You know where we're growing currently, and that growth will translate into margin expansion and therefore, also better cash conversion of the revenues down to free cash flow. So there will be a trend of grinding up the Rule of 40 performance. Now how fast that really goes depends on so many factors. And again, I want to kind of be cautious on that. But it's, of course, it's a journey that we want to consistently pursue.

We had some bumps in the road now with the topics I mentioned before. And honestly, with whatever might happen straight of a moth and some commodities running out of steam and being not available and there might be kind of meltdown scenarios I can't talk about. So that's why I would also refrain from giving you a specific time line because then I would be proven wrong potentially by one of these shocks like we had with COVID or what might happen with escalation scenarios that I don't want to be hung up with.

Okay. Let's do one final question and go to Fred.

Fred Boulan, Bank of America. If I can stay on the AI topic. Can you discuss your AI ambitions? So I think in the past, you talked about that EUR 1 billion-plus opportunity. Can you be more specific about your ambitions, time line, et cetera? I think last year, you had this kind of 5x multiplier with product innovation. So keen to hear your thoughts on that. And then secondly, coming back on the AI commercial model and economics, there's a lot of announcements around AI capabilities for free, pushing FTEs. So can you discuss the economics of building versus running agents? What kind of margins can you make on that revenue stream?

I mean look, let me start and please chime in team. I mean, first, what can we monetize with AI? And I hope this really came across today. First, we have, of course, the platform. Then on the new AI platform, we are building assistance and agents. Plus, we have our AI migration tools where we can deliver a ton of value given the size of this market, which currently is, of course, all going predominantly to SIs. Now if you add this all up together, it's a sizable potential. But think about that, when now Thomas comes with his AI architect to the customer and says, okay, here's the platform. Now we can extend it. Oh, there is a root cause. The master data quality is not so good. Okay, in the platform, let's pull in [indiscernible].

This is -- there's still too much machine learning in. You are actually applying a ton of data scientists. Okay, let's use Prior Labs and RPT 1.5. So I see a lot of cross-sell potential then also in the AI platform together with BDC, where we then can really further optimize how the agents are running our customers' business. And now that the platform is coming to life, that is, of course, for us giving us an enormous scale. And so -- and when you take those 3 components together, given where the solutions are, where the assistance are, I can now say with way higher confidence than 2 years ago that the stuff is mature and it's ready to be consumed and the adoption will go up without any doubt.

Maybe on the commercial model, we have, of course, a very dynamic development on the cost of token. I mean in the prior periods, it was actually a very strong productivity boost, really the bang for the buck you got was tremendous. Now the question is how sustainable is that, given that at some point in time, all the data center investment needs to be amortized. So it's a little hard to speculate on where exactly might the cost of token go. Can you extrapolate from the fast decline in the past? Is the sustainable and of course.

Now the good news is we are not as dependent as others on the development of the token cost because on all the stuff that is not RPT-1 and Prior Labs, we don't do pretraining. So we do inference. So it's a relatively efficient compute, not really spinning a lot of tokens. So I think it also will fit in our margin model on the gross margin. And RPT-1 and prior labs on this tabular model, it is a much more structured finite data set than boiling the ocean of the world's languages and mathematical formulas and ingesting all the books that have been written on mathematics on the planet.

So I think that gives me the confidence that it shouldn't be a major variance on the gross margin development, which we said is probably the part in our margin profile, which is the most stable. And recall, we said we don't want to focus on expanding margin on cloud in percentage terms, but what we want to really achieve is the incremental euros of gross margin from the cloud that we want to maximize.

Great. And this wraps up our Q&A session. Thank you to the Board. Thank you to everyone here in the room, and also thank you to those who are watching online. Looking forward to speaking to you again with our -- at our Q2 earnings in July. And for those in the room, we also have our management reception. So we're looking forward to seeing you there. Thank you all.

Thank you.

Thank you.

Ladies and gentlemen, thank you for standing by. Welcome, and thank you for joining the SAP Q1 Financial Results Conference Call. [Operator Instructions]. I would now like to turn the conference over to Alexandra Steiger, Global Head of Investor Relations. Please go ahead.

Good evening, everyone, and welcome. Thank you for joining us. With me today are CEO, Christian Klein; and CFO, Dominik Asam.

On this call, we will discuss SAP's first quarter 2026 results. You can find the deck supplementing this call as well as our quarterly statement on our Investor Relations website.

During this call, we will make forward-looking statements, which are predictions, projections or other statements about future events. These statements are based on current expectations and assumptions that are subject to risks and uncertainties that could cause actual results and outcomes to differ materially.

Additional information regarding these risks and uncertainties may be found in our filings with the SEC, including, but not limited to, the Risk Factors section of our annual report on Form 20-F for 2025. Unless otherwise stated, all numbers on this call are non-IFRS and growth rates and percentage point changes are non-IFRS year-on-year at constant currencies.

The non-IFRS financial measures we provide should not be considered as a substitute for or superior to the measures of financial performance prepared in accordance with IFRS. Before I begin, I would like to highlight our upcoming financial analyst conference at Sapphire taking place in Orlando next month. We look forward to seeing you there. For those unable to attend in person, I would like to invite you to join our webcast. And with that, over to you, Christian.

Yes. Thank you, Alexandra, and a warm welcome to everyone joining the call. I'm happy to say that we delivered a strong start to the year. Our Q1 results demonstrate the ongoing momentum across our entire portfolio and the continued success of our strategy.

We achieved these results against a shifting macroeconomic environment. And while our business is very resilient, we are not completely immune to external dynamics. Uncertainty remains high, just as it does for every company. Still, we keep on delivering quarter-by-quarter, and we are also working on larger announcements for Sapphire, which will set us up to deliver high-value business AI at scale. We look forward to sharing more details with you in Orlando. Let's begin by looking at our Q1 performance.

Our current cloud backlog increased 25% to EUR 21.9 billion. Cloud revenue grew 27%, almost crossing the EUR 6 billion mark. The strong performance was driven by a 30% acceleration in our cloud ERP Suite revenue, bringing total revenue to EUR 9.6 billion for the quarter. This was a strong increase of plus 12%. Order entry from our public cloud solutions accelerated sharply in Q1, continuing the momentum from Q4.

Public cloud order entry accounted for over 70% of our quarterly volume. And with that, we keep on gaining market share, especially versus best-of-breed software vendors. Gartner Research just announced that SAP grew 15 percentage points faster versus the global enterprise applications cloud market in 2025.

Our strong top line performance translated directly into our bottom line with an operating margin of 30%, up 2.9 percentage points. As a result, our operating profit in Q1 increased by 24% to EUR 2.9 billion. Finally, it's great to see that our growing partner ecosystem performed exceptionally well. Our indirect channel order entry grew significantly faster than our direct channel, accounting for almost 30% of our total order entry.

Now let me turn to the broader macro environment. Geopolitical tensions, mainly the conflict in the Middle East, have increased. The war is already having an economic impact in the region and on many energy-intensive industries. SAP is, of course, not completely immune against these disruptions and the economic uncertainty makes it difficult to predict the impact on our total year results.

But we have seen many, many times in the past that customers are turning towards SAP in moments like these to invest into the resilience of their business. Our strong portfolio, together with our own resilient business model provides a great foundation for SAP in uncertain times. We see this reflected in a very healthy pipeline coverage for 2026.

Let's move on to some great customer wins in Q1. Leading companies from various industries selected RISE with SAP. They included ConocoPhillips in Energy, Thales in Defense, Air Liquide in Industrial Gases and Bristol-Myers Squibb in Biotech.

Moreover, PayPal, as well as the European division of automaker Hyundai and Swiss automotive supplier Aptiv, embarked on the RISE journey. Our software and cloud offerings also continue to gain traction with an important win at defense company Diehl.

Among the net new customers selecting SAP GROW are the superfood brand, OAKBERRY and Adesso, a leading IT service provider. On the AI and data side, we won Red Bull as well as Carl Zeiss, a global leader in optics and semiconductor manufacturing technology.

In addition, Knauf, a leader in building materials, German food company, Hochland and Swedish manufacturer, SKF, selected SAP Business Data Cloud. We also saw many successful go-lives.

Samsung Electro-Mechanics completed the S/4HANA transformation as part of their RISE journey.

Alibaba Cloud and Fonterra, a New Zealand-based agribusiness, both completed their transformation journeys with RISE. We are also proud to have supported ExxonMobil with a smooth deployment of their workforce ecosystem project. This go-live on SAP SuccessFactors now supports more than 60,000 users globally.

All of these customer wins demonstrate the confidence in our portfolio. This brings me to the core of our strategy, business AI. There is no doubt that AI will redefine how companies will run in the future. Here are some examples showing how SAP Business AI is already delivering significant value to our customers.

At Daimler Trucks North America, SAP Business AI helped transform how the company wins contracts with fleet customers. Bid win rates jumped from 10% to more than 40%, delivering a EUR 70 million financial impact within 12 months.

Queensland's Department of Transport and Main Roads uses SAP Business AI to predict road surface issues across 33,000 kilometers. Engineers can now run statewide investment optimization in a single day instead of a week.

Next to the accelerated speed, this predictive AI use case also generates millions of savings. A German manufacturer, Hormann, an AI assistant analyzes complex construction tenders in hours instead of weeks, reducing manual effort by up to 70%; at automotive supplier, Martur Fompak, the invoicing process was accelerated by more than 9x and product innovation time is now 30% faster.

We also see how our AI supports faster and more cost-efficient ERP migrations. For example, our partner, KPMG, uses tool for consultants to accelerate ERP migrations. Project wins are now completed up to 20% faster. EY uses SAP Generative AI app to accelerate how it delivers SAP transformation projects.

AI agents automate key phases from requirements through testing, reducing project delivery time lines by up to 30%. Bosch Digital equipped 1,500 developers with our AI, including SAP tool for developers for their ERP migration.

As a result, Developer productivity increased by 20% with unit test creation now 15% to 20% faster. All of these AI use cases deliver significant customer value today. At the same time, let's be honest, large-scale adoption of enterprise AI is still in its early stages. Also, we at SAP are learning our lessons every day with our customers about what it truly takes to make business AI work reliably, which is key when you run the world's most mission-critical and complex business processes.

Agents often don't have yet the full understanding of business data and processes to deliver highly accurate outcomes. This is needed to deploy at scale and with high accuracy, agentic AI use cases in the most mission-critical parts of our customers' business.

But we are learning fast, and we are very confident that SAP has compared to many other software companies, the right assets to win. Very deep domain know-how about business processes and data as well as enterprise-grade governance and security. SAP's ERP developed over 50 years can also be seen as the institutional brain of every company where data and process domain know-how is getting stored.

The launch of SAP Business Data Cloud last year was another important step to enable our customers to further expand the SAP [ semantic ] data model to non-SAP data. BDC allows our customers to build an end-to-end data platform, which is key for high-value AI. At Sapphire, we plan to announce some fundamental changes to our portfolio to infuse this deep domain know-how into SAP's AI agents, and we will govern the agentic AI layer for our customers.

This foundational change will enable SAP AI agents at scale to deliver highly accurate results to take actions across end-to-end processes in a secure manner. So let me address as well a question many of you have in mind. How will the AI transformation impact the financials of our company? First, SAP solutions as the institutional memory of every company will not disappear.

On the contrary, we expect to continue to gain market share with our best-of-suite offering because now more than ever, a harmonized data and process layer is key to harnessing the power of AI. With the infusion of AI across our products and migration tools, you will see an increasing share of consumption-related cloud revenue in our P&L.

But this shift will happen gradually over the next years with the expansion of business AI in our customer base. The good news is that in line with our system of records, the related subscription revenue will not disappear. Also, it is important to highlight that less than 40% of our 2025 cloud revenue was tied to named users. The remaining subscription cloud revenue is priced via non-seat-based metrics like revenues, memory used and other value-related metrics.

Let me emphasize once again that the ramp of consumption-based cloud revenue will be a gradual evolution and by no means a disruption comparable with the transition from on-prem to the cloud. At our Financial Analyst Conference in Orlando, we are going to show you how our AI transformation will expand SAP's addressable market as well as how both subscription and consumption-related cloud revenue will further drive SAP's growth ambition. Let's now look at how AI influences the way SAP operates itself.

Our internal AI transformation starts with our people and their skills. We are executing comprehensive upskilling programs across the entire organization, so every team can confidently apply AI in their day-to-day work. Our external hiring is highly targeted, focusing on recruiting top experts in data and AI. At the same time, AI will help us to run SAP as a company more autonomously in the future.

We act as our customer zero using our own AI across engineering, support, services and go-to-market with a direct impact on our top and bottom line financials. Let me share with you what we have already achieved with some great examples. In our engineering teams, we are using AI to work more efficiently with Joule for ABAP development and by third-party tools like Claude Code and GitHub Copilot, we are increasing developer productivity already by over 30%.

With AI assistance, our service and support teams are handling significantly higher ticket volumes without a proportional increase in headcount. AI assists 100% of support cases and 20% of our tickets are even resolved by AI, fully autonomous.

Thanks to these efforts, we observed a 12% higher productivity in our support function. At SAP, we have more than 80,000 colleagues in services, and our consultants save with our AI one day per week by much more efficient system configuration and custom code analysis.

This leads as well to faster project delivery and a huge productivity increase. For our go-to-market teams, AI improved our demand generation activities by personalizing and automating outbound campaigns tailored to customer situation.

It saved over 83,000 hours and directly influenced the pipeline with additional EUR 50 million of value. Even better, it helps us to target the right customers, identifying real pain points early and replacing guesswork with focused engagement up to 6x more effectively. As part of our internal transformation, we have communicated a clear goal to achieve a run rate of around EUR 2 billion in efficiencies by end of 2028, and we will share further details at our upcoming financial analyst conference.

Let me now summarize. We delivered a strong Q1 in a challenging and uncertain business environment. Whenever tensions and crisis occur, our software becomes more essential, not optional. This makes us confident for the remaining year. Finally, we are going to make significant progress with Business AI in 2026. You will see this come to life at Sapphire. With that, I hand over to you, Dominik.

Thank you very much, Christian, and thank you all for joining us this evening. As Christian stated in his opening remarks, 2026 is off to a strong start, supported by healthy current cloud backlog and total revenue growth, continued strength in cloud revenue and strong operating profit performance.

These results prove the merits of the strategy we've put in place and the cost discipline in managing our business against the backdrop of an increasingly complex and uncertain macroeconomic and geopolitical environment, now further shaped by the ongoing conflict in the Middle East.

SAP continues to be a valued partner for organizations of all sizes pursuing end-to-end digital transformation. And as we look ahead, SAP Business AI, the Business Data Cloud, as well as the Sovereign Cloud continue to play an increasingly important role in customer conversations and are becoming more relevant in their decision-making and deal activity.

We look forward to showcasing the progress we are making across these areas at our upcoming Sapphire User Conference in Orlando. Taken together, these results reinforce the trust that leading organizations place in SAP as they pursue complex transformation initiatives at speed and at scale.

Now let me provide more details around our financial highlights. Current cloud backlog reached EUR 21.9 billion, up 25%. While CCB growth held up remarkably well in Q1, we continue to expect a slight deceleration in this metric over the coming quarters.

Shortly after the escalation of the conflict, governments as well as other customers and industries directly affected by the consequences in their supply chains and production facilities reprioritized their activities to what one could characterize as immediate firefighting.

Cloud revenue grew by 27%. It was positively impacted by several quarter-specific effects. As these are unlikely to reoccur, we expect deceleration of cloud revenue growth in the second quarter. Cloud ERP Suite revenue increased by 30% in Q1, now accounting for 87% of total revenue growth in cloud.

Software licenses revenue decreased by 33%. Finally, total revenue in the first quarter was EUR 9.6 billion, up 12%. So now let's take a brief look at our regional performance. In the first quarter, SAP's cloud revenue performance was particularly strong in APJ and EMEA and solid in the Americas region.

Brazil, France, Germany, India, South Korea, Switzerland and the United Kingdom had outstanding performance while the U.S. were particularly strong. Moving down the income statement. Our IFRS cloud gross margin in Q1 was 74.6% and non-IFRS was 75.2%, up 0.1 percentage points and marginally down by 0.1 percentage points at constant currencies.

Nevertheless, IFRS operating profit increased by 17% to EUR 2.7 billion. Non-IFRS operating profit was even up by 24% to EUR 2.9 billion. IFRS and non-IFRS operating profit growth was supported by EUR 135 million decline of share-based compensation expenses.

The SaaS Apocalypse debate and the related 28% decline in our share price during the first quarter alone left its traces in that position. While we hedge lion's share of our cash-settled grants, the sheer magnitude of the move in the unhedged portion in combination with related social charges that are not hedged, provided this, I have to admit, unintended relief, adding to continued strong general cost discipline.

The IFRS effective tax rate was 29.1% and the non-IFRS effective tax rate was 29.3%. Both were driven mainly by tax effects related to nondeductible expenses. Free cash flow in Q1 was EUR 3.2 billion, impacted by a payout of EUR 408 million related to the settlement of the Teradata litigation case.

Finally, IFRS earnings per share increased by 9% to EUR 1.66 and non-IFRS earnings per share increased by 20% to EUR 1.72. Now on to the outlook. As you've likely seen in the quarterly statement issued earlier today, we are maintaining our financial outlook for the full year 2026. The outlook reflects the puts and takes we can quantify as of today with a reasonable confidence based on everything we have observed up to this point.

It is based on a scenario of a near-term deescalation of the conflict in the Middle East. Needless to say, a continuation or even further escalation of the conflict and most importantly, the continued closure of the Strait of Hormuz would have the potential to materially derail the supply chains of many industries that are important to SAP.

While as of now, the impact is limited to the governments and industries most directly affected, there could be contagion across supply chains on a global level. This, in turn, could jeopardize business continuity across many sectors and as a result, weigh heavily on customer sentiment and investment behavior globally, ultimately potentially impairing our ability to meet the current outlook.

Given the extremely high level of uncertainty around this, it is impossible to reasonably quantify the impact of such a meltdown scenario. And any speculation at this stage would almost certainly prove to be inaccurate, absent a near-term resolution. What we can say so, however, is that assuming a reasonable resolution and the reopening of the Strait of Hormuz in the coming weeks, mitigation measures on the cost side as well as the expected contribution from the Reltio acquisition, we continue to expect to reach the ranges we have previously communicated for our financial KPIs.

While we were able to weather Q1 basically [ unscathed ] in terms of direct P&L impact, we did already see war-related business impacts year-to-date. For the time being, we expect these to have a limited impact on 2026 cloud revenue, but you should not expect us to raise the outlook upon the imminent closing of Reltio.

We need vast contributions to secure a reasonable level of confidence to reach the previously guided range for cloud revenue. With respect to current cloud backlog, we continue to expect a slight deceleration over the course of the year with a clearly wider range of possible outcomes given the current environment, recognizing that the timing and pace of bookings can still move around depending on how the macro develops.

We did see some impact on half year 1 pipeline and bookings forecast from mid-March, i.e., 2 weeks after the beginning of the conflict. If conditions stabilize, we believe there remains an opportunity to close some of the deals that did not close as anticipated in the first quarter. Please keep in mind that the second half of the year typically accounts for the lion's share of our bookings and visibility remains limited given the [ evolving ] environment.

In summary, while we would have hoped for a more benign operating environment, we are very pleased with our resilience as evidenced by the solid start to the year and the continued progress we are making against our priorities.

The significant market share gains in our cloud business, most recently confirmed by both the most prominent specialized independent research houses, namely Gartner and IDC for calendar year 2025, are therefore solidly sustained into the first quarter of 2026.

While the external environment remains dynamic, we stay focused on supporting our customers, executing our strategy and positioning the business for long-term value creation. We are looking forward to welcoming you to our Financial Analyst Conference in May. Given the myriad of customer testimonials, product and commercial model announcements, we plan for Sapphire in Orlando. We fully trust it will be very much worth your time. Thank you, and we are happy to take your questions now.

Thank you, Dominik. And I would kindly remind you to only ask one question when prompted. Operator, please open the line.

[Operator Instructions]. The first question is from the line of Mohammed Moawalla with Goldman Sachs.

Congrats on the strong finish despite the macro circumstances. Can I just kind of dig in a little bit around kind of your commentary around the environment?

You talked about sort of some sort of impact that you saw, but clearly, we're able to manage that. But as you look into that kind of all-important second half, when you talked about sort of customer decision-making, there have been various kind of commentary out there around the pace of sort of the migration cycle.

Can you perhaps sort of isolate the macro from also maybe more product cycle-specific factors and your visibility around that sort of second half and more broadly over the midterm around how that product cycle evolves?

Yes, I can get started and then Dominik, please add your feedback as well. I mean, first, when you -- the first step actually exclude the macro for a second, I mean, I have seen such crises a lot in my career at SAP.

And what it always starts and leads to is actually that just today, two customers talking to me about resiliency in supply chains, about the intelligence we can provide for them to operate their supply chain. We talk a lot about transportation and logistics. So -- and we see this actually.

We have a strong portfolio there and that combined with the intelligence of BDC and the AI agents we deliver now step by step, actually makes me confident that, that part of the portfolio definitely will see very healthy growth this year.

You also heard me talking about AI. And I really want to be honest here to you. I mean, in the first step, all of the examples I have given to you are real, and we are delivering today significant value to our customers. Now is there also a learning curve for us here at SAP? Definitely.

And I guess every tech company has that because now when you talk about can we deliver these use cases, what I just highlight at scale, I mean, obviously, there is work to do, work to do on the ontology layer. I always tell my team on the product cycle of AI, it's actually great that we are running the world's most mission-critical business process.

You know what, this ERP, what we are actually owning has so deep domain knowledge. And now the task is for the remainder of 2026 is how can we infuse this domain know-how, both from a data but as well as from a business process perspective into the AI agents. And please also, I hope you're going to join us at Sapphire because we will make some very fundamental announcements to really show our customers how we will make this work.

And so overall, from a product life cycle perspective, also from what I see from a pipeline perspective, actually makes me confident but obviously and Dominik, please add your comments. There is no doubt, obviously, that we are, of course, also at some point, also impacted by the geopolitical tensions.

I mean, when you can't actually send your salespeople to customers in the Middle East, I mean, at a certain point, you will see extended deal cycles. And we have a few of those. And obviously, when -- especially in the energy-intensive industries, when this war now continues, when supply chains get further disrupted, I would say we, but as well as every other company will, of course, also see more challenges over the course of the year. But please, Dominik, add your comments, please.

I would like to maybe add one comment on the fact that AI is not only driving the performance of the products once they are in operations, but also helps us a lot on the transformation now. There is really now the first successes where customers are reducing the effort, the cost, the time to implement transformations.

And the big heavy lifting on the transformation is currently still the old ECC to S/4. So sometimes people speculate, well, can customers now wait. But recall, the extended maintenance begins '28, end of 2030, extended maintenance is over. And then the only kind of way to cope with that is basically our famous SAP ERP, private edition, transition option, which is quite costly.

So customers have a big incentive to accelerate, and there are now the tools available. So I think also the confidence of customers that this goes easier than maybe in prior times is increasing. That helps a lot on the migration cycle. Now on the war situation, I mean, I've been in complex supply chains and manufacturing.

And I mean, think about all the petrochemical, food, supply chain and what can happen there. So I think it's virtually impossible to understand what will happen if there is some shortages here and there.

The shortages might also go hand-in-hand with some people building buffer stocks, so there might be panic buying. And once certain products are not available, the product, the end product could not be built anymore. And that's the type of effect where, frankly, we cannot say when it's coming.

What we can say, though, is we don't think that any kind of additional week or months of shutting down the Strait of Hormuz will be a kind of x basis point impact on CCB. That's not how it works. It's a slippery slope where at some point in time, supply chains will be shut down and then we have a massive impact. So it's a little bit of a kind of more binary situation.

The next question is from the line of Adam Wood with Morgan Stanley.

Congrats on that good quarter. Can I just come back, Christian, you talked about the challenges of applying GenAI and the learnings you have to take to get companies to adopt. We're seeing obviously some of the labs really build incredible revenue streams very quickly here.

Could you just talk a little bit about what you're trying to do as you take over R&D? Is this a pace of innovation problem at SAP? Is this needing to get more people into the companies to understand where the pain points are to bring that back into the products? So could you just give us a little bit of an idea of what you're doing from the R&D side and what you're doing kind of consulting in with the companies to get that situation to change? And if there's any help you can give us on time frame to be able to accelerate, that would be really helpful.

Happy to do -- to share some more insights in what are our priorities also on the R&D side for the weeks and months to come.

I would rather talk about weeks. Look, I would say I would rather turn it around and start with the very positive. I don't see one tech company, also not the LLM providers who actually can, actually deliver at scale agentic AI use cases for the world's most mission-critical business processes.

I mean, when I actually started to harmonize data models at SAP 5 years ago to solve the integration challenge, I never thought that 5 years later, actually, we are sitting here in front of 7.3 million data fields in our ERP where you now need to build knowledge graphs that correlate this data to actually solve some of the most complex tasks in the world.

And for customers, and that is also our lessons learned, I mean, we deliver those use cases. But they are -- today, they are 85% accurate. They are 90% accurate. But is this enough when you are touching the payroll, the finance, the financial close, the supply chains of the customer. No, it's not enough. It's not that the customers don't see the value, they see the value, but we have to go the last mile, and that has a lot to do with the ontology.

And so we made some fundamental decisions already a few months ago and how we can also change the architecture of our solutions. You're going to hear some news about our platforms and how can we combine these strengths of SAP. And again, I really want to emphasize, this is a problem which actually SAP is suited best to solve because I don't believe that over this 7.3 million data fields and over 120 mission-critical business process, I don't believe there's any other software company for sure, not the LLM providers who actually can sort that out.

And that's actually our task. We are very confident to do that. And I mean, in these ERPs, there's 50 years of know-how, and this know-how also sits within SAP. And yes, we are going to make that work. And that is the main focus now also on the R&D side. And you can imagine that is also not one job profile who can sort that out.

You need to have the best industry consultants and pair them with the data scientists. You need to have product managers who also then understand how these agents work in a real-life environment.

And when you form these teams, and we call this the all-in on AI program within SAP and bring them together, great things can happen. And that is definitely something what we already changed a few months ago. And since then, we are seeing really great progress. And again, I hope every one of you is also coming to Sapphire to see this in real life.

The next question is from the line of Charlie Brennan with Jefferies.

Christian, can I ask one for you, if I can. I think you very helpfully clarified that the impact of the consumption model will be gradual over time and less severe than we saw with the transition to cloud. Are there any parts of the AI journey that you think will be actually similar to the cloud transition, whether it's in terms of investment costs and maybe margin implications going forward or maybe the need to do M&A to onboard skills that you don't currently have?

Yes. Happy to give some more feedback. And again, I would not compare this transformation to the cloud transformation now. We started 5 to 6 years ago. I mean, especially not from a financial perspective, moving from -- at that time from an upfront license model to a cloud business model.

And let's also not forget, consumption is not new to SAP. I mean I'm very happy that already a few years back, we -- we infused consumption price models also for certain parts of our platform. So the metering, the adoption metrics, also all the incentives, obviously, the tools to drive adoption does already exist within SAP.

So we don't need to transform again certain parts of the company because we have people, we have the tools, we have the systems to make that work. Now I would say what is, again, our main focus and which is then so key to our success, and I can only emphasize this once again, is 5 years back, I learned how complex the master data model is of our ERP. And now I'm learning to know how to not only harmonize data, but how to correlate data, how can such a graph understand billions of correlations. It's not only the 7.3 million data fields.

You need to make sure that the agents understand how to correlate this data and then actually give the LLMs the context they need to sort out the business queries or actually even take actions in some of the processes we are running.

And I guess this -- that is the key part. Now maybe one thing is also important for you is when we actually started to deliver our first AI use cases, and again, lessons learned. We actually started that, oh, we are now running in the cloud, and we can actually plug and play and actually can scale these AI use cases across our customer base.

I mean, again, when you are then sitting in front of a very complex customer landscape, you have hybrid landscapes. You have customized ERPs, you have public cloud solutions. I mean it's not that easy then to first get the agents, the identity and the authorizations, they need to, first of all, understand what data to access, what data to share. Needless to say, the extensibility layer then in the graph to also then make sure that the customized data fields are understood or sometimes customers have very -- oftentimes actually have very individual process requirements.

And today, I had a customer in pharma, and we talked about GxP compliance. And again, it's sitting in the ERP, but the agent needs to understand GxP compliance. I mean -- and this is the stuff what I mean with lessons learned.

But trust me, there is also no other AI agent out there who can actually run such processes for a pharma company today at scale. And that's why I'm saying -- that's why I'm confident, and that's why I'm saying, hey, to all of our engineers, it's actually a good challenge to have.

Maybe on the margin, first of all, I want to reiterate that as of today, we say no reason whatsoever to change the envelope we have given in terms of operating leverage.

We continue to say that our total expenses will grow in that 80% to 90% of total revenue growth. We did say that we want to keep the flexibility, how this is exactly allocated. And even if there might be a slight margin kind of implication from what you alluded to, there's other upsides. Think about the sheer explosion of productivity on development.

I think what is one of the biggest challenges for SAP and has been a challenge for SAP was feature gaps that we took too much time to close and then even some new competitors were sneaking into these gaps.

And now we can really pull in the closure of the feature gaps. I've not seen customers yet to say in this regulatory report process, I would love to develop all these scenarios myself. I said, if you, SAP can do that for us.

But if you are so slow, we won't wait for you. I think that's the discussion we have today. So there's many fold levers on productivity, and this is why we're deeply convinced that we can stick to that kind of corridor.

And don't forget, when we talk about foundation models and transformer models, we are not doing the supercompute-intensive LLM training on unstructured data with fluffy data where you have a lot of data to crunch to come to some results.

We only do that with RPT-1 in the pretraining phase, and that's a highly structured, very kind of efficient pretraining. So we're not expecting the same type of myriads of investments that other people need to spend on to get these models trained because we can build on these models being built.

Yes. And there was a question also around M&A. I mean take our intent to acquire Reltio. I mean we clearly see the need to govern master data now with BDC, not only for SAP, but the agents need access to non-SAP data.

Now again, access to data doesn't help if it's semantically not governed. And that was the reason to hopefully then close soon the acquisition of Reltio. And now with Muhammad and [ Philip ], we are sitting often together with Sebastian and discussing, of course, where in the AI and data space, do we have a few more white spots, both from a technical perspective where we can accelerate certain things or actually from a skill perspective, where we have skills, clearly, with good skills, but maybe where we can also then accelerate certain things because then we have more of those skills.

And this is something what we are now constantly, of course, looking at. And so there is, of course, the chance that we do a few more of those. I would rather call it tuck-in acquisitions because they are clearly not meant to acquire revenue. This is then really clearly focused on the data and the AI space.

The next question is from the line of Ben Castillo with BNP Paribas.

It felt Christian, that the message last time out was that, look, we're selling more product to our biggest customers. You called out at Q4, 90% of your top 50 deals containing AI.

Has anything changed here so far this year if we exclude what's going on in the Middle East and perhaps the firefighting that Dominik alluded to? We heard this week from some IT services partners that some customers may be pausing migrations or perhaps changing the priority of their SAP in their overall spend.

What are you seeing? That would be helpful. And just a quick follow-up. You said that total revenue growth this year is now expected to be broadly stable, but reaccelerate next year. As far as I can tell, you're reiterating the sort of various line items. The only one that maybe is missing is services. Is that what's causing that change to this year's outlook? And if so, why is that?

I can start with your question on AI and then please, Dominik, take the question on revenue. On AI, I mean, yes, clearly, I mean, Q4 was definitely a very good quarter in infusing AI in many deals. I mean, obviously, now in Q1 and in Thomas Saueressig's function, obviously, there was a high, high focus now on making this AI productive, start working on putting those agents into production with our customers.

And that actually progressed really well. Now also a similar pattern actually now in Q1. Now -- but what Dominik mentioned is, I guess, very important, especially now in Q1, which was very encouraging to see many existing RISE customers and to those who are now embarking the RISE journey actually came to us and said, "Hey, we definitely realize that in order to harness the power of AI first, I need to have a cohesive and semantically rich data platform. So I have to do something." With this kind of customized ECC, it's so hard. It will always stay custom, and I cannot build a lot of custom AI agents for mission-critical parts of my business.

I can't do that. And I also don't want to run those because actually, it also has a compliance challenge coming with it. So that is the one thing. And then the second one is what Dominik mentioned is actually what was very encouraging that the AI tooling for ERP migration, we definitely had a very good quarter. They are selling very well, but even more important than selling is also that we see the adoption is coming.

And that is also something which is then related to our partner ecosystem because I, of course, tell all of our partners, I said, "Hey, hey, hey, we are transforming, so are you transforming." So these ERP AI migration tools are a must-have because customers want to see the cost coming down for these ERP migrations now very clearly, which I take actually as a plus for SAP.

And by the way, at the risk of stating the obvious, the fact that the increase of adoption of these AI migration tools, of course, that is reducing the budgets for SIs.

So that is not necessarily a bad sign if that happens, that doesn't mean that the project has stopped. It could simply mean that people are doing more work leveraging these tools. Now on the total revenue discussion, you point to a very important fact. You've seen the services revenues kind of slightly declining.

