Hewlett Packard Enterprise Aktie

Good morning, everyone, and welcome to HPE Networking Investor Day. Thank you for joining us both here in person and on the webcast. I'm Shannon Cross, Chief Strategy Officer at HPE. As you all know, we posted record Q3 results earlier this month. In Networking, for example, we reported both record revenue and orders.

Today, we'll take a closer look at the business and the strategy we have in place to capture new opportunities we see for the future. Over the next few hours, you'll hear from Rami Rahim, Executive Vice President, President and General Manager of HPE Networking about how AI is reshaping the networking market and why networking is considered critical infrastructure in the AI era. We will also talk about how the combination of HPE and Juniper Networks strengthens our ability to compete and win in this increasingly important market. We'll discuss our integration progress and the financial benefits we are already seeing. We'll then invite the leaders of our networking business to join me for a panel. And finally, Rami and I will host a Q&A session.

Before we begin, I'd like to introduce and welcome Dave Rubin who recently joined HPE as our new Head of Investor Relations. Many of you already know Dave from his time at Flex, and we're very happy to have him on our team.

And let me cover our safe harbor statement. This event may include forward-looking statements involving risks, uncertainties, estimates and assumptions. If the risks and uncertainties ever materialize and the estimates or assumptions prove incorrect, our results may differ, perhaps materially from those expressed or implied by such forward-looking statements. HPE assumes no obligation to update such statements. Please find more information regarding our forward-looking statements on our website at investors.hpe.com. This event also includes certain non-GAAP financial information. Such non-GAAP financial measures may have limitations as analytical tools. Please note that these measures should not be considered in isolation or as a substitute for analyzing HPE financial measures as reported under GAAP. Certain financial information featured in the presentation today has been normalized to include Juniper Networks results as of the beginning of HPE's fiscal year 2024.

With that, let's get started. Please welcome Rami.

Thanks, Shannon. Okay. Welcome, everyone, and thank you all for investing your time with us today. So my objective for today is to demonstrate how our new combined HPE Networking segment will accelerate shareholder value for years to come. We're going to talk about a few critical things today: the AI market opportunity and why networking today is more important than ever in the new AI era; how our new combined HPE Networking portfolio and innovation are perfectly suited to capture the large AI opportunity; the significant strides we've made with the integration of Juniper and HPE; the strong financial progress made against our commitments; and finally, I'm going to share a new financial guide later in my remarks.

So the core message I want you to take away is that HPE Networking segment is a key driver of our long-term strategy to create shareholder value through durable revenue growth, margin expansion and strong cash generation. We are executing ahead of plan, with the integration delivering greater-than-expected synergies and expanding into new addressable markets. Our HPE Networking portfolio is not just larger. It's more complete, more integrated and better positioned to compete at scale in large markets benefiting from strong, secular growth drivers. And finally, we're seeing that progress reflected in our results and in our outlook.

So let's just get started. As you know, the AI market is driving major technology transitions, and it's reshaping the technology stack. It has become a strategic core element for how organizations architect and operate their IT to enable scale and competitiveness. So that's why every customer segment is treating the network as core, strategic infrastructure. Now as you can see on this slide, the expected market growth opportunities across networking are significant. And with the combination of Juniper and Aruba, we are positioned to win. We look at our opportunities across the four product categories you see here and our two customer segments, enterprise and service provider, which includes cloud service providers or CSPs such as hyperscalers and neoclouds.

Data center is our fastest-growing market, largely driven by CSPs. And here, we are expanding our addressable market opportunity as AI drives increasing demand for both high-performance scale-out and scale-up solutions. In routing, we're seeing the opportunity broaden as AI moves beyond training and into large-scale inferencing environment. And that's creating demand across both service provider and enterprise networks. This is also an area where we have a long history of innovation and deep incumbency which is a significant competitive advantage for us. Campus & Branch is currently our largest product category. The combination of Juniper and Aruba further strengthens our competitive positioning from both a technology and a go-to-market standpoint.

And finally, we cannot talk about the evolution of AI infrastructure without addressing the increasing convergence between networking and security, where HPE has a unique approach through the combination of silicon, software and agentic operations. The opportunity for HPE is clear. We have the right to win in every networking product category. And we are now even more confident because of the successful integration of Juniper into HPE.

So how is the integration going? In short, it's been a tremendous success. We are moving faster than expected and delivering greater-than-expected cost synergies. Integrating two large businesses into a single networking powerhouse while also delivering strong order growth is hard to do, but we're doing it and delivering quarter after quarter. And the success is driven by the thoughtful integration we've done and the amazing execution by the combined team. Not only have we set a clear strategy but we onboarded all Juniper employees, combined our sales teams and never stopped innovating. This has given us tremendous confidence and excitement for what comes next: an acceleration of profitable growth. Now one of the strategic opportunities in bringing HPE and Juniper together was to combine decades of innovation and differentiated IP while accelerating product development and time to market that will result in market share gains. And we're already seeing the benefits.

For example, we rapidly advanced our self-driving network capabilities, cross-pollinating the best of Aruba Central and Juniper Mist, supported by a new set of campus and branch wireless access points as well as access switches. We also grew our addressable market by expanding our innovation into the AI data center scale up market. We introduced a new data center scale up switch, leveraging HPE's decades of experience in rack scale architecture and liquid cooling technology. And I'll talk more about these products later, but the key takeaway is that we were able to accelerate our innovation through the integration process, leveraging unique IP and expertise. We've accomplished all of this in just over a year, and these are only a few examples.

The second major element of the integration was bringing our sales organizations together. And here again, speed mattered. We called this Sales Day One, and it was completed globally in about 6 months after we closed. We made it a deliberate decision to move early and it is paying off. With our teams bringing Juniper into Aruba accounts and Aruba into Juniper accounts, one catalog, one compensation plan, one set of account rules, carrying all the sellers, carrying a full combined networking portfolio. Now in a minute, I'll talk about the channel opportunity and what comes next in our integration process. But first, let's turn to results and particularly cost synergies.

Now as you may recall, when we closed the transaction in July of 2025, we raised our original cost synergy target from $450 million to at least $600 million in annual run rate savings by fiscal 2028. Those savings come from several areas, which include things like SG&A, go-to-market harmonization and the elimination of R&D duplicate costs. We are running ahead of our cost synergy plan while also spending less on integration through fiscal discipline. Given our progress, we're increasing our fiscal 2028 annual run rate cost synergy target to $800 million.

Now let's talk about what comes next. We have two major integration priorities remaining: bringing our partner ecosystems together; and completing our commercial systems integration. The key milestone for the first one is called Partner Day One. Partners drive most of our networking business, and this milestone will enable all 60,000 HPE partners to sell the full networking portfolio through one unified partner program. Before the acquisition, only around 10% of HPE Aruba Networking and Juniper partners overlapped, and so this creates a significant cross-selling opportunity as we bring the combined HPE Networking portfolio to a much broader set of sellers and customers. Partner Day One is planned for November 1. On that date, we will converge the Juniper and Aruba programs under one unified partner program, HPE Partner Ready Vantage.

The next priority is a little less visible but equally important. It's completing our lead to cash processes and systems. We're building a modern single quote-to-cash foundation that will make it easier for customers and partners to do business with us while allowing us to launch new products and operate more efficiently. The first phase launches in November with integrated quoting and ordering. And we expect to unify our supply chain processes by the end of 2027 with continued improvements coming throughout 2028. Now while there's still work ahead, we have made tremendous progress, giving us confidence in accelerating profitable growth in fiscal 2027 and beyond. As you've seen, the combination is already creating strong shareholder value through accelerated innovation, new expanded market opportunities and our scaled go-to-market reach and higher cost synergies.

So now let's shift to our HPE Networking strategy. At HPE Discover back in June, we introduced our vision for the self-driving network. We discussed how we're bringing networking, compute, cloud management together to simplify increasingly complex IT infrastructure. Today, I want to focus on how we are winning in the market and achieving growth by leveraging the power of the combined HPE and Juniper networking portfolios. We've organized our opportunity around four focused strategic plays. The first go-to-market play is pretty straightforward, whereby bringing Juniper Mist and Aruba Central together, we now have a much larger combined Campus & Branch cloud-native and AI-driven portfolio. We've taken the combined networking portfolio through HPE's significantly broader go-to-market engine, creating larger growth opportunities through scale and productivity improvements.

The second play is increasing services attach rates. Today, services represent about 1/3 of our HPE Networking business. Services deepen customer relationships, add higher margin recurring revenue and improve the quality of our products driving durable earnings. Juniper has a strong history of attaching services to products. And we see an opportunity to bring that business model and the culture to Aruba across a much larger installed base. So by fiscal 2028, we expect to increase services attach rates by 5 percentage points, supporting both recurring revenue growth and margin expansion. The third play is about bringing our data center networking portfolio into HPE's broader infrastructure customer base. This means networking is no longer a stand-alone conversation. It's part of a much broader discussion that can include compute, storage, security and cloud management. And that creates opportunities to participate in larger infrastructure deals across all customer segments.

And finally, the fourth play is about expanding into new market opportunities. One example is AI infrastructure, of course, where we're leveraging HPE's AI data center momentum and innovation to expand into the scale-up networking market. And I'm going to talk more about the AMD Helios opportunity in just a moment. But the broader point is that the combination is creating new avenues for growth, not just expanding existing ones. We're bringing capabilities together to create solutions and address markets that neither company could have effectively pursued on its own. And importantly, we're building a large customer pipeline.

So now let's dig into our specific market strategies and opportunities starting with data center where we're seeing the fastest growth. The pace of change in the data center is quite frankly, extraordinary. AI has fundamentally reshaped how infrastructure is designed and deployed. And the reason is simple, you can invest billions of dollars into GPUs, but if the network can't keep up with the traffic and the performance demands, those resources are not fully utilized. So as a result, networking has become a much more strategic part of the AI infrastructure stack. And we believe HPE Networking is exceptionally well positioned because we have the products to address networking across all layers of the AI data center. That includes scale out and scale up. We expect our data center networking revenue to grow at low to high 50s percent CAGR from fiscal 2026 through 2029. Our strategy starts with innovation and time to market. We have consistently been early with many of the technologies that matter most in AI infrastructure from 800 gigabit Ethernet to 1.6 terabit Ethernet and most recently, with the industry's first liquid-cooled Ethernet data center switch. That matters because it allows us to engage earlier in large AI deployments and establish a competitive position as customers build out next-generation data center and AI infrastructure.

Now AMD Helios is a great example of what our integrated HPE strategy can deliver. We are bringing together Juniper's networking technology with HPE's compute, liquid cooling, rack scale engineering to deliver an integrated Ethernet-based scale-up solution. More importantly, it's opening an entirely new addressable market in networking, either by selling our AMD Helios networking trades as part of an HPE integrated rack solution or independently to other solution providers. We believe Helios represents more than $1 billion networking opportunity over the next 2 years with networking trade orders having already exceeded $200 million. I am excited to share that just this morning, we made an announcement that HPE was awarded a $1.2 billion AMD Helios order from Vultr. It's HPE's first order for the new integrated AMD Helios system, which features our new purpose-built HPE networking scale-up switch and software. This is a great example of the wins we can land working across the portfolio together with our partners.

Let's move to routing and what's happening in routing today sort of reminds me of my early days at Juniper when we were building ever larger routers to keep pace with the explosive growth of the Internet. Today, AI is creating another major wave of networking demand. We see that opportunity developing in two areas. The first is the AI on-ramp, securely and efficiently connecting users, devices and branches to AI infrastructure. This plays directly to our strength of the MX platform and the Trio silicon, which were designed for high scale service-rich edge routing. And when it comes to programmable networking silicon, Trio is truly unique in the industry in its capabilities.

The second is data center interconnect or DCI or scale across. As AI clusters grow beyond individual data centers, customers need to connect infrastructure across buildings, campuses and geographies with enormous bandwidth, with low latency and with high power efficiency. That's where our PTX portfolio and our express silicon truly excel. The important point here is that this is not a new market we are trying to enter. We are applying decades of routing innovation, custom silicon expertise and deep incumbency in some of the world's largest cloud and service provider network to a new wave of AI-driven demand. As a result, we expect our routing revenue to grow at low to high 20s percent CAGR from fiscal 2026 through 2029. Now a good example of our AI data center switching and routing strategy playing out is our recent win with Oracle. We are participating in Oracle's multiyear gigawatt scale AI infrastructure build-out, both in routing and switching platforms deployed across data centers globally. The significance goes well beyond a single customer. This win validate HPE's ability to compete in some of the world's most demanding AI environments and demonstrates the strength of our end-to-end networking portfolio.

So, so far, you've heard how we're applying our strength in switching, routing and custom silicon to some of the fastest-growing parts of the AI market. Let me now turn to our opportunity in the Campus & Branch. The Campus & Branch, as you all know, is our largest networking product category, representing approximately half of our total networking revenue. We expect our revenue to grow at a high single-digit percent CAGR through fiscal 2029. Now the Wi-Fi 7 refresh cycle is one important driver, but the opportunity extends well beyond wireless. As customers upgrade their wireless infrastructure, they often need to modernize other parts of the network as well, creating opportunities across switching and the broader campus architecture. We are entering this cycle from a position of strength with stealth driving automation that truly sets us apart in the industry. Now in just a bit, you're going to hear directly from one of our customers who has experienced the value of this technology firsthand. Importantly, the combination of HPE and Juniper further strengthens our position, with a larger deployed base, a broader portfolio and significantly greater sales and channel reach.

Put simply, we are taking proven technology and putting scaled distribution behind it, and that creates opportunities to drive adoption across the installed base and reach new customers to gain share. Importantly, that leadership is independently recognized. HP has been named a leader in Gartner's Magic Quadrant for Enterprise Wired and Wireless LAN Infrastructure for 20 consecutive years, including 5 straight years with the highest ranking for ability to execute and completeness of vision. The opportunity is clear. We're going to leverage the combined strengths of HPE and Juniper to capitalize on the Wi-Fi 7 refresh cycle.

Okay. And finally, another really important technology trend we're capitalizing on is the convergence of networking and security. Now security vendors are increasingly adding networking capabilities to their platforms. Our strategy is to move aggressively in the other direction, bringing more security into the network, and that's playing both defense and offense. It helps us protect our networking position while expanding our addressable market and the share of customer spending. We have the assets to do just that, including a complete SASE portfolio, industry-leading SRX firewalls, network access control and identity and policy management integrated throughout the network.

Now while some of these capabilities are sold as a stand-alone security product, our primary strategy is to embed them throughout our networking portfolio, including in our Campus & Branch silicon, which enhances our differentiation. As a result, we currently expect security revenue to grow at a high single-digit percent CAGR through fiscal 2029. Look, I believe we have the industry's most complete and most differentiated networking portfolios spanning from the Campus & Branch across the wide area, into the data center and with security embedded throughout. And I am confident that the strength of this portfolio, along with our expanded go-to-market scale will enable us to grow market share in the years to come.

The progress we've made in our integration and the success we have achieved in building industry-leading self-driving networks, would not be possible without great customers. And fortunately, we have one of those customers with me here today. So I'd like to welcome Sajeev Nair of ServiceNow up here to talk to me about it. Welcome. Thank you for joining us. Please have a seat.

A pleasure.

Okay. So Sajeev, thanks for joining us. You have been a customer since 2018, I believe, starting with Mist and then Juniper Networks and now with HPE. You've seen the vision of self-driving networking truly come to life. So maybe start by just taking us back in 2017, earlier, what did the network look like day-to-day? And what was your team just dealing with?

2017, gosh, I think if I have to just simply put it in one word, it was chaotic. If I think about it now, that's how it comes across. The enterprise was not doing anything wrong. It was very typical stuff that you were going and sourcing deep tech solutions, solving a particular problem in a stack. And at the back of your mind, you knew that you are creating another fragmentation in your telemetry that is already fragmented, you are signing up for another control plane that you have to manage and then you're going to be basically trusting the humans to do the magical correlation that needs to happen to solve the problems faster. And the hidden element into this whole thing was the cost that was so high that we did not realize that there was a better way to solve these problems. And we have this thinking in the back of mind that we need to solve some of these products, but how do we go about it?

In 2018, another tiny company walks in, tell us that they are doing something radically different in the wireless space that was -- so Neil and Bob and Sujai in front of us, and they were like, "We're going to do something different," and we said, "Okay, tell us more about it." And they said, "We're going to focus on two things. Number one, we're going to solve something that has been completely overlooked in the networking industry that is end-user experience." We said, "Wow, this is exactly the magic word that we wanted to hear." And the second thing they mentioned was we're going to overlay AI on top of the entire wireless infrastructure. We said, "Okay, that sounds ahead-of-times thinking, but we are on board with this thought process. Let's begin small." And that was the beginning of our relationship in 2018.

That's awesome. We're talking about almost 10 years ago, well before AI was a fashionable term to be used by pretty much everybody in the industry. But we see what you're talking about so often, the consistent troubleshooting, the silos. What were some of the first results that your team saw when you started deploying the technology?

Absolutely. And some of these numbers are going to just blow your mind, because it did exactly the same thing for us. We're like, okay, we are on this journey with this amazing vendor/partner that was so agile and they were ready to listen in, and they were ready to implement and move as fast as we wanted. And as we deployed this stuff in a small way in 2018, 2019 was when the Juniper acquisition -- Juniper acquired Mist, and we said, "Wow, this story is now going to get amazingly better because the goodness of what we saw, the AI and the end user experience that Mist was delivering was now going to be spread on top of the entire Juniper stack." And when that was conveyed to us, we said, "Okay, this is actually opening up a completely different radical thinking in our mind of how we want to transform ServiceNow network."

And COVID happened, and that was the amazing time for me and my team to go and do something strategically different. And we said, "Okay, what are we going to do?" We're going to basically going to rip and replace every single vendor stack that we had in our networking environment, and we're going to basically shift all of that into Juniper switching, Juniper routing. And it worked out phenomenally well because everybody was coming back from home, using Wi-Fi networks, and they said, "Okay, this is not going to be different. My home and my office looks exactly the same." And that was the time when we scaled it up and we took the whole full-stack approach.

So think about it, right, we came from a world of fragmented silos and multitude of control planes that we were managing to now this space where we had Wi-Fi, wired and WAN, all available to us in a single plane overlaid with Marvis AI and the story was just getting brilliant day by day. So 90% reduction in our incident volume, can you believe? 90% reduction. And we had patchings that used to run for hours over the weekend. We fully automated it. We had incidents that took hours to troubleshoot and figure out where the problem was, was taking minutes now. And the end-to-end visibility that we gained and the AI capabilities that Marvis delivered just was game changing for us.

I love that story. And you said you were talking about how excited you were when Mist was being acquired by Juniper. Now we're taking Marvis and applying it across the entire portfolio, and we've made -- already made amazing progress in doing so, so the opportunity is immense. So that's obviously the self-driving network in action that you've just described, Sajeev, with almost instant results. What changed for your team sort of from an operational standpoint?

Yes. This is -- so the time when we met the Juniper team, the company -- the strength of the company today, we were just 1/3 the size. I can tell you that it's the same team that is managing the network today because of the design partnership, the trust, to be bringing the two platforms, the Marvis and ServiceNow that we have done here together, it has just transformed the way the team does the work. So the team that was a nonbeliever in some of these automation and the AI capabilities has fully leaned on that today and focused on doing much elevated work. So as I was talking about patching, I just want to add some numbers to it. We used to spend almost 2,000 to 3,000 hours a year. I'm just pulling out a very small number, I think. Now we spend less than 60 hours a year patching network. When engineers are sleeping, the network is patching itself. So there is an example. One of the examples of self-driving network that we are leaning on and the enterprise is testing.

That's awesome. Thank you so much for sharing that with us. So you've been on this journey with us now for a while, like we said, in almost 10 years. How has the partnership and the technology evolved over that time? And where do you see sort of the momentum right now in that partnership?

I think the -- I would say that we are just getting started because every time we meet there is something exciting and there is something always new going on. And as Rami, you touched upon, in the age of AI, when the data is sitting everywhere and the AI is distributed, whether it is in your iPhones, whether it is laptops, it is no more the enterprise edge, it is actually going to the consumer edge. The role of networking is just more important than ever before. The nanosecond, the microsecond, subsecond latencies that network needs to support for the best-in-class inferencing that needs to happen, I think that network has a huge role. And that's exactly how we see it in ServiceNow that we want to continue to build with you. We love the design partnership. We have brought the two platforms together. We present these stories, and we tell this real -- show these real demos of these two companies and the two platforms coming together, and it has blown customers' mind -- combined customers' mind of how we are able to achieve greater things because network is -- I want to say that when people ask me, what do you do? I said, I run a utility-grade service. What does that mean?

It means that when you flip the switch, the light has to come on. When I turn the tap, the water has to flow. So there is exactly like the working. Nobody needs to walk into the office, thinking about whether my Wi-Fi is going to work or not. It is supposed to work. It is going to work. And it is. If it is not working, it is supposed to inform you proactively that I have a problem. I'm working on fixing myself. Please show up on the other side of the campus, and you will have a fantastic Zoom call that you can take from there.

I love it. The network has become, as I said, the critical infrastructure for the AI era. So the big question we get is around trust, Sajeev, so handing the network operations over to an AI agent is never an easy thing to do from a human trust standpoint. How did you overcome that obstacle and start to get comfortable with taking your hands off the proverbial steering wheel, if you will?

Absolutely. And as I said, my team was non-believers, and there was this hesitation that is this going to -- we found ourselves in crosshairs with all sorts of issues because anything that goes wrong in the company, the first blame was on network. If a Zoom call is choppy, it's the network. If the application is slow, it's network. They said, "Okay, is this thing going to make it absolutely worse and is going to go out of control?" So we had the hesitation, but being an AI company that we are today and being an AI-first approach that we take, we said, "You know what, we're going to lead with this, and we're going to trust on this AI that we have in our hands."

But it has to be essentially a graduated autonomy that we need to lean on. We need to slowly peel the onion, and we need to keep making the progression. It was imperative that we are going to keep marching forward and not look back. So today, the platform, which is AI-driven platform from ServiceNow perspective, from the Marvis perspective, I think we are definitely looking at something that is going to change the way we operate. Personally, we have a bold goal within ServiceNow, and that is 100% autonomous network by 2028, right? And the only way we know that is going to happen is that if we keep our mind and our thoughts open, and we keep working and designing with Juniper HPE team. I think we can make that happen. And I'm super excited. I mean all your updates that just -- you just shared, it was absolutely net new to me. It has just blown my mind and it only tells me that we are partnering with the right company, and I can't be more excited about what we have coming up in the future.

Wow, I couldn't thank you enough for the really inspirational words. We love ServiceNow. We love our partnership. You have made us a better technology provider in the time and through the advice and the engagement that we've had over the years. and I can't wait to take it to new highs together. Thank you so much for joining us. Thank you. Appreciate it. Thank you.

Okay. I love that. So I have talked about the opportunity, the integration progress and our innovation. Let's now discuss how all of this is resulting in financial strength. A year ago, investors were asking whether we could integrate Juniper without disrupting the business. Today, we've demonstrated that the combination is already creating significant value. But this is about looking forward. The market opportunity is expanding, demand is strong, our execution is ahead of plan and visibility into future growth has significantly improved. We continue to see a strong pipeline across the entire portfolio from Campus & Branch driven by enterprise modernization, to Data Center Networking and Routing driven by the massive AI investments. Based on continued strength and demand, we now expect our fiscal 2026 networks for AI cumulative orders to exceed $3 billion. This is up from our prior estimate of $2.5 billion to $3 billion communicated by Antonio and Marie at Q3 earnings.

Now as we continue to look forward, our focus is on accelerating revenue growth. The demand is certainly there, and this is reflected in our orders, which in Q3 grew 3.5x faster than revenue. We've been working through supply chain availability to meet this growing demand. And in the last quarter alone, we doubled our networking supply purchase commitments, which we're confident will help ease constraints as we move into fiscal 2027. So based on strong portfolio demand, the incremental Helios opportunity and improved order conversion, we now expect fiscal 2027 HPE Networking revenue growth in the range of high teens to low 20%. This is up from the prior FY '27 guidance of 14% to 17% communicated at Q3 earnings. We do expect a more back-end loaded year than normal given the timing of Helios shipments and the anticipated contribution from the Oracle agreement.

But revenue growth is, of course, only part of the value creation story. We expect our growth, increased integration cost synergies and operating leverage to drive margin expansion. As mentioned on our Q3 earnings call, we expect our fiscal 2027 HPE networking operating margin to expand to mid- to high 20%. This is up from our FY '26 expectation for low 20% operating margin. We're focused on executing through the current market environment, but we remain equally focused on generating durable profitable growth, which means looking beyond FY '27. Looking ahead, we see continued broad networking demand driven by the AI secular trends I described earlier today. So with that, we're projecting a high teens revenue CAGR through fiscal 2029, with operating margins in the mid- to high 20% range. And we expect to gain market share in Data Center, in Routing, and Campus & Branch relative to the multiyear CAGRs we have provided earlier.

So now I've covered the significant and growing networking market opportunity, I've outlined why our strategy, portfolio and innovation gives us the clear right to win, and I've shared how we're translating these advantages into growth and value creation. We're going to transition to a panel now, so let us welcome Shannon back to the stage to moderate that discussion. Shannon, come on up.

Thank you, Rami. You've heard Rami discuss the networking opportunity, our integration progress and the financial framework that supports our outlook. What I'd like to do next is make it even more tangible. The innovation happening across networking today spans everything from AI data center fabrics and routing infrastructure to security and autonomous operations. Please join me in welcoming the leaders responsible for those areas. So thank you all for joining us. One of the themes you've heard throughout the morning is that AI is changing what customers expect from their networks and expanding the role networking plays across the broader infrastructure stack. You are each leading a different part of the business affected by these trends, so can you please briefly introduce yourself, explain the business you lead and where you have seen the greatest benefit from bringing HPE and Juniper together. We'll start with Praveen.

I'm Praveen Jain, SVP, GM for the data center networking business. I have been part of five start-ups in my 30-year career. And the start-up rigor and the HPE scale is bringing such a big opportunity in front of us that I'm super excited. One such example is the Helios tray or the networking tray, which is on the right side of the room, in the morning, people made comments, it looks like a piece of art, such a complicated technology. And honestly, some of this was not possible without HPE and Juniper coming together. Please find me during the break time, I would explain to you how 1,700 wires are into it, how the liquid cooling technology is playing into it. It's my heart, which has been put into this product. So thank you.

AE?

My name is AE Natarajan. I'm SVP, GM of the routing infrastructure solutions business here at HPE. I'm most excited about this AI opportunity over here. Praveen talked about the right, if you look at the left, the most complex, most compact routers excite me and the opportunity that it actually gives us in terms of what we have there. And least but not the last, getting HPE and Juniper together gives us the ability to scale up and address this market with compute, storage and networking together giving us a total solution across the portfolio and making us one of the biggest companies in the world.

David?

I'm David Hughes, I lead our security business. I joined HPE in 2020 when they acquired Silver Peak. I was the Founder and CEO of Silver Peak. In terms of what's really exciting about Juniper and HPE coming together, I think from a security perspective, first of all, Juniper had a really strong portfolio with SRX, with Advanced Threat Prevention, with Threat Labs. We're bringing that together. And what is really exciting is being able to build an autonomous full-stack end-to-end self-driving secure systems. So embedding that security technology into everything that we do. That's as important if not more important than just the portfolio that I manage.

And last but not least, Sujai?

My name is Sujai Hajela, I'm the EVP and General Manager for the Campus & Branch business. I came to HPE as a part of an acquisition of Juniper, and I actually came to Juniper as a part of an acquisition of a company I co-founded and was CEO, which was Mist. Excitement, it's very simple. It's the self-driving network. And what is even more exciting Shannon, is the coverage and the scale, as Rami had mentioned that HPE enables us which allows us to take self-driving networks everywhere.

Great. Well, Praveen and AE, I'm going to start closest to me here. Before we get started, can you walk us through a couple of terms we've heard a lot about lately in networking? So Praveen, can you please briefly describe scale up and scale out?

So when you connect GPUs inside a rack, like my networking tray goes inside the rack, when you connect these GPUs inside the rack together, it's called scale-up networking. While if you take multiple of these racks, or even stand-alone servers and you want to connect them together, that's called the scale-out networking.

And AE, can you talk about scale across?