And yes, we have made a deliberate decision to invest more in adoption support for our customers, and we're not kind of chasing every hour to bill in the environment where also the migration tools play more and more of a role. So there isn't a [ pivot ] there, which was not visible at that point in time.

So that's one explanation. And also, frankly, we didn't want to kind of -- we want to be honest about that we need to include the Reltio acquisition, which is nonorganic to kind of protect the range, so to speak, in light of the setback we just discussed.

So this enables us to support the guidance for this year. Now why are we then still confident that '27 should accelerate? First of all, the services setback, so to speak, is a one-off happening this year.

So that will be digested. And secondly, if you look at the ramp of our backlog, and we commented on that also at the end of Q4, there is a much bigger increment, so to speak, coming in '27 out of that backlog than we can benefit from in 2026. And based on this, so still on the premise that we are not going down in a meltdown scenario on near East, we think the acceleration is a reasonable assumption.

The next question is from the line of Mark Moerdler with Bernstein.

Impressive to see how well you executed this quarter and how you delivered on the current cloud backlog.

Can you give us any sense for -- other than the risk you've already discussed that there might be some pull forward to make that happen or anything that could create an impact on the future quarters other than again -- other than the macro issues we've discussed? And also, have you had to do anything relating to discounting or contract duration or whatever in order to deliver the solid numbers?

Thanks for the question, first of all, Mark. I mean, I would say, I would call this a very clean quarter. So when I saw the gross margins of the deals we closed, I mean, there was -- actually, it was just to -- in contrast, actually, we saw a very healthy quarter also from the deals closed with regard to gross margins.

We actually -- the pipeline conversion this quarter was also again going rather smoothly, again, except the Middle East. And I really want to emphasize that. I mean, we are not immune against what is happening in the Middle East.

And so no, it was actually from a 360 perspective, a very round quarter. And definitely, there were no further incentives, et cetera.

I guess the main focus, especially also in the field in Q1 now when I see -- look at our top customers was rather to build the pipeline now up for Q3 and Q4. And so that was, of course, another major focus area now in Q1.

The next question is from the line of Frederic Boulan with Bank of America.

Quick question. Do you see any companies reassessing their current migration road map because of new available tools using tools like BDC to extract SAP data and try to build agents outside of the SAP ecosystem?

So there's a lot of concern, as you know, in the market that the agentic layer will capture a lot of value. I mean is this something you see some customers trying to implement already?

I mean, obviously, now did we talk to customers who are trying and who are developing certain agents in correlation or in conjunction with SAP data and process? Yes. I mean the good news is also that we also talked to many customers, I would say, who also then said, hey, we build a few custom agents.

But to Dominik's point, who then actually came back and said, "Hey, when you are building for us this agentic AI use case, there is no reason for us to do this custom or to do this with a large language model provider." So that is the good news. We'd rather see it where we are not yet having an offering out there.

I mean, obviously, this is why watch out for Sapphire where we're going to see where you're going to see the acceleration of agents and assistance come from SAP. But even more important, as I mentioned before, is the underlying platform.

So to really not only deliver a high number of agents, but really also deliver the quality and the accuracy coming with those agents. And so yes, of course, we see those examples, but nothing where I would say now I have tonight a sleepless night because now customers I see already forming a layer on top of SAP. This is clearly not the case.

I mean there's one thing we should all not forget, we talk the lion's share of what we do in SAP is around hard monetary transactions in complex end-to-end processes.

This is a very different business from creating content like writing a software, writing a text and being a 95% accurate service agent on a hot line, that is a very different business. And in my role as a CFO, assurance that these are precise numbers is ultrahigh. And we see a lot of customers in all these monetary aspects where they say, if we can kind of get the rubber stamping from SAP that this will all work and we can trust it, that's worth a lot.

So don't forget that kind of assurance requirement we have in many of the domains we are catering to.

The next question is from the line of Michael Briest with UBS.

Just in terms of -- you mentioned there, Christian, about the value of your data fields and workflows. Do you feel you have enough control and value protection around that?

Are you considering any changes on third-party access to your systems? I think some competitors have limited or tried to charge for that, I mean, digital access, right? So I guess it would be called. And we've elliptically discussed the Financial Times article a few times on this call.

What was the purpose of your messaging there about no long-term gain without short-term pain? Is that a message to customers to move more quickly to cloud and ERP? Was it the workers at SAP? Was it to investors in some way?

Yes. I mean maybe I'll start with the last question first. I mean the short-term pain was clearly, of course, correlated to when you look at the software sector overall and you see how high the pressure is on all of those software companies now to transform.

I mean this creates, of course, some short-term pain. It's not like that when you would look inside SAP that within the all-in on AI program, of course, we need to -- we are fully aware that we need now short-term a laser-focused execution.

And of course, also our employees are asking, hey, is our strategy now gives us -- does it give us the right to win? And we have seen it now. I mean, there's a strong belief inside SAP, and we are very proud about the domain expertise, but there is, of course, also some pain when you look at things which is talked about in the market. And obviously, we are also now engaging very closely with our customers.

Now on the first part of your question is when it comes to the domain knowledge and the domain know-how and how we can protect that. I mean I want to say this very clearly in today's call because I also saw today certain articles out there in the market. I actually was also under my time as the Chief Operating Officer, we actually killed indirect access where there was very bad times of SAP where we said, hey, charging customers for accessing their data.

That will never ever happen today. Customers' data is customers' data and accessing those data, we are not going to charge. But there is a big, big difference now in the cloud world and in the AI world about just accessing the data, which we have no plans to monetize at all versus accessing the IP, the domain know-how sitting in our ERP.

I mean, the [ semantic ] data model, the [ semantic ] process know-how. And that's, of course, something the ontology, the maps, the graphs, this is what we, of course, will actually offer on our platform, but we are going to protect that.

And then last piece, and we will share some further details about that at Sapphire no partner needs to worry as well. I mean we love the partners. We will have an open platform where we also go on, actually SAP agents not coming from SAP.

So we will provide those APIs absolutely clearly. There's only one thing on API, what we already realized. Obviously, when there is mass data egress or millions of calls coming towards an API, we need to start throttling those APIs because otherwise, we are ending up or the customer is ending up in performance issues on the application side.

So these are the things what we are now rolling out. But again, no customer, no partner needs to worry. We all want them, and we want to have an open platform. But please also understand that, of course, the IP of SAP, the domain know-how, of course, is something what we will make available to our customers. But it's, of course, on the other side, something which a great asset to protect.

The next question comes from the line of Jackson Ader with KeyBanc Capital Markets.

Really just one, mostly around guidance. Number one is how much of the Reltio acquisition is actually included in your full year revenue guidance? And the follow-up is, if we are assuming a near-term deescalation in Iran, then what is the reason -- like what is the main reason then for needing this extra inorganic buffer in order to hit your full year guidance?

Yes. I mean on Reltio, luckily, the company on February 10 in a press release, which was referring to $185 million ARR as of year-end 2025, that representing a significant growth acceleration.

So if you think -- we said that the closing is expected to be imminent. Of course, we cannot kind of commit to a specific day, but it should be really, really near term unless there are some surprises here. So we are now end of April, so that's 4 months out of 12 months. So 2/3 remain to do of $185 million.

So if you just take that as a yardstick without being precise, I think you have a good feeling as to how little actually is included in that kind of uptick.

And then let's not forget also, I mean, half year 1 is not volume-wise, not our biggest order entry quarters. But obviously, it absolutely matters for the total year cloud revenue, while a Q4 where we closed the highest order entry is actually, of course, very important for the guidance next year and for the CCB exit rate.

But of course, something what we are going to miss now in March, April and will not come back in the next 3 to 4 months is, of course, something what we cannot just make up on a total year basis with regard to our cloud revenue.

Thank you, Christian. And this concludes our call for today. Thank you all for joining.

Thanks a lot, everyone.

Thank you.

Bye-bye.

Ladies and gentlemen, the conference has now concluded, and you may disconnect your telephone. Thank you for joining, and have a pleasant day. Goodbye.

Okay. Good morning, everybody. I'll start off day 2 of our conference here in San Francisco. Thank you very much for joining us. My name is Adam Wood. I look after European software and payments research here at Morgan Stanley.

It's a great pleasure to have Muhammad Alam with us. Muhammad, thank you very much for joining us. Muhammad is Executive Board Member at SAP responsible for Product and Engineering. So thank you.

Thank you for having me.

Alam, trying to get these out the way as quickly as possible, a few disclaimers. So for important disclosures, please see the Morgan Stanley research disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley representative.

On the SAP side, during this fireside chat, SAP will make forward-looking statements, which are predictions, projections or other statements about future events. These statements are based on current expectations, forecasts and assumptions that are subject to risks and uncertainties that could cause actual results and outcomes to materially differ. Additional information regarding these risks and uncertainties may be found in SAP's filings with the Securities and Exchange Commission, including but not limited to the risk factors of SAP's 2025 Annual Report on Form 20-F.

So with those out of the way, we can get on to more interesting things, hopefully.

So maybe just to start off with, there's been some recent news on your side that you're not going to be renewing your contract with SAP next year when it expires. I wonder if you'd be able to elaborate just a little bit on what the reasoning was behind that.

Yes. I mean, I think not to go into too much detail, but there's a few things going on, on the personal side, for me, which made it a little bit hard for me to commit today a contract extension. And German governance requires, once a contract is coming up for renewal, that, a year in advance, you have those discussions. So I just wasn't at a place personally to be able to commit to that. And we thought it was still, from a transparency perspective, the right thing to communicate both to our colleagues as well as externally too.

But I mean, listen, the way I look at this, we still have more than a year left in the contract. And a year, as you all know now in today's environment is a lot of time to make some pretty rising impact. So that's what we're focused on, both at SAP internally as well as for our customers.

Well, best of luck for the next 12 months, and, yes, looking forward to seeing the impact that comes through.

I was at a dinner last night, and I think we were all asked big theme and the thing we wanted to get out of the conference. I think half of the room said, "How do we understand the moats for incumbent application software vendors?" So that was where I wanted to start off with, the kind of elephants in the room of how you protect what you have already and your moats. So how much risk do you think incumbent software vendors are from GenAI disruption? And what do you think SAP's biggest moats are?

Yes. I mean I think, listen, I mean, let's start with GenAI does pose significant -- has significant impact on existing software vendors. I think that's clear. Nobody is debating that. I think the thing that I feel like we lose in that discussion is not all SaaS vendors, not all applications are created equal, nor do they play the same role within a customer's environment.

So while it is true that a certain class of SaaS applications are going to be a lot more prone to risk, disruption, transformation, I think there is a class of SaaS applications, and from an SAP perspective, we deeply believe we play in that class that are extremely mission-critical for an organization to operate. When you think about financial management, when you talk about general ledger, you talk about treasury, you talk about GL, AP, AR. And then you get into the supply chain side where you really run your shop floor or your logistics environment and things like that, that's a different class of applications.

At least in my mind, sort of the moral equivalent, the way I think about it is, listen, there's a class of apps -- if you think about maybe even Windows, for example, right? It's an application at the end of the day. But it's sort of the OS of what consumers use every day on a device. SAP is very much like the operating system of a business.

There's certainly a set of apps that are built on top of it. A lot of our partners do that, a lot of integrations exist, if you will. But it's hard to go take out the OS, if you will. While you could go build an OS with GenAI, are you going to go now start writing your own OS to be able to run your own devices? That's not necessarily true.

But apps on top of it, sure, and customer applications, those lend themselves a little bit better to be able to go do. And then from an SAP perspective, while, listen, even in the class of applications we work in, there's phenomenal opportunity to create incremental value that now is possible with GenAI, that wasn't possible before, and that's what we're focused on. We do believe that the risk of disruption is different.

And this is where I think the conversation that's happening today out there, is, hey, all SaaS apps are treated similarly, and that's what the initial reaction. But we fundamentally, when we speak with customers, when we look at our portfolio, the reality seems very different.

That brings me really nicely into the next question, which is this idea of there's a set of applications that already exists, the operating system that you talked about. But then there's this amazing white space that this new capability and technology opens up. Maybe first of all, how difficult would it be to replicate that operating system that you talked about? And then what advantage does the operating system give you as you think about moving into the white space and automating things that couldn't be automated previously?

Yes. And again, a very good question. So let's decompose sort of this class of applications and why they're different. That's sort of the crux of the question, right? And if you think about the things that makes this class of applications different is sort of 50 years of logic and business processes that are embedded, not just in the horizontal domains that we talked about, which is finance, HCM, spend, supply chain and so forth. But it's the deep industry capabilities in oil and gas and what we've done in manufacturing and what we do in media and others, if you will.

Now that collective knowledge as well as the data that around it, the semantic understanding of it, the business process definition of it, sort of creates -- and not just that, right? For us as an enterprise player, because our history is we run some of the most mission-critical, largest organizations across industry, the level of sort of enterprise readiness, the governance, the security and the privacy and the fact that we're localized in so many countries around the world, creates this moat, this barrier of entry to say "Hey, I'm going to go build something and put on top of it."

Now that's one aspect of it, right? So there isn't necessarily an immediate threat to say, hey, somebody is going to go build maybe a GL system, localize it around the world and be able to sort of comply with all the statutory. I mean that even exists in parts of our portfolio like Concur. I know there's been some discussions that say, "Hey, is Concur more prone to disruption?"

But if you look at Concur and you decompose that business, it's expense management. And expense management may sound simple to the layperson, but there is a phenomenal amount of statutory requirements around the world that we spend a significant amount of our R&D investment, keeping up to date on a very regular basis. For sure, you can go create a forms-over-data application that allows you to capture expenses, but who's going to go do the compliance to the statutory requirements? Who's going to maintain that over time? Who's going to make sure all of that is integrated across your core OS for it to work?

And that's why even if you sort of look back over the last 1.5 decades, there, from an expense management perspective, you think if it's that's easy, the space will be pretty proliferated. But it's not. Because even there, it's a moat from an enterprise perspective, from a localization perspective. Now that said, that's more of a barrier-to-entry moat. The question is there is now an opportunity to create tremendous value on top of it. And that is what we're deeply, deeply focused on.

The thing that makes me sad, honestly, for the first question you asked me, is this is a once-in-a-lifetime sort of, to me, in a tech space, in a business application space, opportunity to create some phenomenal impact and value for customers and, hence, value back for SAP and our shareholders. And we're deeply, deeply focused on that.

But if you think about sort of the opportunity that exists on top of it, certainly, automation is one big one. There's intelligence that sort of requires that data context, that process context, that context graph to be able to come with automation, to be able to sort of create autonomous experiences as well.

And as you would expect, what we're working on Sapphire for us is around the corner, right? And I'm sure everybody would expect, like what we have coming up is one of our most ambitious launches in terms of now taking the opportunity and what's possible with GenAI and AI in general, putting it on top of the sort of core OS of applications that are unlike any others with its enterprise readiness, localization breadth, industry depth, and creating value, which is what our customers are looking for.

And again, I like analogies a bit so I'll also sort of give another analogy here, which is if you think about autonomous applications in some ways, right? Autonomous not just meaning automation, but autonomous has a significance of intelligence that's baked into it, as well as being able to do things that, frankly, maybe humans weren't able to go do, because now you can do that at scale across a massive amount of data that's very hard for humans to go do.

But if you think about sort of this autonomous -- concept of an autonomous business application, it's not very similar to, take autonomous self-driving cars, right? In sort of that core mission-critical set of applications, can you really go by autonomous software to put on a nonautonomous vehicle, and feel like, "Listen, this is perfect, I'm just going to go make my own car autonomous and drive it around the town." It just doesn't work like that. There's a level of seamlessness in the app, data, agent, think experiences, that really creates phenomenal value that's very hard to plumb on top.

Can you do it in certain business case -- business processes? Yes. But then you have to also be aware of the TCO impact to the customer on the other side. Because what the customer is then doing is saying, listen, for this core area, this -- think of it as the OS of the business, I'm going to go pick something else, a PaaS provider, which everybody, and their brothers today, is building an agentic PaaS platform, right, that somebody can go plumb on top of whatever exists underneath.

But you take on the cost of understanding the data model, which is very hard to go do, doing the integration, figuring out does the context exist, and then taking the risk of the decision and the recommendation that came out to your financial management, to your supply chain processes, to your spend, which are pretty -- you need to be very deterministic. You can't be probabilistic like you can be maybe in some of the front-office business processes, right, in lead opportunity and so forth. You have to be very deterministic here because a simple error could be pretty costly for the organization. You really build, add to the iceberg that exists for organization in terms of maintenance costs, integration and things like that, which, in some cases, even preventing application of the agentic layer.

But again, to summarize, to me, I think this app, data, agentic experience that sort of creates that value seamlessly with the core OS, the systems of mission-critical applications that exist underneath, not just creates value that's unparalleled by just PaaS providers that you can hook into your environment, but it also leads to more deterministic outcomes.

Because we feed it, I think you talked about the moats, and we can go a bit deeper into it, the moats that we have, which is deep process understanding, the data that we have, the data model, the SAP Knowledge Graph and so forth, that we're sort of building into as a post-trained model to work with what you can find off the shelf. And those 2 together then sort of create that value that we believe is very unparalleled.

And I think the confidence that we have, again, I think the debate is raging out there in terms of what does that mean for SaaS applications. For us, what we're focused on, again, as you would expect to hear a bunch of things coming out of Sapphire, the proof is going to be in the pudding. And I think the feedback that we're getting from customers is, hey, listen, I want to go run an autonomous finance organization or an autonomous supply chain or autonomous spend. There's a level of credibility connectedness that comes to make sure that the whole vehicle is autonomous. Not that you buy a layer on top of it and say, "Hey, I'm going to put it on my 1990 BMW with this new software that's there and I'm just going to sit back and let it drive me home." You could do that, but there's a lot of risks doing that.

It's not a car that you would want to get into yourself.

Yes, yes.

No, it makes sense. You've already alluded to this, but I mean, I think that this is a big investor concern, that it's not that we actually replace the systems of record underneath, but rather the user interface shifts to probably a chat-based LLM interface sits above it. There's more interaction there and maybe more value gets created. For people that are running multiple systems of record, investors start to think, well, that could make sense.

How do you think about that risk? How much do you want to collaborate with those companies to enable data access? And how -- the value of the intelligence that you have in the system, how much do you protect that rather than wanting to collaborate?

Yes. I mean it's a loaded question, so let me see. If I forget one part of it, remind me. But the first part you asked is the interaction layer -- yes, the agentic experience interaction layer. Because I do -- I think one of the things we also believe is, for the first time in sort of a long time, the shape of the stack is changing because there's a new layer now on top, which is this agentic experiences layer that sits on top of everything below. And then arguably, everything below becomes more platform, more commoditized and the value shifts up, if you will, both in what it creates for customers and the organizations that can charge premium for it as well. And that's really where the interaction layer happens.

We used to sit at the top as an application layer, I would say. I mean we're still there. The world hasn't completely changed yet. But we see the signs that it's changing. So for us, again, it's not that, as you said, the lower layers are going away, like the application data, the PaaS and the IaaS. They're not going anywhere. They're still there from that perspective.

The question is, for this new layer to work, how does the most value get created? And we believe, again, this agentic layer, the application that needs to exist, because that's where data gets created, that's where action gets taken, that's where compliance happens, that's where statutory requirements happen, if you will, particularly, again, for the class of apps that SAP is in. If you take a small single-lane application that just maybe does an HCM or does some front-office applications, that's a -- I think that's a fundamentally different landscape. I mean we stood on stage -- it was last Sapphire, and talked about, listen, we believe that the best-of-breeds will struggle in this world.

But where SAP's moat exists even in this sort of new-stack layer, if you will, is the application layer becomes, again, a platform, if you think about this, right, because value shifts up. And from an application platform perspective, what organizations are telling us, this sort of sweet message to say, "Hey, at this platform layer, we need finance, spend, sourcing, supply chain, HCM, CX to work sort of seamlessly across." So SaaS becomes effectively a new platform.

And we're one of the very few, you can count on one hand with maybe a couple of fingers, that can sort of go across that breadth of the business processes to provide that SaaS platform to power the experiences on top of it end-to-end. Because if you pick and choose in that SaaS platform 6 different providers, guess what, you have to build the integrations, you have to make sense of the data model yourself. You have to necessarily push the data somewhere else to be able to connect to. And that's resonating with customers because that also allows them to melt the iceberg, right, the complexity and the TCO that they maintain.

Now that said, it's clear to me, certainly as a product leader into SAP, that we have to have a play in the agentic layer as well. We want to make sure that when somebody comes to Joule, they can ask a question about finance or supply chain and spend, from the same pane and get the right answer back through the SAP Knowledge Graph, through the sort of fine-tune model that we talked about, to get the most best response. And then the agentic experience is, again, leveraging that moat, encompassing in that agentic layer for us, sort of creates autonomous experiences. That's very hard to build as a PaaS story on top of it, right?

So that's what we're working on really unlocking, working with customers and unlocking for our customers as well. And sort of this redefinition of the applications, these core applications, as agentic experiences, is, as you would imagine, something that we're working very, very aggressively on. And you'll hear a lot more about that too from a product innovation perspective.

But through this, what we don't aspire to become is to say, listen, we're just going to now play in this SaaS as a platform layer. We do want to play in this agentic layer. But we'll, of course, be compliant with A2A, that if you have other things, as you come understand the SAP data, you come through our agentic layer, if you will. And it then creates this sort of stickiness for SAP's agentic layer through everything else that happens in the organization.

Because while we're mission-critical, we're, in 99 out of 100 cases, not the only application that exists. There's some deep industry stuff, if you take oil and gas, upstream, downstream. If you take pharma, there's other things that happened as well. There are some industries we're actually pretty complete, like discrete manufacturing, it can mostly run on all SAP and things like that as well. But it allows that level of interoperability, if you will, on the moat that we have.

You've alluded to this a few times, this idea of the data, the business context. That semantic business context, how much does that still matter? And the question I get is, can these models not come into the system and infer that structure themselves?

Yes. They can. I mean, I think they can. But again, the difference is: for which kind of application, right? You can even put an MCP server, for those that are familiar with how MCP server works, in front of what I call a single-lane, single-domain application. And it will do actually a pretty nice job in translating the natural language query that's coming in into what is the answer that needs to come back. But soon as the MCP server has more than x number of tools underneath it, the accuracy just goes haywire, you can't rely on it.

And if you look at the SAP landscape, the SaaS platform across finance -- even within one of them, the complexity is so vast, the data model, the table, the entities, the extensions customers build is so massive that it just doesn't work with a simple MCP server that you can put on top of it and think, as you might read in some LinkedIn post, "Hey, this is sort of going to be amazing and game changing."

Now listen, I say that, but this understanding of the context to me is one of our deepest moats. And it's very hard even for us as publishers of the software to sort of solve it in a way where the SAP Knowledge Graph, the SAP process graph, the context graph through the interactions and the behavior of the application, is available at scale for all the agents all the natural language interactions. We've been at this for now 1.5 years, is when we initially said, listen, we're actually getting pretty close to our SAP Knowledge Graph, and this is what we've solved for. And as hard as it is for us to go solve, this is where our confidence comes in, that at scale, to sort of solve for what S4 has, with Ariba, with our supply, it's a very hard thing to go do. And hence, that creates the moat.

So you have the raw data, but it's the relationship of the data and the semantical richness on top of it and then how it's used in the processes, is what creates that moat. And that's what we're sort of building into fine-tune models that, of course, the agents and Joule works on top of it, to again address a very vast surface area of an enterprise that otherwise is not easily understandable.

Now what you could do is take the raw data, put it somewhere. And customers do that today. But then by definition, A, you've taken a lot of burden because you don't have the semantical context. B, you also lose the permissions and the authorizations, right? Because as soon as you go put it into a Databricks or a Snowflake or an Azure or a Google Big Query, then all it is, is sort of flat data that doesn't have permissions and authorization, which is also very critical. All of a sudden, because of GenAI, you're not going to start sharing your payroll data with all of your employees, your financial data everywhere, your supply chain data at scale from that perspective.

So we maintain that as well. So our partnerships, particularly with what we've done with Databricks, allows our customers to maintain the data gravity with the semantical richness within SAP Business Data Cloud, that we launched last year, to be able to then create scenarios that are far more powerful from that perspective.

So this is all about the difference between a point SaaS solution and then the breadth of complexity in solutions that you're delivering the compliance and governance that sits all around it, that really differentiates you from peers.

Again, we move on to data, and that's obviously a big moat for you. You mentioned Business Data Cloud. Could you talk a little bit about how you think about how customers lever the data responsibly and how that works through Business Data Cloud, how you're managing those partnerships?

Yes. I mean, I think when we launched a Business Data Cloud less than a year ago, initially with our partnership with Databricks and then we announced a partnership with Snowflake, Azure and Google as well. And for a first-year offering, it's actually been phenomenally successful. And no surprise to hopefully anybody here because data is the fuel that sort of powers the value that comes from AI.

And the way we've sort of designed the architecture, it allows for 2 things, right? Clearly. One, we want to make sure that we bring to our customers best-in-class data engineering tools that we, at this point, don't feel like it's a battle we want to go win in, and organically build data engineering tools, because that market both become pretty mature and pretty commoditized. So we've brought those tools natively sort of embedded, in some cases, OEM, within Business Data Cloud, maintaining the data gravity, the semantical richness, and our customers can go use that to be able to sort of create value from it.

So we're focusing on the things we add value on, which is harmonize data model across the applications, the connectedness, the Knowledge Graph of it, the semantical richness, partner on the things that have become commodity and data engineering, if you will, for 2 reasons, right?

One, if you want to necessarily -- if you want to create value on top of that data through somebody else's AI tools, BDC is still the way to go do. So we become in that flow of value that gets created by anybody a customer may use. And by definition, that it gives us a level of both attach, stickiness with the value that we provide on top of that data, which hopefully as we just discussed a few minutes ago, is far different than sort of taking raw data out of the database somewhere else and trying to plumb on top of it. So there's value on top of it, and now we're in the flow of value that's being created theoretically by somebody else.

The second part of it is because we now have that data available the way we do, bringing now some of the things we talked about earlier, you can help create agentic experiences and value on top of BDC with our AI platform that, again, covers majority, in some cases, of your core business processes with the right partnerships on top of our platform, where then we actually are the primary partners to create that interaction layer, that agentic layer, and hence, the value for the customer and premium value back to us and our stakeholders.

So in both cases, this allows us to be part of the value chain. In one, of course, you can go do something, but there's BDC and there's other things. In the others, we're sort of core as the agentic layer, if you will.

And again, what you'll find us do, and we talked about this at our SAP Connect event last fall and we've got some good customer examples for it, is what we can uniquely do then, that the PaaS agentic layers are not able to go do, is we can say, we will come up with out-of-the-box agents, right, that work in the seamless app data agentic experience, that you can extend if you have other tools, you need to add some pre/post-logic, because every customer is unique, but you can get to value far quicker in a lower TCO way than to say, "Hey, I'm going to buy this PaaS platform and then go figure all of that out," if you will.

So that then also becomes, on top of the other moats that we've talked about, a value proposition for customers to say, "Hey, the leading provider of business applications is the one that understands my processes the best, industry horizontal, and my extensions, because we know extensions built on our platform can help create these agentic experience, and not just create, but use out-of-the-box AR agents, AP agents, spend agents, supply chain agents, and extend them to fit my need," is pretty phenomenal.

Because while there's pressure, and I'm sure you guys know this, right, I think this is still very early innings for GenAI, right? And there's a lot of board pressure on CIOs and CEOs and CXOs to show value, like what have you done lately in GenAI? And there's a lot of organizations picking a bunch of different things, cobbling something together and say, "Hey, listen, here's value." But the lifetime TCO for that is pretty massive.

What we're doing is saying, hey, we'll help you unlock agentic layer value out of the box with the right extension so you don't end up maintaining the burden of that till the end of time, if you will.

So just to make sure I understand properly on the Business Data Cloud, there's obviously going to be an environment, you talked about discrete manufacturing SAP can run pretty much end to end. I guess in that scenario, there will be less need for people to want to do Business Data Cloud. But where they're working with other systems of record, with other applications, would that be more where you could bring value and say, well, okay, yes, this is an environment now that's more heterogeneous, you can bring different data sets and then build on top of that? Is that the right way to think about it?

No, I mean, I think -- no. No. So if I give that perception, it's not that. Thanks for clarifying. The value of Business Data Cloud is there if you're an SAP shop in whichever form or fashion. Because what it does is it sort of brings the data, harmonize, with the semantical richness, with the authorizations and the permissions intact, with the SAP Knowledge Graph, that you can then go either build or consume the out-of-the-box agentic layer that we provide.

Now what's different, and that's what BDC allows you to go do, is an industry like discrete manufacturing, most of the data would come from SAP, can come from SAP anyway, so it's sort of there. In oil and gas, the SAP data is there and you can zero copy-share with non-SAP data in a seamless fashion. So BDC allows you to be sort of that data layer and app data agents for all customers. In some cases, where there's a lot more data outside, you can do zero copy-share with some of the, again, the data engineering and data platform tooling that are out there with our partnership to bring it together.

That makes sense.

But it works on both flows, if you will.

Okay. We heard from one of the Business Data Cloud partners yesterday, and they talked about starting in the analytical, the OLAP world, adding OLTP, moving into transactions, talking about an application layer. So my question is how do you keep control of that environment from an SAP point of view? Because I imagine some of those partners might also have ambitions to start to build the applications, the agents on top of that. So how does SAP manage that as a partnership, potential competition with those partners?

Yes. And I think a simple way, and I actually found out that this person wasn't wearing a suit, so I probably should have followed his example and not dressed up here. To me, again, you have to go look at the nature of the partnership, right? So with Databricks, for instance, we have Databricks available within BDC. So that allows us the ability of those data engineering tooling with our data within the framework of the data gravity staying with SAP to build those agentic platform. Now if you have other data, you can sort of zero copy-share and build stuff. And same with Snowflake as well, to be able to go do.

So our partnership sort of allows for leveraging the best of their platform within BDC for the value layer to be built. But if there's a massive amount of data that sits outside -- or if customers have already backed and put a lot of data out there, then you can zero copy the SAP data and BDC to that, but then keeps us into the flow of stuff.

And then I think if you go back a few questions ago, I think the piece that, again, it's -- and I know that's not exactly what you're asking. The piece that -- where we feel like isn't necessarily going to be a reality in customers' environment is, because the data is there, somebody is going to start building applications on top of it in the class of applications we work in.

Can they go build custom applications on some of those data platform partnerships? Yes. But then we also talked about the value of what comes out of the box, because there's deep context graph, there's the model that we're taking, there's Rapid one that we launched at TechEd last week that allows you to do data science but in the flow of the transactions. There's enough value proposition there to say, "Hey, what you build with BDC, with our agentic platform for the landscapes that we work in creates most value." But then, of course, there's always other things out there that those platforms can fill the gaps on.

Makes sense. You've actually given us some numbers around the total contract value of BDC. One of the things I'm interested in is, what's the time to value on those projects? How quickly can you go from signing a contract to getting people up and live and generating revenue with them?

So I think -- again, I don't -- Alexandra can sort of provide some of that offline. But more than half of our BDC customers are in usage today. So the time to value from BDC is very quick. It's not like your typical, think of it as an ERP implementation that says, hey, it's going to take x number of months for the value to show.

And that's primarily because the way we've sort of built the product is, if you have the SAP landscape, the data products that we've built start feeding into BDC and then all the tooling is available for you to be able to sort of create value on top of it. And then we can also help you in your modernization journey for BW as well. So with both of those value propositions, we actually see a very healthy amount of usage and time to usage for BDC.

That's helpful. Maybe just to finish off with. I think one of the other things that's happened is we've kind of reduced software companies to just how they generate code. And if you can generate code more cheaply and more quickly, the value goes away. It seems to me that software companies are much more than the ability to generate code. But could you talk a little bit about how you think about that reduction in cost of producing code, how customers will be willing to pay for that in the future? So how does the structure of the industry change in terms of how you monetize going from seats to consumption? How do you think about all of that?

Yes. I mean I think it's an area we're spending a lot of time thinking through, what evolution that we may or may not need to make. I mean, I think if you look at the models that we have today, we have multiple models as well. We have a sort of significant part of value that's just embedded in the application for Joule-based that says, if you're running an SAP application, you can go interact and get value from it. Then there's a more premium capability that, per user per month, that says, hey, we wanted to give you predictability to say, hey, anything that we ship in this class, you can go try and use without having to feel like you need to add for this feature. The consumption is going to be this, and sort of come up with a complicated math.

Because the reality is I think the ROI, just broadly speaking, in general, for some of these things are still to be really proven at scale. So what we don't want to go is get into this debate to say, to unlock this feature, it's this much consumption, so you need to show this. It's about, hey, here's a set of capabilities that's continuing to get better and you can sort of use and consume as much of it if you like. And then we have consumptive measures as well on top of it as well for scenarios that are deeply, deeply more consumptive, if you will.

I think there are things around sort of now more outcome-based measures that we're looking at. We have usage-based metrics even today, not everything. And even our per user per month is user based -- usage-based as well, even in our core applications where, when you think about Ariba, we're talking about the spend that goes through as opposed to the number of users that are actually using the application. So I think as you would imagine, as the industry is, we're going to continue to see where things are going, and then evolve based on customer feedback on what we need to go do.

But the aim is to demonstrate the value before...

Clearly. I mean I think on the AI, I think we could, we at least believe at SAP, we both have the responsibility sort of running again these mission-critical applications to sort of prove and show the value. But then we have the ability to be able to sort of have the value show and then figure out, sort of based on feedback and discussion, what's the right business model in the fullness of time.

Perfect. Well, there's lots more we could discuss. We're bumping up against time. I want to thank you very much for joining us again.

Thank you for having me.

Thank you.

Thank you.

Great. Thanks, everyone, for joining us for the next session. We're delighted to have SAP back at the conference and particularly Sean Kask, who's the Chief AI Strategy Officer. Sean, I think this is probably third year in a row coming back, but we live in interesting times in the -- on the kind of AI front, there seems to be almost an exponential curve in terms of the development. So I wanted to sort of kick off and get your perspective on -- in the past 12 months, what have been the really big step changes that you're seeing. And obviously, the discussion around the application layer has been obviously the most significant. So first of all, just set the stage of what is kind of exciting, what have been the biggest advances and how we see that sort of evolving moving forward?

Yes. Great. Great to be back. So thank you for having me here. Also, given the volume of requests we have coming in from the investor community right now towards me around AI and our team, hopefully, we can address a lot of questions here on math. So that's a great opportunity. Yes, and it's also a good opportunity to reflect. So I think when I was here in 2024, first time generative AI was still kind of new, right?

I think we had released around 10 generative AI use cases in production. We had announced Joule. We had announced AI Core, generative AI hub for building extensions. Fast forward to last year, we had around 130 generative AI features that we had released. Joule had around 1,400 skills in it. So skill is basically mapping a user intent to something that Joule can do, right? Like show me open purchase orders, these kind of things, right? So we actually have to map that in the system to the APIs and all the data tables and things like this.

And I had spoken about some of the innovations that we were working on then, right, that were exciting, right? And we had 3 buckets. So one was agents. So we're just starting to talk about agents. This year, we've now released agents, right? So we have 30 agents released. We have AI Agent Hub that we've released as well for governing agents, right? We have Joule agent builder to extend agents. So within a span of a year, I mean, it kind of went for where we -- I mean, the whole community was kind of defined what an agent even is and is this feasible to actually have it in like productive agents, which moved very quickly.

The other area that I talked about was Knowledge Graph. So this is this neuro symbolic AI, right, where you codify and you represent knowledge and then you provide that to the AI so that it can do its job basically. So we had built a knowledge graph internally at SAP, which is massive. So just in the ERP system, just so you have a feeling for how complex these things are, it's 452,000 tables and 7.3 million fields in the table and 80,000 analytics views and the Knowledge Graph basically links all of these things together, right?

So if I'm talking about a sales order, I know which field and table to look in and that's linked to which business process and all this kind of stuff. That is now also being used and leveraged by Joule. So that is now productive. You can extend it on the business data cloud so that's live. And the other innovation that I had spoken about back then were let's say, alternative foundation models to large language models.

And the one that we're building on and we're building back then, it was still somewhat experimental. Back then, we called it the SAP Foundation model. And basically, what it was doing is a foundation model built on tabular data to do predictions. So regressions, classifications like numerical calculations, which large language models cannot do. We didn't really know if it worked back then. Now it's productive. So we've released that. We called it RPT-1, so relational pretrain transformer, Rapid 1. So that's productive.

We have one big chemicals company who is -- Christian disclosed this in the earnings call. They have around 180 narrow machine learning models that they have in production. So things like auto filling a sales order, predicting delivery dates, these kind of things. They're going to replace all of those with this one model because you don't need to create a data pipeline and feature engineering and train all these individual machine learning models anymore and they get higher accuracy. So I'd also just one last plug. It won a -- or was honored with a spotlight award at the NeurIPS Conference last fall, actually. It's kind of like the top A-Star data science conference. So showing we can play up there where we need to with the big AI providers. That was the stuff certainly, I think, over the last year, right.

Looking ahead, yes. I mean, look, it accelerates. I mean there's much stuff going on. I'll plug the article that we published a couple of weeks ago. So we always make like predictions like top 5 themes. You look that up, it's AI in 2026, top 5 themes. But yes, again, certainly, these foundation models that are not large language models. So things like world models and robotics. So we have some partnerships that we've announced there. We work with NVIDIA on these kind of things, the tabular foundation model. As well as agentic governance is going to become key.

So you're releasing all of these agents into the organization now. They need to borrow from HR, they need a hire to retire life cycle. You need to discover them, onboard them, monitor them, govern them, offboard them, right, have observability, give them access rights, all this kind of stuff. So we've made some announcements around that as well. Sovereign AI, I think, is an increasingly important topic, and we could see the discussion with Anthropic and the U.S. Department of War, I guess they're called now, right? So these kind of things. So yes, I mean, there's just a lot going on. It never slows down. I kind of hope it slows down a little bit sometimes because I am just exhausted. So I'm trying to keep up with it all as you are as well, but it's also really, really exciting and full of opportunities.

So hitting on that point in terms of the pace of it, there has been an accelerated debate around like how Agentic AI might disrupt application software. So what are your thoughts on the current state of play? Like who, in your view, would be like emerging as the relative winners or losers?

Yes. So the elephant in the room, right? So is SaaS dead, and that's more or less the question, right. So a big debate right now. So no, I mean, look, I think there's 2 things. And from SAP's perspective, I mean, one is we acknowledge that this is a disruptive technology, and we're leaning into that, and we're going on the offense in certain areas, right? So AI will certainly disrupt the user interface, right, how you interact with computers.

So you can have intent-driven ERP systems where you ask the system to do something, right? You're giving the intent, the prompt in and then it's going to go and execute and do something, right? So you may not need to click in UI anymore, right? Also, user interface will increasingly contain elements that are generated. So we announced this, we call this Gen UI, generative UI. And we already have elements of that in Joule actually, right?

So if you're pulling up a chart or something like that, it's actually like generated on the fly by the application. And in the future, we also may have, instead of humans talking to SaaS, you may have agents talking to SaaS as well, right, where the human is merely notified. So the UI will be disrupted to some extent. I do think UIs are still somewhat sticky, though.

People -- my wife's friend works in a trading company in Zurich, and she was complaining to me because they've moved off their ECC system that she's been working in 10 years, and she just knew where to click and just loved it, right? I mean these things will always coexist, right? So you always have a UI. And again, the logic of that can still be accessed by agents and generative UI and all those kind of things. So we're leaning into that. We have Joule already rolled out, right, so we have a solid base for that.

Secondly, this AI will disrupt how software is built and maintained, right, certainly. We see how quickly vibe coding and AI pair programming is progressing, right? That opens up a few opportunities. One is -- I mean -- and I think that's driving some of the fears of disruption as well, right? Now for us, we also use vibe coding, right? We've rolled that out to all the developers in the company, and we've seen significant magnificent improvements in productivity to using that.

Now for us to vibe code something, it goes through our 388 product standards and testing before it goes into production, right, okay? So it's like enterprise-grade software. But we can vibe code, and we can also offer that to our customers to quickly build extensions, right? So we have SAP build, for example. We've released a Joule Agent Builder. As a user, you can go in and actually vibe code little extensions to SAP systems, which in our point of view, just increases the value of our systems.

And the third area, I think, is that AI can also disrupt the commercial model somewhat, right. There's this big debate like around what's going to happen with seat-based subscriptions in the future and licensing. So a couple of points there from SAP's perspective. One is the current commercial model that we have for artificial intelligence is actually based on consumption. So we talked about that a couple of years ago.

There was a lot of heavy lifting internally. We built out a capability to actually meter every single one of those agents or use cases that is being run in the system. So if someone goes in and processes an invoice using AI or a delivery note in the shipping yard or something like this, that's metered and charged against these AI units, right? So -- and we basically have a commit-to-consume type model for that.

So you subscribe to a certain number of AI units, and then that's pooled and then you consume that, right? And every AI unit is tied to a business outcome. Somewhere around like messages and things like this, but like where we can do it, we try and tie that to like a business outcome. So we actually have that already there, right? That's actually already in place for us. And the other thing, too, is that I think it's less than 50%. I'm not sure exactly what we disclosed, 50% or 40% of our it was seat-based. So the rest of that is actually based on different business metrics like spend under control, for example, business documents in the system, these kind of things. So there, we're going on offense.

And then I also think there's defensive moats that we have, right? And one is certainly the data that we have. And that's a bit of a truism, right? Like if you own the data, but it's not just the data, it's owning in the data models itself and the contextual meaning around all that data, right? So -- and again, we're exploiting that. So we've built the knowledge graph. We have business data cloud now, right, where we can expose the semantically rich data in SAP and non-SAP systems to be consumed by agents, for example, right?

So we have that data. Customers trust us with it. And that's the other defensive moat that we have is the customer trust, right? But -- I mean, probably the biggest argument I've heard against vibe coding is it's Friday night at 8:00 and you have to close the books and your general ledger doesn't add up. Who do you call, right? Is it your SaaS provider to fix that? Or is it the guy who vibe coded something and tries to figure out what's happening there. So there's a lot of customer trust that we have, obviously, with our systems. So I think -- I mean, I'm biased, but obviously, I think we're in a good position to emerge as a winner out of this whole thing.

So there's a sort of view that horizontal software is particularly more exposed to some of this risk than vertical. But even though you could argue we've had instances of vertical software use cases also getting challenged. You guys are interesting because while you sell horizontal software, you also build for specific industries. Obviously, manufacturing is a key part of your end market.

I mean, when you look at sort of, say, analogies of what happened at the time of the cloud and what's happening with AI, how should we think of kind of your evolution or your ability to kind of navigate this cycle. What are the kind of -- you talked about the data moat that you have, there's process knowledge. And when you think about the disruption of this application layer that's happening, the modules as we know them, the line of business applications are clearly changing.

And so ultimately, from your standpoint, what are the key things we need to look for, for SAP to kind of navigate, is Joule ultimately going to be all end all, but ultimately, for a lot of customers, they may not want to use Joule. They'll say, "Well, I'll use another agent," right? So help us kind of navigate that kind of blurring lines, right, between that sort of application layer in an AI world?

Yes. And I'm hesitant to put us in the box of horizontal player totally, because if you look at what we do in finance, for example, it's quite deep, right? If you want to know how to process an invoice in Brazil and be compliant with Notifiscala and things like that, I mean, we're there. So we are quite deep. And of course, yes, we have a lot of industry solutions as well, like customer activity repository and retail or these kind of solutions. And it's interesting, too.

I mean there's some apps you would think might be disrupted, but they have like messaging apps and stuff. I have a policy of not talking about competitors and stuff, but where they have like network effects and they seem to be pretty resilient to disruption. But I think if I look back at SAP's history, right, was it, 2016, I think we had Clayton Christensen on stage at our big Sapphire event, right? So innovator's dilemma before he passed away.

We have a strong history of actually like disrupting our own business model, right? So going back to like moving from client server architecture, in-memory database, right, moving to cloud and SaaS, like that was basically disrupting a nice profitable business model where we sold licenses and then collected 20%, whatever maintenance on it, the customer had to do all the installations. And I see that same spirit here, to be honest, right? So we're leaning into the opportunity that Generative AI brings for us.

So just wanted to see from your lens, how do you see the landscape evolving overall in this agentic AI world, like in between the incumbents, the new age players, which would be AI native players as well as these LLM providers and also like even companies in-sourcing to an extent.

Yes. And I'm thinking of like Porter's 5 forces model right here, right? So straight up substitutes and rivalry and whatnot. No, look, I mean, so we see the models commoditizing, right? So -- and that's been a trend for a while now where the price per token is just falling and they're pretty much collapsing, right, converging in terms of capabilities, and it's no surprise to them that they -- at least the model providers try to move up the stack a little bit.

And so we see them making their first like PaaS offerings, right? So the OpenAI Frontier, for example, which Gartner, by the way, issued a note 2 weeks ago just telling their customers to be careful of it because they probably can't scale it, but just repeating what Gartner said, but Claude Cowork, these kind of things. So of course, they start to push up the stack a little bit.

From our perspective, the hyperscalers and the PaaS vendors have tried for years to creep up the application stack. And I mean, pretty unsuccessfully, to be honest. I think there's a lot of like -- I would not underestimate the knowledge that's needed to build those applications. So from our perspective, we treat the SaaS -- sorry, the model providers, right, and the hyperscalers as we always have, right?

We treat them as partners. We actively participate in the A2A protocol, for example, MCP, we're like founding members of MCP protocol. We collaborate with them. Of course, they're going to try and creep up the stack a little bit. We do see, obviously, there's around 600 start-ups, I think, that we look at right now in the AI native space springing up. I think they have probably the challenges that a lot of start-ups have, right? It's like just having this like enterprise readiness to be able to compete with us.

So just a follow-up on that. How do you think like incumbents will look like in terms of strategy. Will they look to buy these AI native players? Or will they try to build all these AI capabilities in-house. Where will the balance be more like heavy?

Yes. I think Salesforce and ServiceNow went on a bit of an acquisition spree the past 2 years from AI companies. I don't think the market necessarily rewarded them for that, to be honest. We have not -- I mean, we made an acquisition of SmartRecruiters last year, right, in the HR space that was just to boost our recruitment module and SuccessFactors. But I mean -- for us, it's tricky right now because the stuff is pretty new to everybody. So we haven't seen a start-up where we say like they have some amazing capabilities that aren't available somewhere else, right? We're all kind of cooking with water and the valuations are insane at the moment on these start-ups. So I mean, from our perspective, we monitor them, right? So we work closely with them. We still partner very broadly with a lot of start-ups, but we're just kind of like watching the market at the moment.

Got it. So I wanted to come back on this sort of comment you made around the kind of data moat. I mean there's a system of record, right, which is sort of the single source of truth, pretty kind of hard to displace. But increasingly, what we've seen from the likes of Claude Cowork and OpenAI's Frontier that they want to kind of sit as a sort of abstraction layer on top of that system of record.

In the future, is that kind of a viable option of a direction that we will go into that they just kind of sit on top, but eventually -- because data access is also another key kind of important characteristic here and license agreements, et cetera, play a key role in this. But is there a scenario where the system of record becomes a kind of dumb pipe, sorry to use such a strong word, but -- or will -- because you've got the business logic, the metadata, all that as well. So I'm just curious kind of is this the next big battleground.

Everybody wants to be that abstraction layer, right, and orchestrator sitting on top of other applications, right? I think in reality, the future is going to be the big players are going to have their own agent, their own Copilot, right, that's doing the orchestration in their systems. And that's going to have to collaborate with other orchestrators and other agents, right?

So Joule, for example, again, can access data and context and authorizations and metering and logging and all this kind of stuff in the SAP systems that a third-party agent simply cannot access, to be blunt. And I'm not sure I would want like -- I'm not sure our customers want us to allow this unfettered access to the SAP systems from some third-party agents because I mean that could be -- this is going to be like a security and auditing nightmare, right?

So again, I see a future where you'll have these assistants, copilots, right, that are good in their domains and through protocols like A2A, they will then collaborate together. And I don't see anyone really being able to take over like as the uber orchestrator. I will say as well, we're quite open. So we were one of the -- I think we were the first company. We have a bidirectional integration with Microsoft Copilot that is now GA and Joule, right? So I can be in Copilot and say, help me book my trip, and that will go over to Joule and Concur and then make an entry in my Outlook calendar and vice versa, right?

So we are open in that sense, but I don't see anyone realistically taking over. It sounds funny because -- I mean, like Anthropic and OpenAI, they almost have like dumb intelligence, right? So you can hire your smartest friend to go work in an SAP system, and they have no idea what to do, right? So all that knowledge of how to run a process is like codified in our systems. That's what makes it valuable.

Right. And just sort of following up on that, obviously, there's the BDC, right, which you sort of launched about a year ago. Help us kind of understand how that sort of fits into that sort of broader data model.

Yes, for sure. Yes. So Business Data Cloud, just to be clear, it's a SaaS offering. And it's a little bit different than what's on the market. It is not a data lake, okay? You're not extracting -- I mean, you can, right? So it has components of that. But you're not extracting data into it and then building up a data model and like PaaS, okay, like you would do with other solutions. It sits actually on top of or it integrates with partner solutions like Snowflake and Databricks and BigQuery, right? Because customers have invested time in Snowflake, for example, right? SAP and non-SAP data, external data, right, consolidating that, okay?

And what Business Data Cloud, it sits on top and it operates on the principle of what are called data products. So we release around -- I think we have a target around 500 data products from SAP systems. So how do you consistently define a sales order or a purchase order, right, which can have different data models depending if that's coming from an Ariba system or an S/4 system or whatever. And we expose these as data products and then customers can also then expose data from, again, all these other data lakes that they built up as data products. And that's contextually rich, rich data.

And so now you can put an agent on top of that. So if I'm doing, I don't know, planning, for example, for I don't know, headcount planning, things like this, like hiring planning, I might need to know what are the sales forecasts, right? Like how many people do I need to hire? And that data might be sitting in a Workday system and an S/4 system. And now we can put Joule on top natively with Business Data Cloud with all of these data products. So it just brings together SAP and non-SAP data, and we monetize it, obviously, in a way that can be easily consumed by agents. So it's an important part of it.

So just you talked about Joule and Joule has like seen number of customers almost growing 9x over 2025. So when we look at Joule, can you talk about any of the interesting new use cases that you see emerging? And how is like -- and anything on the productivity, any tangible productivity gains and how you're basically like looking at it in terms of measuring?

Yes. I think we discussed that last year that the adoption curve is pretty unremarkable in the sense it's just a typical adoption curve. So you release a product and then customers look at it and then they buy it and then they go live with it. And so we see this kind of like a hockey stick basically, right? So 9x number of customers adopting Joule.

No, I think the most exciting thing we're releasing now in Joule are the agents. And let me just digress for a second here. So we don't want to get into the number counting game for agents. So we set the bar very, very high for what we call an agent because you could call any like RAG use case, like information search use case, an agent, for example, and I've experienced that in SharePoint, right, where it says I'm your SharePoint agent and the only thing it does was summarize the SharePoint page. And I was like, this is not helpful, right?

For us, agents need to have agency, right? They need to plan and iteratively work through several steps and tools for us to call that an agent, okay? And agents work very well right now, like production grade in these like narrowly constrained type use cases. So some of the agents we've released, for example, is like accruals accounting and finance, right? So when do you post your costs and revenues, right?

And when you do accruals accounting, that might depend on a PDF document with a policy. It might depend on some e-mail chains, some past decisions, right? And it's hard to automate, but an agent can actually iteratively reason through that, these kind of processes, right? And then it has like an explanation of how it reached that decision for the accountant to look in and say, okay, yes, that's right.

And so we think that would save for a medium-sized company that might spend, I don't know, end of a quarter, 12 hours doing accruals posting, they can get that done in 2 hours, for example. Production planning agent, same thing. So you're planning your production, what happens is delivery is late, right? So you don't have the right parts or you get a big order that comes in that's prioritized and then you have to go back and it's like it's optimization problem, right?

We have an agent that will iteratively reason through that and optimize your production planning. And that -- and so you can do this production planning much more frequently now, right? So you can do that like a couple of times a day if you need to, and companies can simply sell more, right? They can deliver more. So release these kind of agents. But definitely, the star of last year was Joule for Consulting. So this won award at the World Economic Forum last month, it's called the MINDS Award for like transformative industry use cases. So we did that together with KPMG as a partner/customer in that.

And look, so one of our top priorities this year is AI-assisted cloud migration and AI-assisted cloud transformation, getting customers faster with less effort to the cloud and obviously, on the standardized SAP landscapes. And so what Joule for Consultants does, it's grounded in all of the SAP help documentation, internal knowledge-based articles. Customers can extend that with their own like what are called business blueprints in SAP world. And it leverages our proprietary large language model that we train on SAP code, so it's called ABAP LLM. And so that can like explain legacy code, it can write unit tests, it can do a bunch of stuff, right?

Long story short, so Siemens is a reference customer there. They said that saves around 10 hours a week per consultant. So if a consultant is working 40-hour a week, that's like a 25% productivity boost. I mean I started in consulting. I think my daily rate was probably 2,000 or 2,500 a day, right, doing SAP transformations. I mean that's like a massive boost in terms of productivity.

Great. So let's open it up to the audience. I'm sure there are questions. So whoever wants to start.

Helena from Millenium. To your last point about Joule for Consultants. Do you see that accelerating or lengthening sales cycles for the SAP core business? And I'm asking because our clients kind of seeing that, oh, I can get the 25% faster, so let's do it or hang on, let's wait, maybe there is a new iteration coming out that will make it even less costly, even faster or something like that.

Yes. So again, the question is the customers wait for AI to basically automate the entire transformation, right, to the cloud. It's not a completely unreasonable question, to be honest, I think customers weigh that with the benefits that they get when they get to the cloud and are able to use more AI. I mean, look -- and frankly, to be honest, what Joule for Consulting does now is kind of like just the tip of what's possible, right? Because we also announced our integrated tool chain, right?

So this is using like Signavio, right, with machine learning to map out the process automatically. We're going to release like data duplication and data harmonization now based on AI, which can be like a huge test automation, right? So there's a lot more coming, right, that will help with their cloud migrations. But I don't -- I certainly don't see any customers who are waiting, right, under the impression that, oh, if I wait a year, then migration costs will be 50% less because AI is going to completely automate it. I think it's more the other way. It already offers quite a good benefit for them to get to the cloud, and they have incentives to get to the cloud because there's more AI that's feasible there.

What are your thoughts on quick commerce specifically on...

So there's been a few protocol -- I think there's 2 or 3 protocols, right? So Google just released the UCP, right? There's agentic commerce protocol. So basically allows agents to shop for you, right, and do financial transactions. No. So we are supporting that in the CX portfolio of SAP and SAP Commerce Cloud. And SAP Commerce Cloud, we're actually opening up MCP servers as well that allow -- because that creates more value for our customers, right? So then it allows agents to more easily find products on the websites. And then, yes, we certainly aim to support that protocol. I think it's a great opportunity.

$100 billion of market cap that's built on top of SAP. And if you do believe that AI does make the marginal cost of R&D in development closer to 0 and historically, you haven't been able to target these sort of upsells and modules that have been done by third parties. Do you see that as a sort of a real opportunity now in unlock? Or do you think you're still far from that and a lot of things have to change? An example of that may be that, you guys did BlackLine last year. So is that something now that's easier to build similar quality in-house because of AI or can get there in the next few years? Or you think it's still very far from the...

I mean everything indicates that AI reduces development by like 20% to 50%, let's say, like productivity to build stuff. And to flip it around, there's also that threat like will there be new entrants challenging SAP? Or will, I didn't completely finish your question before, will customers be more incentivized to build things in-house because it's so easy, right, using vibe coding.

So -- but I think, again, the assets that -- I mean, again, we also apply that, right, to your point. So now can we now enter more markets, right? And I think, again, building on the capabilities that we have and again, all this like the product standards and the enterprise readiness and the knowledge and all that kind of stuff is like a really important base. You need that to enter these markets. And we are -- we are taking some measures to actually do that. So this has been disclosed as well.

We're doing more FTE, it's for deployed engineering. So this is an initiative from the product and engineering Board area, popularized by -- so OpenAI is doing it now and Palantir. But basically, you send engineers to customers, right? In 90-day sprints, you figure out real problems they have and then you try and like build something. And then for us, we want to build repeatable solutions, right?

And that's -- it's not a new thing, but with vibe coding, frankly, and AI, it's a lot more feasible to do that in a 90-day sprint, right? So we can enter all of these markets, and we're going to offer a lot more industry solutions moving ahead to enter those kind of areas where we can certainly build stuff, right? Because again, customers and others can build stuff more easy, but so can we, plus we have these like complementary assets where we have all the customer relationships and knowledge and all this kind of stuff in place. So it has a potential to grow our portfolio.

That's a best of breed? Maybe we carry on. So Dominik talked about this incremental $1 billion of revenue opportunity for SAP from a number of sectors or segments. And obviously, AI offerings, particularly the cross-sell, upsell of these new -- would have historically been line of business solutions is now some of these kind of AI solutions. Maybe talk us through some of the traction you're seeing around this? And to what extent has this really taken off or what we're seeing from some of the other LLMs now that the customers are opting to go down the kind of build route there versus, say, buying something from you off the shelf?

Yes. So I think we've disclosed, I mean, the tremendous attach rates, let's say, right? So 2/3 of the order entry volume was -- had AI attached, like 90% of the top biggest deals had AI and BDC in it, right? So we see -- it's a huge part of the sales cycle, right? Honestly, it's incentive. Now we support customers to also build, right? So that's always been one of the attractive things of SAP systems is that customers like to get in there and for better or worse do their customizations and apparently, they need to run sales orders in the nonstandard way or whatever. But they go in and actually build those.

And so we actually offer tools now, right, like SAP Build, which has -- we have client integration, which is one of these vibe coding apps that they can actually use to build and extend around SAP systems. So of course, we give them that option, and that's always been a big part of our portfolio. But it still needs to be based, I think, on -- you need the standard core.

Got it. Maybe just moving on -- sorry, go ahead, please.

Some of the startups. When you're offering a similar product, what does it -- pricing conversations...

Good point. Two things. So one, for the embedded AI that we offer, right, not the vibe coding, but the embedded AI, right, the -- and we sell these by AI units. These are not tokens to be clear, right? Tokens are part of that margin, but the business value is also part of that margin. So when we release -- again, everyone is always pitching use case ideas to me, and I've seen like -- I feel like I've seen all of them now, right? We put those in our products. And we can build the embedded stuff, right, once, which includes testing it, running an ethics review, benchmarking, building the UI integration, all those kind of stuff, right? We'll build that once, but then we can release that to 5,000 customers. And that's the margin that we take on top.

So the token cost, the model cost is just part of that. And also, again, we have this strategy where we partnered super broadly. We have access to all the Frontier models. We have partnerships with Mistral and Cohere and all these companies, right? We can simply move to the -- we have like some arbitrage, right? We can move to the cheapest, best performing model as we want under the hood, okay?

So that -- in that space, we have the margin, right? And there's a benefit for the customer there. When it comes to pure vibe coding, the offerings that we have are let me put it, it's a little more broader than just the vibe coding part of it, right? So if you're looking at SAP Build, right, there's the vibe coding component of that, which they may undercut us or try to like push the price down. But there's all the other components if you're using SAP Build, right? So like you need to ramp up a database, you need to build a CI/CD pipeline and all those kind of components around it that I think is where we make our margin and can like tell above and beyond what they offer.

So talk about how you guys are protecting sort of cross-sell opportunity assuming that customers vibe code themselves on the edge. So touch on that...

Yes, again. So SAP systems are built to be customized and extended, right? We support that. I think where we have a strength, again, with the broad portfolio that we have and things like Business Data Cloud is a lot of like value-added use cases are not just point solutions anymore. It's solutions that -- or extensions, right, that touch several like business processes and business applications.

Again, so like the planning scenario, for example, that I mentioned before, right? You might need data from logistics system and finance and HR, right? And so we can bring together that data and Business Data Cloud, right? We monetize that. You can do a vibe coding and then extend an agent on these kind of solutions. So I think that just strengthens the whole suite story. And then in terms of customers building on the edge, I mean, that's something we historically support anyway. That's one of the attractive things about SAP systems.

I guess one of the -- part of what's driving all the consternation to SaaS versus AI. It seems to be that a lot of the innovation seems to be happening in the labs start up level. So even things like, for example -- did exist by Microsoft already Copilot and I'm the one who use Copilot.

And I wonder from your perspective, given your seat, do you feel like you have the right team in place to be able to remain relevant and be at the frontier of all of these developers because at the end of the day, switching costs are quite low. And so if someone can use a Claude code versus using SAP's agent builder. Now obviously, there are client's data sets that are more difficult for them, but certainly not a good idea to force your customers to use a product. So I'm just trying to think about why the cadence of innovation has been so slow with the existing AI software, all the software companies seem to be very much behind the curve from a new product and all the headline...

Did anybody install Open Claude in your computer? Oh man, you're brave. Security nightmare. It's impressive though, what's possible with it. Yes. I mean I disagree that the switching costs are that low, to be honest. Again, I do think there are benefits like if you're building SAP extensions, like a lot of our big customers are SAP shops, and they're used to working in the system and they work there because of, again, the enterprise readiness that we have around it, like auditing and all kind of stuff that's needed.

I challenge a little bit about the innovation piece, again, like this Rapid 1 model that we released, this tabular model. I mean that is like, again, spotlight paper at the top AI conference in the world, like huge value. This company, I think what they call it Frontier, I think, just came out of stealth mode, right, building tabular AI models with a valuation of over $1 billion. And we have the same thing basically. I mean, to be honest, even better, I would say, right?

So I think, again, I would challenge that a little bit. I do think there is some innovation coming out of the big companies. But again, we're mostly focused on making things enterprise-ready and productizing. And so -- and also, again, a big part of our strategy is partnering very broadly with companies. So look, we have partnerships with OpenAI, right? We made this announcement. We're hosting OpenAI models on Sovereign Cloud in Germany, for example, right? We have -- we're investors in Anthropic. We just participated in the last -- we're early investors. We just participated in the last round as well, right? So we do stay close to them and let them experiment, okay, and fail and be successful in certain areas. And then when they're successful there, that's where we partner.

In a world where I guess, ERP transitions have been short term, maybe months rather than years, to your point, actually could be managed by AI completely. But what's your outlook for the systems integrator?

Yes. I was always stuck with the -- I've seen every $1 sold in SAP software generates like $6 to $10 in the ecosystem for migrations, maintenance, all this kind of stuff. No, I mean, honestly, to their credit, they are also actively disrupting themselves, right? So they embrace, Joule for Consultants is used by all the big systems integrators, right? They dive into it, the customers ask for it. Their perspective is they can do more projects in a shorter amount of time, right? There's no shortage of projects and customers that need to be migrated. They have trouble finding talent, right? They are pivoting their business models very quickly. To their credit, I mean, they've really been also actively embracing this and disrupting themselves, and I think seeing a lot of growth and success as a result.

Great. Well, I think we're on time. So thank you very much, Sean, for the great insights, and thanks, everyone, for joining.

Thank you. Looked forward to the session. Thank you.

Good morning, and thank you very much for joining us today for our Q4 and full year results press conference. A warm welcome to everyone here in the room, and of course, a warm welcome to everyone joining us virtually. As always, Christian, our CEO; and Dominik, our CFO, will share some brief remarks, and we will then move into the Q&A session. Everyone joining online, please feel free to submit questions at any time. Maybe one disclaimer, as always, unless stated otherwise, all numbers on these calls are non-IFRS and growth rates and percentage point changes are non-IFRS year-on-year at constant currencies. And with this, let's not waste any time. Over to you, Christian.

Yes. Thank you, Monika, and welcome, everyone, here at our headquarter in Walldorf and of course, also to those who are joining us virtually from all over the world. I have actually, from my remarks, I have 2 rather big points. First, 2025, you have seen the numbers. Let me share also some more background on these numbers and then, of course, also the outlook for 2026 and the years to come. And of course, there, I will also double down on the topic of AI. Now talking about 2025.

I mean, first, when you look at the set of numbers, I would say I'm very happy with how SAP once again delivered a very successful year. You can look at cloud and software. We achieved our outlook. And please also remember, in the half year 1, we had a rough start. There were some geopolitical tensions. We -- especially in the public sector, we actually had our challenges to actually do deals. And still, we achieved our outlook for the year. We overachieved and beat our outlook for operating profit and cash flow. It's not only about cost discipline. It's also the way how we transform SAP, how we make the internal processes more efficient, how we're also now applying AI in all parts of the company. I will come later to that when it comes to 2026.

Also in 2025 in Q4, we actually had our best bookings result of the year. So I know there's still some discussions out there on CCB. I will touch base on that in a moment. But actually, Q4 was our best quarter with regard to bookings. We had lower churn than expected and also the discounts we have given actually were pretty stable. So actually, net-net, a very good Q4.

Now again, we started our transformation 5 years back. And we were sitting here, I was sitting here, made a pretty bold commitment about the 2025 ambition we have as a company. There were many doubts out there, but we delivered. The company delivered. I'm super thankful to our 100,000 colleagues worldwide to the customers for the trust because with RISE and GROW, we made a big bet, not only on lifting and shifting our customers to the cloud, but really helping them to transform. And what came out of that is one of the biggest success stories and definitely the biggest transformation in SAP's history.

Now when you deep dive a bit on GROW, I mean, SAP, I know, is known for running large enterprises in the world. And yes, we are very proud about that. But what we also managed over the last years is that actually several thousand net new customers joined from the mid-market. Then the mid-market is actually by far now the fastest-growing market within our customer base. We are expanding our ecosystem because a lot of that will be also covered by our partners. And in 2025, and that is also -- shows the success of our cloud transformation. Actually, our public cloud business was growing 5x faster than our private cloud business.