Absolutely. Scaling now has the third dimension. It's not just within a rack and within a data center. Now GPUs are required to be distributed across data centers. And when you need to connect them across data centers, that is what we call as scale across, in other terms, data center interconnect. Any of those terms that you would use interchangeably for scale across.

Great. And that's what we're calling networks for AI. And so I'm going to stay with AE and Praveen here. So what differentiates HPE Networking? Why are customers choosing us? And how does the Oracle opportunity demonstrate that? Let's start with AE.

Yes. So very interestingly, I talked about AI, transforming the needs of what the network needs to do with AI, which means it is not just scale, it is not just bandwidth, it's not just throughput, it's latency, any of those metrics. But it also means how well you actually have a portfolio with the right tools, the right routers, switches, capabilities to actually build this network out. And it starts with three fundamental principles that we use to deliver anything and everything we built starting from our silicon to our systems to our software.

The first part of it is sustainability, which essentially means we drive the most power-efficient network devices that you can get across the portfolio, whether it be with our own silicon or with merchant silicon. We drive that and we also make sure that these network devices have or use the most -- least amount of space, which means they are the most compact devices that you can get in the market. And these two are very important because AI is hungry. It eats up all the power and space, so you need network equipment that really needs to fit in this environment. The ability for us to do that is important.

The second part of it, which is a slam dunk, which is performance. The ability for AI to actually grow the traffic, require more bandwidth, do all of that stuff, which means we have to deliver 100 gig, 200 gig, 400 gig, 800 gig, 1.6T and beyond and be the first to actually do that. Our strategy to build it with our own silicon and to also deliver this makes it completely possible to build a complete portfolio with the performance that we have. Last but not the least is our ability to actually take AI and put it into our devices. You heard Sajeev talk about automation and self-driving. When you build these complex networks, it needs to work by itself as much as you can. And embedding that makes us bring a huge value to our customers in terms of sustainability performance and automation.

Praveen?

Yes. So I'll start with the breadth of our portfolio like we talked about, we have products for scale-out. We have products for scale across, even on ramp to the cloud through our MX platform, how users and application connect to the cloud. So we have the full breadth of the product, which customers need. But also, I want to focus on relentless execution. And this comes in many forms. As I said, it's about a start-up mentality. We had a startup within this big company. If you look at it, first, we delivered a TH5-based switch ahead of every OEM. We repeated that with Tomahawk 6-base switch, 100% liquid-cooled, first vendor to deliver.

And to deliver that, it was not just taking ASIC and put it on the board and you deliver. There were so many components around it, whether it's cold plates, whether it's liquid cooling technology, whether it was anything else, it was such a complicated piece, and there's a reason why others were not able to deliver, Shannon. We were able to deliver. We leverage the HPE expertise, we went around and talked to every single component vendors and said, "We need to make it happen," and we made it happen. And now when customer needs you, let's say there is a problem. Problems do happen. Customer needs you. We are there at any point, any time of the day because for me, it's the customer experience, which Sujai calls it as customer experience, that's what matters. I believe when I am in their shoes and my network is not working, I will be yelling, honestly. That's what I want to do. That's the reason I'm there. So that was point number two.

But also, if you look at it, again, going back to my start-up mentality, if we are not innovating, let's say I'm executing, but I'm not innovating. This AI is the fastest changing environment. If I'm not innovating, I'm not doing my job. So if you look at the customer like Oracle, we are innovating, co-innovating with them in getting the deeper network visibility. Let's say something fails, network tells you what exactly happened. Why -- where exactly. You don't go need to look and fix, that is my co-innovation track. And last but not the least, company like HPE, the HPE Financial Services, I had no access to that as part of Juniper. Amazing results with combination of all these four factors.

And Shannon, I just want to say good morning, that's relentless execution.

Yes, we're very happy to have them all with us. So I'm curious, AE, and I do think this is one of the underappreciated parts of what HPE Networking has. And it's the combination of what Aruba had in terms of silicon development and then clearly what Juniper had invested in over many years. So from a customer perspective, what does it mean that we have our own silicon?

It's a lot. Rami touched upon it. It says in the AI era, you're basically taking customers and bringing them into the AI world. You're also taking GPUs, which cannot operate by themselves without a network, whether it's scale up, scale out or scale across, and you need to connect them together. And the best way to do that is to have the right tools and our silicon gives us the capabilities to have these right tools and the right devices for it. The best way to illustrate this also is by an example. We talk about building networking gear where we increase our throughput, we increase our performance. So I was working with one of the largest hyperscalers. They spent a lot more money with GPUs, and they put these GPUs in front of them and connected them together. And they were finding with various different network devices, these GPUs were waiting for data, waiting for the network.

What does this mean? This means you need to stop these GPUs from waiting for data and give them the data at the right time and the right way that we can manage. With our own silicon, we have the benefit of programmability. We worked very closely with them to program it in such a way that when they use our networking devices, the traffic management whether it is east-west traffic into GPUs or the north-south traffic across data center interconnects and anywhere else that you would take becomes the most efficient, and it's easily measurable by the GPU duty cycles that you get. That is the power of our own silicon delivered in our networking products.

That's great. So for both Praveen and AE, curious, what are you seeing in the pipeline today? And how is AI changing demand across data center networking and routing? Are these orders AI-driven? Is the interest AI driven? Or is it more a traditional data center refresh? We'll start with Praveen.

Yes. So we are seeing pipeline in both traditional data centers as well as in AI data centers. Let me start with traditional data centers. So first of all, we have a highly differentiated management platform for managing your networks. It used to be called Apstra, now it's called DC Director. But think about this way, where it's not just managing discrete elements. It's trying to create a graph of who is connected to what. So it means if anything fails or anything potentially is impacted by a change in the network, it will tell you because it knows that graph. On top of that, the technology which Sujai brought to the table with AIOps, by the way, tens of years of ago even before this current wave of AI showed up, I said, leverage that technology for us. So we leverage that technology, brought it to data center.

Let me give you an example. So let's say, optics fail all the time in the data center. Let's say that Shannon, you have two options. One option is an optics is failing or failed and you need to rush some engineers because application is impacted, option number one, which is our competitors are doing. Option number two, using my AIOps or this AIOps technology, I will tell you that 2 weeks from now, your optics is degraded 2 weeks from now, it is to a level where it will start impacting your performance, which option will you choose?

I would hope I would chose #2.

So then you should choose HPE products. All right. So that is about traditional data centers. Similarly, in AI data centers across the board, it's about execution and staying ahead of the market. Helios, why I'm so excited. This is our open Ethernet ecosystem where we are able to create a scale-up switch ahead of everybody else.

And somebody might say, "I can create one, this and that." In this switch, we have our own AIOps if you have 1,700 wires going into the switch, do you think somebody -- something could fail potentially or degrade or even if you did the best technology in the market or something might need attention, I'm bringing that AIOps into the same train. That's my differentiation on top of the complicated piece I generated. So pipelines from traditional data centers to the new areas we are investing in, like in scale up we are just seeing growth in all the markets.

That's great. And everyone definitely should go and see the switch you have back there or the trays because it's -- not only is he very excited about it. It's a really cool piece of technology. So AE?

I'm really excited about two different things that AI is bringing to us. One is GPU clusters to cluster connectivity which essentially, what we call as data center interconnect when you go across data centers. Out there, we talked about our PTX12000, which is the most compact router. Think about it. This gives you in 12 slots what our competition would give you in 16 slots or even 18 slots, which doesn't even exist. 30% more ratings and capabilities and throughput that you would need with all of the value that I talked about, with the traffic load balancing with all of that stuff, that is really exciting and interconnecting AI clusters across the globe, whether it is hyperscalers, neoclouds, all of those things is exciting, and that's a huge, big opportunity.

The second biggest opportunity right now is when AI is connecting customers into the AI clusters for AI consumption. And when you do this, no better than our Trio-based MX portfolio, which gives you the security, David, which you use anyway, and it gives you multi-tenancy, it has in-line security built-in and capabilities to program it effectively to make that the de facto on ramp. And it's used by every hyperscaler in the world to get their customers into the AI clusters.

And how are things changing? I mean you can't turn on a TV today or open a newspaper. No, we don't open newspapers anymore -- but look at your phone without seeing something about agentic AI. So how are things changing from a traffic perspective with agentic AI? What are you seeing in terms of customer buying and customer behavior and buying patterns? And how will this influence the next generation of silicon?

Absolutely. This is interesting and important. And everybody looks at AI and say, "Hey, AI is making more demands into the network." Let's take a step back and look at three important things. In your house, you always had an upload speed of X and a download speed of 10x. Now with AI your uploads and downloads are going to be synchronous, which means you need the same amount of traffic because you're sending rich content, images, content upwards for AI as well as for downwards, which means your network has to transform and address that. The second part of it is AI does not allow you to cache. You could take a movie, you could cache it and then you could stream it. It's a lot easier and simpler to handle the networking challenge that way. But you cannot cache because within 10 milliseconds, it gets obsolete.

Last but not the least, we have a new plethora of users that Sujai and I were talking about, the digital user, right? Many of you traveled, you might have jet lag, you might fall asleep, you didn't have your coffee, but digital users don't fall asleep. They are always on, which means they drive the network 7 by 24, every second, every millisecond away, which is important. This transforms the network, and we really need to build the network to address this.

That's -- and I mean, I think -- how do we think about AI inferencing? Because so much is changing, right, within the network within the world over a very short period of time. I do think, ultimately, we're going to look back. I mean, Juniper has been such a phenomenal acquisition for us because we bought it right at the time that this was all inflecting. How do you think about AI inferencing and what it means for your business, AE? Because we're all going to be entrancing closer to the edge connectivity is becoming even that much more important. So how does that play into your plans?

Absolutely. Inferencing is catching on. It is actually more mainstream now, which means you need to consume AI. You have trained these models, you get these models out to the edge and you want to actually leverage them to increase their productivity with agentic AI and everything else coming in. And as you drive it closer to the customer who is -- or the digital user who is using AI, you need the ability to have multi-tenancy, which means, when you get that traffic, you need to be able to take multiple of these traffics and bring them back in. You also need to have built-in security, so one traffic doesn't leak into the other, which is also important.

And I'm going to touch upon something very unique with our Trio. Rami said this, the most programmable networking chips, sometimes he says it's the CPU of networking, where you can program it to anything. I was actually dealing with a customer, and they wanted to actually -- and this is true, where they want to actually have packets go up to LEOs, to orbit, satellites, right? And when they switched from one satellite to another, guess what, you need to be able to balance this really well. Our Trio is programmable, you can actually write code into the Trio with the data plane in what we call as a capsule and make this the most efficient way of handling traffic that goes through low latency but high bandwidth links like satellite links.

And Praveen, we'll come back to you on this one. HPE has such a large enterprise customer base. And Rami talked about the opportunities to cross-sell and we have -- I shouldn't say we because we're now part of one, but HPE brings a wider geographic reach and very, very, very deep enterprise customer relationships. So can you talk a bit about how you're seeing the opportunity to cross-sell with your products across the entire HPE platform?

Yes. Actually, I want to start with this. It's a complete surprise to me how fast we started embracing each of the Juniper part and HPE part. Just to give you an example, GreenLake and OpsRamp and Morpheus, all our products, Juniper products are already integrated. So that's a cross-sell opportunity #1 in front of me. Anywhere you sell OpsRamp or GreenLake or Morpheus, you are there. Then look at the deals like Oak Ridge National Lab, which was a full stack opportunity, including networking, compute, storage. Wouldn't have happened without HPE being present, right? So in other words, that's my cross-sell or the full stack opportunity.

And then I own now as part of this acquisition, the data center part of Aruba business, and those customers, as they need the high-end switching gear, I'm cross-selling the higher-end data centers, which is into that. And before that, obviously, before the acquisition, we were cross-selling into Mist environment. Security plays a tremendous important role, as Rami mentioned, it's built-in security. So when I'm selling a solution, I'm selling, cross-selling with security. So opportunity is tremendous, just we need to just keep going with fire from all angles.

Great. So I'm now going to move down a bit in the panel. And Sujai, before we jump into self-driving networks, can you just give the audience an idea of what's included in our Campus & Branch portfolio at HPE?

HPE's Campus & Branch includes Wi-Fi and our campus-wired Ethernet switching portfolio. Key is to note that it's got a strong time with security, which David is going to be covering in more depth soon.

Great. And David, why don't you give us an idea of secure at HPE because we are a bit differentiated from what some would say are like the pure-play security providers.

Yes. So in terms of what we measure in our security business, we include SD-WAN, SASE, SSE, next-gen firewall and network access control. But what I really want to emphasize is my team's job is not just delivering financial outcomes relative to that part of the portfolio, but to make sure that we have an architecture that extends security edge to edge from the Wi-Fi access point, all the way through into the data center and making security a core part of our self-driving vision.

Great. And so now I'm going to talk a bit about the self-driving network. And I hope everybody saw the commercials that were up there as we were eating breakfast, but beyond that, if anybody watches football and I'm a huge football fan, especially college football. Our marketing team has done a phenomenal job working with the Mercedes team and our F1 sponsorship to come up with some great networking commercial, so please watch for those. But for now with the audience, what does self-driving network mean for our customers? And what differentiates HPE's approach?

Self-driving starts with a very simple philosophy. Up is not the same as good, which means Shannon, our network being up does not mean that you're having a great experience on Teams or Zoom or whatever app you're using. And that is the philosophy, which we are manifesting with the vision of a self-driving network. It is unique in its ability to deliver on that philosophy. Number one, we are the only ones in the industry that measure every user. Yes, every user, every minute across the global universe which is connected to the HPE Cloud and try to understand end user experience.

Number two, Marvis allowing AI engine actually works on troubleshooting or isolating a problem when that happens. And [ Daryl ], who runs our services and support, who is in the audience here, his organization customer success actually uses Marvis, this AI engine to also help with customer support. Again, the only ones in the industry, in the networking industry to do that. And third, guess what? I can identify Shannon, you're having a problem in our Teams call. I know how to fix it because the support team has digitized the solution, and all I need to do now is make it happen autonomously. Welcome to the agentic aspect of self-driving networks.

The differentiation leads to clear benefits. One of the fastest deployment times a customer that rolled the self-driving network across thousands of locations were able to drive their deployment times from years to months. Actually, one customer in manufacturing, Jennifer, you remember this, one customer in manufacturing, went and changed out the network during lunch hours. Welcome to self-driving networks. Number two, Sajeev spoke about this, ridiculous reduction in user-generated trouble tickets you'll hear 90% most of the time. And last but not the least, driving real business outcomes. The fastest check out at a retail store, number of transactions, number of packages shipped, amazing patient experience in a health care organization, welcome to the self-driving networks.

Great. So one of the things we hear from investors fairly often is what's going to happen with Aruba Central and Mist and how does this all come together? So can you talk about how you're going to make 1 plus 1 equal 3?

Maybe 5. But self-driving vision is common to Aruba Central and Juniper Mist. Every customer, regardless of which platform they are on benefit from this vision. We have actually realized products within a year, dual platform access points, Praveen, you're aware of this cross-pollination of AI models coming from Mist to Aruba, all done within a year of an acquisition, and you can check with other vendors in the industry. It's been a decade and they are still at it. And the most important part, the self-driving vision is assuring our customers on the longevity of the portfolio. Frankly, our financial results back that up. Record orders for Campus & Branch and Mist and Aruba, both are doing exceptionally well.

Great. So self-driving is obviously truly unique to HPE, and it's giving us key business benefits. But what are you hearing from customers about trust in the self-driving network? Because clearly, they've got to trust it if they're going to hand over all the keys to the kingdom to Marvis and Mist.

Customers actually trust the self-driving network because Marvis, the AI engine, actually works. That's key. Just to give you a simple stat, in the last 90 days, we ran a check on our cloud channel. Marvis executed thousands of trusted self-driving actions, where the network IT or the administrator did not have fingers on the keyboard. That is critical. Where does this trust come from? It comes from a unique trusted AI harness which has these memory of episodes, which Daryl's team from customer support has inserted into the harness, so we know how to solve a problem, defined scope, clear guardrails and we always provide the customer evidence and the recommendation. And then all we do is run the network on that trusted AI harness.

Actually, it was interesting channel. We recently had our self-driving summit with over 40 large enterprise customers and prospects across verticals and one pattern came out clear. Customers are looking at this network transformation, not just to replace hardware but to fundamentally bring this ability to drive the network on its own. And you would ask, what is this transformation? We heard about digital transformation a few years back. Actually, this is the transformation, which is of a new kind, which is what AE referred to, the digital workforce, the agents. The shift is driven now by a new kind of user. It's not about user, it's not about IoT. It's about the agents that never sleep. Self-driving is the way to keep pace using the same trusted models which we build for humans and IoT to now take care of the digital workforce.

Great. And I think one part of trust is security. Obviously, it's very important. So how do investors -- or how should investors think about our security strategy, David? And what makes our approach different?

Well, first of all, you heard Rami say the network has never been more critical to business outcomes. And nowhere is that more true than with the security related to the network. Everybody understands threats are escalating, particularly driven by AI and defense is critical. You need to use as a cybersecurity team, every tool that you have in the toolbox. And one of the most important tools is the network. Using the network as a sensor to detect problems and using the network as an enforcement point to stop problems spreading. For example, you take a health care organization as a customer example, in a health care organization, there's a lot of east-west traffic with medical devices talking to sensors and much line of traffic going back and forth, many devices that operate with old software. It's easy that for one of those devices to be compromised. What you want to be able to do is detect that compromise rapidly and then you want to be able to isolate the device immediately.

What's really different about HPE is unlike a cloud-driven security company, which is really optimized for traffic going north-south to provide a good experience for users using applications. We have the ability to do things in the cloud, but we can bring everything on-prem, not just to the firewall, the next-gen firewall where traffic exits the organization, but to extend that security all the way down to the access point and the switch port where those devices connect. So we can detect these problems immediately, and we can act immediately. Just think about it. The network is a sense that every single threat coming in, every piece of data that's being exfiltrated out is going across that network. That lateral movement is going east-west, you need to be in the path of all of it. We are with the network. And what's really different about HPE is we're not just focused on the cloud or the next-gen firewall, we're building that capability into the network.

Great. And AI is changing -- I mean we've talked about AI this entire morning so far as we should be. But AI is changing both the nature of security threats and the tools available to defend against them. So how does AI fit into our security strategy?

Yes. I think everyone understands AI is a double-edged sword. On the one hand, we can use AI to find vulnerabilities early, get them remediated, deliver better software. But on the other side, adversaries are able to use AI to find vulnerabilities to be able to move at machine speed to scale their operations by spawning agents. So what is absolutely needed is autonomous protection. Effectively, a network that senses and acts dynamically. In terms of AI, I put it into two categories, AI for security and security for AI. So when we think about AI for security, we're using, for example, AI to detect anomalous behaviors. We're using agentic technology, an agentic mesh to tie together AIOps for the security portfolio with Mist and Marvis so that we have full stack self-driving that delivers fantastic user experiences, helping the networking team, but at the same time, is providing a really useful tool to the cybersecurity organization with autonomous protection.

Switching to security for AI, we are extending zero trust from human identities to nonhuman identities. So securing those IoT devices, securing distributed workloads, maybe running inference, and of course, securing and sandboxing agentic AI for what you call digital users. We need to be able to provide zero trust that covers both human and nonhuman identities. A second example of what we're doing with security for AI is defending traditional and AI data centers. The SRX next-gen firewall is deployed in 9 out of 10 of the top cloud providers. We're also implementing features in the SRX to enforce AI governance and provide guardrails. So AI is driving a whole lot of what we're doing in security.

And what, Shannon, what David is referring to, and to the folks here, this became really real for us recently, Shannon, we were talking about this customer. They had an unfortunate incident, and you'd be amazed, they had no trust on what malicious thing was left in the network somewhere. So guess what they did? They ripped out each and every element of the network. That was the only choice.

But then the CIO and CISO asked us, we're not going to just build the same incumbent network again. And when they heard what David and I are doing together on network as a security sensor, that was a key reason for them to say this is the secure self-driving network we need to go in.

Well, great. Well, thank you so much. We're super excited with what Juniper's team has brought -- well, Rami's team has brought to the table for us, both in terms of the products, the services, the capabilities, the ability to leverage what you have across the entire HPE portfolio, I think it's just been a phenomenal combination.

So our opportunity is very broad, whether we're talking about AI infrastructure, routing, security, campus and branch or autonomous operations, and we're seeing the same underlying trends. Networking is becoming increasingly strategic to how our customers build and operate their businesses. So thank you to our panelists for sharing their perspectives. And now we're going to take a 20-minute break, and we'll reconvene after that for Q&A.

[Break]

Welcome back everyone. This morning, we covered the market opportunity, integration progress, technology differentiation and financial performance. We'd like to open the floor now for your questions. I've got Rami here with me on stage, and if you're in the room, please wait for a microphone, state your name and firm before asking your question. To give everyone a chance to participate, please limit yourself to one question, and we'll come back around as time permits. So before we begin, I'd like to remind everyone that this is a Networking-focused event. The forward guidance we have provided relates specifically to Networking, and we will not be providing any updated forecast for HPE as a whole. So please keep those questions for our fourth quarter earnings call when we will provide a wholesome guide for fiscal '27. And today, please focus on networking. So with that, let's open it up. And it's very hard for me to see, so I'm just going to sort of rely on Jess and Joe.

Joe Cardoso from JPMorgan. Maybe the question I have, and it's a two-parter, so sorry, Shannon. But if I look at the data center outlook you provided today, it implies you're roughly adding $5 billion in revenue through the forecast period. First, can you help frame the contribution between the opportunities there, enterprise scale out, scale up and whether there's any concentration that you guys are thinking around the opportunities or if it's more broad-based? And then as we think about data center mix moving from something closer to 10% to 15% of revenue for networking, to something closer to 1/3. How should we think about the gross margin implications from that?

Okay. I'm happy to address that. So the data center opportunity is the fastest-growing opportunity for us. We've outlined that just earlier today. The confidence level that we have in our ability to capture share in the data center comes down to a couple of things. First is the strength of our technology across every layer, scale up, scale out, scale across. And a second is the ability to leverage the HPE path to market to unlock new opportunities and we are already doing that. Helios is just an example of that. It'd be very difficult for Juniper to do that on a stand-alone basis. But having access to the partnerships that HPE has, the technology across liquid cooling, full rack architectures gave us that capability.

In terms of where we see the bulk of the growth. I think as Praveen mentioned earlier, we're going to see both strength in enterprise and in cloud providers. But certainly, I think it's easy to assume that cloud providers are going to be where the bulk of that growth is going to come from. And that's where the massive investments that are happening are. In cloud, there's always going to be some level of concentration. I mean, for example, we talked about the Oracle deal. That's a big win, but I don't think it's going to be a unique win. I believe there are going to be other deals as well that we can pursue as part of the strength of our portfolio and our go-to-market. Should I address gross margins?

Yes.

Okay. So gross margins, part two of the question. I think we have historically -- maybe this is maybe during the Juniper days, said that data center switching is on average lower gross margin, and that certainly would apply to things like scale out, but also scale up where the capacity is significant. Having said that, if you look at the business in its entirety from a contribution margin standpoint, where you factor in the cost of R&D, the cost of sales, I think it's operating margin neutral for us and certainly would be reflected in our outlook for next year and beyond.

Great. Next question, please. Erik?

Erik Woodring from Morgan Stanley. Just a quick clarification question, Rami. So thank you again for everything today. Just on the kind of fiscal '27, fiscal '28 outlook, can you just help us understand -- you reported earnings 28 days ago. So just where the increased confidence in that guide comes from, is it just Helios? Is it incremental supply visibility that you have? Would just love to get a little incremental color on kind of how you arrived on the outlook today relative to a month ago.

Yes. Well, relative to the last earnings call, I think the one new factor is Helios, right? And so that was -- that is now reflected in the opportunity for both next year and beyond. And as I mentioned, the thing that's so exciting about Helios is that it is net new opportunity. We are taking standards-based Ethernet to a new layer of the data center stack for the first time, so it's unlocking a massive TAM. And if you think about the capacity requirements within the data center, scale out is orders of magnitude greater than scale across and scale up is orders of magnitude greater than scale out. The capacity requirements are stunning in this area, so it unlocks a significant new TAM for us.

Great. Asiya?

Asiya from Citi. Just on that topic, Rami, can you just think about or help us think about what are you envisioning for the market share in these Helios racks? I think you identified the TAM over the next couple of years. And clearly, you have one large other ODM that's a competitor. So how do you envision your market share just in that Helios opportunity?

So it's a huge opportunity. And we didn't assume significant market share gains in our $1 billion or more than $1 billion outlook for the next couple of years, primarily because it's still early. That being said, I think as we go through the year and we start to see more pipeline, more visibility into wins, et cetera, that will give us even more confidence. And at that point, maybe we can talk about market share. It's still very early to talk about market share at this point in time. I'm just excited at the sheer magnitude of that opportunity.

And you rightfully mentioned today, there are a couple of players. There is one networking player and that's us, right? There are going to be customers that are okay with a more of a CM model for procuring the trays and the racks. For us, I think we're pursuing customers, and I believe there are a lot of customers that need a company that can do the kinds of things that Praveen talked about earlier with not just the development of the really complex hardware but the software integration that gives you the AIOps, the troubleshooting capabilities, the ease of management that I think only a networking company can do, and that's exactly what we're doing.

Great. Simon?

Simon Leopold with Raymond James. I want to see some of your insight on how the DCI market is evolving in that reflect back 3, 5 years ago, Juniper was basically part of a duopoly. And today, there are probably four players. So what are you doing to innovate, differentiate and how do you see that particular use case playing out?

Thanks for the question, Simon. I meant what I said on stage, it's fun now participating in this market opportunity because it truly does remind me of my early days when I was a chip developer at Juniper, and we were building more and more capable ASICs performance-based ASICs to keep up with the demand. We went through a long period where there was some doubt about the need for custom ASICs, the capacity requirements that are necessary in the next-generation routers, and we are now well beyond that, and we can't build them fast enough. So yes, it is true that there are more competitors today than I say they were probably back in the early days of the Internet, but the opportunity has exploded. And the need for routing solutions where we have the opportunity to innovate across the entire stack, from silicon to the systems to the software is absolutely there. And our customers are asking us for power efficiency, for performance, for scale, for programmability and also for silicon diversity, and we offer that. And I think that's what's helping us capture more than our fair share over the next few years.

Wamsi?

Wamsi Mohan, Bank of America. Rami, maybe when you think about sort of the opportunity that you alluded to in terms of synergies, you still capped your networking margin outlook where it was consistent, but you also have taken the growth rate significantly higher. Why isn't that necessarily translating into higher margin? Are you investing in particular areas that will kind of play out for higher growth in the future? How should we think about that?

So I'll start, and then Rami can continue. So when you think about the $800 million that we talked about in terms of our synergy target, remember, we started at $450 million when we announced the deal. Then when the deal closed, we took it to at least $600 million and now $800 million. So we're very happy with the trajectory that we're doing there. I think keep in mind, we're giving you a range from an operating margin standpoint of mid-25s -- or mid-20s to high 20s. And so I think as we look at the business, obviously, scale does matter and the growth is going to be there, but we're also investing heavily in the business to be able to drive that growth.

And we also think that we'll see sort of a linear progression over the next several years in terms of operating margins. So I don't know if you...

Yes. I'm happy to add. We were able to achieve these kinds of synergies without in any way impairing our ability to compete and grow. And that to me is the most important thing because the markets are significant, they're growing, and we want to make sure that we can be competitive from both a technology and a go-to-market standpoint. So we're going from low 20s to the mid- to high 20s over the next few years without in any way slowing the innovation train. And the innovation train, as you have seen in my discussion this morning with some of the demos in the back, is really, really strong right now.

The last thing I would just say is that we have been able to do these synergies at a lower cost than was originally anticipated. And as you know, Marie and Antonio are extremely focused on free cash flow, and I'm really proud of how the team has come together to drive that.

Louis Miscioscia, Daiwa Capital Markets. Maybe not focusing that much on the supply chain, but if you look at the growth ranges that you have given, what do you think would be the inhibitors to get you to the high end of the range in comparison? And all the numbers do look very attractive. So congratulations on that.

Thank you for the question. So this year, 2026, it won't come as a surprise to you because I think we talked about it in the last earnings call that we are -- our growth is supply limited. Next year, I anticipate that it will still be supply limited, just less supply limited. Quite frankly, we, including I, underestimated just how explosive the market growth is and also how good our execution is going to be in capturing that as we went through this complex integration process.