And also look at the resilience, what actually SAP in the meantime gained. We have a large recurring revenue share. We actually tripled our cloud revenue over the last year. So definitely, I would say, a huge success story. But we are living in a fast-moving industry. I would say this is probably the fastest-moving industry in the world. And so we can't rest.

Now what we also did when you look at this half moon is actually we put a lot of clarity into our product strategy. I mean we said, hey, all lines of businesses have to come together on one platform. The PDP is now in the meantime the platform for integration and extensibility. We put a BTM business transformation portfolio together, again, helping our customers to do the process transformation to be world-class in enterprise architecture and also just help them to transform on the business side. We launched a lot of new innovations around sustainability, the business network and all these businesses contributing to the overall growth of SAP.

Very important, obviously, is also what we did in the last years around AI and the Business Data Cloud. The Business Data Cloud now produced, in the meantime, over EUR 2 billion of order entry since its launch in January, shows the success, but even more important, shows the strategic relevance because when we talk about AI, we talk a lot about data quality. And for the customers, it's super important to have this semantic layer of bringing SAP and non-SAP data together, and that is also then resulting in the huge success of BDC within the first 12 months.

But of course, we are not stopping here. I mean you have seen our total cloud backlog increased by 30% to EUR 77 billion. I mean, what a number. And that also shows why we are so confident on our guidance to accelerate total revenue growth in the years to come. I mean, with this backlog and the contract duration is around about 4 years. So you can see there is already a lot in the books, which will help us to say with confidence that SAP will be a growth company. The cloud business, when you compare this revenue growth numbers here of 26% in 2025, these are on an average, 10 percentage points faster than our peers, than our competitors, just shows how also SAP is gaining market share. So net-net, also operating profit, free cash flow, Dominik will talk about that. So no need for me to dig deeper. But also there, we beat our outlook, and that speaks for itself.

On -- in Q4, we closed a lot of business, best bookings quarter. Now I can tell you -- share with you a story about all of them. I want to pick 2. And I picked those 2 just to show the relevance of SAP AI in the world going forward. H&M, we all know them, a great retailer. And they came to us and said, "Hey, our business will change a lot as a retailer." And then we prototyped together over the complete year, and we closed the deal in Q4. They wanted to see, "Okay, we are happy with your commerce platform. But in the future, our consumers expect a more personalized shopping experience." So we custom coded for them a prototype on how shopping experience will change. We brought this back into the standard. And they said, "Wow, this is exactly what we need to really address our consumer needs, the consumer trends right in the store online."

Second, we talked about certain things about returns claims management, people ordering stuff, sending it back. How can we make this more efficient? How we can improve the consumer experience? Can we actually propose to the consumer a different good if they are not happy with the one thing? What if a certain good is not available in one store? Can an AI agent help to find the right store to deliver next day or even in the same evening? So -- and we showed them this was the SAP transactional application in the old world. And this is what you get with AI in the new world. And it was tremendous what they have found out on to really personalize the consumer experience to make the supply chain more dynamic, more agile with regard to also delivering the stuff faster to the consumers.

And then finally, of course, they saw all the agents working together also into the back office into finance. And that is what made this deal happen. It was not only the cloud move and get rid of the legacy. It was really the AI embedded in the different parts of our apps, which made this happen.

Fresenius, we did a press release already, super happy about that. We got a lot of feedback, especially in Germany, hey, you were great in patient management, but why do you not deliver the next generation. Together with Fresenius and Avelios, we are now coding on our platform a new patient management solution. And we started to do that. Avelios is our main partner here, and it will revolutionize how much more efficient we can make the doctors and the nurses to spend more time with the people in the hospital, making them more efficient, making more efficient decisions and just also make the whole operations in a hospital way more efficient than it is today. And again, AI agents taking a lot of manual work over what the nurses and the doctors had to do in the past. And we showed this to many other health care customers and they said, wow, this is it. We definitely want to join SAP in delivering the next-generation patient management.

Now talking about the future of AI, talking about the future of SAP. And I know there is a general concern out there in the market about, oh, how will software sustain in the world of AI? Cannot everyone code software? I would say clearly no. Because what we are already seeing with many customers is, of course, they are doing certain -- building certain customer agents for cash flow collection, et cetera, with those LLM providers. But what you always see as a roadblock, and this is now what customers see more and more, and that's why it also explains why we sold 2/3 of our deals with AI. They, first of all, see, oh, an LLM can read when I build a cash flow agent, can read a support ticket. It could be that because of the support of an issue of the customer, customer is not paying. You can read mails, okay.

But what about the P&L data? What about certain sales negotiations, deals in the pipeline? What about certain payment information, which are also necessary for the agent to understand why is this customer not paying? So it always goes together. The LLMs are super good in the unstructured data, but you need the business data and which company has petabytes of data, which we are using to which we are using to fine-tune our AI foundation, this is SAP, and we are using the world's best LLMs for the different use cases, bring this together, have a so-called knowledge graph to correlate the unstructured data with the structured data. And of course, BDC helps to bring the semantical data together for the structured data in the company. And that is the winning formula.

And then the second piece is when you want to change a retailer like H&M, you cannot just go there and said, the IT embed a certain agent in my operations. You have to fundamentally rethink like we do in SAP how will I run a certain industry going forward? How will cash collection work? How will recruiting work? How will workforce management work?

So our product managers are just sitting there using the rich information, knowledge what we have about industries and business processes to really redefine how these agents have to work. An inventory agent as a matter of fact, you can do an inventory. But if the inventory agent has no clue what is happening on the demand side, the inventory agent is not so intelligent, I can tell you. And then, of course, there are a lot of things that, what kind of information can I actually feed into an agent. There are certain security authorization requirements, which all sits in our beloved apps.

Now super important for us is business data, business process, security and trust and, of course, completely rethink how we run those companies, our customers going forward. And so when I think about the future of AI and SAP, I'm super happy that I have our ERP. I'm super happy that I have our apps because without those apps, I wouldn't have the data. And without the data, I wouldn't have an AI. So I know there is a lot of talk about, oh, what can the LLMs take over. The LLMs can take over coding of software, for sure. I mean, because this is unstructured information code. They understand the patterns, how our developers code in the past. But everything related to business data is actually something what SAP can offer, which is pretty unique to us.

So when you think about how will SAP grow its business going forward? And I find it pretty remarkable that we -- on an already heavily growing business, we said we're going to further accelerate our total revenue. Five pillars where we have a clear right to win. We cannot win everywhere, but we have 5 pillars, which are very important for our customers. When you think about SAP and UX in the past, this was not a big success story. I mean we know that Joule cannot take over today every skill of an end user, but we are getting there. And we will not only take over manual work. We will take over analytical requests. We will train Joule also with correlations to understand, not only do analytical reporting, but also give smart recommendations, how to source the best for this good, what I'm looking for, how to actually do inventory planning the best, looking into what is happening on the demand side, what is happening on the market side.

So Joule will not only be connected to an LLM like GPT, Joule will be connected to our AI foundation to get the 2 worlds together. And what it will does is when you think about how often did I sit in front of my desktop or mobile typing into data into SAPs, this will completely change the design, the user experience, the simplicity. And at the end, the productivity of every end user will change.

Second, I mean, this is logic. We are running business processes today, transactions, workflows, complex. We are now embedding not further features into these apps. We are embedding agents. So the agents will take over the features. And the agent will talk to each other. So we are actually infusing across the most mission-critical business process in the world, our agents, and we will train them, again, to also contextualize information because no agent can work in isolation. Otherwise, you are not running businesses.

Very important in that space, in the second space, AI assistant. Not every AI assistant will look the same, for example, the cash flow example. So extensibility is key. So you're getting access in our agent builder to, first of all, understand the process better. And then you can also enhance those agents based on individual needs of a treasurer, of a person in supply chain training and so on. And we have both. We have the tool set for the developers, and we have the low-code tool set for the business users.

Third, industry-specific capabilities already today, super important. I mentioned Fresenius. We had another large deal in Q4 where we could show the customer, oh, you're doing last-mile delivery with SAP. Now we're going to show you how your trucks arrive faster at your stores with AI in the future, how you can improve load optimization of your trucks with AI. So these things are super, super important because here is the value of a customer. This is how customers can differentiate in their industry. These are the main, main capabilities, for example, trade promotion for a retailer, personalized shopping experience, supply chain resiliency in manufacturing, asset management for the navies of the world. These are the things which SAP knows how we run it in the past. And now we were reimagining those capabilities with AI.

Fourth, business data cloud. I mean, again, the biggest road blocker for business AI is today, data, data harmonization, data silos. This is actually what constrains our customers the most. And this is what you have seen in the numbers. BDC is a big success because SAP said, "Hey, we are not a closed shop anymore and really bringing our data together with non-SAP data, BDC and it's only in BDC, we are going to allow you to harmonize SAP data with whatever other business data you have in your company sitting in non-SAP apps.

And then fifth, obviously, this is what is close to our heart for many SAP customers said, Christian, I just do the RISE journey. But guess what? We are paying $1 to SAP. I mean, not exactly $1, but we are paying them $10 more to DSI. I said that is not good. So what we are doing is, I mean, why can AI not take over certain parts of the ERP migration. Think about data migration, think about configuration of the system, think about test automation. So these things are super important. We are doing this together with our partners because they understand as well, hey, in the world of AI, it's not only about putting a consultant to work. This work can be done way easier, way faster and way more efficient. And obviously, when we talk about ERP migrations, I think about SAP, and this is definitely a big focus area for us.

Then coming to our -- I mean, to be credible in AI, we need to use Joule. We need to use our own AI. And yes, does everything already work to perfection? No. But even more important is that we are a role model underpinning our great cash flow results and profit results with the use of AI. And you can ask all of our people, we are pushing this really heavily. So we mean it. So in R&D, code-generation tools, tool for developer, [ APA. ]

We have thousands of developers who already see, oh, now I have much more time on developing those agents and making the agent orchestration work and less about my time producing code. In sales, already in Q4, we did a lot with AI on quoting, on pricing, on packaging, help me to find the best deal for my customer. Help me now to find in the pipeline, the best opportunities for me to close out the year. In HR, recruiting, we made the acquisition with SmartRecruiters, but also on skills, a lot will be handled, and we will work smarter with infused AI into our SuccessFactors solutions and then, of course, into our own HR operations.

So in tech, innovations come at a very fast pace. The most important thing is next to having the right strategy is our people. So AI is, first of all, not only a technology who can run a company smarter, it's also about the skills of the people. So reskilling is a big topic within SAP, and we will double down on that because AI will affect every job, and we need to prepare our people for that. That doesn't mean that we need less people. I want to say this very clearly, but we need different skills. And honestly, there will be a change of the mix of the job profiles going forward. But as long as we post such great top line results, we are not thinking about restructuring, we're rather thinking about how we can reskill our existing employee base to make them fit for the next chapter of our transformation.

Now when we then look forward, and in a second, I will hand over to Dominik, let me just share some geopolitical observations. I mean, SAP is, I guess, by far, the biggest tech company in Europe. But what will be very important for the future of SAP and for Europe is clearly, first of all, talent. So we really need to make sure that we are changing our education system and really our universities give us access to the best talent. That's actually working quite well.

But when you think into every job the next generation has to do, it will change. And then super important, and I'm talking about this since quite a while, especially here in Germany, I see a lot of movement now, the willingness to digitize Germany. But when I think about our home market and compare this to the U.S., oh my God, the regulation, I mean, layers of layers. And that is, of course, something when we are closing deals in Q4 in the U.S., it's FedRAMP, you have clear -- we are not even talking with customers about regulation. They are clear. And here, you -- on the state level, you have regulation, everyone reads a little bit different. The [indiscernible].

Then on the federal level, you find other people who have other ideas on regulations on sovereignty. And then you come to the European layer and then you have layer and layer and layer. And now that is not good for SAP, but think about all of our start-ups where you find the same startup like an NNN in China and the U.S. And so this digital union to come together and harmonize that is of such an essential importance because it's not only about funding and access to capital, it's really about speed and the speed is especially super important for all of the great start-ups we are having.

So with that, I said enough. I'm super confident about our outlook for 2026. Strategy is the right one, and we will also -- you're going to see SAP clearly as a winner in AI. And with that, Dominik, over to you.

Thank you, Christian, and thank you all for joining us this morning. I'd also like to wish you all a happy and healthy year 2026. SAP's strong close to the year reflects steady execution against our priorities. As we navigated a rapidly shifting macroeconomic backdrop at the beginning of the year, we remain focused throughout the year on operational discipline and driving value for our customers in times of unprecedented technological change. Our ability to drive top line growth while consistently exceeding our profitability and cash flow expectations reflects the consistent execution against the outlook we provided at the beginning of the year.

While challenges persisted, we took deliberate steps to reinforce our foundation and align the business for durable, sustainable performance. As a result, we closed the year in a position of strength and the progress we've made has set the stage for continued advancement towards our financial and strategic priorities in the years ahead. RISE and GROW with SAP, both remain core pillars of our transformation strategy, serving as go-to solutions for large-scale enterprises and high-growth midsized companies undergoing complex end-to-end transformations and modernization efforts.

And as Christian just highlighted, AI and the Business Data Cloud are beginning to show real commercial impact emerging as meaningful contributors to customer decisions and deal activity. The combined momentum continues to materialize in large cloud transactions with deal volumes greater than EUR 5 million, contributing a record 71% to our cloud order entry in the fourth quarter. These results validate our role as a partner of choice, trusted by world-class organizations navigating high-stakes transformations and speed at scale.

Now let me provide more details around the financial highlights. The current cloud backlog reached EUR 21 billion, up 25%. Quite frankly, this is a more pronounced slowdown than we had anticipated and more than the slight deceleration we guided at the beginning of last year. Echoing Christian's remark, the outcome reflects a deal mix weighted towards larger transformations, many of which include longer ramp periods or flexible structuring, reducing the near-term CCB contribution. Also further mounting geopolitical tensions have led to many customers putting even more emphasis on exploring sovereign Software-as-a-Service solution options.

While SAP is extremely well positioned in this segment, and we have a significant pipeline of opportunities due to the trust Germany and SAP continue to enjoy on a global scale, it takes longer to negotiate these more complex transactions and also longer to deploy and ramp as compared to plain vanilla offerings done by U.S. infrastructure service vendors. This is particularly true for any state-owned and related entities as well as defense, but starts to also affect commercial customers in certain particularly sensitive geographies and industries.

Total cloud backlog for the year grew 30% to a record EUR 77 billion, again, significantly exceeding our current cloud backlog and cloud revenue growth. Cloud revenue actually grew 26% year-on-year in 2025, again, primarily driven by the strong performance of cloud ERP suite. Cloud ERP suite had another notable year, reinforcing its position as a key engine of growth with an increase of 32% in 2025. By the way, if you want to make that comparable to our U.S. competitor, at a couple of percentage points, if you make this constant currency number, U.S. dollar number, then it would have been 34%. This performance is especially meaningful given the expansion of its revenue base over time, highlighting its ability to scale at a sustainable growth rate, now accounting for 86% of total cloud revenue for the year.

Software licenses revenue decreased by 27%. Finally, total revenue for the full year approached EUR 37 billion, up 11%. Now down the income statement. Our non-IFRS cloud gross margin for the full year continued its upward trend from last year and expanded by another 1.6 percentage points to 75%, driving cloud gross profit up by 29%. In the fourth quarter, IFRS operating profit increased 27% to EUR 2.6 billion. Non-IFRS operating profit was up 21%. Both IFRS and non-IFRS operating profit were growing negative -- negatively impacted by approximately EUR 100 million related to a 2025 workforce transformation. In addition, IFRS operating profit growth was negatively impacted by USD 200 million related to Teradata litigation expenses. For the full year, IFRS operating profit increased to EUR 9.8 billion and non-IFRS operating profit to EUR 10.4 billion.

The IFRS effective tax rate for the full year was 28.5%. The non-IFRS tax rate was 30.4%, which is below the outlook of approximately 32%, mainly resulting from an increased ability to offset foreign withholding taxes in Germany. Looking forward, we expect the midterm non-IFRS effective tax rate to be in a range of 28% to 30%, which is the lower half of the previously communicated range of 28% to 32%.

Free cash flow for the full year was around down EUR 8.2 billion, i.e., at the very high end of our revised outlook range of EUR 8 billion to EUR 8.2 billion. The increase was mainly attributable to higher profitability and to lower payments for restructuring and share-based compensation. This result reflects our continued emphasis on disciplined cash management and operating efficiency building on the progress we've made in strengthening the quality and consistency of our cash flow over time.

We are very proud of the progress we've made this year and the business momentum that contributed to our strong net cash position. As a result, SAP has decided to further step up its capital returns with a new 2-year share repurchase program of up to EUR 10 billion scheduled to start in February. This decision reflects our confidence in sustainable strength of the business and our continued commitment to returning capital to shareholders in a disciplined and balanced way. Finally, non-IFRS basic earnings per share in fiscal year 2025 increased by 36% to EUR 6.15.

Now on to the outlook. As you've likely all seen in the quarterly statement published earlier today, we have provided this year's outlook. We expect CCB growth to moderate slightly over the course of 2026. While some deceleration is anticipated, it is expected to be meaningfully less than what we saw in 2025. At the same time, we see a path for total revenue growth to accelerate, supported by the foundation we've built and the continued strength of our business. And our operating profit outlook reflects sustained operating discipline, driving expense to revenue growth ratio towards the lower end of our long-term operating leverage objectives of 80% to 90%, lower end being good, giving us the opportunity to continue to drive non-IFRS operating profit growth significantly above revenue growth. In addition, in 2026, we expect to generate record free cash flow of approximately EUR 10 billion, supported by continued efficiency improvements and operational rigor. Overall, our guidance reflects a balanced view of the opportunity ahead grounded in disciplined execution and an ongoing commitment to long-term value creation.

With now 2025 behind us, we move into 2026 focused on consistency, clarity and execution. The groundwork we've laid across both transformation initiatives and commercial performance puts us in a strong position to deliver against the guidance we outlined today. While geopolitical and trade tensions have taken a certain toll on our top line performance in 2025, the growing need for sovereignty and resilience also offers unique opportunities for those vendors that could offer technologies and tools to reduce dependencies from dominant offerings.

As the largest non-U.S. software SaaS and PaaS vendor, there is no company better positioned than SAP to satisfy this rapidly growing demand. Our strategy to design a stack, which is not locked into any particular Infrastructure as a Service vendor is a particular asset in that respect. And our decision to keep developing our powerful SAP sovereign cloud infrastructure, SCI, thereby preserving capability to run Infrastructure as a Service efficiently in our own data centers brought us with another now even more valuable option to deploy our SaaS and PaaS offerings.

Despite an unpredictable macro and geopolitical environment, our strategy remains clear and our execution is already driving meaningful progress across the business. Customers are choosing us as their North Star to lead mission-critical change, and we remain committed to helping them move faster scale smarter, become more resilient and modernize with confidence. Thank you.

Thank you very much, Christian. Thank you, Dominik. We have 30 minutes left, and we are going to move to the Q&A session now. Could you please at the beginning, limit your input to one question only. I have a couple of questions here in the tool already, but I want to kick it off here in the room, of course. Heidi?

I have a question related to the topic you mentioned last. You mentioned the better environment in the U.S. as to regulation. And you mentioned your opportunities here given the demand as to more sovereign and resilient infrastructure and solutions offering. But are you facing hurdles there in the U.S., like kind of against the backdrop of growing tensions between the 2 countries and maybe there are some hurdles your competitors are facing here. So they might backfire. Are there any indications for that?

No, actually, the U.S. public sector was one of the best-performing businesses in Q4, and that has completely changed. And those customers are actually less concerned around is the software coded in Europe or somewhere else. They have a clear regulatory framework, obviously, and it has high standards for very mission critical parts of the U.S. government, for example, and still standards for other businesses in regulated industries. So there is not a debate about are you from Europe, are you from the U.S., it's really about adhering to those standards.

And that, of course, when you imagine now applying AI to these parts of the world and to their companies, it's very important because now you can really focus on the business value, you can focus on the technological questions. Here, you can find in Europe customers from the same country, asking you for very, very different regulatory standards because, again, there is really this many layers of regulation and that is something where when we really want to leverage the power of Europe, and I'm all in favor for you. We need Europe more than ever. But then at a certain point, someone has to give up power and say, okay, in order to come to one Europe. We can't regulate everywhere. And I guess that is the biggest difference. Also what we have seen, by the way, in Q4, it was very visible also in all the deal closing activities we had.

Thank you. Let me build on that one. We have one question from writers here in the tools to the same topic. Are your solutions intended to diversify? Or are they intended to replace offerings from non-European providers in the long term?

I don't see it. I mean, Phil sometimes in Germany, we are discussing forever since years now, what does sovereignty mean? And then we are getting very theoretic in, okay, does it need to be a European provider, a U.S. provider. At the end, every little piece of hardware will come at some point of time, either from the U.S. or China, if you like it or don't like it. At the end, it's really about the competitiveness. SAP needs to be more competitive. Our AI needs to be stronger than the ones from our competitors. And then the customers, no matter where in the world will buy that.

Obviously, they will also tell us what the sovereignty standards are. I mean, in India, we are also now going to build a new sovereignty standard with some local partners. We do the same in France. We do -- I mean, that is becoming different. But it's still also for SAP, absolutely manageable because when you think about what did we do in the past, there was less regulatory requirements on data and cloud because cloud was -- 30 years ago was not there. But we always localized our software for over 100 countries in the world, and that's now becoming more, especially with cloud and AI. But we have done that in the past. And now we are doing the same thing, obviously, with other requirements coming towards SAP on the cloud and the AI side.

[indiscernible]

You've mentioned a couple of times how the geopolitical tensions impact the business. So how do you expect these tensions to impact the business going forward? I mean, I know there is an outlook, but what impact do you see in this outlook? Do you plan for a scenario in which the tensions might even escalate and maybe a very special question, I've heard that the next SAP leadership meeting is supposed to take place in Washington, D.C. So is there a reason why you have chosen this location? And do you consider changing it against the political backdrop?

I can take the leadership summit question because it came to my table, I didn't think about the geopolitical tensions when we are making these decisions. But obviously, we should probably, I don't know. Look, the leadership summit, it took place in beautiful road over the last 3 years, and we are a global company. And we love to spend our time here, but I also have to support our customers worldwide. And so we made a simple decision, but a long time ago, let's just make sure that everyone lives in peace, so we do it once in the U.S. We are coming back to Europe and then we go to [ ABJ. ] That is the only thing. And sorry to say, we are still a company who has to support global customers. So we cannot make these decisions depending on what is just happening in the world. I mean, obviously, if there would be a war and otherwise, of course. But at the end, we are a global company, and we have people everywhere in the world, and they want to feel part of SAP. And if I would say to my 30,000 people in the U.S., oh, sorry, I don't come anymore. I mean, what kind of signal would we send? I mean, sorry, but this is how we do it, and I feel we are doing it in the right way. Dominik?

Maybe on the outlook, first of all, I want to emphasize that 2025 was not necessarily an easy year to put it mildly in terms of trade issues, geopolitical tensions. And I find it quite remarkable that on cloud revenues, despite all these adversities, I would call it, we have been able to be really within spitting distance to the midpoint of our cloud revenue guidance. That shows you how predictable that number is by now by virtue of the high share of more predictable revenues. So for the way we now scale the guidance for next year, we have basically assumed the 2025 environment to be the new normal. So I think '25 shows that we have a resilience even if some unexpected events hit us. But of course, we're not embarking any meltdown catastrophe scenarios here in that guidance. But it's, I'd say, a good base to build on because let's all hope that it's not getting worse than what we have seen in 2025.

Okay. Let me continue with questions from the tool because we have a lot of questions with regards to our share price dropped today by 10% for a short time this morning. What is the market not understanding about the company?

So I'm doing this job now since 6 years. I have seen a lot of ups and downs. And I -- when we were meeting here a year ago and the share price looked really great. I mean we had a great one for 2 years. It's not a reason for me to lean back and say, hey, this is now -- this is it. And so we need to make our strategy and we need to drive our execution independent of what the capital market is right now telling us. And obviously, it's not only SAP when you have followed the market in the last 6 months. I mean, they are all our competitors in the SaaS space. I mean, Alexandra, our Head of IR tells me we are in the penalty box. We are in this penalty box because there are questions around, okay, what is the future of software in times where everyone maybe can generate and code apps. I mean I already alluded to that.

When you look back into all of the technological innovations over the last 10 to 20 years, it always starts with -- I mean, these phones here became so powerful because there were better chips, better hardware. And the same is with the LLMs. It always starts with the chips, with the hardware. But I'm 100% sure in order to create value on the business side, you need to move up the stack, and it always happen like that. And what I explained before that these agents need to understand business data. They need to understand business processes in order to deliver the value for our customers. This is very true.

And so while there is, of course, a lot of money now going into the chip and semiconductor space, which I totally get, I'm 100% sure that we are uniquely positioned to win the ways on business AI, and we're going to prove that. And so that's why such a share price today is not nice. But at the end, it's super important that we understand our strategy, that we hear from our customers that the strategy is the right one and that we now are laser-focused on the execution of that. And then I'm going to -- and then we will also see again different times.

And maybe to add some numbers around it. I mean, it's almost like a philosophical war around where the value is created. Is it on the infrastructure layer, which is currently the flavor of the month where everybody is investing. By the way, that's actually good for SAP because we are agnostic and the more money flowing into that, the more competitive that infrastructure will be to run our PaaS and SaaS services on top. We are actually deemphasizing that business. Maybe that will stabilize at some point in time because of the sovereign debate we just had before.

On the other side, if I look at the SaaS and the PaaS layer, which we continue to believe for the reasons Christian mentioned, will be a key layer, we are doing actually great, especially in comparison to competitors. You have seen results of some competitors like Dynamics and ServiceNow over the night. There's others to follow. And if you then adjust to an apples-to-apples dollar comparison, we are actually far ahead of the pack in terms of growth rates. So just to give you some data on SaaS, PaaS. In 2025, we had 30% growth in U.S. dollar terms. So that's what you need to compare our competitors to. And I'd say there are some hovering around 20%. There are some hovering around 10%, some in the mid-teens, but nobody is anywhere close there.

So we have a strong degree of confidence. Right now, that kind of SaaS, PaaS layer is not super appreciated by capital markets. But I think the jury is still out what ultimately will happen. And by the way, we had a similar bifurcation, I'd say, in the last big tech bubble in 2000, where telecoms and fiber optics were going through the roof, infrastructure again because that's kind of rising tide lifts all boats. And I wonder how much dark fiber today is still in the ground, which has never been lit since then. And on the other hand, by the way, the dark fiber, you can still light today, whereas the GPUs you buy will not hold for 20 years. So jury is still out on that topic, I guess.

Thank you. Before I move to my M&A question here from the tool, any questions in the room? [indiscernible]

Can you hear me? Yes. I have a question about the tariffs. How are the U.S. tariffs affecting your business, both directly and indirectly via delayed spending decisions by your customers?

I mean there are no tariffs on software or software services, which is good. So there is no direct impact, and we hope it stays like that because we have, again, customers everywhere in the world and tariffs -- digital tariffs would immediately fire back no matter where are you in the world. And then on the indirect impact, again, we saw in half year 1 2025, that was not great on the public sector. A lot of new requirements came up. We needed new certifications. But we overcome that. And Q4 was actually really good in the U.S. public sector. And yes, so no, today, there is no actually direct or indirect impact. Let's hope it stays like that. You never know. Let's see what's happening tomorrow morning.

So back to my tool. The company plans to start a share buyback program. Is there really no other idea to invest for future revenue?

I mean yes, I knew that the question will come. And look, it's a fair question. But look, I mean, first of all, these share buybacks, what we are also doing with these shares, we have employees, and they -- actually, we are also paying them via our shares. So we have actually employee stock programs, and that resonates really well. And so I mean, there is a mean to it. It's not just about financially buying back shares.

The second piece, obviously, I mean, we didn't do larger M&A over the last years. We didn't need to. I mean, still here, look at the quotes, we are posting the accelerated total revenue will come organically. No many tech companies can say this. And so -- but going forward, obviously, would I now rule out M&A? No. We will at some point, do M&A, but then more for technological reasons, especially in the data and AI space, whenever we're going to see there is a technology out there which can help to accelerate our AI and the data platform, we have enough financial flexibility to do that. So SAP is now after that share buyback not short of money. We have the flexibility to react. And we will react, but not from a financial standpoint, we will react if we find the right technology and the right company we are believing in.

May I add on the financial aspects of that, Christian. First, I want to highlight that SAP today has an extremely strong credit profile. So we have a very good rating, much better than some of our competitors. And I dare say we have managed to base that rating more and more on recurring cash generation. Think about the EUR 10 billion guidance we have put out, EUR 8.2 billion that we delivered in '25. So we don't need to hoard an excess cash pile to sustain that extremely strong creditworthiness. So that's the philosophy. And frankly, we always benchmark investments like M&A against investing in our own shares. I always say, why should we do an M&A if investing in our own shares would give us more value. So this is why we think it's part of the mix. And I think it also is evidence to the success we have in really coming up on the free cash generation massively.

Thank you. [indiscernible].

You said you won't need less people. Does this apply to Germany as well?

That applies especially to Germany because there are -- people here in this part of the world are super well protected, and we are also super happy with these people. We are also investing in Munich, in Berlin, and there are major hubs now in the meantime, Munich more supply chain AI, Berlin, it's a lot about data. And so yes, just still -- I mean, I mentioned some of the headwinds we are having here. We can only always share with our government.

Just look at what's happening in China and the U.S. and we can always agree or disagree with certain things. And if it adheres to our values, I will stay out of that. But what happens on the economical side is they are moving super fast. And when it comes to hiring new people, you have them on board in 2 weeks, matters. If you have to reskill your workforce, there's no one you need to ask on, can I apply now these code generation tools to my workforce. There are way less regulations.

And all of that is a result when we are asking ourselves, why is there not another SAP here in Europe? I mean, you probably can find some of the reasons. Not everything is related to that. You need also great entrepreneurs. You need CEOs who need to make the right decisions. But of course, also the regulatory environment is very, very important, especially for a tech company because this industry moves much faster than any other industry in the world.

So we'll talk about AI in a second here on the tool. Any other questions in the room? [indiscernible]?

How important are deals with the military for your company?

I mean they are as important as every other deal we are closing. I mean we are running a lot of military defense companies all over the world. I mean we are super proud. We have a project going on with the Japanese Navy. We're doing a lot with Australia and so on. So they are part of our customer base as every other customer. And of course, with AI, what we are doing oftentimes there, it's not about war. It's about things how we can make them more flexible, how can we help with AI on asset management, on the maintenance of their fleets, et cetera. So that is actually what SAP is doing. It's pretty similar to what we are also doing for other industries. So yes, they are part of our customer base, yes. And I mean, maybe just from the size of the industry, the public sector is ranked #5 when you -- and we are dealing with 22 industries round about, and it's ranked #5 from a revenue perspective.

So 2 questions from the tool, AI first or current cloud backlog first? Let's start with current cloud backlog.

Yes. I mean the one doesn't come without the other. So the AI is actually part of the backlog. And AI is -- because oftentimes numbers -- people ask, what is your AI revenue? The AI sits within our apps. So the AI brings us the apps. The AI help us to win deals in SaaS. The AI helps us to bring more developers on our platform. So it's a natural part of everything what we do. And with that, obviously, it's also part of CCB.

So what's the question? What's the question on CCB?

Again, explain CCB.

Yes. I presume the question might refer to the fact that we have come in at 25% in actual terms and that we had anticipated post Q3 to come in at 26%. You have to understand that when we forecast CCB, it's about also the granularity of all these contracts. And if you look at -- into the specific composition of the contracts we signed, it was slightly different. So the biggest impact we've seen, and we mentioned that in the introductory remarks is that we had a lot of very large deals, 71% of deals being EUR 5 million or higher. And in these large deals, it just takes longer to ramp because the customers start to start with smaller instances in the company and then tackle the really challenging big elephant, so to speak, in the room later. And so there's a little bit of a kind of back-end loading of the ramps there.

Second point is that Christian mentioned the very strong traction we had on the defense side also on the other side of the pond. And there are sometimes procurement laws in certain jurisdictions where we have very mighty procurement departments that can impose a termination for convenience on the vendor. And then we cannot put it into the current cloud backlog because that backlog needs to be contractually committed and that option to walk is there. Now in reality, that option is, of course, sometimes theoretical because these are deeply embedded systems, which are extremely sticky. So we're very confident that the revenues out of this will come. But technically, we cannot put it into current cloud backlog.

And the last point is what we discussed that there is more and more customers who say, can I really afford to have an off-the-shelf standard plain vanilla U.S. hyperscaler Infrastructure as a Service? Is there a risk that, that might go away quickly for whatever political reasons and look for alternatives? And these alternatives are just about to emerge. Some of them are already up and running. Some are just certified. The certification process takes some time. They also sometimes need to be built. So also from the signing of the contract till the deployment at the customer, it takes time. So these were the 3 factors that actually explain the delta. Each of them not super big, but if they compound together, we talk about that roundabout 1 percentage point.

Thank you. Any other questions in the room? No. Okay. IDC. You are saying that connecting SAP AI to industry-specific processes is critical to winning customers such as H&M. How can SAP move into AI -- move AI into the industry-specific process at scale. What is your vision for that? After all, these processes vary greatly by industry.

Now I have to be careful that I'm not deep -- diving too deep in our industry technological layer. I mean, first of all, there was a certain reason why always customers lean towards SAP to build industry extensions. A lot of data which sits in an ERP needs to be then also flowing through an industry capability. I mean when you do return claims management, it would be good to have the order data from the ERP. If you talk about supply chain resiliency, you need to also understand how do you produce, how to transport and so on. So you always come back to the core. So then to extend that with industry capabilities makes total sense. And that's why a lot of customers also turning to SAP. So we have the knowledge, we have the people also here in Germany, by the way, a lot, who understand these industries extremely well.

Now with AI, we can, of course, completely reimagine how these certain industry capabilities will be done. I mean a machine who needs maintenance, we can actually predict this now way better than with our former asset management solution of SAP because we have agents who are getting demand signals. We have agents which can read out by an LLM then in that case, the machine instruction when something is happening, how to put the machine up faster. We're, of course, getting -- we have the data in our ERP, where are the technical people who can fix the machine. And all of these agents are orchestrating all of that to improve the uptime of the assets of the company. And these are these industry capabilities, which we know very well from the past. And now we have to make sure that we also then co-innovate with our customers the next generation of AI industry capabilities they need.

And so -- and technological-wise, I mean, it's the same like in the LOBs, we need the data scientists now. We need the people who can develop the AI, but we have those people. So now it's about going into this together. And I'm sure, especially this industry AI will be a big growth driver for SAP. Can you standardize this 100%? No. I mean, such an agent will look different even within one industry. One mining company will not exactly do asset management like another mining company or the Deutsche Bahn. And this is where we, of course, have to have the extensibility layer so that customers can go into our agent builder and can see, okay, I want to actually automate that process piece on top of what SAP provided. So this fine-tuning of agents, this extension of agents is a super critical capability as part of our solutions.

If we don't have any other questions in the room, I'll take the final one from the tool combing 2. AI investments. Your peers are struggling to show real AI value. What is SAP's value on AI. And how do you define sales goals in terms of AI for salespeople, if you do not measure AI revenues?

Yes. I mean, first on the value. I mean, I described H&M, I described Fresenius, Avelios. We are doing for other large companies in the world, last mile delivery. So we are doing it already. Now is some of that still to be developed? Yes. But I can say, I speak for everyone in this industry that these things further need to mature. The very important part is of it, do you have the AI foundation? Do you have the data? Do you have the business process understanding? And I can tick like all of that.

Now do we need some time and further investments to make that happen? Absolutely. But we are on a very good track and customers are already seeing the first AI agents, and they are believing in it. Just here in Germany, we had a big health care company, they just removed all of their 120 modules they had for cash flow because our AI foundation came in together with an LLM and showed, hey, we can do this way smarter.

And then last but not least, how do we measure that? I mean when we are going into now the year, I mean, obviously, we review in how many deals is AI part of that. When you sell supply chain, when you sell HR, don't go to the customer and sell it in the old way, sell them the new capabilities with AI and how we can help to transform the customers' business. That's what we are looking at. We are looking at the value proposition and then obviously connecting it to our product road map so that what we are selling can also be adopted later on. And this is how we're going to steer AI inside SAP.

Sales target.

Yes. I mean sales targets, again, we -- the people get incentives, if they're selling value to our customers, we see high adoption and AI is part of the solution. It's not like here is a piece of AI and here is the piece of supply chain software. It needs to come together. And only when it comes together, you're going to see that you also get higher incentives because we want to, of course, sell our customers the future, and that's how we steer it and how we incentivize our people.

Perfect. We're running out of time now. Thank you, Christian. Thank you, Dominik. Thank you, everyone, for joining us today virtually. Of course, also here in the room.

Ladies and gentlemen, thank you for standing by. Welcome, and thank you for joining the SAP Q4 and Full Year 2025 Earnings Conference Call. [Operator Instructions]

I would now like to turn the conference over to Alexandra Steiger, Global Head of Investor Relations. Please go ahead.

Good morning, everyone, and welcome. Thank you for joining us. With me today are CEO, Christian Klein; and CFO, Dominik Asam. On this call, we will discuss SAP's fourth quarter and full year results for 2025. You can find the deck supplementing this call as well as our quarterly statement on our Investor Relations website. During this call, we will make forward-looking statements, which are predictions, projections or other statements about future events. These statements are based on current expectations and assumptions that are risk -- that are subject to risks and uncertainties that could cause actual results and outcomes to differ materially.

Additional information regarding these risks and uncertainties may be found in our filings with the SEC, including, but not limited to the Risk Factors section of our annual report on Form 20-F for 2024. Unless otherwise stated, all numbers on this call are non-IFRS and growth rates and percentage point changes are non-IFRS, year-on-year on constant currencies. The non-IFRS financial measures we provide should not be considered as a substitute for or superior to the measures of financial performance prepared in accordance with IFRS.

And with that, I would like to turn the call over to Christian.

Yes. Thank you, Alexandra, and a warm welcome to everyone joining this call. In countless conversations with business leaders in Q4 and at the World Economic Forum, it became clear. Customers are facing geopolitical uncertainty, macroeconomic volatility, and they would like to leverage AI to make their companies more resilient and more productive, driving growth as well as cost efficiencies.

At the same time, it is very encouraging that more and more customers and partners are turning to SAP to gain real business value from AI. Why? Because they realize that they don't gain value by developing a number of custom AI agents or by applying commodity large language models on top of transactional business applications.

The formula for gaining real value from AI as an enterprise is becoming clear. It's important to reimagine first how AI will change existing business models and mission-critical business processes and to boost process automation and efficiency, AI agents must be embedded in business processes and trained with context-rich business data that is not available to large language model providers.

This is a unique combination only SAP can deliver because our business suite provides us with access to the world's largest volume of business data and we directly infused our Agentic AI layer in the most mission-critical business processes of a company. This momentum of SAP Business AI is also clearly visible in our Q4 numbers.

More than 2/3 of our Q4 cloud order entry includes business AI, increasing by more than 20 percentage points compared with Q3. Looking at the 50 largest deals in Q4, 90% of them included AI or SAP Business Data Cloud. We also saw the number of customers using our AI CoPilot tool growing ninefold over the course of the year.

Now let's have a broader look at our financial performance. Q4 was the best bookings quarter of 2025 ahead of our expectations. This resulted in a total cloud backlog of EUR 77 billion, up 30%. This also clearly shows the underlying momentum of our business as well as our potential in the future.

In addition, we achieved our 2025 financial outlook for cloud revenue as well as cloud and software revenue. A great performance considering the macroeconomic challenges we faced in half year 1. The ongoing transformation of SAP's operating module, combined with applying AI across the company, allowed us to even beat our operating profit and free cash flow outlook in 2025. Some of you might remember, back in 2020, we started our disruptive cloud transformation with very ambitious targets for 2025. And today, I'm happy to say we have delivered what we promised and even outperformed this ambition.

But more importantly, the great technological achievements of our cloud transformation are now giving SAP the why to win with business AI. It is the time to thank our customers, partners and more than 100,000 colleagues worldwide for the trust in our strategy and for making the transformation happen, for sure, the biggest in SAP's history.

Let me directly turn to Q4 number where I expect some questions. Our current cloud backlog. In Q4, it grew 25%. Back in Q3, we expect it to reach 26%. Before I explain the deviation, let me outline that in Q4, new bookings came in clearly ahead of plan. And we saw strong customer retention, low churn and stable discount rates.

This resulted in a 30% increase to EUR 77 billion total cloud backlog, an impressive growth on an already large base. Despite the strong Q4 performance, 2 factors led to deviation. First, we closed a higher share of significantly large deals in Q4 compared to our forecast in October, which is great for the total cloud backlog, but large customers often don't move their mission-critical ERP in the first year. These deals always have more back-end loaded ramps and as a consequence, a limited impact on current cloud backlog in the first 12 months.

The good news, the average contract duration remains stable, so we expect higher revenue contributions from these deals over the next few years. Second, we closed a higher share of government deals in Q4 that included a termination for convenience by law. Such deals are not reflected in the CCB. So while we even overperformed on bookings and are very satisfied with the outcome of Q4, the combination of both effects resulted in a 1 percentage point difference to what we expected in October.

And consequently, a slight shift of cloud revenue from 2026 to 2027 and beyond. Now the great bookings performance included some impressive wins in Q4. Adidas, L'Oréal and H&M Group are embarking on the RISE with SAP journey, Deloitte, Pirelli, RTX, Nokia, Jabil and the U.S. Navy chose RISE too. While Toyota and Daimler Truck are even expanding their ongoing RISE journey. Lockheed Martin went live on RISE with one of their business areas, a further step in our multiyear partnership.

It will continue in 2026 with our human capital management solutions as they keep transforming their workforce. Across all sectors, companies selected SAP growth to transform their business. KPMG, Snowflake and Müller, a large German retailer, just to name a few.

On business AI, new customers included names such as Tech Mahindra, Mondelez, Kirin and Sun Chemical. In health care, Fresenius selected SAP Business AI to sustainably improve patient care. We also saw the Bosch Group select SAP Business AI in Q4 to help boost innovation across all 4 of its business sectors.

This momentum is translating into business AI adoption. Our customers are already achieving impressive outcomes. Just one example. Siemens, consultants can reinvest 25% of their weekly working time into higher-value activities, thanks to tool for consultants. These impressive wins are on just a list of logos. They are the result of our focused strategy execution throughout the year.

Looking back at 2025, we made huge focus. Let me name a few highlights. First, our cloud transformation is in full swing. Customers, representing 40% of our support revenue base have now initiated a move to cloud ERP using device and go with SAP offerings. On the go-to-market side, we established a partner-first approach for the mid-market. And our focus on public cloud pays off with order entry growing more than 5x faster than private cloud in 2025.

Now almost half of our order entry. We have also expanded our software and cloud capabilities and continue to see high demand for our offerings in the public sector. In Q4, for example, we closed a new OneGov agreement with the U.S. General World Services Administration and with the HMRC in the U.K.

Finally, our innovations in SAP Business Data Cloud and Business AI are seeing strong traction. SAP Business Data Cloud secured around EUR 2 billion in total contract value in less than a year since its launch. And as I mentioned earlier, more than 2/3 of our Q4 cloud order entry included AI.

Now let's turn to our growth ambition for the year 2026 and beyond. Over the last years, we have built a solid foundation for future growth, thanks to the successful execution of our transformation. Building on this momentum, we are confident that we will further expand our market share for cloud revenue and accelerate total revenue growth through 2027. This confidence is based on several key growth drivers. First, coming back to our total cloud backlog of EUR 77 billion. This grew by 30%. It outperformed our current cloud backlog growth by 5 percentage points.

In short, we have a significant amount of our future cloud revenue in the books. And given the ramps of the large deals over the next 4 years, we are increasingly building a strong foundation for total revenue acceleration through 2027.

Second, we will continue converting our installed base to the cloud with a multiplier of 2 to 3x, considering our support revenue base of EUR 10.5 billion, this represents a multibillion euro cloud revenue opportunity for us. Third, the vast majority of our cloud customers are expanding their SAP footprint across the SAP business suite.

They now clearly see the value of best of suite over best of breed, especially in the age of AI. So they leverage the best of suite approach not only to run their business processes end to end, but they also seek a harmonized data platform that provides the foundation for high-value business AI.

As a result, in Q4 alone, almost 2/3 of our deals exceeding EUR 1 million involved 4 or more lines of business, a remarkable increase of 25 percentage points, but we see this not only in increased up and cross-selling numbers, but also in market share gains.

Overall, we outperformed the cloud market by 10 percentage points in 2025. Fourth, it is just the world's largest enterprises that rely on SAP. With our business suite in the public cloud, SAP's mid-market business is growing too. We expand in the mid-market, and we are winning new customers through our partner-first strategy and the significant expansion of our reseller ecosystem.

This channel is already growing more than 1.5x faster than our direct business. And we will substantially increase the contribution from our ecosystem. Finally, let me conclude with the growth driver that has the highest potential. And this, of course, to create a strategic relevance by far, our Business AI and Business Data Cloud.

The traction from 2025 is just the beginning. Many customers have seen that an LLM alone is not enough. They need modules, business data and context to build high-value AI use cases and derive business value. So let me explain to you how we drive growth with SAP's unique combination of apps, data and AI. First, Joule, we are reinventing the user experience and the way people work across all our apps with our CoPilot tool. Unlike other digital coworkers, Joule doesn't just access the world's leading LLMs, it also has broad access to business data.

No matter the task, joule is setting new standards for user experience, simplicity, productivity and we'll just redefine the future of work. Second, embedded and extensible AI agents. We are embedding AI agents across the main business processes of every company. Doing so, we are helping our customers to realize efficiency through automation, make better decisions and gain agility across their value chain with connected agents.

To extent where differentiation is needed, we provide a powerful agent builder with capabilities that are unique to SAP such as the access to a semantical BDC data product and the ability to benchmark and infuse agents directly into the business process layer within our apps.

Our pro-code offering serves the needs of our developers while our business users can use the agent builder in a local mode to easily build, deploy and manage AI agents. Third, we have always differentiated ourselves by our deep industry and business process knowledge and coded high-value industry-specific applications. Now we are reimagining these strategic industry capabilities with AI.

Together with some industry-leading customers, we are already redefining mission-critical parts of an industry value chain by developing the next-generation AI solutions like redefining patient care and health care, delivering higher predictability and supply chain resilience in manufacturing. And to run smarter trade promotions and personalize the shopping experience in retail.

Some of these industry AI use cases were already crucial to the success of our Q4 deals. Fourth, BDC addresses one of the biggest all blocks for AI adoption that our customers face today, data silos. BDC brings together SAP and non-SAP data, providing our customers with the harmonized data they need to enable their agentic AI vision. In 2026, we will heavily accelerate the development of data products to enrich the semantic layer for all of our customers.

And fifth, accelerated ERP migration. In customer spend 10x more on their ERP migration compared to what they spend on their ERP software. With our integrated AI-powered migration tool chain in the RISE journey, we are shrinking this ratio by cutting the migration cost and making it faster and easier for customers to realize value from their transformation. Net-net, in all of the focus areas, SAP has a clear right to win.

Together, they'll secure SAP's position as the leading business AI company. Finally, to be credible in the AI market, we have all modeled the use of business AI and kicked off an extensive AI transformation program internally. We are using AI across all functions to unlock significant long-term operating profit potential. In R&D, we see huge efficiency gains as the role of a developer is shifting from traditional code writing toward designing, guiding and validating AI-generated solutions.

In sales, Business AI will help us to better quote, price, identify opportunities. In HR, Business AI will help us better plan and transform our workforce. Every SAP leader across every function is infusing AI into their business and drive the reskilling of our workforce, because AI isn't just about technology, it first needs to be enabled by our people. Altogether, our goal is to achieve a run rate of around EUR 2 billion in real cost efficiencies by the end of 2028, thanks to the internal usage of AI.

This equates to efficiency gains of 15% to 20% of addressable costs, which also helps us to reinvest into our AI road map. So to sum it all up, we closed 2025 with strong momentum. Our strategy is validated by our customers. And we have all the ingredients to win in the age of AI. The groundwork is done, and 2026 will be the year AI delivers enterprise scale return on investment.

And with that, I'll hand over to Dominik.

Thank you very much, Christian, and thank you all for joining us this morning. I'd also like to wish you all a happy and healthy year 2026. SAP's strong close to the year reflects steady execution against our priorities. As we navigated a rapidly shifting macroeconomic backdrop at the beginning of the year, remain focused on operational discipline and driving value for our customers in times of unprecedented technological change.

Our ability to drive top line growth while consistently exceeding our profitability and free cash flow expectations reflects the consistent execution against the outlook we provided at the beginning of the year. While the challenges persisted, we took deliberate steps to reinforce our foundation and align the business for durable, sustainable performance.

As a result, we closed the year in a position of strength and the progress we've made has set the stage for continued advancement towards our financial and strategic priorities in the years ahead. Verizon growth with SAP both remain core pillars of our transformation strategy, serving as a go-to solution for large-scale enterprises and high-growth mid-sized companies undergoing complex end-to-end modernization efforts.

And as highlighted by Christian, AI and the Business Data Cloud are beginning to show real commercial impact emerging as a meaningful contributor to customer decision and deal activity. This combined momentum continues to materialize in large cloud transactions with deal volumes greater than EUR 5 million, contributing a record 71% to our total cloud order entry in the fourth quarter.

These results validate our role as a partner of choice, trusted by world-class organizations navigating high stakes transformation at speed and scale. Now let me provide more details around our financial highlights. Current cloud backlog reached EUR 21 million, up 25%. This is a more pronounced slowdown than what we had anticipated and more than the slight deceleration we guided to at the beginning of last year. Echoing Christian's remarks, this outcome reflects a deal mix weighted towards larger transformations, many of which include longer ramp periods or flexible structuring, reducing their near-term CCB contribution.

Also, further mounting geopolitical tensions have led to many customers putting even more emphasis on exploring sovereign SaaS options. While SAP is extremely well positioned in this segment, and we have a significant pipeline of opportunities due to the trust Germany and SAP continue to enjoy on a global scale. It takes longer to negotiate these more complex transactions and also longer to deploy and ramp as compared with plain vanilla offerings of U.S. infrastructure as a service vendors.

This is particularly true for any state-owned and related entities as well as defense, but starts to also affect commercial customers in certain particularly sensitive geographies and industries. Total cloud backlog for the year grew 30% to a record EUR 77 billion, again significantly exceeding our current cloud backlog and cloud revenue growth. Cloud revenue actually grew 26% year-on-year in 2025, again, primarily driven by the strong performance of cloud ERP suite. Cloud ERP suite had another notable year, reinforcing its position as a key engine of growth with an increase of 32% in 2025.

And I want to make the remark, this is as a constant currency number in U.S. number -- U.S. dollar, the number would be 2 percentage points higher if you want to compare to competition. This performance is especially meaningful given the expansion of its revenue base over time, highlighting its ability to scale in a sustainable growth rate, now accounting for 86% of total cloud revenue for the year.

Software licenses revenue decreased by 27%. Finally, total revenue for the full year approached EUR 37 billion, up 11%. Now let's take a brief look at our regional performance. For the full year, Brazil, France, Germany, India, Italy, South Korean and Spain all had outstanding performances in cloud revenue, but China, Japan, Saudi Arabia and the United Kingdom as well as the U.S. were particularly strong.

Now down in the income statement. Our non-IFRS cloud gross margin for the full year continued its upward trend from last year and expanded by 1.6 percentage points to 75%, driving cloud gross profit up by 29%. In the fourth quarter, IFRS operating profit increased 27% to EUR 2.6 billion. Non-IFRS operating profit was up 21%. Both IFRS and non-IFRS operating profit growth were negatively impacted by approximately EUR 100 million related to a 2025 workforce transformation.

In addition, IFRS operating profit growth was negatively impacted by USD 200 million related to Teradata litigation expenses. For the full year, IFRS operating profit increased to EUR 9.8 billion and non-IFRS operating profit to EUR 10.4 billion. The IFRS effective tax rate for the full year was 28.5%. The non-IFRS tax rate was 30.4%, which is below the outlook for approximately 32%, mainly resulting from an increasing ability to offset foreign withholding taxes in Germany.

Looking forward, we expect the midterm non-IFRS effective tax rate to be in the range of 28% to 30%, which is slightly lower than half -- lower -- which is in the lower half of the previously communicated range of 28% to 32%. Free cash flow for the full year was rounded down EUR 8.2 billion, i.e., at the very high end of our revised outlook range of EUR 8 billion to EUR 8.2 billion.

This increase was mainly attributable to higher profitability and to lower payments for restructuring and share-based compensation. This result reflects our continued emphasis on disciplined cash management and operating efficiency building on the progress we've made in strengthening quality and consistency of our cash flow over time.

We are very proud of the progress we've made this year and the business momentum that contributed to our strong net cash position. As a result, SAP has decided to further step up its capital returns with a new 2-year share repurchase program of up to EUR 10 billion scheduled to start in February.

This decision reflects our confidence in the sustainable strength of the business and our continued commitment to returning capital to shareholders in a disciplined and balanced way. Finally, non-IFRS basic earnings per share in fiscal year 2025 increased by 36% to EUR 6.15. Now let's move on to our outlook.

As you've probably seen in the quarterly statement published earlier today, we have provided this year's outlook by now, we expect CCB growth to moderate slightly over the course of 2026, while some deceleration is anticipated, it is expected to be meaningfully less than what we saw in 2025 in terms of deceleration.

At the same time, we see a path for total revenue growth to accelerate supported by the foundation we've built and the continued strength of our business. And our operating profit outlook reflects a sustained operating discipline, driving our expense to revenue growth ratio towards the lower end of our long-term operating leverage objective of 80% to 90%, giving us the opportunity to continue to drive non-IFRS operating profit growth significantly above revenue growth.

In addition, in 2026, we expect to generate a record free cash flow of approximately EUR 10 billion, supported by continued efficiency improvements and operational rigor. Overall, our guidance reflects a balanced view of the opportunity ahead grounded in disciplined execution and an ongoing commitment to long-term value creation.

With 2025 now behind us, we move into 2026 focused on consistency, clarity and execution. The groundwork we have laid across both transformation initiatives and commercial performance puts us in a strong position to deliver against the guidance we outlined today. While geopolitical and trade tensions have taken a certain toll on our top line performance in 2025, the growing need for sovereignty and resilience also offers unique opportunities for those vendors that can offer technologies and tools to reduce dependencies from dominant offerings.

As the largest non-U.S. software, SaaS and PaaS vendor, there's no company better positioned than SAP to satisfy this rapidly growing demand. Our strategy to design a stack, which is not locked into any particular SaaS vendor is particularly valuable in that respect. And our decision to keep developing our powerful SAP converged cloud infrastructure, the sovereign infrastructure, thereby preserving capability to run infrastructure service efficiently in our own data centers, provides us with another now even more valuable option to deploy our SaaS and PaaS offers.

Despite an unpredictable macro and geopolitical environment, our strategy remains clear and our execution is already driving meaningful progress across the business. Customers are choosing us as their North Star to lead mission-critical change, and we remain committed to helping them move faster, scale smarter, become more sovereign more resilient and modernized with confidence. Thank you, and we will now be happy to take your questions.

Thank you, Dominik. With that, we'll now take your questions. [Operator Instructions]

Operator, please open the line.

[Operator Instructions]

We'll take our first question from Adam Wood with Morgan Stanley.

Maybe if we go to, I guess, the main focus for investors on the CCB at the end of the year. You spoke obviously at the end of the year of 25% being disappointed -- being disappointing, which is unfortunately where we ended up. Could you maybe talk us a little bit through the end of the fourth quarter close? Was it entirely that you had these large deals with later ramps and deals with cancellations clauses in them? Or did you also see some slippage? And maybe if that was also the case, could you help us with what the pipeline looks like going into Q1? Obviously, there was slippage out of a very big Q4, that could mean a very good pipeline for the first quarter.

And then I appreciate you want to give these numbers to the big figure and not to the decimal point. But unfortunately, I think for everybody, the kind of rounding up and rounding down would help. Could you give us any insight into whether 25% and then 26% adjusted for those effects is being rounded down or rounded up?

And then maybe just finally, you talked about slight deceleration. I guess everybody is going to say, well, slight deceleration last year was 4 points. I guess, we could say 3 points organic. Is that the same type of range that people should be thinking about? Because I think that's initially where people would go to.

Yes. Thanks a lot, Adam. So look, let me start with the CCB of 25% as yes, we said in October, 26% would be the target. I mean, first, let me reiterate again. Compared to October, we even had overachieved our bookings plan. And also the churn came out lower than expected, and we also, very importantly, had stable discount rates.

Now why did we end up at 25%? What we have seen is that during the course of Q4, actually, the deal mix has changed. We closed larger deals. And I mean, it's quite standard in -- for a larger customer. Oftentimes in the first year, they are shifting some smaller solutions to the cloud because the larger ERPs, that needs time.

First, to figure out how they would love to run their business processes. We talk about clean core. But of course, there are also a lot of technical things to be figured out before you really lift and shift your most mission-critical system to the cloud. So that was clearly, I would say, the largest factor.

And the second one is actually that we have seen that actually, we performed much better in the public sector. And in some deals, you have a termination for convenience by law. So per se, we are not including this in the CCB. And that was the only mix effect, I would say, we had compared to October.

Again, bookings performance was ahead of plan. Now for some of the larger deals, where there are slippages, I mean, in every quarter, you have slippages. But I would say the execution -- the sales execution was actually -- was really good. I mean some of these larger megadeals what I was mentioning is actually that some of them closed, some of them not.

So clearly, also for 2026, we see actually a better pipeline coverage than when we compare the pipeline now compared to where we were at in 2025. Also, of course, one thing to consider what Dominik also mentioned is sovereignty. I mean in some countries of the world, in some industries, obviously, also deal cycles took a little bit longer. I mean with the geopolitical tensions, customers have more questions around sovereignty and actually, that also then reflects in longer negotiation cycles.

Now for 2026, what means marginal decline or slight decline over the year, definitely not a 4 percentage decline. That is not what we are seeking for. It will be not such a decline like what you have seen in 2025. Also, again, given the strong pipeline what we have now at the beginning of the year, also mentioning -- I mentioned that 90% of deals, which included AI. I mean we see an even better trend for 2026. So you clearly see with BDC and AI, customers are now not only doing RISE to just reimagining their business model and drive to a clean core. We see now in more and more deals that also, of course, AI and BDC is kicking in as another growth driver, also helping, of course, the multiples of these deals.

We'll take our next question from Charlie Brennan with Jefferies.

I'll do 2, if I can. Firstly, everyone is preoccupied with AI at the moment. I think you referred to a EUR 2 billion saving at SAP over the next couple of years. Can you talk to how that's going to flow through the business? Are you going to extract that through natural churn in the organization? Or do we have to think more about, say, another restructuring program? And then as you rewire the business towards a sort of AI age, how much of your R&D today do you think is based on AI-driven tools? And how much of your output today is focused on AI as opposed to some of the core products?

Yes. Talking about our internal transformation, obviously, it's for us super important to also be credible to our customers. And indeed, I mean, all our business leaders already working with actually our product management teams on implementing certain AI use cases. For the EUR 2 billion, that will be reached, obviously, by -- first, we will have a heavily growing business. And with that, we will just under proportionally grow our cost and headcount base. And you will see the efficiencies.

Obviously, in R&D, I mean, this is where the LLM modules alone can do magic on -- especially on the code generation side. But in all other areas, obviously, our own AI foundation is kicking in where heavily also you need business data to build those smart agents who can take over a lot of the work what people are doing today. So today -- as of today, there is no restructuring plan. Obviously, can you rule this out forever? No. But today, I can tell you we're going to achieve that by just scaling our business way more than in the past with AI.

Now on the R&D side, I mean, first, what we did is, I mean, we have great talents in the organization. But obviously, these talents are oftentimes buried in development backlog. So already last year, we lifted and shifted a lot of our AI talents into the work to build the AI foundation to train our foundational module, to build those AI agents. So a lift and shift has already happened. And so for example, the things which are, of course, now becoming less and less is code generation, so we already automated 35% of the code.

And then obviously, that will increase now also again significantly this year. So actually, the profile of a software developer is already changing quite significantly within SAP. Also again, investing and it's not a size of quantity of people, it's really about getting the best people, the best data scientists, the best AI developers on board, to actually build what I just lined out with the 5 growth drivers within AI.

Maybe on the R&D side, let's not forget that I think one of the challenges we used to face very heavily is the enormous rigorous prioritization of what we are actually developing because there's such a scarcity of topics we can go after versus the -- in terms of opportunities there are. So let's not forget that there is ample of wood to chop, so to speak, on R&D to do things that previously we couldn't do using the modern technologies.

And this is also why in terms of financial model, we stick to our 80% to 90% operating leverage. Yes, we're going to be at the lower end of that in 2026. But the last thing we want to have is putting some mortgages on the top line. We really want to push the top line very hard by aggressively investing. And so we want to use these tools to really push the envelope and secure innovation and top line and to be ahead of the pack.

The next question is from the line of Frederic Boulan with Bank of America.

If I can follow up on the AI side. We've seen growing concerns about risks from AI impacting the enterprise SaaS. You made a strong pitch around the kind of software data and agent ecosystem. It would be good to share in terms of AI traction, what percentage of existing cloud customers currently using your AI offering. I mean, it's good to have the kind of 90% of bookings now, including that, but it would be good to see existing cloud customers using the product and give us a range of revenue uplift you're seeing in particular with some of the customers where you've seen earlier adoption.

From a risk standpoint, I mean, a huge amount of concern in the industry about new tools out there. You made a pitch about the relevance of your own models versus commoditized LLMs. But do you see any of your customers starting to use different tools to respond to needs that they were previously addressed by SAP, either in core ERP or across HCM, et cetera?

I can start, Dominik, please also comment. I mean, first on number of customer adoption. As I already said, customer adoption of tool increased by ninefold, which is really significant, starting when I compare this back to 2025 January. And then second, obviously, what we are seeing is, of course, now that we are developing [ new skills, ] more users are jumping on it. Customer is the one thing, number of users. And what we see is actually in the cloud base, a healthy penetration of our AI, around about 60% of our customers are already using our AI actively, 20% are on the way to it.

But also, please not forget, and I know what the market is fearing right now. To give you one practical example -- no, two, what -- in the business world, in order to deliver high-value AI, I mean, let's take code generation. Yes, of course, the LLMs can understand code, can find the patterns, understand how people use the code and actually are super powerful on that.

There is no business data involved. Let's use a German customer. Actually, it started with an LLM to build a cash flow agent. No value because why? The LLM could, of course, read certain e-mail, support tickets and try to figure out why is this customer not paying. But what was completely missing is all the business data, the inventory data, the financial bank statements, the other data on the customer side is there's still a deal ongoing. Maybe that's blocking actually the payment of this customer. So actually, what we did is we used this LLM, we brought it together with our AI foundation and our knowledge graph. And actually, what the customer did, they removed actually over 200 predictive models now, which they built on their own using our AI foundation.

And that example, you can actually replicate to every AI agent in the business world. Second, we won a large deal with H&M. We actually built a prototype for them. I mean -- and that's what they are now going to implement on AI. We show them the personalized shopping experience in commerce, obviously, also using an LLM, but we combine it again with our AI foundation to better understand what did the consumer buy in the past to understand better patterns about what he clicked on the web page, et cetera.

Then we went into returns claims management, also that. We had this industry capability always in our portfolio. Now we are actually reimagining this industry capability with AI, making it smarter, making it more efficient when it comes to returns claims management. So -- and that is a very good example. We closed one of these mega deals I personally was involved in. We didn't win it actually, again, only because, hey, cloud, clean core, that was the thing what we did a year ago. Now we are winning it because last mile delivery. Again, these -- this customer tried to reinvent last mile delivery with an LLM. Again, they were missing business data. We brought it together. We showed them what we can build together with our AI foundation, and they were totally convinced.

So to make this very clear, we are winning deals because of AI. We are not losing deals because of AI. And definitely, these deals are actually now leveraging AI to increase the win rate in Q4.

Maybe to give a little bit of financial color around that kind of winning deals. I'm always a little bit nervous about how people compare our numbers to the industry. We have a constant currency disclosure, but don't forget, we had a massive devaluation of the U.S. dollar. I think over the year, it was about 13% appreciation of the euro. So our constant currency performance in SaaS, PaaS and there's also a different strategy. I mean, we are not investing massively in infrastructure. Actually, you have seen our infrastructure business decline. Frankly, that decline might slow as the sovereign solutions become more prevalent.

But still, it will -- we will largely leverage third-party infrastructure. And there's a spree of investment appetite right now in this frothy business. So we are not very concerned about lack of opportunities to leverage that infrastructure. Now what we focus on is SaaS and PaaS, and we have delivered in Q4 a whopping 27% constant currency growth for that. If you transform that into comparable U.S. dollar numbers, that's above 30%. Now you have seen some of our competitors, Dynamics and ServiceNow report numbers in Q4, which are hovering around the 20s, I think 19% to 20%. You've seen our largest competitors, some of them not even reaching 10%. And so I'd say that is the evidence that we are actually winning in AI as opposed to losing.

We'll take our next question from Ben Castillo with BNP Paribas.

There's lots of positives in here, lots of large deals in the mix, the sovereign cloud opportunity, the high volume of AI in the backlog, record TCB. So that all sounds optically very encouraging. And if we were to compare that against, we ultimately still have CCB growth of sort of 25% in Q4 and still indicating the cloud revenues decelerate this year to come.

I guess could you help us just think about the changing landscape here, that growing mix of large deals in the pipeline that are converting the longer deal ramps. How should we think about that maybe midterm trajectory of total cloud revenue growth into '27 and perhaps beyond? Just help us with how you think about that pace of cloud revenue growth, either deceleration or scope for stabilization?