But we've now -- we are in the process of rectifying that, and we're investing in it. We've doubled our purchase commitments with our suppliers just over -- quarter-over-quarter. We've more than quadrupled it on a year-over-year basis. And the amount of attention and energy within the organization right now, not just within supply chain, across the executive circles, Antonio himself is getting involved to get us the supply that we need is absolutely there, which is why I'm so confident that we can deliver on this outlook for next year. Hopefully, as we get into '28 and beyond, things get a little bit less constrained, but we will see when we get there.

Ananda Baruah, Loop Capital. Thanks for doing this today. This is really fantastic. Appreciate it. With the broadened portfolio and with the amplification of scale out, scale up, scale across, what's a useful way to think about the company's opportunity at neoclouds now going forward, particularly given that it seems like for a large portion of the neocloud customer base, the AI server margin could be shifting upwards as well. You guys have a really interesting bundle. Would love to get your thoughts.

Okay. So it's a great question because if you break out the sort of cloud opportunity in hyperscale and neoclouds, and let's put sort of sovereign clouds in the neocloud category. The hyperscalers will make and will continue to make decisions on different technology components separately. They're going to make independent decisions on compute, on their networking, on storage, et cetera. And so we just have to go in there and compete for every layer of that stack based on the merits, the strength of our technology and our ability to co-innovate and really work with our hyperscale customers at a very technical level, which we know how to do.

As you get into the neocloud, which is the crux of your question, there -- typically, they are very technical. So you do have to engage with them in a very technical way. And obviously, the merits of our networking mattered a lot. But here, the opportunity to go in there with full rack solutions increases because many neoclouds value the simplicity of buying a complete system and buying from a single technology provider, right? Our win with Vultr that we announced this morning would be an example of that. It's the full portfolio that we're bringing to bear to win this, which again speaks to the revenue synergy opportunity that exists between Juniper and HPE. Would have been much more difficult to compete for the broader neocloud and sovereign cloud opportunity as a stand-alone company, Juniper.

Shannon and Rami, thanks for the event. You guys have a broad portfolio. But if I look at the portfolio as a whole and routing in particular, relative to your routing competitors, they have a very pronounced optical business and optical strategy. I'm just curious what Juniper's or HPE's optical strategy is? And how should we think about the CPO opportunity for HPE, especially given that you are pursuing the scale-up market?

Okay. It's a great question. So optics are already a very important part of our ability to compete and to win in the market. Now you talked about routing initially. And in routing, for certain use cases, let's say, when we participate in wide area use cases, optics is typically a separate layer of the network, and there are dedicated optical companies that go and pursue it. And our ability to win is not in any way affected by the optical opportunity.

As you get into DCI, which is a simpler use case, there is an increasingly large market for pluggables, pluggable coherent optics. Here, we have great partners. We, in fact, have engaged with our partners at a very technical level to build named branded HPE coherent optics to capture that opportunity. And that has worked extremely well with us, and I expect it to continue to work extremely well with us. As you talk about CPO, co-packaged optics, I think that is not an immediate opportunity. I think that's going to come about in a different layer of the data center stack.

I think, honestly, initially in scale up, in the next maybe 2 to 3 years. And we will talk more about that strategy and how we pursue that holistically as we get there. But for now, our ability to, let's say, participate in the scale-up market, in fact, the first Ethernet-based networking company to participate in the scale-up market is perfectly good and fine without having the CPO as part of our portfolio.

I think that's George.

George Notter from Wolfe Research. I guess I wanted to ask about sort of the new expectations you guys have for the networking business. How much of that is embedded in -- or comes from pricing versus unit growth? Is there a thought there? And then a second question I wanted to just ask was on scale across. Obviously, your position there, I think, with the PTX. I know there are others in the marketplace that have had a lot of success there. I think Arista, for example, has talked quite a bit about scale across opportunities. But how do you see your positioning in that market? And how do you think about where you fit versus others?

Okay. Great. The problem when you ask your questions is I forget easily. What was the first one?

Yes, I was just doing the same thing in my...

Pricing, thank you. The answer to that one is very simple. It is very much a unit-based growth driver versus a pricing base. We have, of course, increased our pricing to deal with the increasing components that we are having to manage. But for the most part, the vast majority of the growth comes from units.

The second one around scale across and our competitiveness in scale across. I'll just go back to what I said. I think yes, it is a competitive market. HPE Networking prior to that Juniper was by no means unfamiliar with competing in competitive markets. I mean we were up against 800-pound gorillas, and frankly every market opportunity that we went into pursued. But I do truly believe the thing that sets us apart and creates scale across is our ability to build every layer of the stack, starting with the silicon. That is, by the way, why today, we have the densest 800-gigabit Ethernet routing solution on the market. And in this market, density truly does matter.

And maybe that's a good opportunity for you to talk a little bit about the Oracle deal and why you think we won the Oracle deal?

I'd be happy to, Shannon. So we're not new partners to Oracle. In fact, we have been working with them on routing solutions for quite a long time. But I think with Oracle, the thing that they have seen from us is, first and foremost, an ability to keep up with their extremely rapid requirements. And that means from a technology standpoint, hardware, software and also new features and capabilities that they need in order to make their data centers work and to operate.

And I think it was a result of those years of working closely with them, co-innovating with them, demonstrating to them that they truly matter to us as a very strategic partner, we were just given a much larger opportunity. Couple that with the strength of HPE across services and support, financial services, et cetera, it was sort of a perfect combination to go and to secure what is a really huge opportunity for us. I shouldn't say opportunity, win for us.

Great. David?

David Vogt, UBS. Maybe, Rami, can you speak to the opportunity to leverage your campus relationships across the enterprise data center? Is there an opportunity to kind of leverage your historical multichannel solution?

And then I'll give you a second question at the same time. When you think about your silicon portfolio, how do you feel it stacks up against sort of the merchant silicon provider that's the lion's share of the market today? And how are customers viewing your silicon portfolio vis-a-vis what's out there in the marketplace and what's allowing you to take share from that perspective?

So let me start with the second question first. Our -- the customer feedback on our silicon portfolio is amazing. First, Trio is truly unique. It is the x86 instruction set for the Internet. And the level of programmability, the logical scale, user scale that it offers is second to none, which is why I think we do exceptionally well in the AI on-ramp use case. So think here hyperscale direct connect routers that allows them to connect enterprises directly to their cloud, whether they be AI clouds or traditional clouds.

For the PTX, I just mentioned, densest 800-gig density. And that's what matters for our customers, power efficiency, performance and density of interfaces. That being said, you're right, because of the explosive growth of this opportunity, more and more entrants are coming in and more people are investing in silicon. So that just means we've got to do the same. And I'm very happy to report that because of this combination with HPE, the level of investment that we are now seeing in silicon has only increased while delivering these financial results and the outlook that we provided. So that, I think, is going to help us in keeping up and maintaining our leadership position relative to others that do their own custom or to merchants.

The first question, which is around the wireless opportunity and the ability to sort of cross-sell between data center and wireless, 100%. Like I spend a lot of my time right now, along with my team that you saw up on the stage earlier, around thinking around what can we do to leverage the broader sales team that HPE now has to expand our reach. Now some of that comes from the global server and storage sales team. They talk to customers that are buying servers all day long. And guess what, every server has at least 2 Ethernet ports. And we want those Ethernet ports to plug into an HPE Networking switch and then ultimately into an HPE Networking router and into a security appliance.

So that's the first opportunity, but it's certainly not the only one. Even within Networking, we now have amazing technology. Sujai talked about the power of the self-driving network, but we have a far bigger go-to-market team that's selling data center networks, that's selling Campus & Branch networks. And so we have trained all of our sellers to sell the broad portfolio. And we have metrics. We're measuring the performance, and we're starting to see good early signs, but that's a huge opportunity going forward.

Louis.

Louis Miscioscia again, Daiwa Capital Markets. So a lot of things have been surprising us. Obviously, data center has been incredible and strong, but obviously, now servers are kicking in, took a while, storage, too. Obviously, you gave good guidance for Campus & Branch. But could you go deeper into that? I mean, why doesn't that surprise us in a year or 2 of now with a much more material growth in single digits?

Yes. So the market as a whole is growing in the sort of mid-single digits. I think we said 6%. We're going to take share. And we're going to take share because of the strength of our portfolio, our technology, the true uniqueness of our self-driving capabilities. Like don't -- like please when you hear self-driving or automation, AIOps, agentic AIOps from others, you really have to look under the covers to see how much is actually there. And we're fighting a lot of noise in the system and a lot of people that are honestly copying our message.

And the way we're fighting it is simple. We're going to leverage our broader sales team, and we're going to ask as many customers as possible to just give us a shot because when they give us a shot and a shot looks like a proof of concept, they put the self-driving network that's either part of Mist or Aruba Central against anybody else in the industry, we win the vast majority of the time. If we can continue to do that as we have been doing and we start to reap those revenue synergies, the sky is the limit. But for now, I think high single digits is a good place to be.

Anyone? Okay.

Simon Leopold, Raymond James. You've given us some numbers around sort of the AI opportunity. And I guess one of the things I'm trying to get a better sense of is the attach of compute to Networking. So I know this is the Networking day, but I'm trying to get a little bit better context of how we should think about the mix or the ratio of compute attached to Networking and the opportunities to land bundled deals that way.

I think from a perspective of Helios, obviously, that's where we're seeing a big opportunity. And with what we announced today in terms of Vultr, obviously, $1.2 billion deal. A portion of that is obviously Networking. It's a full scale, though, it's a full rack. So what the company as a whole is thinking is we're going to have a balanced approach. Clearly, there's opportunity in racks. There's also a great opportunity to be kind of a merchant vendor of trays. So we will focus on that.

But the one thing to keep in mind is we continue to balance the working capital requirements of doing a full rack because you've got Antonio, you've got Marie, everybody is fully committed to at least $5 billion of free cash flow next year. So we're going to continue to sort of balance that. But obviously, this initial announcement that we made is a full rack, but we've also had some tray sales as well.

Yes. Just to add to that, and I think, Shannon, you rightfully talked about the Helios opportunity. And that's where I think you can expect higher-than-normal attach rates because it's an integrated system. However, we did say there is going to be different paths to market, including selling the Helios Switch tray through other Helios rack-scale providers. And that's perfectly acceptable. I think we are going to sell quite a few trades that way.

Going to the other part, like the non-Helios opportunity, whether it be in the enterprise, neoclouds, et cetera, attach rates are still low, but we want them to be higher. And a big part of our revenue synergies is to make them higher. Praveen talked about Oak Ridge as an example of an end-to-end win. There are definitely more opportunities like that out there to pursue.

It has -- there are a number of things that need to happen. You've got to get the sales team well versed in selling the full portfolio. That's now well underway. You want to create more hooks at the software layer between the different elements between networking and switching and routing. That's well underway. So I expect attach rates will only increase going forward, and that's going to only help the Networking business.

Wamsi?

Wamsi Mohan, Bank of America. Rami, in your comments earlier, you mentioned that some of the incremental growth that you're pointing to this year is going to be more back half loaded. Can you give us any sense of sort of the linearity of how this year might be different from other years given that you made that comment?

I mean the answer is no, not at this point in time. Just think more back-end loaded just because of the timing of the opportunity. It takes time to go from what is a relatively new system that's just about to enter into the market to a ramp with all of these cloud providers. Interest is high, but it's going to be more back-end loaded.

Yes, Victor.

Victor Santiago with Evercore ISI. Rami, you touched on this earlier, but I wanted to ask about the Mist and Aruba Central platforms. Can you help us think about longer term what the year-end goal is here? Do you see ultimately -- or do you ultimately expect Mist and Aruba Central to converge into a single management platform? Or do you see a value in keeping the 2 platforms and continuing to add interoperability like you did earlier this year at Discover?

So it's a great question. And the short answer is, no, they never converge into one platform, and they don't need to. But let me explain a little bit more. The applications, the AI operations, the hardware, the access points, the switching after only a year of integration have already started to converge. Marvis is the AI engine for both. We've introduced our first access points that work with both platforms. We've now added the CX switching portfolio to both Mist and was already there on Aruba.

So in essence, what we're doing is we have these 2 platforms, each is optimized for different deployment models. Mist is purely cloud and Aruba has more deployment options, including on-prem, which is important to many customers. But the experience to the end user is going to be the same, and it only becomes more and more the same in time. The experience to the operator, like you heard from Sujai earlier today, is going to become more and more unified. And there will be less and less duplicative investment that's required on our part to deliver on that unified experience. That's the strategy.

And I have to tell you, like somebody was just asking me earlier, what has surprised you, I knew that it would be somewhat complicated to converge these businesses together. But the speed at which we articulated the strategy and the speed at which we are providing proof points of technology convergence has even surprised me. My team has done an absolutely awesome job of this convergence, and our customers have taken notice. And honestly, our competitors have taken notice as well.

Great. Erik?

Sure. Here we go. So Rami, you highlighted that the unified partner program launches effectively in a month -- a little over a month from now. I'm just curious if you had to kind of like pick your favorite children almost, what products or solutions or even opportunities do you think you can have the most success in that cross-sell opportunity? And like from our perspective, going from 10% to what would be considered a success in 12 to 24 months, what do you want that kind of cross-sell overlap to be where -- versus the 10% that you outlined already?

So first, no, I'm not going to pick my favorite children. I don't have any. That being said, if you're talking about the channel, right, there, I believe it starts with Campus & Branch because much of the Aruba business, which went through the channel was a Campus & Branch business. And so the opportunity to unlock new partners to sell our Campus & Branch portfolio across Aruba and Mist is absolutely there, and that's work that's well underway. We want to make it absolutely frictionless for all 60,000 partners that I talked about earlier to sell that broader portfolio in an enterprise data center.

We want to grow from what is an AIOps amazing self-driving experience in the Campus & Branch into the data center. We can absolutely do it. We've already seen elements of that. Today, again, we haven't even done this yet. It's going to happen in the next few weeks. So the opportunity is still an unlocked opportunity. But when it does happen, we reduce the friction for our partners and our sellers, by the way, to sell the broad portfolio, I expect that the win rate, the cross-sell is only going to grow and contribute to our revenue sharing opportunity. So I'm not going to give you a number because it's still very low right now, and it's only going to go up, and we can probably share more in future events.

Okay. Well, I think we've come to the end of our Q&A session. So Rami is going to make some comments here and I will be back to give you a few logistics at the end.

Sounds good. Thank you, Shannon. Okay. So let me close this out for all of you. I do not think that there has ever been a more important time in HPE Networking. The network has never been more critical to how businesses operate and has never been more critical to how AI gets built, deployed and consumed. So in this environment, I think innovation matters, speed matters, software matters, silicon matters, infrastructure matters and increasingly, scale matters, and it matters more than ever.

But having any one of these things is simply not enough. Winning requires all of them. And that's what excites me about what we are building here at HPE. We have decades of networking innovation and some of the best engineering talent in the industry. We have purpose-built silicon, leading hardware and AI-native software. We have the speed and agility to move with a market that is changing incredibly fast. And now we have HPE's full scale, not just networking, but across networking, services, supply chain, compute, storage, financing, partners, customers as a strategic advantage. So the network has never mattered more. And I believe HPE has never been better positioned to win. Thank you all very much for joining us.

So thank you for your questions and for joining us this morning. We appreciate your continued interest in learning more about HPE and the growth trajectory of our networking business. For those of you who are here in person, we encourage you to visit the technology showcase area, and Praveen really wants you to do that and see firsthand the innovation we've been discussing today. So 2 logistical notes before we close. We do have a bus for those of you heading straight to SFO, and it will be leaving around 11:45. And lunch is available in the back of the room. So thanks for being here, everyone, and this concludes our program.

Well, welcome, everybody, to the HPE fireside chat at the Goldman Sachs Communacopia and Technology Conference. The privilege of having Shannon Cross here from HPE to join us. My name is Kat Murphy. I cover HPE and IT hardware more broadly here at Goldman. We have about 35 minutes for today's session, inclusive of Q&A. You have some safe harbor to read first before we get started.

I'm sure no one's heard this before. So my remarks may contain forward-looking statements. So please refer to our SEC filings included in our most recent Form 10-Q for a discussion of the risk factors that relate to our business. So thank you very much for having me.

Great. Thank you for being here.

To kick it off, HPE reported earnings last Wednesday, raised both your fiscal '26 and fiscal '27 outlook, now calling for 13% to 17% revenue growth, 16% to 20% EPS growth for fiscal '27. Before we dig into some more strategic questions and talk through the drivers of that raised outlook, could you provide a brief recap for the quarter? And any key takeaways that you think are important to level set for this audience?

Sure. We were very, very pleased with the quarter and the strength we saw. I mean our revenue was strong. Our order growth was even stronger. We exceeded EPS significantly from the guide we had provided. We did $1.11 in earnings this last quarter, which was the first quarter the company has done over $1 in earnings. So that was pretty exciting. And we also had a record free cash flow, and that's something that's near and dear to my heart for the quarter as well at almost $1 billion.

So overall, we're seeing very strong demand on the networking side. Our revenue grew 10%. We had 36% growth in order. So very good there. Our storage business actually grew 10%. We saw twice the growth there in terms of orders as well. And we had very strong server growth, especially on the traditional side, and our order growth in traditional server was 75% year-over-year. So overall, demand remains really strong. I think what the management team right now is extremely focused on is converting all of those orders into revenue as we look forward to a really strong '27.

Maybe following up on that point, you've talked about guidance being constrained by supply more so than the demand element, which remains very strong. In your 10-Q, there were some disclosures around $30 billion in purchase commitments, stepping up from $6 billion last quarter. Can you talk about the improving visibility that you have into supply and what's driving some of that opportunity to raise your fiscal '27 outlook?

Sure. Yes. We're very happy with what we've committed to. I would say what we disclosed during the call was basically that our purchase commitments for networking doubled quarter-over-quarter. So obviously, if you see 6 to 30, part of that is networking, but the remainder is focused on cloud and AI. And as you can imagine, that basically is focused on the components that are in the shortest supply right now. What we are doing, we've always signed LTAs.

I think maybe one of the misconceptions that has been out there, I think it's kind of passed now, but was that LTAs were kind of a new invention. They're really not. They've been around for a long time, and our company has relationships with the DRAM manufacturers going back to Compaq days. So very tight there, but we have signed multiyear larger agreements to lock in supply. And that is one of the things that makes us feel more confident in what we gave for fiscal 2027.

Got you. So let's spend some time on the networking portfolio. You raised the networks for AI order target to $2.5 billion to $3 billion for fiscal '26. You've talked about the opportunities across scale up, scale out and scale across. But before we go into each of those in more detail, where do you see the biggest opportunity for that networks for AI number and then in '26 and '27 and then going forward, maybe across those 3 buckets?

Sure. So yes, networks for AI, for those who are keeping score, we started back in October at $1.5 billion by the end of the year, and we're now getting closer, obviously, to the end of the year. And that target is now $2.5 billion to $3 billion. So we have seen significant demand in what was really a nascent market for us even a year ago. So we're very excited about our position and where we stand.

When you look at what we're offering, we're offering basically scale across with our PX router, which has its own silicon. We have the extreme silicon chip there. It's something that Juniper has worked on. We're on multiple generations now. And we believe that, that's a really strong product that's seen extreme demand. If you remember Juniper's history, that was the telco business before. And the telco business, I remember looking back at it and it was like, well, how low can it go? And now obviously, with AI data center and the opportunity there, we're seeing significant demand and hyper growth.

So like that side of it. We have the MX, which is the on-ramp, that's our Trio silicon and the MX router is basically on-ramp to the data center. That's seeing significant demand. And then -- and this is one of the reasons why Oracle was really interested in signing with us. We have our QFX, which is the first 100% direct liquid-cooled Tomahawk 6 Top-of-Rack switch.

And so we're seeing demand across the board. I would say routing is absolutely something that connecting all these AI data centers is in really strong demand in an area that I think we're extremely well positioned. But across the board, we're really happy with what we see in network for AI. And I would expect to see not an update, but more details provided. We're hosting a Networking Analyst Day on September 30. It's in the Bay Area. If people would like to attend, please let us know, and it will be webcast.

Great. Can you talk more about that gigawatt-scale Oracle deal that you announced last quarter? You mentioned the QFX platform as being something very attractive. But any more details on what really drove HPE's win of that opportunity and kind of quantifying the scale of the opportunity?

Sure. So this is different than the relationships that Juniper and Oracle had in the past. I think maybe some of that got lost in translation. This is absolutely AI data center. We're providing the PX, the MX and the QFX switch, basically running the back end of the data center, which this was a competitive bid or competitive contest that we went through. So obviously, Oracle saw value in us. I think there were a few things that really drove it.

One, they were -- I think the interest level in what Juniper can do really piqued up when we announced the direct liquid -- the DLC switch because, again, we're time to market versus the competition by at least 6 months. So that was helpful. It reduces power consumption. It's a smaller device. So it's basically the same thing that you would see in the data center for servers gets transferred over to what we're doing on the switch side. So that was part of it. I think there was an existing relationship between the 2 companies.

I think the scale and scope of what HPE brings to the table from a balance sheet perspective, Oracle has the ability to use some of our HPE FS, which is our financial services, our leasing business. So that was also of interest to them. And I think it's really what's important to us is this is a proof point that we can play in this market. We're talking to a number of other players in the market. And I'm pretty excited about this opportunity for us as we look to '27 and beyond.

Great. Maybe going back to routing. The routing portfolio, you saw revenue accelerate sequentially in the quarter. Can you talk about how much of that, at least in the [ black ] print was related to the core, we'll call it, cloud on-ramp business versus some of these new AI use cases?

The vast majority of it, I would say, is -- I mean, there's core, right, because we all know what core was doing. The growth is really from the new use cases.

Got you. And then back to the scale-out opportunity and talking about the -- really the switching platforms that you're selling into these data centers. Can you talk about the target customer for a Juniper data center switch and how that might be different than if it is different from the target routing customer and some of the use cases that you're going after there?

Well, I think one of the things that's really interesting about the combination of HPE and Juniper. And I think what we're talking about right here is really an example of 1 plus 1 equaling 3, which I have been looking for and trying to find examples of for quite a while in the company, and there have been some, but I think they're really starting to ramp now. One of the big opportunities is to go into and sell to the enterprise. If you talk to Rami, one of the things that they were focused on and one of the uses of OpEx, which dampened margins of Juniper was the need to build out basically an enterprise sales force.

And so combining the 2 companies gives him an opportunity to leverage his technology across HPE's existing channel and partner network. And an important milestone here is that while we combined the sales force back in January, on November 1, we're doing partner day 1. So all the partners will be able to sell both products, and that should be beneficial as well. So I think enterprise is key. From a hyperscale standpoint, Juniper also didn't have necessarily the relationships and the ability to go after -- now they probably would have built it over time, so I'm not saying it wouldn't have happened. But I think combining HPE and Juniper, there are really ability for the combined company to go out and sell some of these products into the hyperscalers in ways that I don't think Juniper would have been able to do on its own or certainly not as fast.

So again, 1 plus 1 equaling 3. But when I think about the opportunity, hyperscale, obviously, is significant. It's kind of like what we think about in server. There's a huge opportunity in AI server, right, for training. But now what we're seeing is this and then we'll get to it. But this big ramp on the inference side that we think is just coming, I think the same thing to some extent on the networking side. Big opportunity in AI data center, but then overall a refresh and a need to modernize data centers to handle AI within the enterprise market as well.

Got you. That's very helpful. Let's round it out with a discussion on scale up. You've talked about participation from a networking perspective in the AMD Helios Rack. Can you clarify like what -- to the extent that you've talked about it, what HPE is doing in the scale-up opportunity for AMD Helios and how that we can think about maybe putting a framework on the size of that opportunity. Now we haven't -- it's not included in the TAM or it's not included necessarily in the outlook, but talking about how investors can think about the size of that opportunity?

Sure. And I think you can also go back to some of what AMD has talked about in terms of the size and extrapolate from there as well. I think HPE will do 2 things with Helios. The first is we'll provide the whole rack, put HPE on it, server technology, obviously, the scale of networking technology, and that's one option. We're also going to be providing to the market trays. So basically, we'll provide on a white label basis, trays into the server competitors of ours, in a co-opetition, I suppose, that are going to be working with AMD on Helios.

So it's a two-pronged approach. I think initially, it will be most likely the full servers. But over time, there's only a couple of us that have been qualified in. It's not that we think maybe the market will stay that duopolistic for long. Who knows if this is really successful, more people will come in. But we're really excited about Helios for a couple of reasons. One, as you look at the -- what we have done on AI server over the past couple of years, we've really focused in on enterprise and sovereign.

And so this gives us an opportunity to become more aggressive in the hyperscale and neocloud space with an offering that has a better margin profile because it will also have our networking IP in it. So that is one area that I think is pretty interesting and exciting. And remember, HPE has a very long-standing relationship with AMD. So we leaned in heavily on this early on, and we're excited about the RFPs and that we're seeing right now.

And then the other side of it is this tray opportunity, which opens up a whole new market for us that I think we'll look at over time. And again, that's something that Rami and the team are working on, and we'll talk a bit more about at the Networking Day. So Helios, I think, is a good opportunity. In terms of size, we're not putting anything out right now because we don't want to get the cart before the horse. We want to make sure we have specific timing and numbers to provide to you. I would assume that by the time we report fourth quarter, which will be late November, early December, we should have more to say there in terms of the rollout.

Great. On the campus and branch side, you touched a little bit already on the synergies of this joint go-to-market from November 1, kind of, combined partner go-live that you're expecting, but we're also seeing strong momentum in the campus and branch opportunity overall from a Wi-Fi 7 led refresh, concerns around security. Anything to share just on how the joint Mist and Aruba go-to-market is shaking out relative to expectations, better or worse and how that opportunity seems to be unfolding?

Sure. So I think the Wi-Fi 7 refresh has been really positive and then adding Mist into our portfolio is really driving a lot of interest from customers. We were pleased with our low teens growth in terms of orders in campus and branch this quarter. We grew our revenue 8%. I think this is not -- obviously, it's a hyper growth level that we're seeing for AI data center in that, but this is a really nice slow -- not slow at that level, but steady business with upside. I think -- when you think about opportunities, we've signed some really big logos. I can't disclose them here, but impressive wins on our side.

And I think from a product perspective, we continue to basically move to converge using the best of both breeds over a long period of time. We're very, very cognizant in making sure that our Aruba customers are super happy and our Mist customers are very happy. And at this point, the response has been positive. So I like what we're doing in campus and branch. And I -- the Mist technology is something that in various forms and where it makes sense, we're going to look at putting across the data center as well.

Got you. Putting it all together, the full year outlook for networking for fiscal '27 was raised from 8% to 12% to 14% to 17%. Talked about all of these demand vectors. You also talked about doubling of network purchase commitments as well. How do we think about parsing out the drivers of that raised outlook across more demand versus better supply?

Yes. I think a couple of things to think about there. One, that does include some contribution from Oracle. We're relatively prudent, I think, in the way we're looking at it, but it does include some from Oracle and then some modest acceleration in the core business. I will tell you, management remains incredibly focused on order conversion. And so I think if Marie were here, she say, if we can do better, we will. We meet weekly, talk about what we can do to improve it.

And what Rami has said is sequentially quarter-over-quarter, it should get better. Those -- the doubling of the purchase commitments definitely help. And so I think that's something that's going to be really important because whether it's server or it's on cloud and AI or it's on the networking side, demand is absolutely not a problem. I think you've heard that from others in the industry. It's basically being able to deliver the customers. And so we're working as hard as we can to fulfill that.

To round it out on networking, you're guiding for margins to expand into the low to mid -- or mid- to high 20s for fiscal '27. Talk about how mix factors into this, maybe some other onetime things that are falling out from '26 to '27 that should get that nice margin expansion within the segment?

Yes. It's mainly synergies and what we're seeing from a synergy perspective, I think it's really important to emphasize that we're not only looking at synergies and ways to improve the new companies coming together. But we continue to invest aggressively in silicon development and new products and leaning into things like Helios and that to be able to drive growth going forward. And I think that it's really important for that because we spent $14 billion on this acquisition. And having spent 20-some-odd years on the sell side, I've seen a lot of failed acquisitions. This one is really going well. And I think that's because the company, the Board, the management team made a commitment to really look at the integration, drive for success, invest where it's required.