Yes. I mean, good question. And for sure, when I'm looking back to October, there is definitely, I mean, honestly, also lessons learned at Q4, and we didn't see this in the forecast. But obviously, when you are then going into Q4, it's not unnatural actually that you have large deals, but you see it in the order entry.

I mean, we, this time, really closed many more large deals. And again, what should you do? I mean, we saw this then in -- during the course of -- actually in December, where customers then said, okay, deal done, and now we are doing the phasing. And the phasing actually, I mean, should we now incentivize our people to keep the first 12 months up, that would be the wrong thing to do. Because, again, it's against the nature of how these transformations work. And also, I don't want to discount the renewal base. At the end, what matters for the company on the mid and the long term is the renewal base because that is what is driving the cloud revenue and the profits actually on the long term.

And so when you look into 2026, I mean, Q1, Q2, Q3, actually, we are not having this larger share of large deals. And that's why we actually see -- we will see a similar pattern. And about Q4, I mean, definitely, when you look into our support revenue base, there are still some larger customers, and we need to make sure that, especially when it comes to the phasing of a deal, et cetera, that we have that right. And -- but again, I would see there is definitely -- you can expect a similar pattern to what we have seen in 2025, I mean, given what is left in the installed base.

Maybe on the cloud revenue in terms of what we see going forward. First of all, we guided cloud revenues now for '26. And maybe just looking back at '25, I venture to say that the accuracy of forecasts on cloud revenues is actually by now extremely high. This is also due to the extremely high share of recurring, more predictable revenues.

And yes, because of all the macro mess being tariff disputes or the kind of sovereign debate we just mentioned, we were kind of [indiscernible] EUR 75 million away from the midpoint of the guidance at the end of '25. Again, such a massive macro backdrop, I feel 0.5 percentage point variance to the midpoint on growth is almost like forward accounting.

So rest assured that our revenue guidance also for '26 is of a similar kind of confidence level. CCB forecast is always a little bit more difficult because as we just learned, the granularity of what's exactly coming and the crystal ball is a little bit difficult. But I also want to highlight how you think about the bridge from CCB growth and cloud revenues. And if you, again, do the math for last year to calibrate your model, so to speak, it will be obvious once we publish our results, you will see the Executive Board compensation that we were thinking more a couple of percentage points decline as opposed to something bigger.

And then the 2 factors that weighed on us was, first, in the first half, it was this tariff debate where the bookings were lower and then the factors Christian mentioned as a surprise towards the year-end. So also there, I would argue that we do have our checks under control, how it kind of translates. And when you translate the CCB of 2025 into 2026, please remember that the transactional dilutive effect is becoming smaller. It used to be close to 1.5 percentage point last year, and it will be diluted to less than 1 percentage point.

So I think the nice thing is about also when you look at the fill rate, I mean, what percent of the cloud revenue is adjusted for currency. There's always a problem that CCB is at point value, which was an extremely high dollar change. The other values are constant currency. If you really depollute that, you see the coverage with CCB for next year is actually quite good. So that also should give you some confidence about '26. And now going beyond that, I already mentioned what slight kind of reasonably means. You can then work from there and assume further dilution of the transactional part. So this is how I would think about it. So it's actually stabilizing in some way.

We'll take our next question from Mohammed Moawalla with Goldman Sachs.

Great. My question was really on the TCB. You talked a lot about sort of delays in recognizing some of this business into the CCB. We saw quite a steep decel in the sort of CCB growth versus a year ago, almost sort of 9 to 10 points. Is there -- I know there's kind of a law of large numbers here, but is there anything you can sort of comment on that's kind of going on here? Because you obviously talked about record number of larger deals. So I just want to better understand that dynamic.

And as a follow-up, could you update us perhaps on BDC and the momentum you're seeing? You've signed up a flurry of many partners. How is that sort of pipeline shifting? And what sort of do you expect in terms of contribution in 2026?

Yes. I can start with the total cloud backlog. I mean actually, you said it really well. I mean at the end, the number is getting, of course, much bigger. When you look at the absolute growth we put on top to Dominik's point, I don't see any other competitor producing similar kind of numbers when it comes to the total cloud backlog, not even close.

And this year, yes, there were larger customers coming. But when you look at the RISE journey when we started this 4 years back, actually, of course, we started with smaller customers, midsized customers. And now there are these mega deals. And that will also continue, and they will just take a higher share in the overall order entry of what we are converting to the cloud with RISE. So I'm actually super proud also given that, I mean, the TCB is always then also, of course, dependent on the contract duration. And that actually was stable. So we are not actually increasing TCB with longer contract duration.

We are actually increasing it by putting real business on top. And that combined also with a lower churn as we are closing more and more healthy business is actually, for me, a super positive sign also when it comes to the cloud revenue development, not only in 2026, but then also for many, many years to come. And then finally, when you look into GROW, I mean, still, it's a smaller business than -- compared to our installed base business. But just last year, I mean, overall, we won over 3,000 net new customers.

And that is, of course, midsized customers, but they will grow over time, the up- and the cross-sell I mentioned of the business suite. So while the share of midsized SME business became smaller because some of the larger transactions happened and will happen, still, this business is going really well, and we are adding a lot of new logos to SAP.

And maybe it's also worthwhile mentioning, if you look at the TCB minus CCB, which is basically the backlog year 2 and the following years, it's actually -- that ratio is increasing. So it gives you also more visibility in the outer years.

We'll take our next question from Mark Moerdler with Bernstein.

Sorry, I had my mic off. Want to -- didn't want to cause an issue. So I'd like to make sure that we're really clear on the CCB and I know it's been a lot of the questions put on it. Can you give some ordering to what you think was the most impactful for why the number was less than the Street might have expected? And can you also give us any sense on the economic impact of these sovereign cloud deals? Does it impact revenue lift multiple or margin in any way, shape or form other than that it may take longer for the deals to close?

Yes. I mean the one factor, Mark, clearly, which changed over the course of the quarter. And again, it's absolutely positive for the years 2027 plus. I mean we closed more larger deals. And then when you think about the phasing of such a deal, I mean, first of all, you negotiate on the business case on the [ ROI, ] the AI use cases. You think about, okay, what are the pillars which are really important. You think about, okay, do we go supply chain first, finance first, logistics, et cetera.

And then during the course of the quarter, obviously, you start also then phasing those deals. And as you then have many larger deals, and this was actually a quite significant shift, we actually saw that a lot of the revenue moved out from the first 12 months to year 2, 3 and 4. That was by far, by far the highest impact we have seen compared to October.

On sovereign cloud, actually, no, I mean the deal margins are almost the same, but what Dominik already alluded to I mean the deal negotiation per se, I mean we are running not only very mission-critical ERP systems, we are also running customers in regulated industries. And they have questions. They say, "Hey, what is happening if sanctions are coming? What is happening if there's an export control coming? What is happening if this AI data protection regulation is now redefined, et cetera?"

So these discussions take just longer than they have been a year ago, and it's a reflection of what is happening in the world. This is not a reflection of a demand issue. Actually, it's good. Look, I mean, I always also start these conversations with, look, when it comes to regulation, trust on one thing, SAP has your back. I mean we are spending over EUR 1 billion on localization, on regulations, et cetera. We are running these businesses in over 120 countries.

So we know how to adhere to all of these also new regulations. And I see this rather as a competitive advantage that we can clearly say, "Hey, look, no matter where you want to do business or where you want to expand your business, SAP will have a software and cloud solution for you in the different parts of the world."

And I'll also add there is the certification step, which is required in many countries. So we see new offerings on the Infrastructure as a Service for sovereign in various countries, almost mushrooming up, I can say. And there are certain requirements in different countries, and we now really see the first country certifying these products. So we are really at the -- in an embryonic phase of what could become something really big, but it's really happening because the capital is flowing there, the certification agencies are getting their arms around it, but it takes some time to groom and mature these projects.

Our next question comes from Toby Ogg with JPMorgan.

Just on the free cash flow guidance, Dominik, of EUR 10 billion, clearly well ahead of expectations and looks to imply a pickup in cash conversion. I know we talked through the year about cash tax, FX and the migration credit headwinds. Could you just help us reconcile these headwinds with the better free cash flow outlook and improved cash conversion you're now expecting?

Sure. I would say the upside is from 2 sources, partially from operational further improvements, which we have matured to a point that we feel comfortable guiding it now, but also from the fact that the delta on stock-based compensation between the P&L and cash is increasing. You've seen it already now in the numbers. You will see that in '25.

So it's a little bit higher. We've always said take the effective tax rate off the non-IFRS operating profit add back around about EUR 1 billion. So now I'd say around about EUR 1 billion plus. So that gives us a part of the upside. And the good news is that's sustainable. So that's the new base basically to jump off. Now on the transformation credits, it's always the game of the overall working capital. It's not the only item there. And from that perspective, for 2026, this is the best estimate we can give today. And yes, we also make sure that we have no mortgages in the future and can stick to that very simplistic formula with the noise that always will be there around the phasing of certain payments, but that's the trend.

Great. Well, thank you, Dominik, and this concludes our call for today. Thank you all for joining.

Thank you.

Thank you.

Ladies and gentlemen, the conference has now concluded, and you may disconnect your telephone. Thank you for joining, and have a pleasant day. Goodbye.

Okay. Good afternoon, and good morning. Fred Boulan from Bank of America Software team. We are delighted to be hosting Muhammad Alam, Head of Product Engineering and member of the Executive Board of SAP.

From a format perspective, we'll go through some questions together before opening up for Q&A. For Q&A, you need to e-mail me your questions, usual e-mail, and I will go through them.

So before that, I will go through a quick safe harbor statement. During this fireside chat, SAP will make forward-looking statements, which are predictions, projections and other statements about future events. These statements are based on current expectations, forecasts, and assumptions that are subject to risks and uncertainties that could cause actual results and outcomes to be -- to materially differ. Additional information regarding these risks and uncertainties may be found in SAP's filings with the SEC, including but not limited to the Risk Factors section, SAP's '24 annual report on Form 20-F.

Great. So welcome, Muhammad. Thanks a lot for taking the time to be with us today. It's been precisely a year since our conversation in the same format in our virtual tech trip. I guess perception software has gone full circle since then. So very good timing to catch up. So maybe, first of all, if you could start with a quick intro on your role SAP, your background and what's your vision for the SAP product suite?

Yes. No, sounds good. Thanks for having me here, Fred and excited to be back again. So again, my name is Muhammad Alam. I am the Executive Board member for product and engineering at SAP, and I'm responsible for effectively all our products, our engineering, our design, our product strategy from applications and our platform product as well.

In terms of our vision for the product suite, the applications that we have at SAP, it's relatively simple. We believe, and we've got a very simple point of view that the application data and AI flywheel is what generates the most value for customers, a seamless application, data and AI layer applications where users work on a daily basis, which create the data, data that's harmonized already across the end-to-end business processes, which allows AI to reason over it could be able to come up with the right recommendations and those recommendations then get embedded back into the applications that users use on a daily basis. And this flywheel creates the most value.

If you try to do this application data, AI flywheel in a disparate way, that means from an application perspective, you have to take the data somewhere else. You have to then harmonize the data, you have to make sense of the data once you've done that, you have to then sort of apply a disparate AI layer on top of it. Think about all the security, the privacy and the permissions that you have to apply to it, again, in a bespoke manner. And then once AI generates the recommendations, you have to plumb that back into, again, a disconnected application layer.

So the more this application data and AI layer comes together seamlessly, it creates more value for customers, not just in a vertical manner but also in a horizontal manner, meaning that if you have an app data AI flywheel from one vendor on procurement, let's say, and a different vendor on HCM and a different vendor on finance, that still isn't optimizing for the global optimum for the global Maximum, right, because you still have 3 different flywheels there. Where it creates the most value is the data across your business processes, finance, spend, supply chain, HCM, customer experience comes together, pre-harmonized across, which is what we provide with AI that can reason across the whole, not just in silos, you get the biggest outcome and the most optimal outcome.

And that, from an SAP perspective is what we're focused on in unlocking for our customer to make sure the application layer is one of the most comprehensive and complete in the areas that we play in. We're making sure that in each of the areas, we're best-in-class. And for each of these areas, they're integrated out of the box seamlessly for our customers, which has massive TCO benefits but then the data layer gets harmonized. And then, of course, the AI then reasons across the whole, providing a global maximum as opposed to a bunch of local optimum, if you will, that otherwise then customers have to manually plumb together to stitch the story, which while it can happen, it will still be from a value perspective, far less than what it would be if it just comes seamlessly across those 3 layers. So that in core is our application strategy. And our product strategy in each one of those 3 layers, we want to make sure even though we are very comprehensive, because I said, we have finance, supply chain spend, HCM, CX and then we have the data layer and the AI layer that we know customers don't only live in just an SAP-only world. So at the application layer, we have our business technology platform that allows you to both integrate with non-SAP applications as well as extend and build your own custom applications and ISV solutions.

On the data layer, we've put partnerships with Databricks and Snowflake and Business Data Cloud that allows you to bring in or zero copy share out to other data assets in your organizations, again, harmonized with the data products and the data model for some of your most mission-critical applications such as finance and supply chain and the AI layer with Joule and our agentic capabilities, you obviously cannot just reason over SAP applications and build agents but you can connect them with non-SAP applications and build agents that work across non-SAP applications, too.

So we want to make sure we obviously are best-in-class in the SAP landscape that we play in but not just that but that we can expand into and provide the holistic needs for our customers. So hopefully, that makes sense.

Great. So maybe -- I mean, you've covered all those things. Maybe at a high level before we go into some of those items in a bit more detail. But how do you think AI will change or transform the way customers -- the kind of product that they want looking at it on a multiyear view?

Yes. I mean I think it's -- I mean this is sort of the biggest question out there, right, as to how -- what -- in everybody's nonexistent crystal balls, what would the world look like from a customer [indiscernible] perspective. And we actually outlined a very -- again, a very simple point of view that resonates and it's informed by a lot of customer conversations. The way we think about this is there's sort of at least 3 large steps you can argue. Today, what we want to do and enable our customers because largely, if you think about customer organizations, it's humans, it's people that are running those organizations and there's defined role constructs and functional responsibilities that people are working on.

So we want to make sure, first and foremost, we're making those roles, those people more efficient, more productive and smarter with our AI. So we've outlined this concept of AI assistance in Joule that brings together end number of agents or AI capabilities but let's say, for your accounts receivable agent or your controlling associate or your customer service agent or your planning associate and supply chain. We want to make sure that there's an AI assistant available for them that makes them again super productive, far more efficient, and they can understand and reason over things that they wouldn't have been able to do as humans because the data that now everybody has access to is massive. So AI assistance is first and foremost.

Let's make the people a lot smarter, faster and efficient. As then these people develop more trust in the AI and the AI becomes more complete in terms of the functional things it can go do for that role, we step into the world of autonomous execution, right? So as the role -- as the person becomes more confident, they can say, hey, this set of customer service -- customer service requests that are coming in, I'm actually okay with the AI agent taking them, reasoning over them, finding out what the right response is coming up with the response, sending the response to the customer, waiting for feedback and closing out the case because now my trust level is so high that I can let that process run effectively in a touchless manner or in an autonomous execution.

So we believe these assistance, AI assistance over time will lead to autonomous execution in different functions of the business, and we're seeing proof and results of that already. Customer service is a great example. We think about that on the financial close side. We look at some of the stuff on the supply chain side where you can take a demand signal, understand what the impact on the supply is going to be and come up with the right request for how do you need to make sure that you can meet the change in the demand, if you will. That's the second part. So make humans and the people better, then you get to autonomous execution.

And the third part is deep research. Like today, there's so much data. You need to be able to reason over that data, uncover things that would take humans either a very long time or be very hard for them to be able to go do to come up with the right strategies and the right recommendations to either increase top line, reduce costs or be more resilient. So this combination of AI assistant, autonomous execution and deep research is what I think organizations over time are going to continue to rely more and more because it makes their function more efficient. It allows them to do more through autonomous execution and uncover the deep 10x, 100x opportunities through deep research, if you will.

Okay. Excellent. The big debate out there has been around Agentic AI and a lot of concerns on how if you prove that logic, that can replace scale SaaS. You've been advocating the kind of opposite but it would be great to have your kind of your vision in terms of how Agentic AI will potentially impact SaaS and more specifically SAP.

Yes. I mean I think, again, this is one of those bigger debates, if you will, that are happening through the course of the year as well. And our point of view, again, is very simple and informed by customers that SaaS, listen, I think business applications will be reimagined with AI and experiences. There's no doubt about that. Now the question is, what would that reimagined set of business applications would look like? And we believe that there's at least 5 patterns, and we'll see more patterns evolve in terms of how the applications would change. And most of these patterns that you will see, there isn't -- there isn't a place where the application itself goes away because where does the execution happen? Where does the compliance and the regulation and all of that stuff happen. But these patterns range from you have an application today, this is pattern #1, where now AI is deeply embedded already.

So anything that you do is enhanced and enriched by AI to make you more efficient. A simple example of this would be take your SuccessFactors, your HCM application. Now you need to do performance feedback or job description, AI can help you do that far more effectively if you want. The second pattern is agents. Agents will automatically be part of these applications that allows you to do certain functions and make those roles that we just talked about far more efficient. So on the finance side, for instance, the application is still there but there are agents that can help you do accruals management in a much more continuous manner and effective manner than what humans would be able to do. Agents that can do dispute management when an inquiry or dispute comes in for your open AR or AP and come up with the right recommendation. So the application gets enhanced by agents that are just available.

Now the third case is autonomous execution. As we discussed, let's say, customer service, for an example, the application is still there because you need to sort of work through have the knowledge base, have the cases come in and all the warranties and things like that. But an agent can look at the case and be able to go resolve it all the way to the end in an autonomous fashion. Now a certain percentage of cases, a human might still need to touch because they're more complicated and things need to happen.

The fourth pattern, it becomes a little more interesting, which is what we call the app-less experiences, right, where the application is still there but does a user need to touch the application. We believe that in certain classes of application, that will probably go away. So let's take expense management in Concur, for example. Do I really need to ever go into a Concur expense application to complete it? Probably not because an agent can see your receipts coming in from your credit card provider, understand the policies, reason over it, sort of prepare the expense report because it knows when you traveled out, when you traveled back and all the things that happen in the middle.

It may ask you a question proactively through Teams or something that says, hey, the dinner you took. Did everybody attend that was in your calendar, so I can complete the expense report and you say, yes, and it asks, should I go submit it like a human assistant would and then the agent can go submit it. If there's more complicated, sure, you can get into the application and do it. So we do believe there's going to be some app less experiences like that. But again, that doesn't mean the application doesn't exist because there's tremendous amounts of regulatory compliance per diem requirements that are different by region that need to exist as well.

So that reasoning engine still needs to exist. The agent just creates a far more better experience for you as a user. And then you maybe get to the fifth pattern, which we're calling the no-apps experience where you can either effectively by put or generate an app or get the information you need in a conversational experience without needing an app in the middle. We do believe this pattern we already see as well. But this pattern, we think and we're seeing largely applies to where people were already building low-code, no-code applications. So it will impact that area first or custom applications. So if you're going to go build a custom application, do you need to go build it or can you generate it through bite coding, use it, discard it and then keep going as well.

So we'll see all patterns, and we're seeing all of these patterns in our applications as well, and we're building that, too. So in most of these, as you can see, the application still exists, if you will. Now we do have a point of view. Like if you go back to the product strategy part, right, Fred, we said, listen, these app data AI flywheels can't exist in silos in a single landscape because that effectively means somebody needs to then still bring them together. So what we do believe how the SaaS landscape will change is we will see that the best-of-breed players as they're called today struggle to find relevance in the future. Because if you're just a procurement application SaaS provider, and certainly, you can put some AI on top of procurement but you're leaving a pretty long last mile for the customer to take the data still somewhere else out there, match that up and harmonize it with finance and SCM and others, apply AI with its own custom security model that cuts through your application because a user never only lives in procurement, right? They do multiple things.

And where a customer can get a broader end-to-end business process suite, which has, again, best-in-class capability with data already harmonized, it would just make sense for a customer to go do it. So we do believe there is going to be a disruption in SaaS but it's going to lead more to first the stand-alone single lane players, as I call them, they will struggle first to find relevance because the TCO for the customer to now take them on, integrate them, pull the data out, apply AI, make the AI consumable is just not going to be worth it from that perspective when the capability differentiation isn't going to be there as well.

Now the third dimension for this is you can argue from a business model perspective, users and user-based licensing. And if you think about us from an SAP perspective, less than half of our cloud revenue is actually user-based. Most of it or a larger majority of it is more outcome-based. And what we are going to continue to see the shift as we move from AI assistance to autonomous execution to more value-driven stuff that the shift to outcome-based is going to be more and more. And outcome-based it could be a bunch of business document metrics or it could actually be value generated in terms of value realized in terms of efficiencies or savings as well.

So those are the things that we're also continuously evaluating. But from a journey perspective, just specifically on the business model with the world of AI and agents, we do believe it's going to shift more over time to outcome-based. We're already less than half from a user-based perspective. We don't think users are going away because a lot of organizations have a lot of work to do. So it might still be the same users but more outcomes. So it will shift more to outcome-based.

Okay. I mean that's a good -- very good question we get a lot around the business model. I mean you're touching it. So I'm just going to just maybe go a bit deeper on that. So if you think about traditional SAP users, I mean, some of them are still seat-based, especially in the legacy on-prem world. But if you think about -- I mean, when you have this conversation with clients on that kind of change in business model, is this something they're open to? Do you think there's an outcome where it's actually a net positive from a discussion perspective? And the concern out there is that you will need -- if your products are so much better and more efficient, you will need less users of SAP, right? So we'd be keen to understand how you see the equation playing out.

Yes. I mean I think -- listen, I think there's no organization that have come to me that says, hey, as long as the value is there, I'm okay with the value or an outcome-based measure or metric. I think the premise is can you create the value? And we believe we're in a very strategic spot with having core finance systems that we run, core supply chain, core spend and others that end-to-end, we can create the value. And as you spoke about the value, then sort of an outcome-based business model resonates from our perspective, generally well with customers. It doesn't necessarily matter if it's more users or less users as long as the outcomes are there.

So I think in general, the conversation that needs to be had is what's hype versus what's value? Can you realize the value from it or not? And as long as the value can be realized, there's good precedence that customers are willing to pay a percentage of that from a price and a cost perspective to whomever is sort of generating that value. What we see today broadly in the market is, again, the talk of the AI hype and the bubble that's there and what it's going to lead to and the billions of agents, like none of that is a reality for most customers, if not generally a large number of customers. I probably stop sort of saying all customers.

Now it's -- when you talk about the hype without realized value, then it becomes a tougher conversation. But as long as there's confidence that you can create the value, as we just discussed in the 3 questions, we haven't seen pushback on it.

Yes. Great. Maybe talking about the product and the kind of business AI vision that you've been kind of laying out. Can you update us on where you are on the road map in terms of agents, I mean, off-the-shelf agents you've been pushing? Any metrics around traction and adoption of the business AI proposition?

Yes. I mean I think I'll share, I think some of the metrics that we've shared publicly here as well and just some more broader statements. I think today, we have over 34,000 cloud customers that use SAP Business AI. We have shipped over 400, what we call premium or embedded AI use cases. We have about 40 Joule agents that we've shipped and over 2,100 Joule skills to our customers. Now I'll sort of put this a bit in context. And then we've also outlined now these rule-based agents for your accounts receivable, your accounts payable, controlling and things like that as well.

I think I'll say this here openly, I guess, to this group, is 40 agents might seem when folks are talking about hundreds of agents or thousands of agents, 40, like is that super impressive or not. But this is where I think going back to the hype versus reality, the noise that's out there. I mean 600 agents are great, but the question is who's really using those 600 agents. So we're taking a lot more customer-driven, customer-first pragmatic approach to say, hey, the agents really need to create value that is landing and we're working with customers to deliver as opposed to let's just call everything an agent, identify things and say we've got 1,000 agents that as a customer, you struggle with to say, well, how does it relate to my accounts receivable department or my accounts payable department, if you will.

That's why I think the feedback we get is, listen, I think you're probably one of the more pragmatic ones to say 40 agents but these 40 are the ones that are hardcore agents that deliver the value that we're working with customers to be able to deliver on top of the embedded cases that I don't know if you want to just go rebrand them as agents, but those obviously also provide acceleration and value from an efficiency perspective to our customers or the 400 embedded use cases and so forth as well. We also now at our SAP TechEd event talked about our new RPT-1 model that allows you to do some predictive functions and capabilities on tabular data, if you will, in a, I'll call it, a generative AI manner, if you will, which historically would have required a lot of data scientists and machine learning folks, if you will, in an organization.

So we're bringing the best of the generative, the tech space to now the tabular data with the RPT-1 SAP foundation model that we talked about, that reasons over numbers to again, create more value and adoption from an AI value and adoption perspective from our customers too. And then the Agent Builder, which allows customers to build their own agents outside of the ones we're shipping is also something we're working with many customers on now, and we believe it goes GA here at the end of the month.

Excellent. A question we get a lot on Gen AI is what's your approach around monetizing this? And maybe you can split it depending on the type of patterns you kind of laid out earlier but it would be great to understand what is your commercial approach, the different models you've been testing. I mean I know there's been a lot of discussion in the industry about how we get customers to reward us for the value we deliver. So keen to understand the approach you've been taking and anything you can share in terms of kind of what you've seen from a -- specifically from a customer standpoint in terms of monetization?

Yes. I think there's 4 or 5. And we can go through them. I think there's some super simple to understand. I think the first one is there's a set of AI capabilities that we believe are just now part of the core application because it realizes this vision of application data and AI seamlessly bring together into a singular experience. Those are just part of the application that if you have one of our applications, you get those AI capabilities and you can get value from it. So that's part of our core applications.

The second set is and this is where we did a shift that we announced with Sapphire that's resonating well with customers is, we said, hey, there's a set of agents and AI capability that based on the workload, if it's finance, spend, supply chain, HCM and others, that you get in a per user per month basis where you can use as much of it as you want. And we're going to continue to add more AI capabilities and agents to our finance per user per month AI package, if you will, and you can use that across. So that's simple, that gives predictability to customers to say, hey, if you're using our spend applications or finance and others that the innovation that we've shipped and the innovation that we'll deliver is going to be part of this. And majority of our out-of-the-box agents are part of this as well.

So from a customer perspective, it gives them predictability. It gives an ability to see plus. It allows us to continue to drive innovation without having to charge for every single innovation for every customer or a customer to worry about, well, this new feature is it going to cost me more or charge me. It could be like a runaway usage.

Now the third part is there are certain cases that are more consumptive in nature. So for that, we have AI units. It's more of a consumptive monetization package that says, hey, the more AI extraction you do from documents, the number of documents, it obviously is more of a consumptive one in nature, and you can obviously go forecast that out based on the documents that you want to go. So we've got AI unit-based model as well for the consumptive type scenarios that don't lend themselves well for any of the first 2 models.

And then we have custom AI that you can go build as we talked about with Agent Builder and business technology platform. And of course, it's custom and it follows a consumptive AI model from that perspective. And then what we're -- those are the models that we have that seem to be resonating well with our customers, both from a simplicity perspective and then creating value.

Great. So now maybe moving on to the competitive position. So you've kind of explained why you're saying that kind of street approach will differentiate SAP. We keen to hear how SAP is differentiating around AI integration. In particular, I would be keen to hear your thoughts around your direct competitor on the ERP side with OCI. I mean do you think this is emerging as a new type of competition from that standpoint? Or do you think it's just a different debate and the approach you have in terms of application layer will enable you to remain and to keep that differentiation?

I think specifically as it relates to Oracle, I mean, I think our point of view is a lot of where Oracle's focus and growth is coming from is from the infrastructure side and the partnerships that they're sort of driving there. While where our core focus continues to be on the end-to-end sort of cloud applications, end-to-end business processes and the industry depth and this app data AI layer that we create. Now they have that as well. But for us, this is our core focus. We're not getting distracted by now building data centers or on the infrastructure. We have that in strategic areas but we also have what we believe is also an advantage for us through our business technology platform that we can -- we partner with hyperscalers and are able to deploy our stack and our app data AI value layer in any hyperscaler in the regions that are wherever the customers' choices, if you will.

And that gives us both a level of flexibility and an ability to then really focus on creating value at the top of the stack as opposed to going to the infra layer of the stack from that perspective. So that continues to be our focus, and we believe at the application layer, it's a different debate, as you said. I think in the infrastructure one, our point of view is clear.

Now what I would say I think from a competitive differentiation perspective, I think you alluded to what we discussed already that if you think about -- we'll see a level of, I'll call it, commoditization or simplification in the application layer. We believe, as our customers are telling us is it's going to be towards more of the suite players as long as they can deliver the best-in-class capability integrated out of the box because the value in the AI then gets realized more easily. So the best-of-breed players will struggle is what we're seeing and what we think will happen from a competitive perspective. Our value proposition, our differentiation there is very clear. Like if you want to build a supply chain plan, of course, you need financial data. If you need spend, you need to be able to look at it holistically.

Now when you look at against suite players, I think the things that also differentiate us from that perspective is our business transformation suite, right? So if you look at Signavio that we have that looks at the process of our customers and what you can generate both from a process mining and now an agent mining perspective that we've talked about as agents and innovation that we're landing, it gives customers this frame that thinks about -- because customers think about their businesses from an either industry lens or a process lens and Signavio gives us the significant advantage that on top of the application, we've got the world's leading process mining and process modeling situation that with the right agentic experience and the right agent logs can give you better insights as to which agents are working well, which ones are not.

That gives us, again, an area of differentiation. If you think about with our digital adoption platform with WalkMe that sits across, again, not just the SAP applications, but your entire landscape. And when you look at some of the computer use AI that's out there and what you can do with Joule action bar, it gives a level of differentiation from an AI perspective that certainly has the deepest integration with SAP applications but through WalkMe, it's any application and can transcend all of that, if you will. So we believe that's, again, a very different -- sustainably differentiating area for us because WalkMe is also, as hopefully most of you know, one of the leaders in the digital adoption platform.

You think about LeanIX, which already has a view of your enterprise architecture and your landscape being the AI agent hub alongside understanding your data, it gives us, again, a significant advantage to say, hey, not just agents with SAP, but we have a full view of our entire enterprise landscape and architecture and where we're leveraging agents and how. Now like all of these things alongside you look at Business Data Cloud and our partnerships with Databricks and Snowflake that gives us sort of the leading data platform and data warehousing capabilities with our data and what you can do. I think creates a level of differentiation that we believe adds up to a very differentiated value for customers.

Okay. Great. I've got a few more questions, but I also have a number of questions from the audience. So I'll take those later. I don't worry, I have them. Maybe moving on to the cloud migration. If you think about your current SAP ERP customer base, can you give us an update on where they are on that journey to the cloud? And what's the strategy to accelerate that migration?

Yes. I mean I think, listen, I'll share some of the numbers that I believe Christian and Dominic have already shared in the past that from a -- our peak support revenue was $11.9 billion in 2022. We still have roughly over $10 billion, $10 billion to $11 billion in 2025. So we still have a massive runway left for ERP cloud conversion with revenue uplift sort of coming in the years. We see the pace to migration to the cloud continue to increase. So the growth in going into our RISE and Grow offerings continues to increase. We already see about 1/3 of our ERP customers have initiated ERP cloud journey. And then from there, obviously, there is a significant upsell, cross-sell that happens too. So that's sort of where we stand from -- maybe call it from a numbers perspective.

But then we're also investing significantly, as we just talked about, you look at our business transformation, how do we help customers get to the end state in a predictable manner. We're also investing heavily in AI-driven migration tooling like Joule for Consultant, Joule for Developers as well as migration tooling that help make this cost and effort for our customers to get to the cloud landscape in a much more simple and predictable manner, not just obviously, the tooling, the transformation suite and AI, we're also investing heavily from an SAP perspective with architects, enterprise architects that help our customers through that journey with the right decisions on the enterprise architecture side, on the data architecture side and on the business process side as well. So we're -- we've got a heavy focus in supporting our customers through this accelerating migration journey that we see of our installed base.

Great. And maybe touching on Business Data Cloud. I think you -- that was launched for general availability in April. You've been talking about a very strong traction and pipeline on that product. So it would be great to have an update around the adoption, the ambition for the product and maybe help us understand the economics for SAP with Databricks.

I mean at some level, and I think I might get into trouble. I wish I could share some actual numbers here but I'll sort of leave this to our Investor Relations and Dominik and Christian to share at the right point. But we do see a very significant uptake of Business Data Cloud, both the pipeline as well as the customers we have. And not just that, the usage of the customers we already have in the 9 months is all very encouraging for us as one of our fastest-growing products in the recent past. If you will. I think one obviously, data point that you can look at from -- as a validation of the success and product market fit it's finding with our customers is the partnerships that we just continue to announce. And the ones we've announced, I think, are in numbers, far less -- we've announced quite a few, far less than the numbers that we're already working with other partners to go announce because the partners are also lining up to say, hey, listen, we really want to be a part of this. So Obviously, Databricks was our leading partner, and we've got a special relationship with them as part of their business data cloud.

But then we also announced our partnership with Google BigQuery as well as Microsoft Fabric from a zero copy share perspective and then a deeper partnership with Snowflake as well, which has not just the zero copy share but also the ability to use Snowflake as a solution extension solution on SAP as well because we've got that feedback from a lot of our customers. And then we're working, as I said, on many more partnerships, too, on the data side, not just data platform providers but enrichment providers and providers that have data assets like we announced the Adobe partnership that brings sort of the marketing and the customer experience data that we have with the financial and supply chain data that we have. So that continues to also go really well.

In terms of the economics, I mean, I think the way our customers see it and the way we've sort of positioned the economics of Business Data Cloud is, in my mind, very simple. And I look at this both as a TCO reduction and a TCO shift from services to SaaS. And I'll explain what that means. And that's why it's resonating so well. From a TCO shift perspective, it's not that the customers aren't spending significant amounts of money in taking data out of SAP applications, either through tooling that they've bought or SIs that they've engaged and they engage significant either internal headcount or SI headcount to harmonize that data then across SAP and non-SAP applications. And not just that, that's -- you can argue a onetime effort but it's not because you have to then ongoing continually maintain that data asset in this disparate manner.

So the value proposition for Business Data Cloud is to say, listen, all of this stuff that you bought bespoke, disparate tooling had a lot of humans and SIs harmonizing the data, managing the governance of it is now available as a SaaS service. So the data products from SAP applications are now available, managed, governed with semantical richness that you otherwise just wouldn't have been able to go do anyway in the previous way and the quality -- and I guess the harmonization of it across it, like all of that is now as a service. So on one hand, it's a shift from what you were doing, call it, in the services or a tools-based heavy manner to a SaaS experience that now comes with SAP, if you will.

And if you look at it from a TCO perspective, they -- the value is far higher. And I would argue, in most cases, the TCO, ultimate TCO from a customer spend perspective is actually lower for them to be able to get a product that's higher value, if you will. So that's the economics that's driving it. It's a shift of, call it, services to SaaS that sort of increases the TAM, if you will, of where we go out. And in a way where it also expands that to say now with our partnerships with Databricks and Snowflake, and we can also have a larger data purview with zero copy share for non-SAP data or non-SAP data in Business Data Cloud as well as some customers are doing because they'd rather have a single data platform to really then drive the AI layer on top of it. So it also obviously feeds into the multiplier effect of AI because data is the fuel for AI, if you will. So hopefully, that provides some context.

Okay. That's very clear. Maybe one question around the upsell journey. So I think one interesting data point from the Sapphire was that you've pretty substantially upgraded the upsell expectation you have when customers switch to the cloud. And I think there was -- the big delta is product innovation, which is really AI. But you still have that kind of more traditional modules like BTP. So it would be great to understand like a concrete case of where you see those upsell as customers move to S/4 in the cloud. What's going on in terms of upsell journey, most popular modules that are being consumed and how you see that -- those cohorts developing once they've moved to the cloud?

Yes. I mean I think it's -- again, the story is very simple as well. Cloud ERP as sort of the leading anchor leads to an upsell from -- let's go back to our app data AI. It leads to business technology platform as that natural extension platform to make sure that you can meet the unique needs of the customer. So BTP continues to be a big one, both from an SAP build perspective and integration suite. And then certainly, business data cloud, right? Because as you move into the new environment, you need a way to be able to sort of have the data in a managed governed manner so you can apply AI and then AI on top of it becomes, again, not just a natural upsell but we're seeing that now as the bigger pull as well, increasingly as a reason to go to the cloud as well, so you can unlock more value too. So those 3 continue to be the very logical ones, if you will, as an upsell.

But then certainly, from a cloud perspective, we also see significant expansion into this end-to-end business process context as well. So we have our supply chain applications. We have Ariba on the procurement side, where we announced the next generation of Ariba built natively on BTP, seamlessly integrated to our cloud ERP. That's getting very positive feedback. We've got SuccessFactors. And we're seeing great momentum on our customer experience applications as well. So that continues to be, again, a cross-sell capability there from that perspective.

And then certainly, as part of the journey, we talked about our business transformation management suite as well, which is, hey, as you go through the journey, LeanIX creates significant value for you in understanding your landscape. Signavio gives you that process map of your organization and WalkMe gives you that adoption platform. So this whole story sort of comes together in a nice way from that upsell and a cross-sell standpoint. And in each one of them, as we talked about again at the very beginning, if you step back and think about it, right, in each one of these, if you go back to that simple frame of app, data and AI, right, because apps is where you have our cloud ERP, all the adjacent line of business applications. You can also put our BTM suite there as well, LeanIX, Signavio on the data side of the business data cloud with SAC, with what we're doing with Databricks and others on AI.

You have a significant opportunity in what I would call workloads that we may historically have not played in as much. So on the application side with build and BTP, you can build non-SAP applications. That's obvious. On the data side with our partnerships, we're seeing non-SAP data is one that is coming together in certain customer workloads to say, hey, we need a data platform that's cohesive. On the AI side, you can use AI now to stitch together not just AI and SAP applications but AI across your broader function because generally, the core of an organization is what SAP powers, right? It's your finance or supply chain or spend, but there's a lot of adjacent applications.

So we're seeing AI being sort of this umbrella thing that pulls even more usage and capability through Agent Builder and others in non-SAP workloads, too.

Great. Question -- I got a couple of questions from the investors. I'm just going to go through them. Firstly, in terms of M&A, any capabilities that might be missing in the overall value chain? And in particular, I mean, there's a bunch of questions around the merits of bringing some partners in-house. And so maybe you can help us understand to what degree, for instance, integrating SOLEX makes sense from a data standpoint, data understanding, data monetization.

Okay. So I think I'll answer it a couple of different ways, right? I think the first part of the question would be, I think we're always looking at strategic M&A targets, if you will. We just recently completed SmartRecruiters. So in our HCM portfolio, we felt, hey, this is an area we needed to bring in a best-of-breed player that our customers are asking for more innovation, and that has landed really well, and we're seeing some significant positive momentum there. Similar to that, we continue to evaluate across all of our domain areas, if you will, and where do we need to go shore that up. We have also done some interesting partnerships as you think about in the customer we launched a loyalty management solution but that is partnered with a provider that we've OEM with significant expertise and SAP integration and data capabilities on top of it that's resonating very well. It brings us the maturity of a product with the customer base that completes our story on the commerce side, and that's again resonating really well for us.

On the -- and we don't -- we're looking at M&A not only just across sort of the application side, we're obviously looking at it again, just continue to use the frame of app data and AI. If there's things on the data side that we need to go do besides the partnerships that we're doing, obviously, for us, continuing to bring in the world's leading data engineering platform with Databricks as an OEM seamlessly into BDC is resonating very well with our customers as we see. We also saw that in this space, there were a lot of customers that had already bet on Snowflake as a platform. So they wanted that option as well. They get the full value prop of BDC.

And this is where our openness is what drove us to say, listen, we want to bring this because it's a customer choice already, Snowflake into a fold as well. And we did that as a SOLEX because obviously, then if customers have that option, they can go with Snowflake as well, but Databricks continues to be deeply embedded part of Business Data Cloud. And then the zero copy share solutions as well in AI, we've announced significant partnerships as well. We announced one at TechEd with 8N that comes together and brings their agent building capabilities with us, too.

So I think -- I guess that's how I would answer it. If there's any specific aspects, I'm happy to go deeper into it. We also did in the procurement space, if you notice, we've made a decision early on to say, hey, contract management as a space has got a lot of sort of best-of-breed domain level innovation that's happening, and this is where we partnered deeply with Icertis, and we've now launched a SOLEX solution with Icertis and not just that with our next-gen Ariba application, we're actually providing a seamless singular experience on top of Icertis and our procurement platform that from a customer perspective, they wouldn't be able to differentiate whether they're using Ariba or Icertis but they get the value and the power of 2 organizations innovating aggressively in each of the domains as well.

So we'll continue to look at how do we meet the customer needs and the gaps in the white spaces that we have and where there's opportunity to go to.

Very quick follow-up. When you have a SOLEX agreement, do you actually have access to data or it stays at your partner? If you're offering?

So on the -- so it depends on the nature of the SOLEX, if you will, and how the integration with the applications have set up. I guess that would be the short answer. There's no single if you will. It depends on the nature of the integration that we've built with the provider.

Okay. A few more questions from investors. I'm going to go through. I know we have taken almost an hour already but a question around how SAP is using AI internally. I mean, I guess it's a big topic from Sebastian and is part of the operation but keen to hear from your standpoint on the engineering product side, et cetera, is there further you can do in terms of OpEx and cost management after the kind of big efficiency ramp-up we've seen in the last few years. I mean, looking forward in your organization, it looks like the velocity of innovation on AI is accelerating. To what degree you can leverage that internally?

Massively. I mean I think we've already realized significant benefits, as you can see in the ratios that we have out there, and we continue to expect the ratios to generally continue to trend in the direction that they have been in the recent past, if you will. But alongside that, I personally know and we're working towards that, that there's a massive uplift in throughput that we can still continue to derive with all the AI toolings out there. So we are looking at every -- effectively, you can say at some level, it's a little bit of a hyperbole because you can't really do every single one. But every single sort of new tech that's out there that affects software development, not just in development, but product management, design, UA, QA and how can we bring that in and scale that to our 35,000, 40,000 colleagues in product and engineering to increase the throughput, if you will.

And alongside that, we also continue to look where do we optimize, where do we focus the right skill sets on. So I expect our throughput from an innovation perspective continue to skyrocket, if you will. And at the same time, our ratios continue in the trajectory that we've sort of articulated with you all.

Cool. Next question around your mid-market offering grow. It's been also kind of a bit of a new growth area for SAP, not really the kind of core of the proposition historically. Can you discuss a little bit where you are in terms of the product, the availability, the go-to-market on that, the momentum you're seeing in growth?

We're seeing a really tremendous momentum in our -- I'll call it our S/4HANA public cloud solution, our SaaS solution, if you will, that has now finance a supply chain package, supply chain premium and finance base and the uptake that we're seeing is it's, again, one of our faster-growing areas, if not the fastest. I don't know the actual numbers in my head to be able to lay that claim here definitively. But what we do see, of course, is a significant uptick in net new from it, which is what the design principle was in the grow motion in the mid-market space. But we're also seeing a very real interest from what I'll call the upper mid-market or lower enterprise that says, hey, we really want the value proposition of a true SaaS cloud ERP, and we're willing to go through the business process change and reengineering to be able to benefit from having a SaaS cloud ERP application.

So we're actually, from a product perspective, continuing to add a lot more capability. But then at the same line from a go-to-market perspective, we're also spending a lot of energy and investments to continue to, if you will, fuel the machine and meet the demand that we see out there alongside significant partner motions and ISV motions that sort of completed the tool. This is an area, of course, we're also seeing significant amount of AI adoption, too, because the app, data AI comes seamlessly together. So things about -- think about the agents or the AI experiences in the core ERP applications that are just there natively for you enabled in the next update without you having to do anything. We're seeing some very significant usage there.

Excellent. Well, I think we're up on time. Look, I think you've given us a very upbeat assessment of your, I guess, perspective around cloud -- well, the whole AI side in general, BDC grow. So lots of areas of growth for you. So exciting space, a lot of partnerships. I mean, you mentioned Microsoft Fabric, et cetera. I mean there's a lot of innovation around BDC.

So look, I mean, thanks a lot for giving us some of the time, Muhammad. It was a pleasure to have you on our field trip. And yes, I mean, talk to you soon. And for our audience, we -- stay tuned, we have -- we start again in 5 minutes. Thank you very much.

Thank you, Fred. Thanks, everyone.

Okay. Good morning, everybody. Thank you for joining us today. I'm delighted to have SAP's CEO, Christian Klein, with me. Before we start on the fireside chat, just a quick safe harbor statement to read through. So during this conversation, SAP will make forward-looking statements, which are predictions, projections or other statements about future events. These statements are based on current expectations, forecasts and assumptions that are subject to risks and uncertainties that could cause actual results and outcomes to materially differ. Additional information regarding these risks and uncertainties may be found in SAP's filings with the SEC, including, but not limited to, the Risk Factors section of SAP's 2024 annual report on Form 20-F.

Well, with that out of the way, Christian, welcome to Phoenix today. I want to start with a look back. So it's just over 5 years actually since you announced RISE. A lot of analysts and investors, myself included, saw it as a brave move with quite a lot of risks around it. But I think it's clearly been very successful. Casting your mind back to your expectations in November 2020, what do you think has been the biggest single factor that's gone in your favor or you've executed perhaps the most well upon?

Yes. Thanks a lot. First of all, Michael, for having me. I mean, being in Phoenix at that time of the year is actually not a bad thing, I have to say. Yes, look, 5 years back and now sitting here with some more quay has -- I mean, definitely, it was a ride. I would say, indeed, it's -- it was a very successful ride. I mean the biggest achievement, I mean, you probably are not surprised is technology-wise, it was pretty clear. The product strategy was pretty clear. I mean the co-CEO model broke back then, but I was super clear that SAP can't compete only with this many, many best-of-breed competitors when we are running businesses end-to-end that we are smarter than others to understand how to connect an omnichannel sales to the warehouses, the inventories, the commercial models of our customers. And so that was pretty clear.

I would say that the biggest achievement was around on the people and the cultural side. I mean, to turn such a company around who was then successful for 50 years in coding a very successful ERP, you have to communicate a lot. And not only the investors were a bit skeptical back then, but also, of course, also people, they needed a belief. And when you have to start and share price is not doing that well, I mean, obviously, there was skepticism back then. But we turned it around. And the good piece is there is no debate anymore about cloud. I mean, think about our customers. I just said this morning, a customer and he thanked me 3 times that I forced him to the cloud because now he gets all the AI, the last-mile delivery capabilities of our supply chain in the cloud on the latest release. So at the end, I would say the cultural change was definitely the biggest achievement we actually had at SAP.

Okay. And if we think back to the financial targets you communicated around then, I think EUR 22 billion in the cloud, that included Qualtrics. So if we take that out, actually, you're heading a little bit ahead of that. But implicitly, maintenance or support was somewhere below EUR 9 billion, and it's maybe 20%, 30% above that today, which is quite an achievement. Intuitively, that would suggest you've been more successful in cross-sell and upsell, but the migrations have gone more slowly. Maybe you can elaborate.

Yes. For sure. I mean, there were some lessons learned. I mean, first of all, it's great to see my former CFO, Luka Mucic tells me question. I mean, look at 5 years and how close we really or even nailed all the targets we have set ourselves. This is HANA. This is our analytics. But now kidding aside, I would say there are 2 factors to it, why we are also seeing still the sticky support revenues. First, I mean, the world has changed. I mean you see the war in Ukraine, the cybersecurity attacks. There's a lot of new regulations coming and our customers understand really well to get support from SAP is a high asset. So there was actually the churn is the lowest ever. I mean there is hardly no churn. I mean they really appreciate when a company says, hey, in over 100 countries in the world, we're going to make sure that you are running a compliant business.

Second, clearly, on the cloud side, I mean, the BTP was a lonely island back then, a platform without any apps. And imagine that happens at SAP, that tells you something. And now there is actually a large developer community and not only an ABAP community, also the next generation is now joining, building, we come, I guess, to that in a second, also now AI agents on the platform. So also the multiples, what we gain out of RISE was way higher because we see better upsell, better cross-sell, first of all, in the platform business. But then second of all, I mean, of course, now we are seeing when the customer on the RISE journey 2 or 3 years, I mean, they definitely expand now the SAP footprint.

I said it before, an ECC on-premise customer of SAP, I mean, despite having an ERP, didn't necessarily use all of the modules in an ERP on-prem. So now we are expanding from finance oftentimes, we landed, went into HR. Payroll is anyway SAP, even if you buy your stuff from Workday, guess what, is an SAP payroll behind. So we are going into ECC, and now we are going into the supply chain. And when you then think about AI agents who need to collaborate cross company because this is only how you get really extract high value out of that. That is, of course, now another lever in the RISE journey to really bet more on SAP line of business solutions than probably going with a best-of-breed vendor here in some areas.

Okay. And I talked about sort of a 5-year look back. We're not quite there yet. So on the last earnings call, you sounded very excited about the health of the pipeline for Q4 and beyond, and you would be disappointed with your words if you only exited the year at 25% CCB growth. Obviously, a few weeks left, but can you give us an update on how your conversations are going with customers? And thinking beyond Q4, I think an important question, given the sort of imminency of the RISE completion, do you ever believe that current cloud backlog can accelerate again?

Yes. Look, yes, indeed. I mean, it's actually a good timing that you're asking the question when after my gym session in the morning, there was the forecast call worldwide. So I got the latest. And you see I'm smiling. I mean, yes, indeed, I was confident, and I'm confident about the quarter. I mean we know usually when we see the pipeline, when we have the material that we can execute now, of course, Q4, there's always a big swing. I mean there are 3, 4 mega transactions which we never had before, deal size also proving the value of AI because these customers would never ever now sign up for such a large end-to-end transaction on RISE if we would not have our AI and data strategy, our act together. Now of course, is this quarter nailed? No. So that's why also, I guess, Dominik rightfully said, hey, there is still a swing in the number.

Now I said 25% would be a disappointment, and I stick to that so clearly and we're going to want to make this quarter work. We have the material. There was also some commentary around where I said we pulled some deals forward. Now that doesn't signal that we are then running out of pipeline next year. I mean, with a EUR 10 billion maintenance revenue, I mean, there is obviously enough pipeline. I would say the growth trajectory then after Q4 and the guidance, what we're going to pull out is -- has a lot to do with data and AI because we need to see that we are not building -- that customers are not building an agentic AI layer on top, but they really use our agents, that they use our software to have agent orchestration to have really agents serving end-to-end business processes. And that is, I guess, the most important factor. SAP will, for sure, not run out of pipeline next year. That's for sure. I mean, just look at the installed base, what we are having.

And then last piece, also this quarter, I'm here just touring through U.S. Last week, I was in MENA. There are a lot of new start-ups, also midsized customers who are now leaning towards SAP. They want to expand their global footprint and also the net new customer business is actually running pretty well. I mean the focus we put gave the team, to our partners, to the ecosystem to build out more capacity to hunt more net new mid-market customers via the channel, not via our semi-expensive salespeople. I was definitely the right choice. So we are also seeing there's good growth there. And we are just getting started. So for me, the mid-market is also a big growth engine also in the years to come. So Q4, yes, I'm still confident. And let us work the last few weeks and then at earnings, we will know more about it.

And as you look forward to sort of 2027, 2030, does that catalyze a reacceleration potentially? Or is the business too large now to sort of sustain that?

Yes. I mean, acceleration of total revenue, as I said, definitely. I mean we will deliver so that the total revenue will accelerate through 2027, and I would even say beyond that will clearly be the case. Now on the CCB side, it's really about now the magic thing, but now we need to prove. And in my eyes, we will prove it. I see the AI use cases now that we not only tell our customers to transform with SAP on process automation, on efficiency, on and on but also go into the data platform. I guess you have Oli here as well. I mean ask him, I mean, the pipeline is phenomenal because I have not seen one AI use case in the B2B sector who delivers high value only with a large language model. I mean you always need to somehow -- without data, I mean, there would be no AI at SAP and so -- and I would not have data without software. So this discussion, I'm somehow not getting because I feel we are in a super strong position.

And we are building now AI use cases, and we are committing to someone now in Q4 where we are saying, hey, we actually will be much better in predicting your volatile demand in a -- for semiconductor company in the future. This agent on sales forecast projection will talk natively out of the box to your inventory to see do you have the right products, the right chips at the right place, at the right warehouse? Should we ship it that way? Should we ship it that way? And then talk to the procurement agent to tell the procurement agent, hey, I'm running out of stock, you're better now going to negotiate the next order with this and that supplier. And these are the use cases. And if we build that use cases, I mean, I definitely have no worry that also the CCB will deliver strong growth rates, but the numbers are also getting bigger. So it's fair to say what we always said is that there will be a slight deceleration, but I guess a lot will now really depend on data and AI.

Okay. I mean I think you've almost answered the question I'm going to come in with, but just to see if there's any more you want to add. I mean, a theme over the last few months has been this death of software, death of SaaS at the hands of AI. What do you think investors are misunderstanding if they subscribe to that?

I mean looking at what happened at the stock market in the last 3 months, obviously, as a CEO, you would not do your job if you wouldn't ask yourself, what is happening there on the infrastructure? And is there not a way to get back into it? And I can answer that question clearly, no, no, no. We will not go back to the infrastructure business because our strategy was much -- was really successful in the last years. And we are convinced that over the time, let's see how long this will continue this flywheel of infrastructure chips, building data centers. But we are convinced that all of our customers are looking at SAP and say, "Hey, give me the best, the smartest agent for manufacturing automation, give me the best agent for demand and supply chain inventory optimization." And these are the cash flow optimization. I mean we have many. And that is, I guess, what we have to focus on.

And with that and with our software, we have the business context. I mean, there is no way I can see it. I just had H&M with me last week, and we showed them the future of retail in our experience center. And we literally said, okay, you can shop online, you can go into the store and rebuild the store and then not demo, real live. Then the agent came and said, in commerce, do you have this dress? Is this the right size for you? Yes or no. Click buy, oh, sorry, it's not available in the store next door, but we can ship it to you maybe to this store or we deliver it to you in 2 days. So that is really the sales agent sitting in our commerce. But this sales agent needs to understand the material workflow into the inventory and that is what the software does. So the software gives me the data and the software gives me the business process context and the workflow.

No LLM, no other AI will sit on top of SAP having the understanding of our business -- of the business processes of the customers. And then lately, obviously, you also then need to make sure when we then show them how we can even solve some of their challenges in the supply chain functions and the fulfillment functions of the company, then we talked about how do we deliver it standard. We're going to train it. We're going to optimize these agents going forward. And actually, what we're also doing, making sure that the identity and the authorization rights also is managed because somehow not everyone in H&M should see every piece of every financial data or customer data. So that is actually why I'm a big believer that our software and our data what we are having gives us a huge advantage when it comes to business AI. I could not see a business AI without software and data. I just can't.

That's very persuasive and I think a lot of investors would sort of agree with me there. I mean there's a maturity issue here perhaps. And I think the financial ambition that Dominik communicated is around EUR 1 billion of AI-related revenues in 3 to 4 years. Given what you've just described and the size of the business, that feels potentially cautious. What could make you more ambitious in your expectations?

The AI units and how much will be consumed, it's going well, by the way, especially in the last 3 months with -- now with Perplexity being GA. We see a good uptick in tool usage. Is it now EUR 1 billion? Is it EUR 1.5 billion? Is it EUR 2 billion? Is it EUR 3 billion? At the end, I'm going to win against Workday if my HR agents are the best. I'm going to win against Coupa inside out if our procurement agent is well integrated and orchestrated with the financial agent or with the supply chain agent. So for me, when we are doing next week our portfolio around with Muhammad and Philipp, I'd rather going to look at, okay, what do my Chief Revenue Officer telling me what they need on AI to beat all of these competitors. So for me, I would be -- actually say, I'm totally fine if it's EUR 2 billion and not EUR 3 billion as long as I'm winning market share in the apps. And the apps will become more and more an agentic AI layer for our customers.

And so for me, this really comes together. And then on the monetization, it's a good question. We rather also follow the strategy, by the way, that I'm saying, infuse this AI units in every deal, make it work, show them the use cases, the customers, make them adopt. And I'm sure the next HR deal will come to us anyway so that we land and expand. So I'm rather not saying, let's just overprice it like hell because everyone believes AI is the next big thing, just rather land and make us win in the whole category. And that actually works. In every customer, when we delivered a successful agentic AI user scenario say, I want more. Let's go and look into from HR core, we go into learning.

Let's go into recruiting with our latest acquisition. So let's put on the recruiting agent. And so that is, for me, I guess, Michael, at the end, much more important. And when I then look at the financial plan of SAP, I look at CCB, I look at how the LoBs are performing, how is the platform performing. And I want to see how is AI contributing literally to every deal, yes. I don't get the deals anymore through the door if AI is not changing the game on commerce for H&M. If it's not changing the game for supply chain for the U.S. Army. And that is, I guess, more important than now discussing, will it be EUR 1 billion, EUR 2 billion. I believe it could be more. But at the end, I want to win in all of these categories with AI.

Okay. I mean turning sort of the lens inwards, I think the margin improvement over the last 18 months has been pretty phenomenal. Can you talk about how AI has contributed to that? And what are the still largest opportunities internally in leveraging AI?

I mean, first, it would not be true to say now to sit here and say it's all AI. I see some statements out there where I sometimes wonder. But anyway, so I -- to give you the truth, I would say, when you look at the margin progression of SAP, I would say there are 40,000 developers using code agents, they become 20% more productive. We would have loved to see that at 30%. Let's see Joule for Developer, GitHub, WinFIRST, a few others, we are just getting there. And I'm see -- you're going to see more productivity gains. On quoting, pricing and deal approval, heavy usage. I mean this is a no deal in Q4 will not be touched by any AI -- will be touched by every AI use case we are having in our quote-to-cash process, and that helps a lot to accelerate also the handling the managing of this huge volume what we are having in Q4. So overall, I definitely see that we have, what we always said, a triple-digit million amount of efficiencies already through AI.

Next year, you're going to see it in the guidance. Obviously, we have high ambitions on margin and cash flow. So this is where we definitely also now just infused further AI use cases across SAP. So all of our managers have now the task, tell us which AI use case you are implementing next year and don't ask, first of all, for headcount. You don't ask headcount, ask AI. And then IT and our product owners will help you to drive more efficiencies. And that is actually pretty promising. We are finalizing this now in the month of December. And you can definitely expect that some of the good performance you have seen because of AI will further accelerate. I would say the other half of the good progress of SAP is just discipline. I mean we -- as you know, we removed a lot of overhead functions inside the company.

We reduced hierarchies, decision-making. I ask for a bit more entrepreneurship to take further risk and not having 10 people around you to tell you what you shouldn't do, but rather take decisions and move forward as the industry is moving fast, and that will continue as well. We -- I will not allow the company now to get lazy on efficiencies also without AI. I just want to see that we are driving as a company also further efficiencies, for example, expanding our sales channel via partners. I want to see in development that we are not going crazy in 5 new categories. Let's also see what can we co-innovate. We have great SIs. We have Databricks. We have Snowflake, we can build data products together. So that is something what I really would also want to see out of the product organization that we keep this discipline, focusing on the core where we are good at and then also expanding our portfolio via very smart partnerships. And that is -- that these are big efficiency levers as well.

Yes. I mean coming on to that, you recently added both Microsoft Fabric and Snowflake as partners on BDC. How is that affecting conversations with customers? And it's, I guess, 9 months since you announced the sort of launch. What's the uptake of that been and maybe the impact on the rest of the business?

I mean for me, BDC really moved the needle. And now it becomes pretty evident that in every AI use case we are building and we are then deploying with our customers that BDC is actually a prerequisite. And BDC not only for SAP data, there is oftentimes, of course, in many of the AI agent agentic scenarios, what we are then deploying at the customers, you also need non-SAP data. And BDC is the common nominator now for not only doing 0 copy, which we already did with Datasphere, but really to harmonize the semantics of data. And it's a good sign then when I was sitting here 2 years ago talking about Databricks that immediately Snowflake, Microsoft, Google, they all reached out and said, we want to be part of that. And I want to underscore BDC, and Hasso always told me that Christian, the biggest crowned jewel of SAP is our data.

And you see also, by the way, with a few lawsuits, which I don't understand. But anyway, that we are sitting on a crown jewel, and that is our data. And with the BDC, we are not just exposing our data everywhere, we are keeping the semantics within BDC. So when you are not under BDC and when you are not signing a contract with SAP, you don't have access to the semantics. And that is super important to also protect our IP, to also protect our unique differentiation, and that is the mission-critical business data, which we are having in our apps. And that's why software will not go away. Without software, I don't have data and without data, I don't have AI. I repeat this now in every session, Michael.

Okay. That's a good point and I try to remember that one. So just moving back to the installed base. I mean, you mentioned at the beginning how AI is helping the conversation around RISE. There's a fairly large cohort there who went to S/4 on-premise, and I know you have a transformation and incentive program to help them on the journey to the cloud. Can you give us an update on how those customers are evolving?

Yes. Actually, the journey actually is very promising. I mean you know some customers who joined us really early in the game to mention some large brands, BMW, Exxon, they are now a heavy also customer of BTP. They are now really progressing on clean core. I mean companies like that, which we can almost run like a SaaS, landscape is actually huge. So there are no upgrades anymore. These are updates, and that is huge for them. And now it's, of course, on us to infuse our agentic AI knowledge, which is much easier when the landscape is clean, and we can just plug the AI agents in and almost play without doing a lot of groundwork on the customer side. Now what is, of course -- what the customers are asking us, and I guess this will be a big market for us. And I guess this is not so much seen in the public yet is they're all telling us, Christian, these migrations are much more expensive than the software itself. I said yes, that we should change that equation. And they are now asking for AI tooling, for data cleansing, for data migration, for system configuration, et cetera.

So we are investing heavily in this AI tooling for the IT migration of legacy ERP into our modular public cloud ERP because we understand the data models on both sides, so we can clean and we can migrate with AI tooling. We can give this intelligence to AI and also on system configuration. I mean, we know our system configuration tools pretty well. So that is something what we are investing. And what we also want to monetize because it's real money. At the end, it's a win-win-win, yes, for the customer, they can go faster to the cloud and into the future with us. We actually win because we have those tools and we can accelerate the RISE journey, and we hopefully get higher multiples on the RISE journey because customers don't need to spend so much money on SI and services. And last but not least, I would assume also the ecosystem is actually a winner because they now also are with us in building this AI agent in this agentic AI tools to accelerate the customer's journey and their business model will change also quite drastically in the next years for sure.

Just a reminder, if you want to ask a question, I have an iPad up here and you can submit through the conference app. But just we're almost up on time. But just on M&A, we obviously discussed at the beginning or earlier around sort of that is dead. How does that influence your M&A strategy? Are you looking -- SmartRecruiters sort of AI-first business? Is that where investors should expect activity to come?

Yes. I mean SmartRecruiters, yes, you can say, oh, why they are still buying software. I mean I have 8,000 recruiting customers sitting on a legacy platform and before they go to Workday, there was no way that we can catch them if we are not having a modern platform. and SmartRecruiters, by the way, had already also pretty good AI. We are now harmonizing their AI platform with us. But this was for me also a tactical move to give our existing SuccessFactors a path to the cloud or to a modern cloud platform. And then second is, of course, it opens up now cross-sell opportunities. Going forward, when we will do M&A, I mean, needless to say, we are focusing a lot on data and AI. So we have this narrative that in the future, you're going to work with our software and there are agents supporting you along your business processes and you can press a button and you can build your own agent.

So there must be a native agent builder built into the software where you see the business process logic and you say, why do I need to put so much work into this business process. And here, you can build your own agent. We need to further leverage what we can see on the data layer with the Business Data Cloud platform. We had a good start. But everything what helps SAP to accelerate our road map on agentic AI and on the data platform could be a potential acquisition target. Nothing what we have to do again out of growth, acquiring growth, nothing what we have to do right away because we believe organically, we won't get there, but acceleration in such a fast-moving market could be a reason to go for the one or the other M&A.

And you mentioned context of the data that you have, the consistency across the whole platform is important. So what is the future for stand-alone solution extension partners? Obviously, one of them has been mentioned in the press or newswires recently. You partnered with these companies historically. They fill a niche need, but they're not exclusive to SAP. They have their own management teams, their own strategies. Do you need to fill these gaps now organically? How do you sort of ensure full control of that data end-to-end?

Yes. Good point. I mean we are working with all of our partners now also how to connect them to tool. That works now better and better. So also there, we have a strong API layer. They actually get some pieces of our AI foundation built into their products. We're actually asking them to do. That is not any more option so that they are really connecting to our agentic AI layer to our AI foundation. And then second, no, I mean, the partnership will not change dramatically. We are relying on the SOLEX partnerships. I actually love those because, again, we cannot code every piece of software by our own as long as they are integrating natively into our stack. And now in the future, they need to natively integrate also into our agentic AI layer so that there is clear differentiation against some of the competitors, yes, when customers really want to run an end-to-end business processes with one of those partners.

Okay. And then I think we've got time for one last one. You mentioned BTP earlier as one of the drivers of the sort of success of your 2025 plan. I think a couple of years ago at Sapphire, you highlighted the ambition to take that to maybe 1/4 or 1/3 of the business from maybe 15% today. And I know you're not going to give numbers -- but can you give a sense of progress on that ambition?

Yes. No, it progressed really well. I mean today, actually, the BTP is a native part of our stack. So when you're going to sell, I mean, we completely replatformed Ariba, we completely replatformed our CX portfolio. BTP is, of course, the platform now which wants these businesses. What you get is out-of-the-box integration, out-of-the-box extensibility, and you're going to see that in the number. I mean BTP is getting now a natural boost just by being the underlying platform, native platform of those apps. And we -- you're going to see that also going forward that now with agentic AI, we want to build the native AI agents into the solutions. But actually, our AI foundation is actually sitting on BTP. So BTP will see in my eyes also in the years to come, accelerated growth also based on the fact that everything what we do around building AI agents and also then deploying it standard into the solutions is coming via BTP.