And I think that's important because legacy tech companies, which we used to be, now we're a growth company, so we're going to change our sweatshirts or whatever. But these kind of companies had a tendency of buying companies and kind of I don't know, kind of forgetting about them or letting them live on their own and then just moving on to the next thing. And when there was a downturn, cutting investment dollars in that, and that's absolutely not what's happened here. And I think it's really important because we want to make sure we're good shepherds of shareholder dollars.

Let's switch gears to cloud and AI. We'll start on the traditional server side. You mentioned orders up 75% year-over-year in traditional servers. Can you articulate for this audience your confidence in the durability of demand here? And I'll ask the question of units versus ASPs, but anything you can share just on what is underlying the confidence in the growth in demand?

Yes. I think there's a few things. And I mean, for those of you who know me, I've been around quite a while. And I truly believe there's something different here. Like there is a new TAM developing. And this is different than some of the other -- oh, they're going to move it on-prem because of whatever from a cloud perspective. I think there are key drivers here that are actually creating incremental demand beyond what we've already seen.

I would say the first is because customers need to invest in AI technology, they need space in their data center. So I do think that there's some refresh going on in the older stuff running databases and whatever just because you can consolidate servers. And when you consolidate servers, you save on space, power, et cetera. So there's a bit of that. But more importantly, what we're seeing and what we're hearing from customers is this idea of being able to contain token costs and security of data.

So from a token cost perspective, if you bring -- and we're drinking our own champagne here because my group is actually working on this. We're putting out effectively what you call an AI factory or what have you. Ours, in this case, is PCAI, which is our private cloud AI hardware, putting open-weight models on top, having an intelligent router that basically says, if you're a cutting-edge software developer, you can go up to the -- hit a frontier model and spend for tokens. If you're me sitting there doing sort of basic corporate stuff, I never need to leave the cloud. So our costs in that case are just managing -- buying the hardware and managing the hardware as opposed to paying one of the token providers.

So we're doing that or the large LLMs. So that's something that we announced, for instance, that we were awarded a $3.5 billion enterprise deal after the end of the quarter. And we wanted to highlight that. I mean, the number is great, but more it's this concept, and we're seeing many of these -- maybe not as big, but many of these opportunities come through from an RFP perspective, where large retailers, large technology companies, large banks, large pharma, they're all looking at this and going, wow, our token costs are going to ramp substantially. How do we manage it? How do we control it? And I've talked to transformation officers and CFOs across the country, different industries who are all working on this.

So I think this is absolutely a new reason for people to be buying hardware, and I think it's very sustainable. The other thing I would say is that there's absolutely concern about data sovereignty and security. And you don't necessarily want all of your data going up and somehow becoming the sum total of human knowledge. So from that perspective, people are really focused on gating things off to the best of their abilities. So it will be -- I think it's interesting, but in terms of how this market is going to develop. It seems to us, and maybe this is anecdotal, that it's ramping faster in the U.S., but it's ramping -- but we hear things -- again, HPE is in like 160 countries and everywhere. When I talk to some of the country managers, you hear anecdotes of more demand out there, but we've really started to see this ramp. And I think it's sustainable. And I'm pretty excited about it because it's cool to be in hardware again or at least in server again.

Talking more about that $3.5 billion enterprise hyperscale deal where you were engaging in some on-prem use cases. Similar to the Oracle question, what drove HPE's win in that particular deal? And maybe more broadly, when thinking about going after this enterprise and even sovereign AI opportunity, how much of your sales motion is led by the breadth of your portfolio in both compute storage and networking?

It's a good question. I think in terms -- and I have actually asked why we won this, because it's curious. I think we won it because we can do services, we can do installation, we can maintain it. My speculation here is that maybe some of these companies are getting a little stretched with all the growth, everything they have to do. And so they wanted a really real trusted partner that can come in and build a data center for them and help them out. So to be clear, this was a hyperscaler, but these are traditional server margins.

So we're very happy about the development of this business because, obviously, we all know that AI server tends to be a very low-margin opportunity. This is not. And the other deals that we're seeing out there are similar. So I think the interesting thing about HPE when it comes to our customers, one, we are like the Fortune 200 in terms of the customer base, Fortune 500. We have a really marquee set of customers. I'm sure our competition has good customers as well, but I was personally surprised when I came from the sell side and saw who we work with because it's pretty impressive.

So from that perspective, I think our existing customer base will be a really good area to mine as they go and they look at more inferencing options. I think that from our product perspective, we are the only one that can bring networking, server, storage, services. We have a professional services arm, HPE finance. So we have the ability to finance things. And again, if you're a Fortune 200 customer and you want to lease something, we're going to be happy to do it because you have really good credit quality.

And then we have this enormous channel. And that's another thing I think that's underappreciated at this company. The channel that we have originated out of Compaq and has been expanded over the years. It's a global, deep, very close relationship channel. And I think that's something that's also going to be really beneficial. Because one of the things that as you move a little bit down from maybe the super big customers can manage all of this stuff once you get it out there on their own. When you move down, they're going to need partners, they're going to need help and solutions. And I think what we have will be good. In terms of cross-sells, I think that's still something that we can candidly do better, but it's an area where between networking and server, we're definitely starting to see some cross-selling.

Let's touch on the AI server opportunity. You've been very focused and deliberate about maintaining margins in that opportunity set with a focus on enterprise and sovereign, but saw some nice orders in the quarter, $2.4 billion in the quarter, almost $7 billion in backlog. How should we think about the mix potentially changing now that you're getting more scale in the business and you're seeing some attractive opportunities both from an engagement with the hyperscale and neoclouds on the AMD Helios side? Like is there an opportunity to change your mindset on the AI server framework?

I think it will be really interesting to see how it plays out. I do think Helios provides a margin uplift when you combine in the networking and the server technology. So I'm excited about what we can do there. I think that overall, we are maintaining a balanced approach. We're not -- we won't -- we still bid for a lot of these, but if the margin is really not there, we don't need empty calorie revenue. I don't think investors necessarily want that. And we're really focused as well on driving our free cash flow, so you have to manage your working capital. And I think people probably saw we took our working capital -- or sorry, we took our free cash flow target this year up to $3.75 billion in fiscal '26 at least -- and at least $5 billion next year.

So this company is a cash-generating engine. It's a Board decision, obviously, in terms of return on cash flow, but -- or return of cash, but the bias has always been to give it back to shareholders. And obviously, we've said in fourth quarter, we're going to get back at least 75% of our free cash flow through share repurchase and dividends. So I think that's also a very important balance that we look at in terms of how much working capital do we want to tie up. For instance, you saw our inventory went up this quarter. That's positioned because we do expect AI server revenue to be up in fourth quarter. And we do have some big deals. We expect growth in fiscal '27 in our AI server business. You've seen some of -- they've been press released that we had some big wins from, for instance, the U.S. government and some of the labs.

Talking more about those U.S. government and lab wins and we'll call it the sovereign opportunity more broadly, really where the legacy of Cray seems to have a competitive advantage. You mentioned earlier that things seem to be ramping faster in the U.S. and internationally, but HPE also does have a very significant international footprint in EMEA in particular. Can you talk about your outlook for the sovereign AI buyer, maybe both domestic and international to come to market and how the profile of that type of buyer might look different from the neocloud opportunities you may have walked away from earlier?

Sure. Well, to be clear, when I talked about the fact that things were starting more in the U.S., that was really the inferencing. So the traditional server purchases this idea. There was an article in the FT, by the way, today talking about Latham & Watkins that are bringing on -- I don't know if they're our customer. I've asked, I haven't heard, but they are buying NVIDIA GPUs, putting them on-prem basically in a colo and managing it themselves so that they can reduce token costs. So just that's something that I think has got legs and it's kind of starting in the U.S.

From a sovereign perspective, we do -- we work with sovereigns all over the world, and we have very close relationships stemming from Cray and frankly, from HPE as well because HPE in Compaq server business had long-standing government relationships. I think from a margin perspective, if you think about the continuum, the more you do for a customer, the higher your margin. So if you're with a hyperscaler wrapping metal around an NVIDIA stack, margin is probably not that great. As you go up and some of these sovereigns are pretty sophisticated, a lot of them need a lot of help. And so that's kind of how you think about the margin there.

We've -- I think we've got several areas where we're talking to people, whether it's on Helios, whether it's on our existing NVIDIA kind of racks that -- especially in Europe, where there's a lot of interest. I mean governments are looking at wanting to have their own clouds within their own borders, whether it's for defense or a government kind of activity or it's literally to help jump start some of the start-ups in that within their own countries. So there's a lot of demand out there. There's probably some questions about where the dollars come from, from a government perspective. But in general, we see that as a really healthy market.

Okay. I'll ask one more, and then I'll open it up to see if there's any questions from the audience. Just talking about your cloud and AI segment level margin outlook for fiscal '27, guiding to 13% next year. You delivered 17% this quarter, but maybe some onetime items in there. Can you talk about the various impacts that are driving whether it be mix, whether it be supply, anything to consider when talking about that margin framework?

Yes. I think, look, if you look at the company, we did a 40% gross margin this quarter, pretty high. There were some deals, some things that came through. This is not really a component pricing situation. It was more like composition of deals in that, that benefited us this quarter. I think as we look forward, there's always a balancing act between sort of units and ASPs and then margin, which flows through to EPS, which flows through to cash flow. And in cash flow, you also want velocity, right? Because these -- whether it's the server business or it's a PC business, which we obviously don't own, you can really speed up your cash conversion cycle. So you want to make sure that you're driving a sufficient amount of units through the model.

I think we've -- we're definitely kind of leaning when we see growth next year to be more of a combination of ASP increase in units as opposed to this year, it's been definitely driven a bit more by ASP. So that can have a little bit of pressure on margin. The other thing I would point out is that we do expect our AI server revenue to be up year-over-year, and that does have a lower margin profile than traditional server.

Now stay tuned if we keep seeing the kind of growth rates that we're seeing in traditional server and the -- some of these deals that are coming over the pipe, maybe there's some opportunity there. We do continually strive to drive for the most gross profit dollars we can as a company.

Great. Any questions? We can get a microphone over here?

The $3.5 billion [indiscernible]

Oh. I'll give you a microphone.

That CPU server with a CSP, but it's on-prem?

Yes, it's for their internal usage.

Okay.

Yes. So just think about -- just like us and we put -- I mean, we use PCAIs, they're using more of a traditional server approach, and it's for inferencing.

Do you have any examples of what they're using this PCAI for or rather the compute for internally just to contextualize?

I don't. I mean it's an order that they placed. I mean I would say internally, what we use ours for is basically my chat doesn't. If I use ChatHPE, it doesn't go outside. It just stays internal, so I don't actually incur any token cost.

What would like the duration of a deal like that be because that's a big deal over a year, or...

I mean, it's just a server deal, so it's a fairly quick turnaround.

Okay. Are there more of them like that in the pipeline?

Well, I don't know if there are $3.5 billion ones, but there are several large enterprise customers with multi-hundred dollar deals that are out there that we're looking at. I mean it's pretty -- the amount of investment dollars that people seem to be -- again, it's early, but I think it's going to -- I think there's legs here. The amount of money that people seem to be wanting to put into on-prem is pretty impressive.

And is it all CPU, is it not GPU?

I believe it's predominantly CPU.

[indiscernible] margins.

They were traditional server margins.

We have a couple of seconds here. Do you want to plug the Networking Day again and let this audience know about any expectations for...

Yes, I think we're going to have the Network Investor Day, September 30 in the Bay Area. And basically, we want to explain who HPE networking is now. We'll have Rami, we'll have all of the direct reports. And I think it will be a good opportunity for people to ask a lot of the same questions you're asking now and hopefully get answers from the subject matter experts as opposed to me.

Thank you, Shannon.

Thanks.

Appreciate it.

Good afternoon, everyone. Asiya Merchant here. I'm part of the tech hardware, tech supply chain research here at Citi. Happy to see everyone. It's day 1 of Citi's Technology Conference, the afternoon portion here.

Very excited to have HPE. We have Marie Myers here. She is the CFO of Hewlett Packard Enterprise. We have a few other folks from HPE as well in the audience.

Before we kickstart, is there any safe harbor? We're good. I think -- okay. All right.

So this is a fireside. We have a bunch of questions here. I'm going to leave some time towards the end for investors. I just request that you please do raise your hand so we can bring the mic to you.

So Marie, you guys just reported a very, very strong quarter. And you guys are guiding for fiscal '27. You did guide for fiscal '27. You upped that guide now for fiscal '27 as well. Your order backlog, you normalized orders, like you said, was up 42%. You have very strong backlog.

One of the questions that we're asking is like -- as you step back and you think about the current demand environment and the fact that there's so much -- so many supply constraints as well still, what gives you confidence that even as you look into fiscal '27 that this is durable demand that you're seeing in your order book and it's not just customers coming to HPE or coming to one of your peers and saying, "I have an order, take us?" And so there is durability to this demand and not just people signing to make sure they get something?

Yes. No, first of all, thank you for having me at the conference. And as you just said, we posted, frankly, one of the best quarters I've ever had in my career. So I was delighted to have to present a set of numbers like that, and we did actually have beat and raised and we guided up our '27 numbers as well.

And in terms of the demand profile and what gives us confidence, we are in an environment where demand is just outstripping supply at levels that were probably unprecedented in the industry. And as we reported, our order book on networking was actually up 36% year-on-year. We're seeing a very durable sort of super cycle going on in networking, where it's refresh of networking, combined with new momentum, particularly since we closed the deal with Juniper around areas like data center.

And then on the Cloud & AI space, our orders were up over 70%, almost 75%, 76% on a year-on-year basis. And what we're seeing in that side of the house is just really strong demand in terms of data center modernization and refresh. But I think what's really interesting is the whole conversation around the data center has moved from just a pure refresh. You have folks that have got pressure on power, cooling. And the economics are pushing really hard on how do I really manage this environment, but you've also got new workloads like AI coming into data center.

And we did disclose in the quarter, a very large enterprise customer where we saw a very significant inference investment. So we're starting to see these inference investments really pick up pace in the enterprise. So that's what gives us the confidence. And that's why we felt really it was the right time to go ahead and reguide the framework, which we guided for 13% to 17% on revenue and EPS actually growing faster than revenue at 16% to 20% and at least $5 billion cash flow, which was, I think, a really great number for us.

Right. Especially given that you had an Investor Day not too long ago, where you were sharing targets, which were much lower than that. Right. Okay. And just -- you just talked a little bit about the constraints, right? So you wouldn't have raised your revenue guidance for fiscal '26 and fiscal '27 if you didn't have the supply, and that's been something that Apple is talking about supply. You talk to any of your peers as well, they're talking about supply constraints.

So just -- you obviously did have some supply agreements. Tell us how you feel about -- I mean, your purchase commitments went up. Like how confident are you of having the supply that's needed? And is that a gating factor? Like if you could get more supply, could your guide be even better?

Well, I'd say the guide that we actually gave was really anchored around the supply that we have confidence in getting for '27. So that's why we thought it was opportune to come out with a revised framework because we knew we would have -- we could see line of sight to the supply for the guide that we gave.

Now we did also disclose, in fact, we just filed our Q the day after earnings. Our purchase commitment level is at an all-time high, actually over $30 billion. So purchase commitments are up both on the networking side and on the Cloud & the AI side of the house. And those are multiyear, what we call long-term arrangements, LTAs. So that really gives you assurance of supply. Not necessarily price, but at least gives us the confidence that we could see supply.

Now obviously, the more supply, given the order demand, then there could be -- if the strength was there in terms of continued supply, then that would help us translate a lot more of that order book into revenue.

Okay. All right. Is outside of supply, is there any other factors that we should think about where HPE could track towards the high end of the guide that you guys have shared now for fiscal '27?

Look, I'd say there's puts and takes as always. I think Antonio did mention, and I do want to sort of follow up on this point that the Helios stack, which I know we're going to talk about here later is not included in our fiscal '27 guide. It is early days on Helios. We're super excited about it because it has a combination of both our server. But most importantly, we have a networking tray in Helios. And it's early days in terms of customers sort of trialing, looking at the workloads and POCing. So more to come in that space, but we're super optimistic on Helios, but none of that is included in our '27 guide.

Okay. And then just in terms of visibility, right? So your order book, like is -- are customers willing to sign orders that are maybe extending beyond what they traditionally have done? I don't know if it was a few quarters, it was up to a year. Are you getting better visibility from your customers? And are the duration of the orders extending a little bit beyond just a couple of quarters?

Yes, I'm just thinking about -- I think it depends very much on the customer segment. What we're seeing -- and it's probably a much more recent trend -- is that a customer will place an order and then potentially come back shortly after that and actually increase that order quite significantly. So -- and most of that we're seeing so far to date has been around AI workloads specifically.

So I wouldn't say there's substantial change in customer order profiles or patterns. But we are constantly reviewing our backlog to ensure that there is no evidence of double bookings or what I would say, sort of placing orders that potentially could be canceled at a later point in time. So we feel really confident in the quality of our backlog that we don't see any evidence of those types of things going on.

Okay. And gross margins, I mean, they were very robust. And -- I mean in the quarter -- sorry, in the third quarter, you guys even talked about it. And you said that we should expect a little bit of moderation here in your fiscal 4Q. But the pricing environment is still very robust, right? None of your peers are necessarily doing any discounting. So just help us, why margin should perhaps moderate or maybe there was a bunch of factors that led to margins being so robust in Q3?

So I think I commented on the call that our gross margins were 40% Q3, which actually was a record for the company. And it was a confluence of factors that really brought that gross margin out, and it was one, obviously, the pricing discipline that we've shown as a company. Secondly, the mix of deals in the quarter, particularly on the AI side, we did not have a large shipment of AI.

So going into Q4, I think I commented that we do expect to see an increasing mix of AI in the quarter. And then we had some benefits that came in with respect to the synergies that also helped to buffer some of the gross margins as well. Then going forward into Q4, we do expect not just a more nuanced environment in terms of pricing. We've come off some several months now, very aggressive pricing curves. We expect that to sort of moderate. We do still expect pricing increases to happen into '27, but the pace and the acceleration of those prices, we expect to slow down somewhat as we get into '27.

Now after Juniper's acquisition, you now have Campus & Branch, you have data center, you have routing, you have security. And you are now competing against very large competitors, right? I mean, you have especially on the networking side.

And so just help us understand where are you seeing opportunities to gain share relative to where you guys -- relative to Juniper on their stand-alone or where HPE used to participate with your Aruba portfolio?

Look, I'd say there's opportunities across the entire stack that you commented on. If we look at Campus & Branch, obviously, the combined portfolio makes us a very formidable player in Campus & Branch today. We have been going down a very deliberate path to integrate which was known as the Mist platform that came from Juniper with the CNX platform from our HPE legacy Aruba side.

And I think the Mist platform has exceptional AI capabilities. And we've been able to successfully integrate a lot of that actually into the CNX platform or moreover, continue to build off the Mist platform. In fact, we leveraged a lot of the data out of CNX into Mist to build even a more powerful AI engine. So I'd say we're continuing to leapfrog that AI baseline that we built in Campus & Branch.

And I see that as an opportunity for us to really potentially win more and more business. I actually went in the labs with the team a couple of months ago. And I'm a bit of an AI geek myself, and I was completely impressed with just how far we've gone in a very short time since the integration with that platform.

On the data center side, we were first time to market with a liquid cool switch Tomahawk 6 QFX. Nobody else in the market was able to do that. Shortly after the deal closed, we took the liquid cooling technology out of our server business and actually applied it to the Juniper technology. And so we had absolute time to market. We beat many of the names that you've mentioned. And I'd say that puts us in a prime position to win some of those large deals.

Now we just closed and announced a deal with Oracle in our earnings call. And clearly, that I think gives us a really good sort of toehold into the data center at a good proof point that we can become, I think, a very formidable player in the data center over time. And that deal in itself leveraged the strength of the entire portfolio you mentioned, it's a combination of routing. So switching and then services. Our services portfolio was actually probably one of the hidden jewels. It has a very high margin rate, very, very sticky. And then we also leveraged our financing actually to pull that deal.

Okay. Good segue into the next question. So how should investors think about the Oracle deal, right? I mean, the Oracle deal has routing, you have Express silicon in there. I think you mentioned software, you have AIOps. It's a multiyear, it's a gigawatt-scale agreement. So when you think about the revenue ramp and how much of -- how should we think about the timing of the revenue ramp over the next couple of years?

And could this be a proof point for additional, I don't know, neoclouds or sovereigns to come to HPE? Are you thinking about additional hyperscalers here as well coming to HPE for these kinds of deployments?

Yes. No, I think I'll start on your second point, which is absolutely yes. I think that's a great proof point for us that we could absolutely compete against the best of the best of this space and actually win a big deal like this. And as you said, it is one of the largest AI infrastructure build-outs. So we're delighted to be chosen by Oracle to be a key partner to have that level of confidence in our ability.

And what I love is it tests the whole portfolio. So I think it is absolutely a great example of opportunities that I hope we can continue to win. And I would say there is strong interest in our portfolio for the reasons that we just discussed earlier.

So in terms of the deal ramping and how to think about it, we just announced it, so it's fresh off the press. We see '27 very much as sort of the early stages of just getting started with that deal, and that deal is multiyear in nature.

So we do have an Investor Day coming up September 30th, and we'll provide some more details and insights about how we think around the timing of that deal. And then as we get up into our Q4 revenue guidance at the end of our Q4 earnings call, we'll give you some more specificity about the guides.

All right. When we talk about the networking, I know margins and -- both the top line growth rate as well as margins have been key focus for investors as you guys are -- walk us through sort of when you think about margin expansion in this segment, is it just revenue that's growing, let's say, mid-teens here? Are you talking about better order conversion that's underpinning that? You're talking about Juniper synergies that's underpinning your margin guidance? What should investors be looking for as you're looking at that margin expansion story within networking?

Yes. First of all, I'd say we're really pleased with the performance of the business. We did actually close 10% revenue year-on-year in Q3, and we posted operating margins at 22%. When we first bought Juniper, the deal itself was only predicated on cost synergies. So we guided investors in the Street to actually $600 million worth of synergies by the end of '28. And I'd say in terms of signposts, one of them is absolutely, are you on track to meet that.

So the answer to that is absolutely. In fact, we're performing better than we expected in terms of delivering cost synergies. You can see that in the operating profit that we guided for the remainder of 2026. And actually, we guided up into '27 for mid to high 20s in terms of operating margin.

And if you think about it, what's different between '26 and '27 is the fact that we'll have a full year of those Juniper synergies in '27. So that's absolutely helping to drive some of the improvement in the rate that we expect in '27.

Secondly, let me just comment about revenue. Obviously, very strong order performance. I think I mentioned 36% year-on-year. We're seeing better conversion in Q4 and then even better conversion into '27. So we guided for 12% in Q4 and then up to a range of 14% to 17% getting into '27. So we're expecting revenue to continue to convert at a better pace.

As I mentioned earlier, we have secured purchase commitments both for our networking business and for our Cloud & AI business. So we do expect to see that improvement in order conversion take place picking up in Q4 and then more so into '27.

We haven't spoken a lot about revenue synergies. I know Rami did at Security Analyst Day speak specifically to cost synergies and where we saw those coming from. As we get into the Networking Investor Day, there will be more to say in this space. So I'll just put another plug in for September 30, where I expect Rami will cover a little bit more of the success we've had here. But I would look at the data center deal and say that's a good example.

Okay. And then talk about AMD Helios. I know it's kind of -- it's a big opportunity. You guys are looking at scale-up now as well. It's not incorporated in your guide initially. I know you talked about early days. But just -- what should investors be looking for? I mean, are there any qualifications? What is it that's kind of going to help investors understand the size of this opportunity? And what could HPE's share be in scale-up? Could it be equivalent to what you have in scale-out? When we think...

I'm personally super excited about Helios because it gives us an opportunity to really get out there and participate in a segment of the market that we potentially haven't been as competitive in. What's great about Helios, it incorporates a networking tray. And then obviously, we work with AMD on the server architecture.

But frankly, by having the networking tray in the stack, it obviously gives us a much better margin profile on these types of deals. So what I would say is where are we in terms of looking at these opportunities.

Right now, we are in POC stage with a number of different customers. And I'd say we've got some strong interest. Customers are obviously testing those workloads because they have to evaluate performance. It's a new stack. Many of them have been accustomed to working on their existing architecture, their existing stacks, so they're happy to gain confidence that they can see these workloads really performing at the levels that they want.

I expect that once we get through that stage, we'll start to see those proof of concepts convert into orders and then obviously into POs. Once we see that happen, then we'll obviously bring that to you in terms of the guide. But this is very much, much more into '27 and beyond. It's still early days is what I would say. But so far, we've definitely got some strong interest from a number of different players around the world.

Okay. And I know Rami is going to talk a lot more about it on networking, but margins, and we talked about synergies. I mean, I know you were a big factor in driving some -- you're talking about those synergies even earlier on. Just what's gone better than expected as you talk about, whether it's channel, whether it's go-to-market, whether it's products, stuff like -- what has gone better? And what are the -- is low-hanging fruits done? Where are you guys now kicking it up so do you get to better synergy?

Yes. Like I said, I think really pleased with the performance of synergies. What went faster was probably just the core integration. I mean, in terms of just mapping a lot of the sort of overlap between the two organizations went really, really quickly. What I would say that very successful was the sales and go-to-market integration. That's usually one of the tougher sides of large deals like this I've been involved in a number of transactions in the past.

And you can sometimes get it wrong in the sales space, but the mapping that we did of our accounts and the actual sales day 1 that went live a couple of quarters ago was virtually very seamless, and we didn't see confusion in terms of customer account mapping, region mapping.

And the one that was probably the most -- probably had us the most focused going into the deal was the U.S. because that's where we knew we had potentially much more overlapping accounts. But so far, that's gone extremely well, probably exceeded expectations in terms of the pace and the sequencing.

And then I'd say on the product side as well, too, we've had -- you could have expected to be more sort of ambiguity around road map consolidation, et cetera. But for the most part, customers, I think, have been very happy with the way in which Rami and the team have navigated the road maps as well.

So pleasantly surprised. I always believe in deals like this, it does come down to people and culture. And the two companies, if you sit in the room with the team from HPE or the team from Juniper, it's really difficult to detect which company people came from.

I mean, the cultures are very, very similar and have a very strong ability to blend very quickly. And I think that, that was one of the probably underestimated parts of the deal that the new team would assimilate so quickly, whether that was technically or even out in the field in sales.

And I had a chance, like I said, to sit down with the Mist team. Some of the AI team from the HPE side is working hand-in-hand side-by-side, and you wouldn't have had a clue who would had worked for which company. Having worked on Compaq and Digital and all these other deals in the past, I must admit, I've had quite a few different experiences. So this one actually surprised me. And I think it's been a huge advantage for us getting into this stage of the transaction.

Okay. And then looking ahead, like sort of where is more of the focus going to be in year 2?

Year 2 is heavy going on IT systems integration. First year, we probably kept a lot of the dual systems sort of side-by-side. As we get into next year to '27, we're going to do a hardcore system integration. Big deal for me in finance because we'll map a lot of the financial architecture. So my team will be under a fair bit of pressure going into '27, but that will be the year where we get a lot of the system integration done.

Great. Switching a little bit to Cloud & AI. Orders were up quite meaningfully. I think you said 70% plus, 75%. Clearly, there's a lot of demand. I mean, I know enterprises are updating, modernizing. There's agentic AI, there's inferencing. But there's also pricing that was probably a large element of that. How should investors think about this unit and pricing demand versus -- unit demand versus pricing as you're thinking about into fiscal '27?

Well, I'd say, look, in terms of the relationship with pricing and units, clearly, pricing has played very importantly into the overall top line and bottom line results. But we can't -- units are also incredibly important. We're in an environment where today, demand continues to outstrip supply. And we don't see that changing into '27, which is what really underscored the confidence in the '27 guide.

What I would say in terms of just the quarter itself, I know we got some questions about, well, how do you see unit volumes sort of growing between Q3 and Q4. We were -- we announced that very large deal, the $3.5 billion inference deal, which is clearly, inference, traditional compute as an indicator of the sort of order strength and the volume strength that we see out there. So I do expect there's a little bit of a timing mix issue in terms of Q3 and Q4. We do expect to see unit volumes continue to accelerate even as we get into Q4 and beyond.

And clearly, as I mentioned earlier, some of those pricing curves that we've seen throughout '26 will start to moderate in '27, but we still expect to see an elevated pricing environment in '27.

Okay. And then talking about that deal just because it was a pretty sizable deal. Just how -- what differentiated, like what went your way? What led them to say, we'll give this large deal to HPE? And just given the size of this deal, how should we think about both from a margin perspective and working capital to support as these revenues ramp for this deal?

Yes. So maybe I'll just put up the working capital first. So this deal is traditional CPU. So it fits inside our normal working capital paradigm, which -- for me as a CFO, I really like this model because we get tremendous leverage because that has negative cash conversion cycle. So absolutely no impact in terms of degrading working capital. In fact, incredibly positive in terms of cash flow generation. Margins, right in our framework in terms of how we think about the business.

What I would say about this deal, in particular, why we won this deal, really twofold. First reason is they wanted a trusted partner, somebody that could manage a large-scale deployment like this that are very -- in a relatively compressed time frame that they could trust and that had the know-how that could handle the architecture requirements and also provide the service and support. We're like a one-stop shop for them because we brought all of those elements of the deal to bear.

And then secondly, the other piece, obviously, in an environment like this is you've got access to supply. So we were able to secure all the components required in that deal in the time that the customer wanted. So in these environments, you've got to be able to do it all. The customer wants to have product delivered in time and also have the know-how and the support, knowing that you're going to be there to kind of handle some of the complexities of potentially doing an install on this size and scale.

One of the topics that we've been focusing on is just this whole enterprise AI adoption. And not just on the Cloud, but on-prem. How are you thinking about the TAM that's available to HPE -- as the TAM growth that this is seeing? And how are you thinking about HPE share within that expanding TAM?

Well, I would say that as a company, we're incredibly well positioned for that expansive TAM that we believe is going to continue to grow. The reason for that, we've always been a company that participated in the hybrid space. And it is very much our DNA and legacy in terms of inference and on-prem.

And in fact, we did comment on the earnings call. Antonio and myself made a deliberate decision to actually build our own AI factory internally on-prem because we felt that given the rising token cost environment, we knew that we could have a significant amount of token costs by actually making this conscious deliberate investment. This investment, we believe we're doing it on our own architecture called PCAI, which is Private Cloud AI Architecture.

And we have sort of seen enough use cases now to see that we can save up to 60% on token costs versus, say, the Cloud. So there's very compelling use case environments, not just based on the cases themselves, but also in the economics. So we see this whole space really becoming very relevant for us and is an area that we feel like we're very well positioned to compete in.

Okay. And then these AI deals storage, right? It's a smaller portion of your Cloud & AI, and you do have a GreenLake offering there as well. Just how much storage is being attached now to these AI workloads based on your own experience, what you're experiencing in-house as well? But as you're looking at this Cloud & AI, to what extent are you seeing storage attach to these workloads?

So I would say that storage is definitely one of the parts of our business that continues to accelerate and should be a very strong beneficiary of this AI build-out. We have seen 10% revenue growth on a year-on-year basis. And actually, we saw even faster order growth. Orders grew over 20%. So that gives me confidence that we're starting to see what you mentioned in terms of storage becoming a more important part of AI build-outs.

I'd say it's still early days as companies, you're really talking about large enterprises that are starting to move in this direction. So companies are starting to consider storage as part of these deals. But it's an opportunity, I think, well into '27 and beyond.

But for us, storage growth has been something we've been very, very focused on over the past few quarters. We did actually build out a whole new platform that I think really provides relevancy for us today in the market. It's the Alletra 10000 platform. And we've seen now, I think, 7 quarters of consecutive growth in that platform. So we know we've got the right platform at the right time in the market today to really help serve customers.

Okay. When we come back to the Cloud & AI, there's a little bit of mix element there that tends -- especially on the AI side of things, I mean, that tends to depress margins, larger deals, GPU stuff. But you guys have also had a lot of experience with your past acquisitions, Cray, et cetera, with liquid cooling. We had one of your peers talk a lot about reliability and services attached that is growing the margin of these AI systems. How are you guys thinking about the margins for your AI business?

We've been very intentional, I think, around where we wanted to play in the AI space. And we've said quite deliberately that, first of all, we had a pricing framework that we wanted to use to help guide our decisions around these deals. So we've stuck very closely to our pricing framework, which frankly, at the end of the day has been really about building margin.

And secondly, what we felt as a company that we were much better positioned to play in both the enterprise and the sovereign space. And the reason for that is exactly what you just said is that given our heritage as a company, we had so much experience around liquid cooling technology and services that this is an area where we had, frankly, better margin attach. And we've actually seen that play out.

And the liquid cooling expertise, I might add, was one of the early beneficiaries of the Juniper deal, where I mentioned a moment ago that we were first to market with a liquid cool switch, which is QFX. We're able to leverage that liquid cooling heritage and apply it to a switch, which was something that none of the competitors were able to do.

So yes, absolutely, enterprise and sovereign, they're about 60% of the book of business that we do today, and we do see that as a natural sweet spot for us as a company. What I would add is, given the discussion we just had about Helios, as we start to see Helios become more mature, we do think it's an opportunity, though, to revisit some elements of the market that we perhaps didn't play in because now with the network tray in the mix, once again, we have a very strong margin profile to play with. So I think Helios will help us to open up some of that market that potentially was not as accessible to us in the past because of the margin profile on working capital.

Okay. Talking of working capital, I think that $5 billion in free cash flow is a pretty strong number. Yes. But inventory has also come up quite a bit. And so just walk us through like, as you're thinking about this high free cash flow conversion now, despite the higher inventory, what's driving that confidence that you can convert those orders very quickly into revenues, collect on those revenues despite the higher inventory and working capital needs?

Yes. So maybe I'll comment on the free cash flow drivers and then I'll talk a little bit about what's going on in inventory. So on free cash flow for '27, first of all, really, really excited about the fact we're able to give numbers of at least $5 billion is what I would say for '27.

Underscoring that is really two things. One, we've had restructuring for our synergy programs in our numbers for the last couple of years, and we'll see that start to taper off as we get into '27. So we've got less restructuring.

And secondly, just the tremendous earnings growth. I think I mentioned in the guide that we're guiding EPS to 16% to 20% on a year-on-year basis. And you're seeing that drop straight through to free cash flow conversion.

So that's really what's underscoring the tremendous performance in cash flow which Antonio and myself are very, very focused on.

In terms of the inventory comments that you made, right now, we're seeing elevated inventory levels, I think, across the whole industry. Obviously, with higher commodity prices, you're going to see all the tides are sort of raising at the same time.

So that's what you're seeing in terms of passing through predominantly some of the impact of the numbers on inventory. I'd also comment that we've actually got some AI deals that will ship in Q4. So some of that is just moving through inventory. It's more a timing issue.

But I would expect inventory levels will continue to remain more elevated than what we've seen prior to this whole commodity super cycle. And partly that's due to the valuation of the inventory, which is just grossed up due to the prices of memory, et cetera.

Okay. The Oracle deal also had warrants associated with it. I mean, it was -- the details were in your Q. You do also have -- on the offset side, you do have a lot of share buybacks that you're going to now accelerate given the free cash flow generation. So just on the warrant side, like what sparked warrants? Like why do warrants? And maybe you can give us like a rationale? Should we be expecting more warrants as you get perhaps more deals coming your way?

Yes. Look, I'd say we're starting to see warrants become more common in these types of transactions. And I think that was really nothing more to read into it apart from that. And secondly, I'd say what we did do is we attached the warrants to actually sort of like the stage gating in terms of the initial projects. So they're only sort of going to vesting based on the volumes associated with the infrastructure investments.

So it's not all at once. It will be stage gated over the multiyears that we talked about. And then in terms of cash flow and our capital allocation framework, we actually pulled up our share repurchasing into Q4. So we commented in the earnings call that given the fact that our leverage is now actually at 1.8. We're way ahead in terms of our timing. We said we'd get under 2 by '27. We actually got there this quarter, so a year and a quarter ahead of timing.

We actually announced that we will start to use the principles of our capital allocation strategy, which was to buy back at least 35% -- return at least 75% of our cash flow back to shareholders by share repo and also dividends. We're going to start that up in Q4. So what was really good is now we're able to really pull up the tenets of that capital allocation framework much earlier due to our leverage ratio.

Let me see if there's any questions from the audience here. There's a lot of changes that are happening on the technology side. Customers are dealing with reference architectures, changes in those reference architectures. You have a lot of chip providers as well with multiple -- like when you look at these inventories, you look at your order book, you're buying inventory to support that order book, how do you prevent or maybe there's some terms in these agreements so you don't deal with maybe some obsolescence risk here with the inventory that you've purchased?

Well, I'd say we've been incredibly judicious about assessing the inventory that we buy and ensuring that the reserves are appropriate. And I think we have a very strong process around our excess and obsolescence risk management. So at this point in time, I don't see any particular issues given the current environment with excess and obsolescence.

Okay. And then there is -- I know you're just digesting still working through the Juniper acquisition. But as you're sitting here, lots of free cash flow here. I know a lot of it is going to be returned to the shareholders.

But as you think about your portfolio, you have the Neocloud opportunity, pretty significant. I know you guys are strategic about where you want to invest: the sovereigns, of course. How are you thinking about all these various opportunities that are out there? And to invest in and relative to using the excess free cash flow to return to shareholders?

Well, I think we've been very diligent in our framework and excited that, frankly, that we have the opportunity in Q4 to accelerate. We're bringing up our framework to buy back at least to return back to shareholders at least 75% of our free cash flow. What I would say is that we are still -- like where you started the question, we're still in the early days of the Juniper acquisition.

We have plenty to do, I might add. So I think we're very pleased with where we're at and the performance of the transaction so far to date. But plenty of opportunity to continue to accelerate the business, as you correctly said, both in the networking space and in the Cloud & AI space.

So happy with the portfolio in terms of where it's at, and we'll be very focused on closing out this deal in every way we can and really bringing forward the power of the two companies that we put together.

Great. The last few seconds here, Marie, like what do you think is underappreciated about HPE's story?

I think the '27 story is incredibly strong. The free cash flow that we're posting is record levels for the company. Certainly, the numbers of at least $5 billion, I believe, are incredibly important. What you're seeing also is just the leverage in the model. As we continue to grow, we posted revenue of 13% to 17% in terms of growth, but we posted EPS growth of 16% to 20%. So you're seeing earnings power actually outperform revenue.

So you're starting to see just the benefits of the work that we laid and the seeds that we laid with a lot of the synergies and cost programs. So leveraging the model, operating at scale, winning in key places like data center and inference. That really should set us up very nicely for '27 and beyond. And posting cash flow numbers that are, I think, very noticeable.

Citigroup Inc., Research Division

Yes.

Thank you.

Thank you. Thank you, everyone.

Good day, and welcome to the Third Quarter 2026 Hewlett Packard Enterprise Earnings Conference Call. [Operator Instructions] Please note this event is being recorded.

I would now like to turn the conference over to Mr. Shannon Cross, Chief Strategy Officer. Please go ahead.

Good afternoon. I'm Shannon Cross, Chief Strategy Officer for HPE. I'd like to welcome you to our fiscal 2026 third quarter earnings conference call with Antonio Neri, HPE's President and Chief Executive Officer; and Marie Myers, HPE's Chief Financial Officer.

Before handing the call to Antonio, let me remind you that this call is being webcast. A replay of the webcast will be available shortly after the call concludes. We have posted the press release and the slide presentation accompanying the release on our HPE Investor Relations web page.

Elements of the financial information referenced on this call are forward-looking and are based on our best view of our business and the external factors affecting us as we see them today. HPE assumes no obligation and does not intend to update any such forward-looking statements. We also note that the financial information discussed on this call reflects estimates based on information available at this time and could differ materially from the amounts ultimately reported in HPE's quarterly report on Form 10-Q for the fiscal quarter ended July 31, 2026.

Figures used in verbal remarks are rounded for ease of discussion. For more detailed information, please see the earnings materials as well as disclaimers relating to forward-looking statements that involve risks, uncertainties and assumptions. Please refer to HPE's filings with the SEC for a more detailed discussion of these risks. For financial information that we are showing on a non-GAAP basis, we have provided reconciliations to the comparable GAAP information. Please refer to the tables and slide presentation accompanying today's earnings release on our Investor Relations website for details.

Throughout this conference call, all revenue growth rates, unless noted otherwise, are presented on a year-over-year basis. Unless otherwise noted, all financial metrics and growth rates discussed today are non-GAAP and and EPS refers to non-GAAP diluted net earnings per share. Certain financial information featured in the presentation today has been normalized to include Juniper Networks results as of the beginning of HPE's fiscal year 2025.

Antonio and Marie will reference our earnings presentation in their prepared comments. We will also be disclosing records for certain financial metrics during the presentation. Please refer to our end notes in the presentation while reading these statements.

With that, let me turn it over to Antonio.

Thank you, Shannon. Good afternoon, everyone. Our strategy is proving itself again this quarter. We delivered another set of record financial results which demonstrates the durability of our profitable growth momentum and disciplined execution across the company. We exceeded all our company-wide financial commitments, achieving record results across revenue, gross margin, non-GAAP operating profit and earnings per share. AI has become a multiyear growth driver, expanding demand across our HPE portfolio. Customer demand in the quarter accelerated across both business segments with orders growing faster than revenues. We booked more orders than any prior quarter in our history, resulting in a record break in backlog for the company. Supply constraints continue to affect our ability to fulfill the increased customer demand. We are collaborating very closely with our partners to secure additional multiyear supply agreements. We're also providing our customers with alternative product configurations and deeper planning interlocks to better forecast supply availability.

In fiscal Q3, HPE delivered record revenue of $12.2 billion, up 34% from a year ago. Our HPE revenue growth year-to-date has risen about twice as fast as it did over the same period last year. HPE non-GAAP gross margin was a record of 40%. We generated record non-GAAP operating profit of $2 billion, 2.5x more than a year ago. Non-GAAP earnings per share was $1.11, another record and the first time we achieved more than $1 in non-GAAP EPS in a single quarter. Our outstanding operating results translated directly into stronger cash generation, resulting in our highest free cash flow ever for a third quarter at $958 million.

Last quarter, we updated our fiscal 2026 outlook and introduced our initial fiscal 2027 growth framework. Thanks to our record results, record orders and record backlog, we are raising our outlook for both fiscal 2026 and fiscal 2027. Marie will discuss the details shortly.

Before I hand over the call to Marie, I want to provide some observations about the market and our business segment performance. Also want to note an important milestone regarding our Juniper Networks acquisition. In August, a U.S. Federal Court approved our settlement with the Department of Justice, saying, "It serves the public interest." We are pleased with the outcome, which reinforces our confidence in the long-term value of bringing these two great networking portfolios together. A year after closing the Juniper Networks acquisition, our integration plan and cost synergies remain ahead of schedule. The business performance continues to strengthen through expanded innovation and strong execution. The enhanced ability to compete is already driving more innovative networking solutions for customers and higher profitable growth for shareholders.

Order bookings and revenue for our networking products and services reached record levels despite supply constraints, which limited our ability to convert the higher demand into revenue in the quarter. Campus & Branch had record revenue as customers modernize aging edge infrastructure and deploy AI-driven network operations. Orders were ahead of revenue, demonstrating the differentiation of our self-driving networks and the versatility across multiple cloud deployment models. Routing and data center switching demand accelerated in the quarter with orders substantially ahead of revenue and our backlog at its highest ever. Our backlog reflects strong customer demand from hyperscalers and neo clouds for our routers, switching NII-driven operations software, as they continue to increase their AI cloud CapEx infrastructure investments.

Growth in our backlog shows strong customer demand is running ahead of available supply. We expect to convert more orders into revenue in Q4, which give us even greater confidence in sustaining our networking growth in fiscal 2021.

Today, we announced an expanded collaboration with Oracle to accelerate gigawatt scale AI infrastructure. Oracle will deploy HPE Juniper networking routers and switches across one of the largest AI cloud infrastructure build-outs. HPE is uniquely positioned to support Oracle with a network for AI portfolio, which is one of the most comprehensive in the industry.

SASE and Security also contributed to our growth. Our strong performance in firewall, SD branch and branch SRX reflects growing customer demand for networking and security that operates as one converged solution. I hope you will attend our upcoming network in Investor Day later this month to hear more about how our networking strategy, innovation and business momentum are giving us even greater confidence in the opportunity ahead. While our networking segment continues to strengthen through our thoughtful integration, our cloud and AI segment continues to perform exceptionally well in a very supply-constrained environment. We delivered record revenue, operating profit and operating margin.

AI is beginning to inflect beyond early proof-of-concept trend in deployments into a broader enterprise workflow transformation opportunity. Customers are increasingly investing in new agentic AI applications and AI inferencing, requiring accelerated computing infrastructure, secure data storage access and enterprise-grade cloud management. Our comprehensive cloud and AI portfolio is perfectly positioned for this market inflection. We continued to experience strong demand across traditional servers, AI systems, storage, private cloud solutions and GreenLake cloud services.

The server product category drove the outperformance with high demand for our traditional servers and AI systems. We saw strong demand from large enterprises, neo cloud service providers and sovereign customers. We expect demand to remain exceptionally high as our pipeline remains multiples of our backlog. A fundamental shift in the server business is becoming quite clear. The way customers value their IT infrastructure is changing. Their focus is not just whether a server can run AI workloads, but also how it can enable entirely new business workflows using new AI applications.

We are seeing AI-related enterprise initiatives receive higher levels of investment than traditional IT projects. There is more top-level executive engagement in making those investment decisions, including company boards, which are championing AI technology to unlock further business transformation potential.

Storage had a standout quarter with record revenue. It benefited from our decision to focus on our own IP offerings by delivering a modern multi-data protocol platform for the AI era. Customers are beginning to evaluate where the AI workloads can be run most efficiently with the best secured data management and the lowest cost per token.

Organizations are still in the process of modernizing data storage environments while preparing for the next generation of AI-driven workloads and applications. They want data to be closed to their AI infrastructure and one control over sensitive information. These trends are driving strong demand for our HPE Alletra MP storage solutions. We are confident our comprehensive data storage value proposition will continue to accelerate this momentum in our storage business.

Our private cloud AI platform allows customers to manage enterprise AI applications and AI agents, while keeping control of data, government security and operations. Demand for PC AI is strong from enterprise customers that want to optimize the token economics of large-scale AI deployments on premises. That is rapidly scaling our order bookings and the size of our customer base.

Green Lake remains one of our greatest differentiators cause. It allows customers to manage infrastructure and software through a secure hybrid cloud operating model regardless of where our traditional and new AI workloads reside. Customers are broadening their utilization of our GreenLake cloud services, expanding their existing usage by consuming our new software and intelligent cloud services, which increases our net retention rates.

In Q3, the number of GreenLake customers grew 18% to 52,000, up from 44,000 a year ago. HPE Financial Services continues to deepen our customer relationships and offer a meaningful competitive advantage, which has become especially important as more customers look for financing options to help with their AI investments. As a result, HPE FS generated third quarter records in financing volumes residual value and return on equity.

In closing, HPE delivered another outstanding quarter, exceeding our company-wide commitments and demonstrating the ongoing differentiation in our strategy and its execution. As we look ahead, the same underlying drivers of our performance give us confidence in continuing to deliver higher profitable growth, higher cash flow generation and higher capital returns for shareholders. I want to thank our team members for their focus and strong execution so far this year. Our amazing talent and culture has set our performance apart this quarter and in fiscal 2026 to date.

With that, let me turn the call over to Marie. Marie?

Thank you, Antonio, and good afternoon, everyone. We delivered another strong quarter, reflecting accelerating demand for AI and solid networking momentum, together with disciplined execution across the company. The demand environment remains robust as orders continue to outpace revenue. Investment in AI infrastructure is increasing at a rapid pace with enterprise spending focused on genic AI workloads and AI inferencing. Importantly, this opportunity is broadening across use cases, customer verticals and geographies, reinforcing the value of HPE's expanded portfolio and our ability to provide customers with integrated solutions across the enterprise technology stack. We remain focused on executing against strong customer demand navigating a dynamic supply environment, managing mix and input costs while driving operating leverage. This discipline is reflected in our financial performance, supporting durable, profitable growth in fiscal 2026 and 2027.

Let me walk you through the results. Revenue of $12.2 billion increased 34%, exceeding the high end of our guidance range with order growth up 42% on a normalized basis, led by demand and traditional service, AI systems and networking. Gross margin exceeded 40% driven by disciplined pricing in traditional servers and increased networking mix. Going forward, we expect our gross margin to moderate toward more historical levels driven by the growth in AI systems and the normalization in traditional service, offset by the growing mix of networking.

Operating expense was up 17% sequentially due to higher variable compensation reflecting our record financial results. We expect operating expense to decrease in FY '27 as variable compensation normalizes, and we see continued benefit from Catalyst transformation efficiencies and Juniper integration synergies. Operating profit was $2 billion, up nearly 40% sequentially. Operating margin of 16.2% expanded by 290 basis points sequentially, driven by gross margin expansion and operating leverage. EPS was $1.11, well above the high end of our guidance. GAAP EPS was $1.06. We delivered Q3 free cash flow of $958 million, driven by strong operating profit as well as collections.

Now let's turn to our segment results. Networking revenue of $2.9 billion was up 10% on a normalized basis, consistent with our outlook. Orders increased 36%, about 3.5x faster than revenue. Order growth was broad-based across the portfolio, led by AI infrastructure-related investments in data center switching and routing and strong demand for self-driving networks in campus and branch. Networks for AI demand accelerated in Q3, with orders reaching a new high of $700 million, up triple digits. Our portfolio and competitive position in scale-up and scale-out and scale across strengthened by the recent launch of our direct liquid cool Tomahawk 6 based switch at our differentiated PTX and MX routing portfolio.

Our pipeline continues to increase with further acceleration expected as we bring the Helios platform to market. We expect networks for AI to be a meaningful growth engine for the company. Cumulative networks for AI orders were $2.2 billion, surpassing our FY '26 target. As a result, we are increasing our year-end target to $2.5 billion to $3 billion. To meet this order growth, we have more than doubled our networking purchase commitments quarter-over-quarter. Within networking, Campus & Branch revenue grew 8% on a normalized basis. Routing revenue growth accelerated to 23% as we benefit from increasing demand for our on- and off-ramp AI network infrastructure. Security grew 12%, while data center networking revenue declined 6% due to shipment timing driven by supply constraints.

Order momentum was much stronger across most product categories with data center switching and routing, up high double digits and Campus & Branch up low teens. We remain focused on improving order conversions to drive faster top line growth and greater scale. Across customer verticals, enterprise revenue grew 12% and service provider grew 5% on a normalized basis. Enterprise growth was driven by strong demand from large global accounts prioritizing network modernization across Campus & Branch and data center switching. Networking operating margin of 22% was in line with guidance, reflecting disciplined execution and the early realization of Juniper synergies, partially offset by higher variable compensation.

Moving to cloud and AI. We delivered fiscal Q3 revenue of $9 billion, up 25%, exceeding our outlook, reflecting strength in traditional servers as higher average selling prices drove server revenue to an all-time high. Our disciplined pricing and increased scale drove operating profit above $1.5 billion. We were pleased to see operating profit growth accelerate, up 61% sequentially and triple digits year-over-year. Operating margin of 17% was up 460 basis points sequentially, demonstrating our ability to scale our business profitably. Server revenue growth of 35% accelerated sequentially as strong ASP growth in traditional servers offset supply-constrained unit volumes. Orders increased strong double digits year-over-year, reflecting robust demand from large enterprise, sovereign and cloud providers.

Our supplier agreements now multiyear, in some cases, ensure us the capacity allocations we need to reduce lead times, improve at that log conversion and drive higher new order growth, supported by our historically highest level of purchase commitments. We see enterprises increasingly moving from AI pilots to production deployments using traditional servers for genic AI workloads and inferencing. Examples include a global financial services firm, leveraging AI for market analytics and trading insights and a large retail customer deploying on-prem agentic AI workloads to lower public cloud AI token costs. As evidence of this strong growth, we are pleased to report that after quarter end, HPE was awarded a multibillion-dollar server deal with a hyperscaler customer specifically designed for inferencing, supporting our view that demand for inferencing and agentic AI workloads is building.

AI systems orders of $2.4 billion increased over 30% sequentially, reflecting broad-based demand across customer segments. Enterprise demand more than doubled, reflecting increasing overall infrastructure spending as AI initiatives have become board-level priorities. Our AI Systems backlog increased 14% sequentially to a new high, and our pipeline remains multiples of our backlog. AI Systems revenue for the quarter was almost $1.6 billion. We expect AI Systems revenue to improve sequentially in Q4 given timing of backlog conversion.

Storage revenue increased 10%, driven by strong order growth with higher ASPs at a favorable mixture towards higher-value owned IP and private cloud. PCAI orders increased triple digits in Q3 as customers are adopting our AI factory platform to support agent AI and inferencing initiatives. Alletra MP orders and revenue increased strong double digits year-over-year. We see robust growth potential for our X 10(k) object and file system, broadening our AI solutions portfolio to address the rapidly expanding unstructured data market. And finally, financial service revenue was roughly flat year-over-year, and the business continued to generate a return on equity exceeding 20%.

Turning to our integration and transformation initiatives. We are making strong progress in building a more efficient company as we are running ahead of plan on multiple projects to lower our cost of sales and operating expenses. Juniper synergies capture remains on track to achieve our $600 million annual run rate savings target by the end of FY '28, with integration costs tracking better than planned. Last quarter, we highlighted the growing contribution of AI-enabled process simplification within Catalyst. Since then, we have expanded both our AI and operational simplification efforts across the enterprise. HPE is now deploying an internal agentic AI platform built on our own private cloud AI, open source and open weight models, leveraging intelligent routing that sends each workload requests to the most cost-effective AI model.

According to our own internal analysis, our PCI offering can reduce token costs versus the public cloud by up to 60%. Routine task stay on-premise, while frontier models are reserved for the most complex work.

Moving to cash. We delivered operating cash flow of $1.6 billion. Free cash flow totaled $958 million in Q3. As a result, we are raising our free cash flow target to at least $3.75 billion for FY '26. Our cash conversion cycle improved by 1 day from Q2, driven primarily by a decrease in days receivable due to more favorable billings linearity within the quarter, along with stronger collections. This was offset by an increase in days of inventory due to higher purchases in anticipation of future shipments. Inventory ended the quarter at $11.8 billion, up year-over-year and sequentially, reflecting higher commodity costs and targeted purchases to support increased orders and increased backlog. In Q3, we returned $324 million to common shareholders, including $189 million in common dividends and $135 million via share repurchases. We received gross proceeds of approximately $1.4 billion after closing our H3C transactions and use cash on hand to retire our term loan. Consequently, we exited Q3 with a net leverage ratio of 1.8x, below our target of 2x.

We completed the sale of our Telco Solutions business last month and intend to retire $1.25 billion of notes maturing later this month. We plan to return at least 75% of our free cash flow to shareholders in Q4.

Turning to guidance. We are increasing our outlook on the strength of our Q3 results and confidence in the durability in demand. We expect Q4 revenue to be between $13.9 billion and $14.8 billion, reflecting continued strong demand across both segments. We expect networking revenue to grow 11% to 13%, driven by order strength and improved supply chain conversion. We expect networking operating margin to improve modestly quarter-over-quarter driven by top line growth and Juniper synergies.

In cloud and AI, we expect revenue to grow 60% to 72%, reflecting sustained demand, higher ASPs and traditional servers and greater AI revenue conversion. We expect operating margin to moderate sequentially to a mid-teens rate. We expect Q4 total operating expenses to decrease sequentially by a low single digit due to lower variable compensation expense and increased catalyst transformation efficiencies and Juniper synergy capture. We expect our operating margin rate to decline sequentially driven primarily by a higher mix of AI systems in cloud and AI and pricing. As a result, we expect EPS between $1.20 at $1.30, a GAAP EPS between $1.12 and $1.22.

Based on our Q3 results and Q4 outlook, we are raising our FY '26 EPS guidance range to $3.75 to $3.85. We are also raising our GAAP EPS range to $2.93 and $3.03. We now expect FY '26 free cash flow of at least $3.75 billion. Given the demand strength and sizable backlog we saw at the end of Q3, combined with some large deals we signed post quarter close, we are updating our fiscal '27 framework and now expect consolidated revenues to grow 13% to 17%. Networking revenue growth of 14% to 17%. Cloud and AI revenue growth of 14% to 18%. Company operating profit growth of 14% to 18%. Company operating margin of 14% to 15%, supported by a modest decline in operating expense.