Okay. Well, we're up on time. It's a full agenda today. So Christian, thank you very much for your insights and investors for your attention, and good luck with the last 3, 4 weeks of the quarter.

Yes. Thanks for having me. Thank you.

Okay. Good morning, everybody. My name is Adam Wood. I look after Software Research in Europe for Morgan Stanley. It's a great pleasure to have Dominik Asam with us this morning, the CFO of SAP. Dominik, thank you very much for joining us in Barcelona.

No, thanks for having me here.

It's a pleasure. So let me get the safe harbor statement out of the way. So during this fireside chat, SAP will make forward-looking statements, which are predictions, projections or other statements about future events. These statements are based on current expectations, forecasts and assumptions that are subject to risks and uncertainties that could cause actual results and outcomes to materially differ. Additional information regarding these risks and uncertainties may be found in SAP's filings with the Securities and Exchange Commission. including, but not limited, to the risk factors of SAP's 2024 annual report on Form 20-F.

So with that out of the way, let's maybe look back a little bit over 2025 and there's been -- I would say there's been good and less good elements. So there was a strong start in the first quarter despite the tariff disruption. And then I think you gave a pretty confident message at Sapphire back in May. I think then in Q2, we saw some real impact from tariffs, maybe also in Q3, but to a lesser extent. And so there was a more cautious message like in September. Q3, though, there was a much more confident message from SAP. Could you talk a little bit about framing the year from your point of view? What changed in October with the third quarter and fourth quarter data?

Yes. And if you go back, to the beginning of the year on the forward looking CCB growth, there was a debate about where should we end the year and we never give a precise guidance, but we said that CCB growth should be slightly down at the end of this year, part of which was the effect of the WalkMe acquisition, where the prior year's comparable would come in Q3. And you've seen that effect also in our numbers. So that was the planning hypothesis. And I say that's still reasonably intact at this point in time. Now where we have seen the variance to our initial thinking was that some deals we wanted to close end of June, we're actually slipping over the turn and the issue is, of course, when you have slippage at that point in time.

You are suddenly confronted with the summer period, which is a little bit slower. So many of these are actually then slipping towards the end of Q3, September. And that's exactly what happened. We hoped it wouldn't, but it did happen. And the impact on that on the CCB growth end of the year is, of course, not very dramatic or not there at all, but on Cloud revenue growth, if you -- sorry, on Cloud revenue growth, yes, in 2025, if you're not closing a deal by end of June, where we would still have 4 months of effective revenues from that signature and you're slipping into end of September, you're suddenly down to one month.

So on that kind of portion of shift, you do lose a lot of Cloud revenues. And when I say a lot, what I talk about is saying that rather than hitting the midpoint of the cloud revenues, we are more gravitating towards the low end of it, which is 0.7% variance. So it's not that we talk about huge numbers here. But the CCB growth is still what is important basis to jump off for next year. We always say there's probably around about 1 percentage point of attrition from transaction revenues and some other effects, but the lion's share of the growth for next year is, of course, embarked already in the CCB growth and on that one, we continue to stick to what we said at the outset of the year that we are seeing probably like kind of slightly down, including that effect from WalkMe and there's a little bit beyond that.

But that's how I want to frame the discussion. And now in terms of very tangible outcomes, while in Q2, we've really seen some slippage. In Q3, it was more that it was a back-end loaded quarter, but we did have a very good closing. You've seen it in a very strong CCB growth where the only effect actually in terms of sequential decline was the M&A stuff, and the underlying was actually 0, which is super good. Now we have to see how Q4 shapes along and Q4 has a lot of material what makes it a little bit harder to predict this time around is that within that kind of pipeline, there are some extremely large transactions, which might kind of meander around the turn. If we can close them this year, it can boost the CCB growth if they are slipping into next year, while it doesn't have necessarily a huge impact on Cloud revenue growth, it does have an impact on CCP growth.

So that's the overall situation. And then there's very specific stuff like our Institutional Business in the United States. We have not huge, but it's meaningful enough to move a little bit the needle as was kind of frozen in the first half. And now you've seen that we have signed a deal with the IDIQ deal, which is a framework agreement for procuring up to EUR 1 billion of TCV, and we've already filled that with life with one contract with the U.S. Army. And that's good for that specific customer, but is also a strong signal that we can share with other customers about the confidence of the U.S. government and our solutions even in these times where maybe a German vendor in the United States would be regarded a little more critically. But I think it shows that how strong the product offering is that we, as a foreign vendor, are still considered as a prime vendor here by the U.S. government.

So we look into the fourth quarter because you've even talked about some deals from '26 coming potentially into Q4 from a pipeline point of view, very relaxed, but it's about the execution now of closing it.

I think it's more about the kind of lack of predictability. Of course, when you have seen that slippage, you get a little bit less confident and you say, do I need to extrapolate that slippage also a little bit longer. Now when you see it's getting a little bit more robust, then frankly, you have a little bit more confidence that we can kind of close it towards the end of the year. So that's what we mean about kind of the meandering around the turn of the year, which is always hard to predict. But our sales force is highly incentivized to make sure as much comes in as possible in Q4. But that doesn't mean that then the pipeline was depleted.

We always found a strong pipeline and also entering the new year, but we have a strong seasonality that's very well known. Despite the cloudware you'd think that theoretically, it shouldn't be as seasonal as license sales, but it's still a bad habit that has been continued in the cloud era versus the software era.

And if we say looking back for a second that, that slippage from the second quarter, you've obviously made a lot of changes from a restructuring point of view. Changes in go-to-market? Are you very comfortable that, that was a kind of tariff volatility disruption issue and not because of changes in your go to market?

Look, it's hard to really isolate these things. Of course, it doesn't make your life easier when you're transforming things, but also the alternative. I mean there are so many ways to sell these solutions in a better way. And AI is also playing a role there. So I was just over the last couple of days going through all these transformation projects. And when you are transforming the company, it's always a little bit more turbulent and then maybe employees are a little bit less confident. So I couldn't completely rule it out, but what's the alternative. We are pretty sure that everything we do here is ultimately giving us more performance.

So even if there was a little bit of a headwind, it's very worthwhile I'm kind of sustaining that or working against that because really the only constant is change, and we review every function of the company. What we try to do a lot is because we are a large corporation ourselves, is to drive AI within our own shop, that's the best way to invite customers and say let's do some tire kicking you come trust, we show you what we do. And given that we have already a pretty clean tech stack in comparison to our customers, they tend to be quite impressed by what we can do already. And then that is another -- and then we say cloud. I mean, if you want this, yes, yes the RISE or the GROW journey you have to embark on to make your own operations more efficient. So this transformation, I always use proverb of Shakespeare actually, too much care kills the cat.

So you have to take these risks these days because if the cat is waiting for too long, the mouse is going away or is eaten by the other mouse -- other cat. So we don't think we have a choice. We have to tackle everything that's not working properly, everything that is not fully optimized yet and this is spinning the flywheel. And yes, when you do that, there is a disruption in the system, but I don't see that as something. There has to be new normal. For us and our customers is the same with AI, nothing will stay as it is.

So obviously, you alluded to it, there's a very lively discussion on where your CCB ends the year. I want to bring Gen AI into the discussion. If we ask Joule, what would Joule tell us that the CCB is at the end of the fourth quarter? .

That's a trick question. Joule will probably say, I'm not allowed to tell you exactly.

It's very intelligent Gen AI.

We have relevant, reliable and responsible AI and responsible means the compliance rules would need to be respected. So if tool is properly trained, it would not volunteer a number, it would regret later. So it would be a very smart agent, of course, and a reliable agent. And the truth is, if you have these big white elephant deals, some of them can move the needle, right? And that's why it's a little bit -- the good news is the Cloud revenue growth next year, in case you work to close something in January, it's not a huge impact yet. It's only an impact as we've seen when the slippage is longer, like we've seen now in the summer period.

But that slippage -- I mean think about U.S. government. That was just -- it was just there were some negotiations around this IDIQ, which took a long time. And in a normal environment, that negotiation wouldn't even exist. And by the way, I think they do it with job there because they're really consolidating demand, they bring them procurement rigor into things. And so I understand I'm not criticizing them. They do exactly what I would do, if I run such a complex procurement organization.

And then think about these manufacturers where all the supply chains are getting s****** up. I mean that they are not having 100% of their bandwidth now on the next digital transformation, but are just firefighting and figuring out. I mean, do I now launch a factory in another place because U.S. is getting more difficult with tariffs and so forth, that's a lot of the distraction. So I'm actually -- if I then see that, that we are still hitting our Cloud revenue band guidance. I mean I'm not too worried honestly.

If we move from the backlog on to the revenue growth, I think your -- the low end of your guide as you've talked about for Cloud revenue implies maybe around 24% for the fourth quarter. I think there's a little bit of nervousness with investors that for next year, again, on our math, you pre need around 24% to be guaranteed of the acceleration in top line. Could you just help us why you're comfortable with potentially ending the year there, but then keeping that revenue through...

Any single quarter in Cloud revenues can have some noise in it, for instance, it's also true by the maintenance. By the way, the year-on-year growth numbers if you had some special effects in the prior years, if you think of Q4 last year, one special effect I remember, and I think we communicated that back then, was that we had one relatively sizable customer that came out of financial distress where we didn't book the revenues for a while because even when you're paid because of the financial distressed situation, we have a very conservative way of looking at that and say, the insolvency administrator might say, that money was not properly given to SAP, and then there was a big refinancing package so we could get some catch-up on the revenues, and that gave us a little bit of a boost.

And that is now, of course, making our commerce more difficult for Q4. So I would always encourage you to look at the more kind of 12-month sliding averages and the CCB growth in some way is such a sliding average because it integrates the layers of bookings. So I'd say the CCB growth is the more meaningful number to jump off. And then we should see a very, very roughly 1 percentage point or so from transactional dilution and -- so don't -- there are some quarters which are strong on Cloud revenues, there's a little bit of noise around the trend line. So -- the most important number I would look at is really CCB growth exiting...

That's really helpful, thanks Dominik. We've called SAP the cloud conversion engine. One of the really important inputs that goes into the calculations there is how much of the base has moved. I think when we were speaking last year, if I did the maths around what you were saying, I was getting maybe as low as 15% of the base by value has fully shifted. But then we've heard numbers around 30%, 40% are on the journey. Could you just help us a little bit with how much of the base is moving and deadlines.

I mean the way I look at it is that I look at the maintenance base we have. And then I look at who of these customers in our maintenance base are already on the rise and paying already some form of cloud revenues. So these are basically the customers on the journey. Well, I don't see that anymore is the people who have already completely migrated. But honestly, that's not a huge amount yet because many of the customers have long journeys, especially the large enterprises, which are more meaningful in our revenue mix, and that number, so that kind of hybrid number is about 40% right now of the maintenance in ERP.

So you see the full maintenance base you strip out what is a high triple-digit million for some other stuff, which is not ERP-related, so there's about EUR 10 billion of ERP maintenance and roughly 40% of that is already on the RISE journey. The remain to do of that is 60%. The good news is that if you look at the software maintenance numbers, and you then say who of these guys is already subscribed to S/4, it's the lion's share of them. So this is why our confidence that these guys will ultimately end up in our shop is high because they have already chosen SAP as the future vendor with S/4. And so the remain to do is actually very significant. And you're right that, that kind of 40% of hybrid customers, they have not fully materialized their revenue opportunity in cloud yet. This is why we probably come to this 15% plus whatever. So -- but it's, of course, the share is increasing, but the remain to do is still very significant. Now if you think about that EUR 10 billion maintenance base -- maintenance revenue base in '25 roundabouts, in ERP, you can say that more than half of that is ECC and older technologies.

And there will be not so many people who go into kind of customer-specific maintenance because it's quite expensive. So we have to assume that, that half will be pretty much done. So you have to assume that there's a slight deceleration of the decline in maintenance revenues, but that's also good for the conversion story because these then will be converted in 2 to 3x in that kind of ballpark, and then put us on the journey to be able to cross and upsell that and of new cloud customer base from the installed base.

So there's a lot of gas in the tank as we say, for the conversion story. But also I want to emphasize that we've spent quite some time in Sapphire to highlight net new opportunity in growth and then upsell and the cross-sell. And we have a going as far as saying that kind of 2 to 3x can actually double over a, say, a 5-year migration period because we are driving the suite. We are driving the upsell from new functionality and the best of suite approach. And then, yes, we bring AI to bear and either we monetize it directly with tokens or we gradually increase prices to reflect the higher value added of the software.

So all these things are then kind of boosting that initial move to the cloud to something more meaningful over time. And I can see that in the numbers every day. So my confidence in that story is very high. So it's not a one trick wonder. We have several aces up our sleeves. I will say these 3 really new opportunities, we have very little revenues to start with, but it can become very meaningful. It's the AI story, it's the whole data story where it's not only BDC with Databricks now. It's also a Snowflake now. It's with BigQuery. You can monetize that kind of data platform. And then there is the move to tackle a smaller customers.

And I have a high confidence that each of these opportunity in isolation would give us EUR 1 billion plus revenue pie and now when exactly that kind of EUR 1 billion plus can be crossed on each of them, not 100% clear yet, but it will happen within a 5-year forecast planning horizon. So if you add that EUR 3 billion incremental revenue minimum and then you compare that to whatever revenue base we will have at that point in time. And you see it's actually a very sizable increment in terms of revenues, which then can kind of take the relay so to speak, when the cloud migration story in 2030 is maybe kind of going down and is constrained to then only the S/4 conversions. We still have then the S/4 customers being converted.

And also there, we try to convert them as quickly as possible because any S/4 customer on-prem cannot use our AI. And so the gap -- and by the way, it's also interesting for the migration story, when I -- it's very important. It's -- with AI, it's getting so different that before you could tell the customer, well, I go the first step as for on-prem, and then I go on cloud and then I do all the AI stuff. That is getting more difficult now. Why? Because the AI makes the process in the cloud even more different.

So that the blueprinting of the processes is already different. So there's a significant incremental cost to do the detour of on-prem to the cloud. And so most of the customers are now moving straight from ECC to S/4 cloud and not go over the kind of on-prem detour.

That brings me really nicely into the next question. I mean when we speak to partners and to customers, the feedback we get is we love the vision, and we love the Gen AI that is not the problem. The problem is we're on an ECC heavily modified system and the journey is challenging. Can you talk a little bit about what you're doing to try to make that journey easier for the customers?

I mean it's true that the product shift to S/4 is a pretty radical shift. So basically, it's a new process. They want everything. It means you have to reinvent the company, so to speak. And -- but it's also a big opportunity to get the complexity out of the system and reap significant benefits. But it's more complicated journey. The lift and shift to the cloud is not the big issue actually. So that is something people have to do. It's like any platform investment is -- when the base in your house needs to be renovated, that's painful, it's hard to do, but you have to do it sooner later. And then the question is, is there a better solution for them so they would run away from SAP. And luckily, as I said, most of these guys have already subscribed to S/4, but it is a journey.

Now there are, of course, new tools, which make the journey less risky, cheaper, and that's all about AI in the transformation journey. So if you have to refactor old ABAP CO to ABAP Cloud to do your extensions in a clean core compliant way, you can automate that to a large degree today. There is this tool for consultant tool where basically all the experience of the best consultants on the planet is at the fingertip of every consultant use it. I think we have recently announced that Deloitte is also now subscribing to it. That is our business transformation suite. You recall that these M&A projects were all in that area.

It's about making the process analysis very surgical, facts and figures driven and really measuring the performance of the process today versus what the target stage is and how much money you can save if you go down that journey, it's about the complication of identifying submarine software in the company, doing the enterprise architecture management and seeing what solutions are out there. And then it's the adoption. I mean, we are all creatures of habit. We find it hard to change tools and change the way we work. and WalkMe is a perfect tool to do that. By the way, LeanIX is really a blockbuster. We have quintupled customer count last year on that product. CIOs love it because it's like a radar you put on your enterprise architecture, and you can see everything that's happening. So if somebody builds a small AI bot suddenly, you will find it. And then you can figure out, can we use it elsewhere? Or should I beat it because it's a dangerous thing that is easy to attack or whatever is not compliant with our security requirements.

So we have [ venture ] to say by far the most powerful tool suite to do exactly what you say, to drive down the cost of these journeys and to also reduce the risk that the outcome will not be exactly what customers think. And it's super important because we still have a lot of wounds from some SIs having worked on projects and something went wrong and then -- of course, the reputational damage is not necessarily with the SI, but is in many cases with us. So it's super important to make it safer, so to speak, and more predictable.

Can we talk about cloud multipliers again. You've talked about the support to subscription journey being a 2 to 3x multiplier. Maybe first of all, why are we at the low end? What gets us to the high end? And then at Sapphire, Christian talked about even 4 to 5x. How do we go from 2 to 3 to 4 to 5.

Yes. I mean the question is really on the initial signature of the deal. To what degree are we already able to convince customers to embark more content than just the bare minimum of S/4 into that journey. So are they also adding already a procurement solution there or some consolidation software on finance and that drives that kind of 2 to 3x. Now the upsell is really -- we had a nice slide in Sapphire where we show the different sources for that kind of doubling on top of that. It's simply the growth.

I mean, contrary to what some people think it's not that the prime driver of our growth of metering is like the seats. It's of time also just stuff like revenues, which are growing. And then is, of course, the cross-sell to add more applications, and we are very aggressive commercially to say, look, guys, if you consolidate more on our platform, you can do that. Then there is the value add with AI, which is coming in. And then also when you move from the old version of SAP to the new version, that new version has more functionality and the customer is also willing to pay more for that. So it's not a huge challenge to double that over, say, a 5-year horizon because if you break it down then in the kind of net renewal rate over these 5 years, if you compound that at a kind of reasonable rate, that's where you are. And yes, this is quite well secured. And is the other big pillar besides the pure conversion that has been giving us a nice uptick on growth in the cloud.

That's helpful. I think the other surprise for me at Sapphire was the scale of the net new business that you talked about, EUR 2 billion with cohort is growing at 20%. Could you just talk a little bit about what's changed in the mid-market from a product go-to market and the opportunity there?

I mean, it's all about the maturity and the go-to-market engine for a product that has not been really existing before, which is a product where you can start kind of if you want to spend EUR 40,000, EUR 50,000 per annum that's the entry ticket to buy GROW with SAP and then to make it easy to implement for young, growing companies or smaller companies and also build an indirect sales channel which was not our forte in the past, and that is still in the making. So I think the best is still to come on that front. So we only have these ducks in a row now. And now you see the inflection actually that now the public cloud is actually in terms of growth rates, gathering momentum and on the bookings, the growth is actually higher on the public cloud now than private.

So it's really kicking into gear. One big advantage we enjoy with this tool is that we don't have like a small company solution anymore and a large company solution so we can go to customers and say, even if you are now a smaller tech company, you don't need to replatform because we can run from small companies like a Mistral AI or something like that, all the way to a giant company like Schneider Electric with manufacturing. And that was a huge development effort because we have basically matured and developed ECC over decades. And now we have to do a full kind of native -- and now it's about the maturity of the product, really one, two after the other is dropping that we can suddenly go for higher hanging fruit and have a more complex product that is also fulfilling the needs. And once you have that in that scalable multi-tenant deployment form, you can also get bolt-on small customers easily.

So it was a start from scratch basically before we had a tiny different tool, but it was not the focus of SAP, now we have that, and it is a focus of SAP also the large customers. We want to mature this solution so high that with the Clean Core journey from RISE, over time, we can also migrate these people to GROW because that will then be the brave new world where all the upgrades we are doing with AI and new functionality are -- the customer doesn't even notice anymore that there's some work.

We all do that for them. On the RISE journey, there's still some deployment work for upgrades, so they do it probably once a year. But on GROW, we can do it really quarterly, and that means we have the velocity of innovation, and we finally shed that impediment of showing our innovation to the customer later than others. And that's the flywheel we are really excited about.

So you've talked about a lot of the drivers, Cloud Perversion, Business Data Cloud, GROW. We've got a whole list. When we think about CCB growth in the midterm rather than just this year, could you help us -- how should we think about -- is it still a deceleration? How should we think about that momentum?

If you look at the PaaS and SaaS layers in our segments, I've seen some steps, and I think they're right from the research houses that SAP has been printing on latest 12 months like 29% U.S. dollar-based growth there. And our competitors -- the market is like mid to high teens in that bucket. So -- and yes, you can look at our direct competitors, Workday, Oracle on ERP and well, we're growing probably about twice as fast. So for the question boils down to how long Dominik do you think you can sustain kind of growing twice as fast as the market, and we have to be a little bit humble on that and say, look, at some point in time, the air will get thinner and you might gravitate more towards a normal X percent plus market than double.

So I do see actually that this can gradually decelerate. But it's not a problem because we have that mix effect, which is propelling our total revenue growth I always do the math, no matter whether you look at '24 growth rates in each of these markets, cloud ERP suite, extension suites and the maintenance and the software and the lack of services. And you use the same growth rates for each of these buckets in '24 or first half '25. You can do the same conclusion that where we're able to sustain these growth rates for the next 3 years, we would see a 3 percentage point acceleration in total revenues every year.

So it means I can actually afford quite some deceleration and still accelerate the top line. So I mean I'm not expecting compounding 3% growth for the next years that would bring us into high teens type of total revenue growth. And that's not in the cards to be very blunt. It's -- by the way, the overall market, if you look at the sum of software services and cloud is, of course, not at the growth rate either, it's much lower than that.

So the truth is somewhere in the middle, and it's all about mitigating that slight decline to the lowest possible number. And -- so you have to continue to expect that kind of slight decline. We have been peaking. I mean, so far, it was an acceleration. I think indeed, we have been peaking now. But we have still a fuel in the tank from the conversion, which I mentioned. And this is why contrary to my habit of not giving long-term guidance, I'm willing to say that total revenue growth should accelerate in '26 and '27 every year.

Now in January, we have to then see what we can say for '26 specifically and how much it will be. And I would also go one step further saying, look, there is no logical reason why they should suddenly be a cliff and something should end at that point in time. The fundamental drivers of the growth story are very intact. And yes, at the end of the decade, there might be a slower conversion story. But by then, these 3 new opportunities should have really, really meaningful size and can kind of more than offset, hopefully, that kind of conversion normalization, I'd say.

Maybe from top line on to profitability. You've talked about the aspiration to be a Rule of 40 company, being pretty hard on yourself by using a free cash flow margin. Could you talk a little bit around the levers you still have on profitability? And then what year would you expect to be a Rule of 40 company?

I'm a fundamentalist. And for me, discounted cash flow, the way I learned in the business school is still important. When I was walking over the hallways there, and I've seen 25 years long Stanley Conference, I was thinking back 25 years ago, that was 2000.

At that point in time, people started to say cash flow doesn't matter too and then infrastructure was everything. And I was an investment banker at that point in time, and I was able to get more money out of an IPO of a fiber optics company than the enterprise value was 2 years later, fee is higher than the enterprise value 2 years later. So I'm confessing fundamentalists. I care about free cash flow and the growth thereof. And of course, we have to -- I also say that growth in our business is more important than just the margin.

So I don't want to take mortgages on the future, but we have decided for us that we really want to shine on the past half layer. I gave you the market share gain numbers we had. Infrastructure, we will do selectively on the sovereign side, there is more and more customers who are nervous about a kill switch in the system, so they might not want to put their stuff in U.S. hyperscalers anymore, which is a pity because our U.S. hyperscaler partners have great products, and they are super cost competitive. So the fundamental model in terms of margin expansion and cash conversion should not change. So we are still seeing that 80% to 90% total expense growth versus the revenue growth we need to activate a lot of AI scenarios to really get there.

The cash conversion, we will not change massively. There might be some opportunities to bring customers to the cloud that previously was stuck on-prem because of legal constraints. I mean look at Germany. We are just in the last innings of getting certification by the PSI to move defense customers to the cloud, stuff that we've done in the U.S. for 7, 8 years already.

So that's where we are going to invest. And if there are other local partners that can do the infrastructure business for us at lower cost, we're happy to also embark them on our platform. So it's not our differentiator. So -- we think that from that perspective, while it's not the flavor of the month in certain stakeholder groups these days, we still continue to stick to our guns there and do what we have done in the past. Does that makes sense?

Absolutely. Would you give us a year or is that too much?

Year for...

Rule of 40.

No, no, no. Sorry, you were coming from the Rule of 40. So I was just spending time why the cash flow is so important to me and the growth of it. Look, I don't want to speculate on when exactly it's happening. I think we are on the right trajectory. I think you can do your own math with the kind of framework I gave you in terms of the margin expansion, the acceleration of the growth. I mean it's been about -- I mean, it's not so easy to hit it in '27 if you do the math.

So -- but when exactly, we don't want to be more specific, but we should see '27 revenue growth higher than what we see in '26 and higher than what we see in '25. And the margin will expand within 80% to 90% rule and the cash conversion will be operating profit, non-IFRS operating profit, minus 30% tax, plus EUR 1 billion for stock-based comp, which is noncash. That's it. And then you can do...

We can build our model. Maybe if we -- the financials, it's amazing this year that we've had this kind of flick in August up until that point, it was all about how positive is Gen AI going to be for you? How do you monetize it? How do you get benefit internally to how disruptive is it going to be for your business? Could you talk a little bit about how you think about this? Is this more of an opportunity or more of a threat to SAP?

I think it's more of an opportunity. The reason being that if you think about what is really needed for AI, what are the key ingredients is good data, reliable, relevant data. And where is that data generated? Actually, sorry to say it's in the app. And then the question is, once you have used that data and turn it into insights, our app is not there to write a report. Our app is there to trigger transactions in companies. So you have to go back to the app.

So the starting point is the app, the endpoint is the app. Now of course, there is a lot of data management and a lot of AI in between these days. But we have the full position that the endpoint and start end is in our system. And we have now created a flywheel ecosystem that fills the middle. We have embarked on our platform. Now data platforms, data engineering platforms like Databricks, like Snowflake, like BigQuery, which means there is a monetization opportunity for us when people start using these, the meter is spinning. And that is actually some incremental revenue, too, because before, frankly, for lack of alternatives, we didn't monetize it at all.

If you think about software, on-prem software, you sell a software. So all the connectors are simply programmed by the customer. They just change the code and do the APIs they need, and we did some APIs ourselves. And as we didn't have a monetizable offering, we were allowing people to really make us a patient stupid data. And now we have a tool that at least we can monetize it. Now that's one thing. It's also a big benefit for the customer, not only because they use it, but also they can avoid SI costs because it's plug and play.

So the money -- the waste we drive out by creating that preconfigured integration with these vendors can be shared between our customers and the 2 vendors who are integrating. So there is a value pool we can tap and then the meter is spinning. And on top of that, what many people forget is the initial idea of SAP was the guy, the founders, they looked over the shoulder of the customer, and were just seeing the same customers are doing the same thing. So why don't we do a standard software for that. We were limited in that approach in the past to our own data in the SAP system. Now with BDC and the other things we have the connector to the unstructured data. I always bring the example of a bottling company.

You need weather forecast, which is not in the SAP system for that. Now you can still speculate should every bottling company on the planet to its own kind of forecasting software to decide how much bottles they need to order and when stuff needs to be where? Or is there an off-the-shelf SAP up, which does that. And we see that a lot of customers would like to just activate something that's out of the box because there's always risks when you hold yourselves there's uncertainty about cost, there's uncertainty whether it works.

So the fundamental story of SAP simply looking over the shoulder of the customer and trying to do it better than what they could do on their own and using the economies of scale that's fully intact. So -- and we have 25 years of domain expertise in these 25 -- sorry, 50-plus years in 25 verticals we are catering to. We have that functional expertise in accounting, in controlling, in procurement, in travel. And that's the unfair advantage we can use on top of this platform.

Let's assume this platform were to some degree, democratized. It's actually not because there are still some features we keep to ourselves like the Rapid One foundation model or the knowledge graph that is really proprietary to SAP to build even better AI than others. But let's even assume that wouldn't exist and it's all democratized. I mean there is also a whole position to develop these apps, and we have the biggest customer base doing exactly that. So why should I have any inferiority complex that I'm not able to also be very strong in everything that's in between with everything we have today.

So this is why it's more an opportunity to come back to your initial question than a threat, but it's also true, we have to be very quick on our feet. So it's all about speed, execution. And that's what currently is our key concern to make sure that we are really rapidly grabbing these opportunities. And this is also, while some don't like it, we employ a lot of people very rapidly. We have some shed about 10,000 people last year. You've seen that the headcount is going up. So we've higher more than that again. But these are exactly the new skills we need to win that rate.

And I do believe we have an edge on the recruiting there. Why? Because these people -- the people who are interested for enterprise applications. They can either work of some of these generic platform providers, which have very little clue about what's happening in the customer or they can work at specialized start-ups, which are currently eaten alive.

So if you would build a small niche-y SaaS index, in the NASDAQ, you would see that performance is not very great. And these people have a lot of stock-based compensation. And this is also the reason why we didn't acquire any of them because already 5 years back, we said you are too expensive and then we reconvene 3 years later, and we say you're too expensive. And that was at half price compared to 2 years ago.

So the niche-y guys are under pressure. And then the other opportunity is you do the own development in a company. But if you are developing that stuff in one enterprise, you're doing it from one enterprise. If you joined SAP, that's the pitch to the people who join SAP, think about you're doing a super thing for procurement and you don't do it from one customer. You do it for tens of thousands of customers. Is that much more interesting.

So we are pretty strong. And I'd say one of our advantages vis-a-vis our customers is that we probably find it easier to hire because also some of the industries of our customers are under pressure, and we're still a nice house in a nice neighborhood in digital, where there's also some opportunity on stock-based compensation for these people. It's not always going through the roof, of course, but over the cycles, SAP has been a decent story there. So I think we are well positioned. But of course, it has to ultimately be executed, and that's the key focus. So the good news is in our hands, not that we need some magical things from outside to have.

And if we think about the Dual Copilots and the Dual Agents that you're putting in, could you maybe just talk a little bit about the functionality that you're delivering to customers and maybe also the monetization path there?

Yes. I mean you take my specialty, which is, of course, finance and the adjacent functionalities like procurement and so forth, we are also in charge of SAP to do that. Every customer, including ourselves, is currently building these agents to get stuff done on consolidation. I mean, very tedious work where -- I'll give you another easy example, cash collection. If you compare our accounts receivable days outstanding, you will figure out they are too long in comparison to our competitors.

And there are some things you can change about that and the AI is a good example. I mean you can done people to death with AI. You basically if you are fighting against the bot, customers tend to have always good excuses why they don't pay. And one classical one is I can't recognize the invoice because the purchase order I have is a different one. And I investigated that as an example, as an agent and said, why do we have these many wrong purchase orders? The truth is when we sign a contract, these contracts are complex. They have hundreds of pages, and then customer puts in a purchase order and they make -- almost every purchase order has some mistake.

And when we receive the purchase order from a customer, we don't even look at it today or we did not even look at it to date. Why? Because think about this tedious work who wants to kind of screen the contract and compare it to the purchase order hundreds of pages, no human being in the shared service center will do it, and they don't have to. Why? Because it's not a transaction. That's the kind of just FYI document, and we don't do anything with FYI documents.

Now I can build that bot that's doing nothing but comparing literally millions of these purchase orders against the initial contract. And the second we get our own purchase order, sending it back to the customer. And then when I send the invoice next time, I have no dispute. And if I get EUR 100 million, EUR 200 million, EUR 300 million out of this because I can accelerate the DSO and you apply the cost of capital of that, that's a much higher amount of money, I can save.

So I can give you -- I just yesterday was presenting to my colleagues the myriad of agents, we are now activating over the next 3 years at SAP, and the way we do it is we do that in our own shop, we then bring customers to us and look at it and say, look, this is what we do. And this is part of the story where we can decouple the total expenses on the top line despite the fact that most of our expenses are personnel expenses, which are inflated every year by merit hikes.

So it's very broad, and we should make a big effort at Sapphire to show you more of these things. So please, everyone is cordially invited to join us there. But it's really in the making, and we are at that kind of tipping point where it's coming from PowerPoint to real products and where then with the forward deployed engineering, which we do not only with customers, but also with our own groups internally, we are debugging these things all the time and make them better and better.

And this is how we then commit to very heavy productivity goals. I mean you think about finance, we are growing the top line more than 10%. We are going down market with smaller customers. We have a regulatory thing. Yesterday, you have seen the EU has ruled Omnibus, but big customers are not affected. So we have to do all the regulatory burden of the EU. Basically, you said. We understand it's kind of bureaucratic monster. This is why we give relief to small customers, but the big ones, they don't care, they can pay it anyhow.

So there's a regulatory onslaught. So I would say our workload in my department is probably increasing high teens every year. And I want to keep the cost. So I need massive productivity gains. And this is -- in other industries, it's a different problem there. The margin is eroding. The top line is stalling and they need to drive productivity.

That's amazing that you can use your company as a showcase to customer...

Of course. We've done that also with the standardization S/4 before and they came to our shared service center in Prague, and this is -- this tire kicking is super important for customers to get confidence.