Networking operating margin in the mid- to high 20% range, clouded AI operating margin of approximately 13%. EPS of $4.40 to $4.60, which implies growth of 16% to 20% versus the midpoint of our FY '26 EPS outlook and free cash flow of at least $5 billion. Importantly, this framework builds on our higher FY '26 guidance, pointing to a significant improvement in our fiscal 2027 outlook.

In closing, Q3 was an exceptional quarter for HPE. We generated strong financial results, raised our fiscal '26 and fiscal '27 commitments and achieved our leverage target more than a year ahead of our original plan. Demand remains ahead of revenue, and our backlog is a record. At our Juniper integration and Catalyst initiatives are delivering ahead of our FY '26 plan. As we head into the final quarter of fiscal '26 and look ahead to fiscal '27, we are executing from a position of strength. With durable demand, strong margins and the operational discipline to sustain both.

With that, I'll turn the call back to the operator to begin Q&A.

[Operator Instructions] And our first question for today will come from Katherine Murphy with Goldman Sachs. [Operator Instructions]

It was encouraging to see the momentum across the total portfolio, the record orders that you mentioned in the quarter as well as the raised fiscal '27 outlook for 13% to 17% growth. First, can you talk about what's giving you confidence that the current demand represents a sustained infrastructure cycle rather than customers pulling forward spend? And then as a follow-up, can you quantify how much of the raised fiscal '27 outlook is related to the new hyperscale inference and Oracle deals that you highlighted versus improved outlook within the remaining business.

Well, thanks, Kathy, and good afternoon. Look, our guide is informed by what we see in the market, and the market is telling us the demand continues to be exceptionally strong. So there continue to be large build-out for AI cloud. And obviously, we participate that in a very disciplined approach. Although networking, we continue to see significant demand for our routers and data center switches, which you saw we had record-breaking orders of $700 million this particular quarter, and we expect that to end between $2.5 billion and $3 billion for the year. And when I think about 2027, in our guide, we have not included the AMD Helios opportunity at all, which is going to start ramping sometime end of this calendar year in 2027. So demand is exceptionally strong. And then on the cloud and AI will give us the confidence is the acceleration in the enterprise. The enterprise clearly has hit an inflection point. And that inflection point is driven by the deployment of agentic AI and AI inferencing. And what we see that is because the number of use cases. And we see that ourselves. Just to give a perspective, we have more than 1,200 use cases in our company, 300-plus in production. And we continue to learn how to do that and accelerate the pace. We see that now in the broader enterprise market across multiple verticals. And the reality, that's going to favor our traditional server and storage business in our private cloud stack because they don't need huge amount of GPUs or even CPUs for that matter. What they need is a very tight capital infrastructure that ultimately brokers the cost of tokens that ultimately allows them to do what they need to do. So the number of tokens on-premise is growing very, very rapidly. So that's what informs us on the durability of this demand. And we see that in our pipeline, because ultimately, you guide yourself about the pipeline and how much of the pipeline you could convert first in orders and eventually through revenue. So that's what gives us the confidence to provide the guide that we guided for 2027. Marie?

I think, yes. So just in terms of the guide itself, as you know, we guided to 13% to 17% of revenue for the total company. In terms of networking, the Oracle deal plus the core, the beginning of the Oracle deal is in the [ 14% to 17% ] that we guided for the networking growth. And in terms of the hyperscaler deal, some of that as well is included in the cloud and AI, which we guided to [ 14% to 18%. ] So that's how you should be thinking about the revenue. Once again, this puts and takes in all -- there is more AI revenue in cloud and AI as well as you get into '27, so just bear that in mind.

The next question will come from Amit Daryani with Evercore.

Congrats on some fairly impressive numbers over here. Antonio, I wanted to just ask on networking though, the organic growth of 10% looks a little light relative to what I think your peers are seeing right now, I think, relative to what you perhaps expected. But your orders at up 36% is really strong. Maybe just talk about what's driving the gap over here? And how should revenues begin to catch up in orders? If you just spend a little bit of time on that, that would be helpful. And then I didn't hear you folks talk about the Oracle announcement a lot. Maybe just help us appreciate what are you providing them? Is it scale out, scale across, just provide a little bit more color on that deal because it seems like a fairly important thing on the networking side, at least.

Sure, Amit. The order momentum is super strong. In fact, as we said in our prepared remarks, our orders are growing 3.5x faster than the revenue. And so what it has limited us is the availability of supply. And we expect that supply to become more aligned to our order bookings as we go forward. Starting Q4, where you can see where we go on the revenue side from 10% in Q3 to 11% to 13% in Q4 and then eventually to 14% to 17% in the full 2027 year. And so that's our focus is really on the supply availability. That's why Marie said that we have doubled the number of commitments in terms of inventory. And we are working with our suppliers through that. And the reality is that we will see continued orders ahead of revenues, but we expect that to kind of close a little bit as we go forward. So we don't -- we expect an asset ratio of revenue as we go forward, but supply will continue to be the constraint. Now within that, Marie talked about core. So core for us is the Campus & Branch, which always has represented more than 50% this quarter was probably 50% of it. And that had double-digit order growth, low double-digit order growth, but we posted record revenue, and we expect that to continue to improve as we go forward because we have a terrific value proposition with our self-driving networks, and the versatility of the both Mist and Aruba Center platform. And then on the Oracle side of the equation, is an expansion of what Juniper used to do. But now we are doing at gigawatt scale and it's going to be a very set number of deployments on a global basis, and they're going to use both our QFX switching products, which is based on the Broadcom Tomahawk 6 and our software and our AI Ops. So think about that as a scale out. And then our scale across, which is our PTX routing platform, which uses our own silicon, which is a major differentiation for scalability, which is our Express 5 silicon, which we designed now a couple of years ago. So on a combined basis, this is a multi-gigawatt on a multiyear basis.

Your next question will come from Aaron Rakers with Wells Fargo.

Congratulations on the strong results. On the traditional server side, I guess the question I have is, I think you talked about a lot of the growth being driven by the ability to pass through pricing relative to unit growth. So I guess my question is, as we think about AI moving into the enterprise environments more prolifically, how would you characterize the installed base and the upgrade opportunity that you're seeing associated with that? And should we start to think about unit growth accelerating on top of the ASP pass-through? And then secondarily to that, the hyperscale deal, I think in the past, HPE has been pretty clear of you'll be opportunistic on AI opportunities and maybe some of the larger hyperscale. Has this changed strategically at all? Are you going after some additional hyperscale deals more actively going forward?

Thank you, Aaron. Maybe I'll start with the latter because it's a very important use case. It is a hyperscaler customer, but think about them as an enterprise customer who is going to use our AI inferencing for their own internal usage. So our strategy has not changed from a selling a large amount of infrastructure for them to serve like it used to be in the past, the cloud business. This is about a multibillion-dollar AI inferencing for their own internal usage as an enterprise customer. It just happened to be they are labeled as a hyperscaler customer, okay? And so that's one takeaway. Second is that look, units will modulate as supply becomes available, right? And so we -- as Marie said in her prepared remarks, we were obviously -- we still, like everybody else, working through the supply availability. But what we are very excited about is that the acceleration of traditional service and storage, by the way, in private cloud because some customers go server only, some go with server attached to -- storage attached to servers and some are using the full stack like a private cloud AI, which in many ways, the AI factory that we co-engineer with NVIDIA. And in that case, right, it's about the growing of the AI deployment on-premise. And so over time, right, it's going to become how large those deployments come. And ultimately, whether you serve only or you go private cloud, that will drive units, but it also would come down to the conversion of the units based on the supply availability as we navigate 2027. But right now, as we said, right, we expect continued exceptional demand into Q4 and 2027.

Your next question will come from Joseph Cardoso with JPMorgan.

Congrats as well. Maybe if I could, I think, Antonio, you mentioned that you're not embedding the Helios opportunity into the fiscal '27 outlook. Can you just touch on the rationale behind that decision and what's keeping you on the sidelines from introducing that into the framework. And any thoughts on how we should think about the magnitude of upside that you could introduce to the networking revenue and margin outlook when and if that gets introduced into the framework?

Yes. Now thank you for the question. Well, I mean, we are working very closely with our partner, A&D. And we expect that infrastructure to be available for ordering later in this calendar year, and we're working together on a very large pipeline, which obviously this infrastructure is designed for large AI deployment at scale, particularly for training and then obviously, it can be used as well for inferencing. But these are a concentrated number of customers in the end that they need that level of infrastructure, not different they're using today with NVIDIA NBLs 72 and 144 of [indiscernible]. We felt that we wanted to see a little bit more as the schedule firm up and then eventually start deploying these capabilities. And then as we go through the sub single quarters, we're going to share more about how that's happening. But the opportunity is pretty massive. When you look at the size of that deployment can be as a market okay, not HPE, tens of billions of dollars, okay, tens of billions of dollars. The difference this time is that, for us, the scale-up tray switches or HP Juniper. And let me be clear, we are going to sell HP Juniper scale-up switches beyond just embedded in HPE Helios rack. So we can sell it to anyone for that matter because obviously, customers will want sometimes different compute vendors, which is totally fine. But that opportunity is in the margins of the tray switch and the opportunity to see broadly beyond the HPE as a partner with AMD as we go forward. So once we see a little bit more of that, Marie now will share more. But this is not in the 14% to 17% growth that we just shared with you as a partner networking only.

And I would just add that once we get to the end of Q4, we'll give a fulsome guide for '27 as well.

Yes.

The next question will come from Wamsi Mohan with Bank of America.

I was wondering if you could share any more details around your $3.5 billion inferencing deal that you signed. What exactly is part of that? And can you talk a little bit about the economics around this? And are you changing your approach to incremental hyperscale opportunities?

Yes. So obviously, we can't talk about the customer, but it's a hyperscaler customer, but is acting, as I said, to Aaron as an enterprise. We are not selling 1 were used to refer as a Tier 1 infrastructure. You recall that during the cloud days, we are now selling that type of infrastructure. We are selling traditional servers for AI inferencing that they will use for their own internal usage.

The next question will come from Asiya Merchant with Citi.

Can [indiscernible] down a little bit on supply constraints. Where do you see them most acute right now? And kind of your expectations on the supply constraints easing or how you're thinking about your agreements long-term agreements that you've signed to source the supply.

Yes. Thank you for the question. The supply constraints, generally speaking, continue to be the same, right? So obviously, on the commodity side, DDR5 is a great example of it. The DDR4 for the older generation, NAND in the flash drive space. Those have been consistent themes now for 3 quarters since the beginning of 2026. And then there is an other set of parts underneath that they are constrained by wafer capacity. And so we expect that to continue to be the case because ultimately, you have to solve two problems. One is clean room capacity to turn wafer into more available supply that going to -- that should improve some in 2027 because we know our partners continue to invest in clean room capacity, but ultimately, structurally, this will be solved with wafer capacity because in the end, you need the wafers to turn parts into actual supply of products. And so that's the challenge we're all navigating through. And the wafer capacity affects other parts, right? But in the memory space, you also have another trend underneath that obviously is driven by the technology shift. We had DDR4 to DDR5, that's understood, but then you have traditional DRAM moving to HPM. And that HBM demand is super high because it's driven by the GPU and the better memory that comes with it. So this is why you have to look at this wafer capacity, clean room capacity and then eventually the mix of what type of memory will be used and demanded as we go forward. Now in the traditional server, we use DRAM. We don't use HBM. And so that's where we are focused very extensively. Once you buy a rack scale architecture, you come with HBMs and therefore, once you get that server tray with the GPUs, the memory comes with it. So this is where we need to navigate through, but my expectation personally after seeing the exceptional demand that we see in the market and talking with our suppliers, some of them we signed already multiyear LTA agreements to lock our capacity. We decide how to use that capacity. Then it tells me this is going to last for a longer period of time, which obviously will have consequences on cost and pricing. But so far, I think HP has been very effective in managing that process.

The next question will come from Erik Woodring with Morgan Stanley.

Congrats on the nice quarter here. Marie, your prepared remarks on -- you mentioned gross margins, and you mentioned a normalization in traditional server margins, I think, looking forward. Can you maybe just elaborate a bit on what that means? Like why would you see normalization in server margins if demand is as strong as you're referencing in your unit trajectory should seemingly improve as you get better supply? Like is this -- have you benefited from low-cost inventory now that's starting to uptick in your billing materials. I'm just trying to understand your comment on exactly what you're trying to message there.

Yes. No worries, Erik, and good afternoon. So maybe I'll just start up by giving you some context on the quarter, what drove those margins and then how we think about them going into Q4 and into '27. So I would think about Q3 more as a confluence of everything coming together at once. Obviously, you saw the impact of strong revenue scale, and that played through in terms of leverage higher gross margins. We talked about our disciplined pricing. You heard Antonio talk about how we've been very diligent around pricing in this constrained component environment Don't forget we've also got the benefits of programs like catalysts that have been flowing through as a put and take. And then there was a mix of deals in the quarter that also impacted and really frankly benefited us in terms of our Q3 rates on what I'd say with respect to servers. As you get forward and you look forward into Q4 and then this does carry forward into the guide that we gave into '27. Bear in mind that the mix of deals, specifically in AI will ship. So we do expect to ship more AI revenue, specifically in Q4, and you can see that our in number, inventory number went up. So we're positioning ourselves to ship some larger AI transactions, and we expect to have a bigger mix of that also into '27. And then I just double down, it's actually the mix of deals even inside of traditional server will also moderate as we go into Q4 and into '27. So that's how I'd be thinking about the margins. Obviously, we're pleased with the guide that we've given, but I think at this point, we had a great quarter in terms of a confluence of all the factors coming together.

The next question will come from Tim Long with Barclays.

A two-parter, if I could, on AI networking side. First, nice win with Oracle. As you mentioned, had been a pretty big Juniper customer. So just curious if that win can propel any other use as a reference design or reference case for other either hyperscalers or neoclouds, there's obviously a lot of new networking opportunities out there. So I'm curious if you think that larger, more profile, AI win can do that? And second, Antonio, you mentioned the scale across with the custom silicon. I'm curious what you guys are hearing on the importance of having that customer silicon. Do you think that was important for the wind? And similarly, is that something that can help drive even more scale across as that's becoming more important for the AI companies?

Yes, Tim, thank you. The answer is yes and yes. I mean, yes, because obviously, it proves on the first part of your question, that we have a scalable set of products that deliver the performance with the AI capabilities that now everybody is looking for to drive these self-driving kind of operations. And we have embedded that across the entire portfolio, not just in the Campus & Branch, but also if you look at our routers are amazing what they can do in optimizing the bandwidth using AI. But also, we were first time to market with a 1.6-terabit. And time to market here is a very important aspect of competing and winning in the market. So Rami and I, spent a lot of time with the team how we continue to stay ahead of the curve and be first time to market with these latest technologies. So we were definitely first time to market Juniper was with a 1.6 terabit in air cooled. And we were the first 1.6 time to market with the direct liquid cooling. And then on the route -- and so we hope that, that's going to drive a significant amount of interest, and we have a lot of conversations with customers, right, to leverage this portfolio. On the routing side, look, it's very hard to do what our silicon does. There's only 2, 3 players who have done this at massive scale. And it's a source of differentiation. What our express silicon does for the routing is unique. And if you think about the PTX, I'll give this example, right, perhaps not in the eye space. But if you take the latest PTX product, which is 3/4 of a rack and you take New York and London, all 60 million people watching a Netflix movie concurrently on the platform. The platform can manage it. Think about the scale, right, that these products can do. And that's very hard to do. So that's why as these gigawatts and gigawatts of infrastructure gets deployed, which we think by the end of the decade will be over 270 gigawatts. You have to connect all of them. And therefore, you need a -- at the end of that pipe, you need a router with that level of capabilities.

The next question will come from David Vogt with UBS.

Maybe a combo for Antonio, Marie. So I think Antonio, if I think about your '27 outlook, you're taking the revenue up by about $4 billion relative to where we were 90 days ago. Can you kind of dig in on the supply chain? I know you talked a little bit about it before, but what improved on the margin? Was it just securing more purchase commitments what gives you more confidence that you have enough supply chain relative to 90 days ago to kind of hit that target? And then Marie, for you, same kind of question. You're taking that number up by about $4 billion, and it looks like the free cash flow drop through is pretty impressive, incrementally $500 million flows through at least, which is better than your current conversion on the business. Can you help us understand kind of what's going on with the free cash flow conversion from the guidance raise?

So yes, the first part of your question, the answer is yes because obviously, we put numbers out there if we have the ability to fulfill it, and that came through the work our supply chain team have done to secure these multiyear long-term agreements that locks the capacity. And obviously, every 90 days, we can actually adjust what type of usage in the capacity within that capacity, we want to get out of it. And so your math is absolutely in the range, and that's why we are confident in this guide because at one end, you have the demand for it, and then you have the supply to fulfill it. And that's why this is a durable profitable growth. And on the cash flow, Marie?

Yes. No, look, first of all say, look, really pleased with the guide we gave of at least $5 billion, which is up 33% year-on-year. And as you correctly pointed out, one of the biggest drivers of that improvement in the rate is really the fact that we've got a lot less restructuring as we go into '27. Just remember that the programs that we had like Catalyst and the Juniper synergy programs, really we had the sort of, I'd say, the brunt of the restructuring in this year in '26. So as we get into '27, we start to bleed that down, and it will be a bit of a tail left on the Juniper synergy plan, which will honestly bleed off by the sort of end of Q4 of next year.

I think the other thing, Marie, that's important to understand is that as we accelerate the growth in networking, which obviously 14% to 17% is an acceleration compared to 2026. The working capital demand in networking is significant lower because it's a faster turn to revenue once you get the inventory on hand.

The next question will come from Mark Newman with Bernstein.

Digging a bit more into the service side. You reported server revenue up 35% year-on-year. Obviously, you've got some AI servers in there. So I think if you take that out, it implies that traditional servers growing a bit faster than that number. Just wondered if you could -- and you said orders for traditional servers up 75% year-over-year. I wondered if you could clarify for us like how much of this growth is higher ASPs and richer configurations, which you mentioned on Slide 8 versus unit growth. Is there any significant unit growth in terms of units of CPU cores or any kind of metric like that? Or is this exclusively pricing and configuration? And related to that, the orders being stronger growth than the revenue, can we ascertain from that that you're significantly supply constrained and should we -- how long would that supply constraint last? I'm just wondering in terms of projecting out like should we see further acceleration of server growth from here as supply constraint alleviate? Or is that a supply constraint going to remain at similar levels going forward?

Yes, you put a lot in that question, but I'm going to simplify it for you a little bit. Look, we expect in Q4 units to strengthen because of what we see in the market. And obviously, in 2026, the unit has been constrained by the supply availability, which means a lot of the growth came through the ASPs, but as we go into Q4, we expect units to strengthen. And a lot of that will be on the back of the AI inferencing in agentic AI that we see. Supply will continue to be constrained, which means we're going to continue to run into high backlog as we go forward. But to the question that was asked earlier on, we factor that in, in our guide because our guide reflects what we believe the supply availability against the backlog and the demand will be. So I will stay focused on the guide. I will stay focused on the fact that supply will continue to stay constrained, but the demand will continue to be exceptionally high. And so we expect the cloud and AI segment to grow 14% to 18% on revenue. And that's a very strong growth, and it will continue to be led by traditional servers, higher conversion on AI systems as we go forward, particularly in Q4. And then ultimately, the storage business because, obviously, the storage business helps on the profitability side because of the margin structure. But look, as we said in our remarks, storage grew twice as fast as the revenue that we posted in the quarter.

Yes. And I'd just add, Mark, that we actually -- it was a raise on cloud and AI revenue actually for '27 to get to the 14% to 18%. So I think that just illustrates the strength of the demand that we're seeing out there.

Operator, we'll take our final question.

Our final question will come from Matt Niknam with Truist.

I'll echo the congrats on the great results. Maybe more of a high-level question. I'm wondering, Antonio, if you're seeing any incremental hesitation or pushback from customers with regards to demand appetite, just in light of some of the bigger pricing actions aimed at offsetting higher memory costs. And maybe on a related note, if you can speak to where incremental budget to invest in IT infrastructure and HP products are coming from at some of your larger customers?

Sure. No, we don't see hesitation I will say at the beginning of this hyper cycle on the cost, obviously, there were a little bit, I mean, shocked, and they're trying to navigate through the timing by focusing on understanding the trends and looking at the spot market and the like. Once they understood that, I understood that, particularly in AI, you need to go faster they figured out, look, waiting is not an option, but they are getting smarter about where to land their budgets, and how to optimize for these tuck-in economics, which Marie talked about. Look, when you do it on-premise, and we do that ourselves and we share the number, we can see up to 60% cost benefits on a tuck-in basis. And so no, I don't see hesitation on this point in time, and that's why demand continued to be exceptionally strong. That's very clear. In terms of budget, look, budgets overall are going up. Look, in our case, we can talk about it. Of course, there is prioritization within the budget to replace that all the infrastructure to invest more in AI is a balanced approach. But in our case, we are growing the budget to consume more tokens because we are very aggressive in deploying AI as a part of the catalyst transformation. Obviously, we do it with governance, rigor, return on invested capital and all the things. But in the end, it's an add-on, okay? It's not the substruction of something else. So I was talking to a large customer yesterday, which is in the financial sector. And he told me, yes, we are going all in, and now we have done the math, and we believe it will be better suited for us to build an AI factory on-premise so that we can improve the agility of deploying AI with the cost and controls, particularly in financial services with the compliance, right, that regulates that vertical. So we see that momentum continue. And I believe 2027 is going to be even stronger for enterprise because they become more confident in what they are doing and one win takes -- leads to another way, right, on how this has been successful. Because in the end, business process transformation, it's workflow transformation. It's not just technology for the sake of technology.

Antonio, would you like to...

Yes. No. Thank you for your time. I know you have a lot of earnings to cover I will remind you of the network in Investor Day, please attend if you can. We're going to share with Rami, our view of the future. Clearly, super excited about the Juniper acquisition. It has been a huge success. And we're just at the beginning. You saw some of the wins that we just announced. The runway ahead of us is enormous, whether it's new wins to Tim's question early on or with further customers or the Helios opportunity. The fact that everybody was concerned about integration, this was executed very thoughtfully. And the fact we're growing 3.5x faster than revenue shows you that we have the right portfolio at the right time with the right talent. And then in the cloud AI, I think we are delivering operating leverage. I mean, the execution there has been excellent. But our strategy is very intentional. We are leading with networking. We are at the core becoming a networking company. And the rest of the portfolio is there to serve the customer needs by driving the profitability and ultimately generating more cash, which has been a huge success so far, and it will be a bigger success in 2027. So thank you for your time today.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.

Good morning. Wow, it's a big room. It is great to be here with you. HPE Discover is where we showcase what is next and shine a light on the ambition and innovation shaping our industry and our lives. Today, we are witnessing one of the largest technology platform shifts in history. Workloads and applications are moving from being driven by end users, but now being driven by both end users and AI agents, agents that will fundamentally transform how we design and build how we serve our customers and how we operate our businesses.

I have always been drawn to how things work, how systems are built and how they evolve over time. In fact, if I had not become an engineer and a CEO, I would have become an architect. Architecture like engineering teaches you to think in systems to build for today and for the needs of tomorrow. You don't design a building around a single room. You design a structure that allows the whole system to flow and adapt over time.

Architecting for AI demands the same focus and discipline. Fundamentally, AI is only as strong as the data foundation beneath it. If the foundation is not robust, nothing else holds. Across networking, cloud and AI, HPE is delivering the essential building blocks that make your AI-ready foundation possible. networking to connect to your infrastructure and workloads at scale, cloud to enable you with a hybrid operating model to run your workloads and applications where they belong and AI to turn your data into intelligence and put it to work. Architecting for AI starts with your network.

For years, HPE Aruba Networking has helped you deliver secure connectivity across campus, branch and the edge, creating a digital on-ramp that connects your users, devices and data. With the addition of Juniper Networks, we have extended that leadership into the data center and across the critical networks, connecting the AI era: scale up, scale out and scale across. With our combined HPE networking organization, our goal is to deliver the best user and operator experience possible. We will do this through our next generation of secure self-driving networks across every domain. They fix problems securely before they impact the experience.

We make new rollouts faster and easier and fundamentally transform how you manage your network. Whether you are on the HPE Aruba Central or HPE Mist, you get the full benefit of our accelerated innovation on both. Nobody is left behind on the road to self-driving. But the phrase self-driving may have caused some confusion with the Mercedes Formula 1 drivers. Let's take a look.

[Presentation]

I had the opportunity to talk to both drivers this past weekend, and they told me they have fun doing that video. But the reality is that they really are very, very keen and interested to understand technology. So congratulations to the Mercedes team for a fantastic year so far. Every AI architecture needs more compute. In the AI era, your servers need to operate more efficiently than ever.

With HPE ProLiant Gen12, you get the performance to run everything from enterprise workload to AI inference in a much smaller, more efficient solution. As AI moves from generating content to taking action, the demands on the compute are increasing. Agentic AI requires fast orchestration, continuous evaluation and real-time access to data, which shifts more pressure to the host CPU. That is why we are expanding our ProLiant portfolio with the new HPE ProLiant Compute DL394 Gen12, powered by NVIDIA Vera CPUs. Vera provides the low-latency memory access, bandwidth and coherence required for agentic AI, reinforcement learning and other CPU-intensive workloads. It does so with the security and ease of management you expect from ProLiant. As AI moves beyond the data center, compute needs to move with it. We recently expanded our ProLiant Edge portfolio, bringing secure AI-ready compute to rugged and distributed environments so that inference can happen closer to where decisions are made.

AI also needs to reach the always-on systems at the core of digital business. With HPE Nonstop, we are bringing AI-powered fraud detection and autonomous operations into high-volume payment processing. This helps financial institutions detect fraud faster, automate compliance monitoring and keep transactions moving safely.

The most advanced servers in the world are only as valuable as the data they can access, which makes storage a critical part of your AI foundation. The HPE Alletra storage MP X10000 makes your data accessible, context-rich and ready to continuously feed your AI data pipeline. Across the full life cycle from ingestion, training, inferencing continuous learning, the X10000 now supports native file and object storage on a single architecture. It is also the first object storage platform validated through NVIDIA-certified storage for enterprise AI.

And for the mission-critical applications that run your business, the Alletra Storage MP B10000 continues to deliver. The B10000 is the fastest-growing all-flash block storage array in the market. Sitting across the top of the stack, of course, is software. As AI advances, enterprise environments are becoming more distributed and complex. This find virtual machines, containers, AI infrastructure and, of course, public and private clouds. At the same time, rising virtualization costs are pushing many of you to look for more flexibility and choice.

With HPE Cloud Ops software, we have brought together HPE Morpheus, OpsRamp, Zerto and our broader cloud portfolio capabilities into one unified operating experience. This enables you to modernize on your own terms while simplifying how you provision, observe and protect your hybrid multi-vendor environment.

Security and resilience must be built into every layer of the stack of your architecture. As AI changes the speed and the scale of cyber threats, new risks continue to emerge. Resilience is no longer just a set of isolated tools. It is a converged strategy across your systems, your data and your network.

With HPE iLO Silicon Root of Trust, we provide secure attestation from the silicon to cloud, helping verify that your infrastructure is trusted before your workloads run. And with Zerto and Cyber Resilience Vault, we help protect your critical data so you can recover faster and minimize disruption.

Networking and security are also converging. As AI becomes more distributed, the network often is the first place to see what is happening across your enterprise. With Zero Trust Architectures and integrated SASE, the network becomes an active security layer, enforcing policy, detecting threats and reducing risk from edge to cloud.

We bring all the elements of your AI foundation together with our GreenLake Cloud. GreenLake gives you a unified cloud-native experience across your entire hybrid estate with the flexibility to run workloads across public and private clouds, colos and of course, at the edge. With GreenLake Intelligence, we bring agentic AI to hybrid IT operations helping you see across environments, act faster and continuously optimize performance. From simplifying network operations to streamlining virtual machine migration, GreenLake Intelligence makes your infrastructure more adaptive, more autonomous and easier to manage. So you can spend less time managing tech and more time managing and advancing your business. So let's take a look.

[Presentation]

It's an amazing advancement we brought with HPE GreenLake. But by the way, what you saw here is just the beginning. In the Discovery showcase, you can see how GreenLake Intelligence brings agentic AI operations to life. So I will encourage you to go and experience yourself. Architecting for AI takes more than technology. It also requires the right people, processes, and partnerships. Our services team is here to help you through your AI transformation. With the HPE Financial Services, we help you modernize with confidence and better economics, including lower upfront capital investment. And with our IT life cycle management program, you can retire legacy multi-vendor technology and turn that value into funding what is next.

This week is an opportunity to explore our full stack AI foundation firsthand with demos, sessions and in conversations with your peers. Discover is one of the few moments where we have the full power of the HPE community together in one place.

During the rest of our time this morning, I want to elaborate on two core tenets of our strategy. First, the networks that are at the heart of today's AI data center build-out. And second, how we are enabling your transformation into an agentic enterprise. In the AI era, the network is both the essential enabler and the main bottleneck for performance. Nobody understands what is at stake more than our customers who are at the leading edge of AI, customers like Vultr, the world's largest privately held hyperscaler. Let's take a look at what HPE and Vultr are building together.

[Presentation]

Thank you to the entire Vultr team for being such a great partner. I'm excited to share that this partnership continues to expand as we work together with NVIDIA to support Vultr's next phase of growth. What Vultr is building at hyperscale points to a truth that every architect knows that there is always one core element of your infrastructure that touches everything.

With AI, that core element is the network. The performance of your entire architecture depends on it. Every bite, every token, every decision, all of it crosses the network, which is why today, we are bringing the HPE Juniper network into our AI data solutions, enabling more efficient, high-performance AI environment. Whether you are a hyperscaler service provider on neo-cloud or a large enterprise, you have more choice in how you connect and secure your largest AI investment.

Let me show you how it all comes together, starting with a model training use case. An AI data center is designed for one purpose, turning data into intelligence as quickly and as efficiently as possible. At this scale, performance depends on how tightly compute and networking work together. For customers building AMD-based systems with Helios, we are introducing the industry-first HPE Juniper networking scale-up switch, purpose-built for the AMD Helios architecture. The QFX5250 brings scale-up performance into an open Ethernet fabric designed for AI at scale. It connects 72 GPUs into a single rack, delivering 260 terabytes per second of aggregate scale-up bandwidth. You get the low latency performance required for large-scale AI workloads with the openness of standard-based Ethernet SONIC OS support and Juniper AI automation.

But one rack is just the beginning. The largest models train across hundreds, even thousands of racks operating as a one cohesive cluster. Multiply a small delay across hundreds of thousands of GPUs over weeks of training and your network can mean the difference between training a new model in 90 days or 30 days. Think about that. It is the difference between chasing a breakthrough or making one. That is why the scale-out network is so important.

The Juniper QFX family is built for this next generation of AI scale-out connectivity. Our newest addition to the portfolio is shipping today. The QFX 5250 is the world's highest performance 100% direct liquid cooled ultra Ethernet transport rated switch. Powered by Junos OS, it moves data across massive AI clusters. It achieves this through low latency congestion control and operational simplicity required to keep hundreds of thousands of GPUs working together.

Increasingly, AI data centers are expanded beyond single sites that find multiple data centers and regions, sometimes hundreds or even thousands of miles apart. That puts new pressure on the network between these environments. That is where the HPE Juniper PTX routing family excels. PTX is built to carry massive volume of traffic across the data center interconnect core and edge networks that connect today's distributed AI infrastructure.

Our PTX 12000 series is an ultra-dense routing platform designed specifically for AI fabrics. It enables 800-gig routing, 1.6 terabit-ready scale and ZR/ZR+ coherent optics to connect data centers across sites without compromising performance. And to protect that connection, we have our HPE networking SRX family, including our most popular firewall, the SRX 4700. It is one of the fastest quantum-safe firewalls on earth, delivering up to 1.4 terabits per second of security performance in a single rack unit. It helps you secure modern data centers without slowing down the application and AI workloads you depend on it.

But when we talk about a complete portfolio for AI training, we support every layer of a modern networking architecture from scale up and scale out to scale across and secure data access, all in a single coherent architecture that is secure and fully automated. With the introduction of the HPE AI grid with NVIDIA at GTC in March, we extended this integrated network even further. Built for service providers, the AI grid combines NVIDIA accelerated computing and AI networking, including Spectrum-X, ConnectX and BlueField with ProLiant Compute and Juniper routing security and unified orchestration across the full stack. Together with NVIDIA, we are enabling a wide range of new real-time AI services from conversational agents and interactive media to hyperpersonalized experiences across hospitality, health care and retail.

The real value of AI increasingly comes from inference when intelligence moves closer to your users' applications and data. That requires a network built to extend AI to the edge locations like regional data centers and service provider sites where the Juniper MX family of edge on-prem routers is top of the class. Our MX301 brings the proven performance and flexibility of the MX family into a small form factor, 1RU, power-optimized platform. It is purpose-built to move AI inference out of the cloud and closer to where the data is processed for inferencing so we can accelerate decision-making.

Powered by Juniper's sixth-generation Trio silicon has near infinite flexibility to meet your networking needs today and into the future. To build your inference environment, you also need a high-performance switching. That is why today, we are introducing the new HPE Juniper networking QFX5140 inference switch, purpose-built for distributed AI deployments. Also in 1 RU, the QFX5140 delivers up to 16 terabytes per second of switching capacity, connecting GPUs and inference infrastructure with the AI optimized load balancing and end-to-end congestion control to maximize performance. The 5140 gives every edge location, the local intelligence to host AI workloads closer to where inference is needed for faster AI responses and better experiences.

Look, the bottom line is to win in the AI era, you need a network built for the full AI life cycle from training at the core to inferencing at the edge. AI is also transforming the demands of the campus and branch networks. They still need to securely connect people and devices. But now they also need to support AI-powered workflows that depend on real-time access to data without compromising speed, security and reliability. That level of complexity cannot be managed through reactive troubleshooting alone. Your network has to see more, understand more and do more.

Self-driving networks move IT from reactive troubleshooting to proactive assurance, understanding experiences, identifying root causes and resolving issues faster. In the race to self-driving, HPE continues to lead. In fact, we were recently recognized as a leader in the Gartner Magic Quadrant for both wired and wireless LAN the 20th consecutive year, positioned highest in the execution and furthest in vision. What matters most is what this capability means for customers like the Milano Cortina Winter Olympics, where HPE helped deliver flawless network performance across a very complex environment, spanning 15 venues, hundreds of miles apart.

HPE Mist adapted the network in real time, helping ensure seamless secure connectivity for everything from broadcast, live streams to event operations and client engagement. Every moment could be viewed by millions. I hope you watch the Olympics, while organizers operate with confidence, knowing the self-driving effort was working behind the scenes to maintain a rock solid performance.

Today, we are extending this self-driving experience across our Aruba networking portfolio with 2 new announcements. First, Aruba CX switching is coming to HPE Mist. You gain AI native assurance, faster troubleshooting and automated operations across your campus and branch environments. And second, which is what we talk about cross-pollination, right, we're running. Marvis actions is coming to HPE Aruba Central. Marvis is the first network assistant in the industry to bring conversational AI to networking. So your network can move from reactive to self-driving with AI native operations that are continuously improving.

So you can see how much progress we have made with Juniper in such a short period of time. We are really proud of the progress we're making for you, our customers and our partners. But across industries, customers are making the switch to HPE networking and discovering that the self-driving network is a quiet network because it just works. That is the experience we want every one of you to have. If you are considering a change from your current networking provider, we know who those are. I encourage you to start with a single site or even a single floor. Experience what a self-driving network can do in your environment. You will be amazed by how simple the experience is. And I will ask you to not miss Rami Rahim's general session later today plus our four networking spotlights throughout the week to see how our self-driving capabilities are coming to life across every single domain.

We have talked about the networks that make AI possible and how HPE is delivering the next generation of self-driving networks that are self-healing, self-protecting and self-optimizing. Now let's turn to the next major shift in the AI era, the rise of the agentic enterprise. AI is no longer just a tool for finding answers. It is a critical part of how work gets done. Agents now reason across data applications, models and workflows. They help you make decisions, automate processes and are increasingly taking action on your behalf. Soon, IT will be responsible for thousands of agents that are part of your enterprise workforce operating across every function. But today, much of that innovation is still happening in local clients in the hands of developers and small teams, often outside formal IT oversight. That speed of adoption is exciting, but also creates a real challenge, the shadow cost of an agent workforce that now must be managed at scale we have never seen before.

Agentic AI demands a new set of enterprise requirements. Agents need to be secure and governed with clear guardrails for what they can do, what systems they can act on and most importantly, what data they can access. They need to be trained with trusted enterprise data because the agents are only as good as the data and context behind them. And they need to -- they need also infrastructure that can scale as demand grows without runaway cost.

Well, we introduced our HPE Private Cloud AI 2 years ago. We gave enterprise a Turnkey AI factory that simplified AI adoption and provided more control. It brings AI to your data, not the other way around. So today, we are enhancing private cloud AI for the next generation of agent workloads, helping you govern agents, ground them in trusted data and scale your inference initiatives.

Let's unpack what is new, starting with agent governance. You can now register agents built in any framework and wrap them with security controls that protect API calls, identity and encryption with zero code changes required. A new 3-tier identity model verifies the user, governs the agent and enables human approval for sensitive action. Today, we're also announcing a new capabilities for secure agentic operations with NVIDIA OpenShell and NeMo Cloud. OpenShell provides a modern and active run time for advanced private AI agents with policy enforcement built into how agents run. Each agent operates in its own isolated environment with guardrails for what data it can access, what systems it can interact and what actions it can take. And with NeMo Cloud, you get an open source of reference stack and blueprints for govern agentic workflows, helping you move faster while maintaining the control and accountability enterprise AI requires.

As agents operate across production environments, they also need a new class of operational risk, introduce a new class of operational risk. And that's why we are bringing Zerto to your agentic enterprise. If an agent makes a mistake, Zerto helps you quickly roll back to a clean state, reducing downtime and helping you protect your business. Governance in your agentic enterprise is paramount, but governance alone is not enough. Your AI agents are only as smart as the data you use to train them. Traditionally, that data requires custom preparation for every use case and months of building the right AI data pipelines, but not anymore.

Private cloud AI helps make that data you already have ready for agentic AI -- enterprise for agentic AI. With a govern data layer and integration with the NVIDIA AI data platform, you get a unified way to access, prepare and manage enterprise data across your existing environments.

Now with Alletra storage MPX10000 as the storage layer for private cloud AI, you can build on a high-performance data foundation designed for modern AI. The X10000 adds real-time metadata enrichment in native MCP support, so your agents and applications can retrieve the right data and context faster across structured and unstructured data. That means less custom integration work and 7 to 12x -- months faster time to value compared to what you normally do, which is yourself builds the whole environment.

Once you have governed agents and train them with the right data, it is time to scale across both agentic AI and your broader enterprise inference workloads. Private cloud AI can now serve larger models across multiple systems with multi-node inference, so capacity grows with demand. A new unified gateway simplifies access to frontier and open source models. This gives your team one unified API for model access with centralized credentials, budgets and policies.

We're also expanding private cloud AI with new configurations that scale up to 256 GPUs, including the new ProLiant DL394 with NVIDIA Vera CPUs designed specifically for inferencing. And for long context workloads, we're also introducing shared KV cache capabilities that reduce the need to recompute context over and over. This delivers significant cost benefits to first token and massive performance gain in compute capacity. With private cloud AI, you have now the foundation to build your agentic enterprise with confidence. And what is make us stronger is the ecosystem we have built around this.

We continue to expand the HPE Unleash AI program, our curated ecosystem of validated partners, blueprints and orchestration frameworks for private cloud AI. With more than 60 partners and hundreds of use cases, unleash AI helps you find trusted solutions for scaling AI across your enterprise from securing agents and models with partners like CrowdStrike and Fortinet to expanding where you can deploy AI with Digital Realty and Equinix. Private cloud AI and the unleash AI ecosystem help you move from AI ambition to real-world impact faster.

Take examples of that. For St. Jude's Children's Research Hospital, that means bringing AI closer to doctors and researchers, accelerating life-saving discoveries while protecting highly sensitive medical data or for Blue Star operations, the business behind the Dallas Cowboys, it means reducing lower-value work streams and advancing strategic decision-making across football and business operations. And for the Ryder Cup, it means being able to power real-time event intelligence from crowd management and concessions to volunteer assistance and operational planning. In fact, the Ryder Cup organization are now leveraging the private cloud AI to turn the next event in a massive success by really using a digital twin approach so that will help them architect the 2027 tournament experience.

But look, these are just a few examples of how we are giving you a faster, more structured path to AI adoption that will transform how you run your business. And there is more to come. Tomorrow, Fidelma Russo will share additional news in her CTO general session and go deeper in how our latest cloud and AI innovations help you build a new operating model for your agentic enterprise.

AI today is about moving faster from ambition to outcome, accelerating time to token, reducing execution risk and ensuring your environments are ready to perform from day 1. Our AI factory solutions are designed to do exactly that with validated architectures, agentic operations and enterprise-grade support. They also meet you where you are designed for your unique operating models, governance needs and scale. For enterprises, I share how private cloud AI is a secure and govern prepackaged AI factory for your agentic enterprise.

For model builders, service providers and neo-clouds, our AI factory at scale is built for large multi-tenant AI environments. And for government, regulated industries and sovereign entities, our AI factory for sovereigns enables you to deploy AI aligned to your local data, security and compliance requirements.

Across our AI factory portfolio, our deep collaboration with NVIDIA helps you build the latest accelerated computing platform like NVIDIA Vera and Vera Rubin. NVIDIA Vera CPUs is on our latest ProLiant servers are powering now agency workloads across enterprises. In supercomputing, NVIDIA Vera and Vera Rubin architectures are advancing our Cray portfolio for both HPC and AI. And in AI factories at scale, NVIDIA Vera Rubin NBL 72 is driving the next frontier of rack scale solutions. Compared to NVIDIA Blackwell, Vera Rubin NVL72 delivers AI training with 1/4 of the GPUs and AI reference -- inference at the 1/10 of the cost per million token. So think about that, the massive gains you can get to get to that token faster. So whether you are building for the enterprise or training frontier models, HPE gives you a path to build and scale on the latest NVIDIA accelerators.

As AI scales across more users, more data and, of course, critical operations, trust must be built into that foundation. That is why we are making confidential computing standard across the full HPE AI portfolio, helping protect sensitive data, models and workloads while they are in use. With NVIDIA confidential computing, AI workloads run in trusted execution environments that are a hardware-protected layer of security across the stack. And for organizations operating in a most sensitive environment, we are taking that trust foundation even further. Our sovereign AI factories now include defense grade security hardening, federal compliance readiness validated encryption standards and global data protection requirements all built in. So if you're in defense, government or financial services, this is the sovereign AI architecture you have been waiting for.

Architecting for AI requires looking ahead, anticipating and designing for the constraints that will shape the future. There is one challenge we all need to overcome, not just for our industry, but for our society and our planet, and that is power. Every model, every workload, every agent depends on power because at its core, an AI factory is doing one thing, turning electrons into token.

The U.S. is on track to have a 19-gigawatt power gap by 2028. That's roughly enough electricity to power 16 million homes. And data centers are expected to account for nearly half of the U.S. electricity demand through 2030. One customer, Siemens Energy is tackling this challenge head on, helping build the energy infrastructure, the AI era requires. They are doing it by applying AI to their own business with HPE helping deliver the AI foundation across networking, storage and compute. Let's take a look.

[Presentation]

I want to thank the Siemens Energy team for working on such an important challenge. We are proud to support your ambitions. Initiatives like this underscores a bigger point. As AI scale, the future will not be defined by compute alone. It will be defined by how efficiently we can power it, cool it and connect it. That is why research become so critical. For 6 decades, HPE Labs has helped shape the future of enterprise computing. Your next-generation infrastructure will need to operate with much greater intelligence, efficiency and transparency. Today, our researchers are applying AI to improve AI systems themselves, making them more scalable and more sustainable. This is where HPE has a unique advantage. We engineer the compute the complete architecture from compute, servers, obviously, networking, storage, software and security. And we apply that system expertise across the full stack.

With innovations like GreenLake Intelligence, we are developing predictive self-driving intelligence that can learn workload patterns and place data where it needs to be before an application asks for it. Across your broader data center environments, we are using AI to improve resource management, identifying idle patterns and reduce energy and water consumption without compromising performance. We also advanced in the next frontier of computing through our work in quantum with initiatives like the Quantum Scaling Alliance and our work in distributed quantum simulation, HPE is helping bring quantum out of the lab into the real world.

You can see that future taking shape right here at Discover, where a quantum chandelier sits along an original HPE Cray-1. It is a great reminder that how far high-performance computing has come and whether it is heading next.

As network in HPC and AI quantum converge, progress will depend on how effectively we bring these technologies together at scale. Yesterday, we took another major step forward with the announcement of an expanded industry collaboration to advance hybrid quantum. Together with these leading companies, we are building a full stack hybrid quantum platform that extends our world-class HPC and AI infrastructure and moves Quantum closer to real-time and real-world deployment. Quantum advancements like this are accelerating the path to faster, more efficient solutions for most of the world's complex scientific and industrial challenges. These are the challenges that inspires our HP Labs. Ultimately, it all comes back to one simple mission to advance the way people live and work. It is a guiding force behind our innovation, our people and our long-term strategy.

Today, we have covered how architecting for AI starts with your network and how we can help you transform into an agentic enterprise. You have seen the powerful outcomes that results from people with bold ambitions that are matched with the right technology and the right partners because none of this happens alone. Our partners help HPE bring these ideas to life with expertise, reach and execution customers depend every single day. They help us extend our impact, bring innovation closer to the customer and communities we serve. Many of our partners have generously sponsored this week. Discover is only made possible because of you, like us, believe in the vision and the power of pursuing it together. So I want to thank all our sponsors and all our partners. We appreciate you very, very much.

We take tremendous pride in knowing that our innovations are a catalyst for your success, driving outcomes that propel new opportunities for you and your customers. As you experience all Discover has to offer this week, keep these things top of mind. First, architect deliberately. The choices you make today will define your success tomorrow. Second, start with the network, make your network the core foundation of your AI and cloud solutions. And finally, choose HPE as your partner to bring the full stack to help you build your AI future with confidence.

This week is an invitation to think bigger, to move faster to architect the future you want to lead and for the world. And remember, you don't have to build this future alone. We are here to provide the intelligent foundation so you can move boldly, live with purpose and unlock your ambition. Thank you very much. I hope to see you on the floor. Enjoy the rest of the week.

Thank you for joining us at HPE Discover's Networking General Session. Please welcome to the stage, Executive Vice President, President and General Manager, HPE Networking, Rami Rahim.

Thank you. Good afternoon. Hello, everyone. Welcome to HPE Discover Las Vegas. My very first Discover Las Vegas, could not be more excited about being here. Okay. Thank you. Thank you.

So I got a story for you. A few years ago, in San Francisco, actually, San Francisco became home to one of the most famous engineering cautionary tales in modern construction history, and it was all about the Millennium Tower. This is a stunning 58-story luxury skyscraper in the heart of the city, beautifully designed, technologically advanced, built to be truly iconic. But over time, something unexpected started to happen. The building began to sink and then it began to tilt, not because the structure above ground was poorly designed, but because the foundation underneath it was built -- wasn't built to handle the long-term realities of the environment around it. And as the demand on the building increased over time, the weakness underneath became impossible to ignore.

So right now, companies everywhere are racing to build intelligent applications, autonomous operations, real-time experiences and entirely new business models powered by AI. But AI places enormous new demand on the infrastructure, massive data movement, constant inference, real-time responsiveness, explosive scale. And if the underlying foundation isn't designed for that new reality, eventually, the strain starts to show up.

So to succeed in networking today, we have to think differently because AI is changing everything. One of the clearest messages from this morning's keynote was simple. AI is reshaping every part of the enterprise. But none of that happens without the right foundation underneath it. And that foundation starts with the network. The network is no longer infrastructure sitting quietly in the background. It's become a strategic platform for how organizations operate, innovate and scale. Why? Because the demands on it are exploding, more users, more devices, more applications, more data and entirely new expectations for real-time experiences across every industry. From digital payments, connected stadiums, health care, research, media and AI-driven services, the network is what makes those experiences possible.

That's why leading organizations are treating the network as core strategic infrastructure, not just to keep up, but to unlock what comes next. Now the ones that embrace this shift will be better positioned to innovate faster and to compete more effectively. The ones that don't will increasingly find themselves left behind. But the good news is this, while AI is placing unprecedented demands on the network, AI is also becoming the answer to how the network adapts, scales and withstand that pressure because the old model of networking, static, manual, reactive simply cannot keep up with the speed and complexity AI introduces. What's required now is a network that can learn, a network that can predict, a network that can optimize and heal itself in real time. In other words, the future of networking will not just support AI, it will run on AI.

Now we see this in two really powerful ways, AI for networks and networks for AI. First, AI is changing how networks are operated. As environments become larger, more distributed and more dynamic, manual operations are just not going to keep up anymore. That's where AI for networks become a true game changer. The payoff is significant, better uptime, better user experiences, fewer tickets, faster remediation and more time for IT teams to focus on strategic work instead of constantly troubleshooting.

And second, AI is redefining what the network itself must deliver. So the reality is this, AI innovation can only move as fast as the network allows. You can have massive compute power and millions, if not billions spent on GPUs. But if the network introduces latency and bottlenecks and instability, you're limiting performance, slowing down outcomes and giving up ground to the competition. That is why the network has become essential infrastructure for the AI era. And none of this works. None of it works effectively at least, unless security is built into the foundation itself because the network is now both the connected fabric for the business and unfortunately, increasingly one of the main pathways that attackers use to target it. That means the network has to be a core part of the security strategy with AI-driven anomaly detection, automated response, role-based access and enforcement and Zero Trust approach that helps protect users, applications and data everywhere.

And just as importantly, it has to deliver that protection without adding friction that slows users down or piles more complexity onto IT teams. That is the real shift here. Security and user experience can no longer be trade-offs.

Now all of these point to a new era of IT, one where self-driving networks are no longer optional. They are essential because AI scale infrastructure cannot practically be operated manually. The networks of the future must be able to sense, to learn, to optimize, to protect and heal themselves in real time. And let me be clear about this. HPE has made more progress than any other company in these areas. We are bringing together AI native hardware, software, silicon, security and agentic AI Ops into a closed-loop system that operates at speed and scale that humans alone simply cannot match.

The result is a network that delivers better performance, stronger resilience, simpler operations and better user experiences, and that's what the self-driving network is really all about, moving IT teams from manually operating infrastructure to accelerating the business.

So today, I want to explore what this next era of networking actually looks like and how secure AI-native self-driving networks are solving real problems for our customers. Now some of these customers are going to be joining me on stage to share how their organizations are navigating real-world problems today. And you're going to be seeing demonstrations of how HPE, HPE networking helps organizations overcome these problems and move at the speed of innovation with confidence. Because you know what, the organizations that modernize their networks now with the right architecture and the right intelligence and the right operational model are going to be much better positioned for what comes next.

With that said, enough from me. Let's hear from the people out there building and operating these environments in the real world. So please join me in welcoming my first guest, CIO of the Ohio State University, Rob Lowden.

Rob, welcome.

Thank you. Thank you very much.

Rob, thank you so much for joining us at HPE Discover Las Vegas. Maybe just to get started, introduce yourself and tell us a little bit about your role. You're responsible for managing a very large campus, thousands of students, tons of connected devices and what a great school Ohio State is. So tell us a little bit more.

Well, thank you. Absolutely.

I think there's some alma mater in the audience.

Any grads out there. We have 650,000 living alumni. So there's got to be a few of them here. So thank you for the opportunity to share a little bit about the Ohio State University. Not only is it the namesake land grant university, small city within Columbus, the capital state of Ohio. We have 66,000 students and 8,500 faculty. We're the fourth largest university in the country. And networks matter a lot to us, as you've already alluded to. So, on our Columbus campus alone, we have 22,000 HPE wireless access points, indoor and outdoor across numerous acres on campus. We have 15 colleges. We have a hospital. We have a comprehensive cancer center. And maybe a few of those folks out there know that we also have a football team.

I have heard, yes. And I think there are some rivalries out there that people care a lot about. Higher education institutions are probably some of the most demanding network environments around because it's pretty much everything you just stated. What are the biggest challenges and demands that, that places on your infrastructure?

Absolutely. So one that's unique to the hoot and holler that we just heard, we have a stadium. Some of you might have heard of it, the Horseshoe, not the Horseshoe Casino, although similar to the Horseshoe Casino, the house always wins at Ohio State.

So at the Horseshoe, we can host 100,000 participants viewing us crushing maybe that institution to the north that we don't refer to too often. But that creates a unique networking challenge and fan experience, ticket, check-ins, everything being electronic. Last fall, we hosted a team from down south, and they were ranked #1 when they came to Columbus, but maybe they left a little lower. And that was the largest broadcast in NCAA history of any sporting event. So there's a lot on the line.

During an event there, there can be another 100,000-plus people outside the stadium. So it's not uncommon for a home day game for us to have 200,000-plus people around that. With the partnership with HPE, as we speak, engineers are working together, collaborating on refreshing our WiFi network there. And it's critically important to us, not just to ensure that, that fan experience is there. 200 yards away, we have a $7 billion hospital system, comprehensive cancer center, 2,000 beds in that facility. And that's going on next to 200,000 people enjoying the game. So we're putting in, I believe, the single largest implementation in the country, 2,000-plus wireless access points HPE, Juniper Mist. I'm super excited about.

Let me ask you, with that kind of complexity, I would imagine this is where AI Ops fits and self-driving automation, yes, leading question can be quite useful. Would it -- just comment on that a little bit.

Absolutely. AI Ops is something that we absolutely are excited to have in place. We have 100,000 managed devices where expectations for us on bring your own devices is substantial, adds tens of thousands of more. Throw in a game like that, we need AI Ops that's provided to crunch all of that data in real time. And we've seen this, in effect, move us from scenarios where problem resolution can take hours sometimes, and we just can't afford that with the various operations going on at the institution. So it's literally moving us from absorbing that threat intelligence, applying the AI Ops and giving us answers to resolution, and we're seeing things being resolved in minutes versus hours.

That's awesome. Look, Rob, we so thoroughly enjoy having you as a customer. We've become a better technology company as a result of having you as a customer. Thank you so much for joining.

Thank you, Rami.

Okay. I am pretty sure that what Rob just shared resonates with many of you because what he described isn't unique to one university or one industry. We hear the same thing from customers everywhere, more users, more devices, more applications and now AI adding an entirely new layer of demand and complexity. And at the same time, IT teams are being asked to move faster, simplify operations, strengthen security and deliver a flawless experience. And that is exactly why the self-driving networks matter. And if the self-driving network can handle an environment as dynamic and demanding as Ohio state's, it can handle just about anything.

Now you all know we deliver those AI-native autonomous capabilities through 2 industry-leading Agentic AI Ops platforms, HPE Aruba Central and HPE Mist. Each platform brings unique strengths and serves different customers and is trusted every day by organizations around the world. And let me be clear about this. We are committed to innovating and innovating aggressively on both of these platforms, including bringing the best capabilities from each platform onto the other. That means both platforms continue to get stronger and both are here to stay.

So to show you what these self-driving capabilities look like in practice, please welcome Sunalini from HPE's campus and branch business.

Hi, Rami.

Hello.

Hello, everybody. Hello again. Welcome, welcome. As you said, we are innovating in both platforms to deliver a consistent self-driving experience. This is possible due to two things: Microservices and a common Agentic AI framework. With micro services, we are able to develop self-driving innovations once and deploy them on both the HPE Aruba Central and HPE Mist platforms, much like a single app experience is delivered on both iPhone and Android. So no matter what Agentic AI Ops platform you are on, you are going to be enjoying the benefits of a self-driving network. And with a common Agentic AI framework, we are able to accelerate self-driving capabilities at a much faster pace on both platforms with trusted actions that deliver measurable value.

Our Agentic framework, Rami, is unique in the industry because of these key foundational pillars. First, we use real live experience data, every user, every minute, which HPE has uniquely validated against real customer support cases and enriched with digital twins to maximize the efficacy of our AI-driven insights. Second, we have an API-first approach, which means all the data is available via our APIs. This makes our MCP server and tools extremely powerful, delivering agentic automation at scale. Third, a powerful set of AI agents and skills analyze all data sets and apply reasons from HPE Marvis Minis, which serve as digital twins of user experience to agents analyzing packet captures, logs, knowledge-based articles, security vulnerabilities and all the support data and signals we get. They all work together to proactively diagnose and autonomously root cause problems impacting users without inundating operators with data.

And fourth, not the end yet. The fourth aspect of what makes us unique, models take the analysis and curate it to understand and analyze post connection issues. For example, Marvis has a large experience model or LEM, as we call it fondly, that identifies the cause of bad Zoom and Teams calls and predicts future problems to prevent them when possible. We bring all of this together into the Agentic framework, resulting in a system that is continuously observing, reasoning and analyzing with autonomous actions that optimize user experiences. This is a self-driving network.

So Sunalini, most vendors talk about AI assistant and agentic AIOps, but still rely on reactive, human-driven troubleshooting. But if humans still have to fix the problem, where exactly is the self-driving in that? So can you provide us an example of why real self-driving network operations actually matter?

Of course. We all know that today's networks need to cater to high-density requirements while also meeting the performance expectations of every user and application. But most networks are not designed to meet intermittent surges in peak traffic. Think of an all-hand in an office building, a crowded classroom during a popular university lecture or even this very room right now where thousands of you are eager to see the best networking solutions ever. Operators try and accommodate these situations with a combination of static boundary parameters and on-demand changes. But often, that just isn't enough.

Let me show you a self-driving network powered by real Agentic AI and how that handles this problem.

They can't wait. Let's do it.

If we look at Marvis, we see that all the users in this office building are currently happy. But was this the case all of last week? No, it wasn't. So what happened? How did unhappy users become happy? Let's take a look.

Last week, Marvis detected that over 6% of user minutes were bad, which may not sound like a lot, but it impacted hundreds of people. Marvis has a self-driving action for dynamically fixing capacity issues. This action was enabled and was able to autonomously fix the problem by enabling dual-band 5 gigahertz. This reduced the fleet utilization from 90% to 54%, enabling a better experience for users that were unhappy. But how did Marvis know what to fix and when to fix it? The unhappy minutes in the past week triggered various models and agents in the HPE Agentic framework to reason and analyze and root cause the problem.

For example, one model analyzed the service level experiences. Marvis Minis agents were activated to test the network using digital twins and other skills went into effect. Based on the reasoning and analysis, Marvis determined that the wired and WAN was not the problem. The wireless network was, but what in wireless wasn't working? Coverage was fine, roaming was fine, wireless capacity was bogged down on a few APs that were functioning at 90% peak utilization. This is when Marvis went into full self-driving mode. It changed the RF parameters and validated the service expectation of the users to ensure they had 100% satisfaction.

Rami, this wasn't manual tuning. This wasn't chart an error. This was a network optimizing itself to deliver the best user experience. And this is available right now in HPE Mist.

Today, like right now?

Today, right now.

Amazing. So let me just think about what I just saw there. The network identified the issue, right? The network understood the root cause, it determined the right action and resolved the problem automatically before any user even had a chance to complain about their experience. No emergency troubleshooting, no IT team scrambling to diagnose, no help desk tickets. I mean that sounds like real value for IT teams and also for end users, Sunalini, right?

Absolutely.

But that's just one example. I suspect you've got more.

Well, let's find out.

Okay.

We heard from our Mist customers that Marvis Actions, which automatically identifies network issues and proactively resolve them is mission-critical to daily operations. So that's why HPE Marvis, our industry-leading AI engine is coming to HPE Aruba Central. It has all the simplicity and all the impact that customers have come to expect from Marvis. This is experience-first AI in action, and it is a perfect example of how we are able to develop self-driving innovations once and deploy them on both the platforms, Aruba Central and Mist seamlessly, thanks to our Microservices foundation and the common Agentic framework. That's how we are bringing Marvis to HPE Aruba Central right into the global NOC View.

Behind the scenes, Marvis does all the heavy lifting, correlating logs, alerts, signals across the entire stack. Your morning cup of coffee view will now showcase end user impacting issues across wired, wireless and SD-WAN with recommended actions. Missing VLAN, MTU mismatches, negotiation failures, they're all coming to Central. In addition, what we call the Marvis Trust list is also coming to HPE Aruba Central. These are actions that you can choose to be fully autonomous. When enabled, Marvis not only finds the root cause, it fixes it for you.

Imagine a security camera connected to a wired port in a bad state. No camera feed, that's a real problem. With Marvis in self-driving mode, that port is recovered automatically and expeditiously and the camera is now working again. That is the power of bringing Marvis into HPE Aruba Central, not just more visibility with highly accurate actionable recommendations, but a better end-user experience, thanks to self-driving capabilities.

Okay. That's like a pretty awesome example of just how quickly innovation can move when we bring together the best of HPE Aruba Networking and HPE Juniper Network, bringing Marvis into HPE Aruba Central is a -- not a little, a huge step forward. That being said, Marvis truly is the AI engine behind the self-driving network, and now Marvis will be available across everything, both platforms. So our mission is simple. It's not easy to do, but it's simple to say, bring the best innovations to both platforms so that every customer in every industry gets the same powerful self-driving network no matter which platform they choose.

Okay. Now Sunalini, we did this by promising both software and hardware cross-pollination. You just demoed some compelling examples of common software. What can you tell us about hardware? Have we made any progress on that front?

Yes, Rami. We have made quite a bit of progress. A few months ago, we made a commitment on common hardware by announcing the first dual-platform access point, the 723H, which is now generally available. Within the first year of the Juniper acquisition, not only have we cross-pollinated AI models and Agentic frameworks, we have delivered an access point that works with both the Mist and Aruba Central platforms. And on that same theme today, I am super excited to announce that we are doing something very similar with switching. Our world-class HPE networking CX portfolio, which previously was supported by Aruba Central will also very soon be supported by Mist for day 0, day 1 and day 2 operations.

It's a big deal.

It's a big deal. Let me give you a sneak peek of what you can expect in just a few -- few weeks, sorry. Let's start by showing how easy it is to onboard the HPE networking CX switch into the HPE Mist platform. It starts with a simple scan of a QR code of the CX switch from within the Mist installation app. Once onboarded, the devices show up in the inventory and they are ready to be configured with templates that enable you to centrally define and apply consistent configurations at scale like pushing VLANs and port profiles to hundreds of switches.

With telemetry coming into the Mist platform every minute for every wired client, combined with high-efficacy AI/ML models from Marvis, we can measure pre-connection and post-connection wired SLEs and not just a port is up. For example, the successful Connect SLE measures pre-connection experiences and the bandwidth congestion and throughput SLEs will give us insights into post-connection experiences. With the SLEs, the IT administrators can quickly identify where an experience is bad, including interface anomalies and other issues but we don't stop there.

All the goodness we get from Marvis Actions will also be available for CX switches. This includes proactive recommendations for missing VLAN, MTU mismatches, bad cables and more. And the best of all, a self-driving trust list will also be available, starting with the ability to autonomously fix stock ports to remediate wired clients in a bad seat. This enables real-time closed-loop self-healing, reducing mean time to repair, eliminating manual troubleshooting and delivering great user experiences at scale.

This is huge because we all know the network being up is not the same as users having a great experience. So for our CX customers, no matter what agentic AIOps platform you choose, you get simpler operations, actionable and proactive recommendations and a real self-driving network.

Let me just pause here and say like, what makes this so powerful Sunalini is not just what you just saw, what all of you just saw, but how fast we made it happen. In just a few short months after the close of the acquisition, our teams came together to deliver real software and hardware cross-pollination. Honestly, I could not be more proud of you and the team, Sunalini because this is -- yes, yes, clap for Sunalini and the team. Mad respect because this is exactly what innovation at scale should look like, moving fast, bringing the best ideas together and delivering value to customers with speed. And the result is incredibly powerful.

You all have a self-driving network with the flexibility to choose the platform experience that works best for you with the confidence that your investments are fully protected. And Sunalini, the market, I think, is taking notice of this, right?

That's right, Rami. The best evidence is the new Gartner Magic Quadrant for wired and wireless access published just 4 weeks ago. Hold the applause, it shows HPE as a leader, applaud. And even more so, we are furthest to the right for vision and highest in the ability to execute. In my humble opinion, it is clear who the overall leader is. What do you think, Rami?

Look, I could not agree more. Thank you so much. Thank you so much for joining us again.

Thank you, Rami. Thank you, everybody.

I really think that says a lot. Last year, the industry looked at HPE Aruba Networking and Juniper Networks as two separate leaders. This year, the market is recognizing what happens when you bring those strengths together. One team, one vision, one innovation engine and most importantly, one clear direction towards AI-native self-driving network.

Now of course, this doesn't just apply to wired and wireless networking. It also extends into security, which happens to be the next topic I'd like to talk to you about today. And to help me tee it up, I'd like to welcome from the Royal Bank of Canada, one of my favorite countries, by the way, [ Marlin Grumman ].

A lot of people right?

A lot of people.

Please have a seat.

Thank you.

Thank you so much for joining us, Marlin. I know you're at RBC, an iconic bank. Just maybe tell me a little bit about your role at the bank.

Sure. My name is [ Marlin Grumman ]. I run -- I'm Senior Director, I run engineering and automation and by extension, AI, fifth largest in North America. And just sort of lay the context of -- so there's difficulty from a threat perspective, about 4,000 endpoints from the SD-WAN and the Edge Connect. But if you took the entire scope of the threat landscape, it's probably double or triple between cloud, trading, every business line. So it is quite extensive, and it's extremely difficult for a legacy bank. And I'm not going to -- I don't mean that in the sense it's truly legacy, but very, very difficult to do.

I would imagine you're dealing with massive volumes of not just data, but sensitive data. So security, I suspect, is a bit of a consideration for you, right?

Yes, it is job #1, I think, for the most part. I don't think -- I think if you're a bank and you're regulated like we are in 29 countries, you have all of the, I'm going to say, the environmental from a regulatory perspective, you have to deal with for every single country. It is a very difficult process. So security for us is job #1. We actually don't have any other job other than protecting our client data. And not only that is a blueprint to actually how we operate. We feel that it's our competitive edge. You have 13 million clients, you're in lots of different countries. That data in itself forms the basis for decisions and all exactly our competitive edge. So that we protect with everything in our [ power would be ]...

So test me on this. One thing I think every malware has in common is that it has to use the network to do security work. So I always said that an effective security strategy must also leverage that same network to better detect and to better enforce policy. I mean, what do you think of that?

Absolutely. I think like we think about this a lot, and this is primarily to sort of keep the services up. I know you talk about self-healing and -- but for the most part, in general, you troubleshoot at the network layer. That's the first -- that's the only place that you can get some immutable evidence to be able to identify what's going on. So we've always lived on that mantra. So for the most part, how we sort of detect and manage troubleshoot, protect is all at the network layer.

So being able to sort of identify when somebody is knocking at the door, that's very important. And the only way you're going to really see that is whether it be lateral movement in the network and so on and so forth. So you want to make sure that, that side of it has -- is fully covered part of the resetting the stage for the 4,000, at least what we can identify and speak about clients is the Edge Connect and that platform using CPI engine helps us sort of create, I would say, a persona or a personality for a user. Anything outside of that becomes an anomaly. And that's something that allows us to now be able to go look, something is going on and so forth. But it's part of the intel that we need to be able to sort of shore up our threat intel.

Makes sense. I know you're working on some big AI transformation projects. I mean maybe just share with us a little bit about how these AI projects are geared towards network operations in particular?

So as you know, I know it's public knowledge, we are AI first and AI everything right now. We've pledged to our Investor Day shareholders that we will generate $1 billion in revenue just using AI. So we've embarked on this incredible effort. So the AIfication, I guess, if you want to look at it, of the entire bank. So that's a significant effort. The operational side of it, I think is sort of table stakes sort of -- I don't want to say it's the easy piece, but it certainly stretches all the way through every business line, AI being sort of the horizontal in which you kind of build the framework.

But the operations side is near and dear because that's where we actually can make a difference at least on our side. And we're sort of extracting telemetry. We're building the harnesses and creating a framework to be able to take that data you take the Connect, for instance, that's one place we also collect data to be able to sort of put into whether it be a [ SIEM ] or through some sort of [ MITE ] framework to be able to sort of mitigate the threat response but -- or create a threat response. But the whole idea is that this will be an AI everywhere, collecting data, mining it, normalizing it and vectorizing it and using that data to be able to help us solve the AIOps problem.

Thanks so much for sharing that with us. Fun fact for you. Everybody, I built my career at Juniper and now HPE. Many people think that was the only job I've ever had. Actually, my first job was serving ice cream downtown Toronto. And the very first paycheck I got, I deposited at RBC.

I have to tell you...

So I am proud to have you...

Absolutely. Absolutely. And I have to tell you, I hear it a lot from a lot of people. We're sort of that cradle-to-grave kind of bank. We want to make sure that you come in early and stay forever.

Thank you so much for joining us. Appreciate it.

Thank you, Rami.

What Marlin just described is really the reality for IT team is pretty much everywhere in the era of AI. Increasing scale, more complexity and a nonstop wave of evolving security threats, which is why networking and security can no longer operate separately. Security has to be built directly into the network. And that's exactly how we designed the AC self-driving network because the truly self-driving network doesn't just optimize and heal itself, it protects itself. So to show you what that looks like, please welcome product leader for SASE and security, Madani.

Hey, Rami. Good afternoon, Las Vegas. I'm excited to be here.

We're excited to have you. We're excited to have you. Okay, listen to me Madani. Attackers are already using the network as their weapon of choice, as you all know. And with the AI making threats faster, smarter, more sophisticated, defenders need to use the network as part of their defense. That's why security can no longer sit beside the network. It has to be built into the network. How do we help the good guys do this?

Well, Rami, it starts with a Zero Trust approach to network security. This assumes that no user or thing is trusted by default and requires continuous verification of every access request. This would be like a Blackjack dealer making a check a background check before ever dealing a single card and then kicking it off the table in the case of cheating. Wait, I think they actually do, do that. We successfully implementing a Zero Trust approach requires 5 core elements: visibility into all connected users and things, policy-driven orchestration, ubiquitous policy enforcement, real-time detection and automated AI-driven responses.

Now you do have to have the right mix of protected -- protective layers in the network to deliver autonomous protection for the strongest defense possible. And HPE has a full security portfolio, which includes firewalls with industry-leading efficacy and performance, NAC with comprehensive access control with consistent enforcement across every type of device or user on the network, SSC with agent or agentless deployment options, supporting a broad set of applications with intelligent routing. And last and certainly not least, SD-WAN with integrated application performance, which is -- and security, which is optimized for any environment.

So okay, we clearly have all the core security pieces in place and all of the elements of a winning hand, you might even say we have a full house. I'm sorry. But ultimately, what's important is how these capabilities come together, Madani. And nowhere is that more important than in SASE, right? Where networking and security operate as a single system to securely connect users, applications and data everywhere. I think you have a pretty interesting announcement to make on that front.

That's right, Rami. I am very excited to say that we have now combined our EdgeConnect SD-WAN and our SSE into a unified SASE orchestrator with one console, consistent Zero Trust policy and AI-driven operations for simpler, faster and more secure connectivity.

Very cool.

Now what do you all think about that? But Rami, I can't just come to stage and talk about it. I need this audience to actually see it and to demo it live.

Okay. Let's do it.

So let's check it out. All right. The new SASE orchestrator is a powerful solution combining both networking and security under one umbrella. Here, you see the dashboard page. This is where an administrator would get sort of a quick snapshot of everything that's going on within the environment for the last 1 hour to up to 7 days. But this isn't the special sauce.

Let's go ahead and check out the business intent overlays. To me, that's really where EdgeConnect shines. This is a great example where EdgeConnect has been self-driving for many, many years already. It autonomously delivers the best quality of the experience while also handling the worst kind of links for any type of application. Now let's jump out of this a little bit and talk a little bit about security. We'll navigate over here and check out the global firewall policy rules. This -- what's great about this feature is the fact that these firewall rules are written here, but automatically distributed across the entire SD-Branch -- SD-WAN fabric. Next up, let's take a look at the SSC policy rules. This is the new capabilities that are part of the unified orchestrator. Now being in Vegas, Rami, Blackjack is on my mind, which is probably okay. But visiting a gambling site on my company laptop, probably not so much.

All right. So let me show you how quick and easy it is to ultimately deploy like a web filtering policy for our ZTNA users. I'm going to attempt to type up here in front of a live audience, and hopefully, we'll get this right. I'm block gambling, come down here to the destination, and I'm going to look for gambling again. We've made sure that it sets a block and I'm going to go ahead and save it. Now what's going to happen is it's going to push this out. As you can see, essentially added that policy rule, fairly easy, even a product manager can do it. I'm going to go ahead and apply it. And all at once, we're now pushing out that policy for all of our users. And again, this policy could -- making these changes is all done directly here from that one place.

Now, that's what's really exciting about the fact that what used to be done in two different products is now possible in the single SASE orchestrator.

Super simple.

Super simple. Now before I leave, there's one more thing I want to take a look at, which is the new SASE Copilot, where ultimately, a user can ask a variety of different questions. We also have some pre-populated ones. And ultimately, what's great about it is it allows them to resolve issues and minimize risk to their environment very quickly.

I think this is a pretty big deal, right?

Absolutely.

So our SD-WAN and SSE solutions are already like excellent on their own. But now we're taking things a major step forward, bringing SD-WAN and SSE together into a deeply integrated solution. What else you got for us?

Well, I've got a few more demos, I do want to tell this audience that this is coming out just later this year. All right.

Let me actually -- before you go on, it's clear that SASE Orchestrator is like a great step towards agentic AIOps and Agentic SecOps with a network that is not just self-driving, but also self-protecting the Madani, right? But security is also critical to AI itself. Organizations need a way to harness the power of AI without compromising their data. So what are we doing to help customers on this front?

Well, Rami, we are all over it. Our AI aware firewall lets you safely embrace AI by giving you governance over how AI is used with real-time visibility, #1 rated efficacy and threat detection and simplified security operations. That means that you can protect sensitive data without slowing innovation. Let's take a look.

All right. So here, we have the Security Director Cloud, which manages our SRX hardware and virtual firewalls. Up at the top, we have a new capability, the Security Director Copilot, which lets our administrator ask different types of questions and can also select pre-populated ones. So this little exercise, I'm going to try my luck again at typing. And let's see here. We're going to show the latest threats to my environment. All right. We'll go ahead and do that.

And now the Copilot is going to spend a little bit of time thinking, doing a little bit of crunching of data. And what's happening behind the scenes is that we're pulling information from all the SRXs that are in this particular topology. We're also leveraging threat intelligence from our HPE Threat Labs. And ultimately, what we should see here any moment now as it finishes up its thinking is that it's going to give us some very specific insights about the different types of threats that we're seeing, as you can see here.

I'm going to go ahead and scroll up just a little bit here. We can see a variety of different pieces of information, user names, source IPs, [ destinations ] and so forth. But we really get some real detailed information in terms of things like what the threat is, industries that are being targeted and what countries. Now at the bottom, we also have a great set of recommendations, right? It's not just about showing the information, it's also being able to help supercharge our administrators.

All right. Let's go ahead and we'll jump out of that. Now for this next little segment, I'm just going to put a little bit of background on what we're going to do for my last demo. Taking a look at some of the policies that we have here. And I'm looking specifically around some AI governance rules that we put in place. Now there are 3 types of rules that we've implemented. One is for sanctioned AI apps, which are AI apps that are allowed by the organization. Unsanctioned AI apps, which that's just a fancy way of saying we're going to block it and tolerated AI apps, which are AI ops that have some guardrails. Now sanctioned and unsanctioned is pretty straightforward, but I would say that tolerated is a little more nuanced, and it's about having the right types of guardrails in place.

So let's just take a quick look at what that is. So here, we have a few different types of things that we're looking for. We're going to be looking at upload protection rules, prompt protection rules and last but not least, some keyword protection rules. Now let's see this in action. So I'm going to jump over. Now we've switched over to another view. This would be the end user view, if you will. I happen to be here on hp.com. And you know what, I'm going to go ahead and try to go to ChatGPT. Immediately, I get a block message, right? This is one of the apps that we've actually flagged as not being allowed. It's an unsanctioned app.

Now let's go ahead and try Claude as an example, same thing, just to make sure that everyone kind of saw that happen, we'll go to Google, and we'll bounce back, same effect. Now let's see what happens when we try a tolerated app. So I'm going to go over here to Gemini. And remember those guardrails I spoke about, Rami.

I remember clearly.

Okay. So I'm going to attempt to upload a corporate file, a company report, probably has some information, maybe it's an internal content. I don't necessarily think I should probably uploading it into this tool. And as we can see, Gemini is struggling a little bit. It's attempting to pull this file. But what's happening behind the scenes is our firewall is going and blocking this. And so ultimately, what we'll see here in a mere few seconds is that lo and behold, the upload fails.

All right. Now let's try a different example. I'm going to grab some text here that has some unique words like restricted, secret, encrypted. I'm just using this to facilitate my typing. And what I'm going to do here is I'm going to say, "Hey, let's summarize this information." I'll go ahead and paste that and let's go submit lo and behold once again, Gemini struggled. Can't figure out what to do, can't process this information because I was doing what he needed to do. All right. Now as I mentioned before, this is a tolerated application with some specific guardrails. And so I want to basically show to the audience that I haven't completely crippled this.

So let's see what happens as I say, summarize HPE Discover, and I go ahead and submit that. Look, already, we see a little bit of a change in behavior here. It's searching the web. It's going to figure this out. And yes, it is our flagship annual conference. So very cool. As you can see, how we would use this technology in a real-world case.

That's awesome. That's really, really cool. And I would imagine it's super, super powerful, right, Madani.

/>

Absolutely. And to me, this is what makes this solution so robust, so powerful, if you will. It proactively detects threats, simplifies operations with guided insights and enforces granular real-time controls to safely govern AI application usage.

You know what I really like about our AI firewall is that it gives customers the ability to see, govern and protect how AI is being used across their entire organization without slowing down their businesses because at the end of the day, nobody wants to choose between being secure and moving fast. And what you just saw were two powerful examples of something bigger, right? At HPE, AI, networking and security are no longer separate domains. They are converging into a single intelligent self-driving system. Madani, thank you so much.

Thank you, Rami. Thank you, everyone.

Okay. I want to shift gears a bit and talk a little bit about something that sits at the very core of networking. You know what that is?

Routing.

This is what Juniper was originally built to do. And honestly, routing has never ever been more important than it is right now because no matter what kind of network you are building, AI fabrics, data centers, campus environments, WAN, cloud connectivity, security architecture, service provider network, blah, blah, blah. Keep going, routing is foundational to it all. It is the connective tissue. There are some routing fans here, okay. It is the connective tissue for the modern infrastructure. And as networks become more distributed, more dynamic and more AI-driven, the demands on routing are growing dramatically.

So to help me kick off this important topic, I'd like to welcome the Director of IT at Sentara Health, Tom Johnson.

Tom, How are you?

Great, Rami.

Thank you for joining us. Tell us a little bit about yourself and also about Sentara Health for those that are not familiar.

Absolutely. So I've been with Sentara Health for 29 years. And during that time, I've witnessed and been a part of many major technology transformations. We are one of the largest not-for-profit integrated health systems in the U.S., Mid-Atlantic and Southeast. Over $14 billion in operating revenue, 35,000 employees, 12 hospitals in Virginia and Northeastern North Carolina. We have over 200 connected sites with 400-plus total points of care. We have a health plan division that serves almost 1 million insureds in Virginia and Florida. It all keeps us busy.

So I would imagine with that many different locations, employees and then patients and insurance business as well, you have a lot of sensitive data that you are dealing with on a daily basis. How does that impact your infrastructure and networking decisions?

Well, we moved massive amounts of data across the organization, and it directly impacts patient care. That's why our network has to be resilient, secure and always available because when data is delayed, care is delayed. Our clinicians rely on real-time data for decision-making at the bedside. And that requires instant, reliable delivery everywhere. Areas like radiology, where images from some of our systems are always pulled fresh with no caching. Latency simply is not an option.

Okay. So get it. You mentioned latency as an important consideration or requirement for your network. Now with many AI applications that require that, I get it. But what kind of AI applications are you typically deploying in health care? And what does that mean? What else does that mean for your network?

There are a lot of opportunities for AI. One example where we've seen real success is using ambient AI to capture patient conversations. It generates the clinical notes and even helps identify potential care gaps. It allows clinicians to spend less time focused on documenting and more time focused on the patient. But for those kinds of capabilities to work, our network has to deliver that data in real time, reliably, securely and without latency. So clinicians, they can trust it at the point of care.

So we were talking backstage and discussing this a little bit. But looking ahead, I know you have ambitious growth plans and digital transformation plans in particular. What can you share with us about how this is -- how HPE is helping you on this journey?

Well, I could spend a lot of time talking about our initiatives. Health care has got tons of them. But to enable them, our infrastructure has to be ready to grow with us. We need to be able to expand our operations and embrace new technologies while maintaining performance at scale. And HPE is a trusted partner that has given us the solutions we need now and into the future. And we love the AI capabilities that this gives us. You've heard about some of them today already. But we would love to see that extended across all the networking domains.

I hope you're paying attention to the keynote that I'm delivering here because that's exactly what we are delivering, my friend.

Excellent. I can't wait.

Listen, thank you so much. You are on an important mission, and we're proud to support you on it.

Thank you, Rami. Appreciate it. Thank you.

As Tom showed, the network sits at the center of everything we do and routers are the backbone. In fact, our routing solutions power some of the largest cloud providers and service providers and enterprises across the globe. If you access the cloud today, guess what, you use an APE Juniper router. It's that simple. What enables our routing solutions to deliver exceptional scale, flexibility and resiliency is the way we engineer them from the ground up, purpose-built silicon, purpose-built systems and purpose-built software, all designed together as a single architecture. That end-to-end approach allows us to optimize performance across every single layer.

And you see that engineering philosophy across the entire routing portfolio, our ACX routers built for enterprise and metro access and aggregation, our PTX routers with industry-leading density and power efficiency and our MX routers, my personal favorite, purpose-built for limitless flexibility across demanding edge environments and our AI native software and self-driving capabilities that automate the entire network life cycle. This is far more than just a routing portfolio. It is the infrastructure foundation of the modern digital world and the engine powering the AI era. So to show you more, I am excited to welcome the product leader for our routing infrastructure solutions, Katrina.

It's great to be here at HPE Discover. And I'm so excited to show how we ensure our customers' networks deliver the experience that we promise at scale without routing and complexity because you don't need to be a routing expert to manage your routers. With Mist and Marvis AI engine, we make routing operations simpler, easier and more intuitive. Many of you guys have the horror stories when it comes to launching new applications. The escalations, the sleepless nights, the weekends in the situation room. There's got to be a better way, right? There is, and it starts with HPE AI native routing.

Let me give you an example. So it's Friday afternoon, and our network operations manager is sitting down at his desk, he's ready to go home. He works for a leading health care provider. And on Monday, they're going to do a big launch. They're going to put virtual AI agents deployed into every hospital, office and clinic in their whole network. It's going to manage everything from patient care to hospital triage. If the network falls short, patient care will suffer. And the staff is the one that picks up the slack. So how can we help our network operations guys with this? Let's go ahead and take a look.

Yes, I'm excited to see.

So the network is ready and everything is connected to miss. There's just one problem. Until Monday, there are no users. So let's see what we can do with Experience Twins. And everything looks green, which that's great, but what does all this mean? So effectively, Marvis turned the routers into digital twins, generating synthetic application traffic into the network just like a real user would, detecting degradations in real time without a single truck roll or single user being impacted. That's the power of Marvis. So if we revisit our engineer, he's sitting on his couch, he's trying to have a good day and lo and behold, notification. The experience has deteriorated. The latency has increased from 80 milliseconds to over 200.

Let's go to Marvis and see if we can fix this quickly before it ruins our weekend or worse, the launch. So red does not look good. But what does it mean? We've got a bunch of failed tests across two separate KPIs. So there isn't an obvious root cause. So what are we going to do about this? Let's ask Marvis. So why has the latency suddenly increased? And Marvis has a clear read on what happened. It analyzes the network in real time, and it gives us an answer in plain English. The latency has increased because the traffic has shifted to a less optimal path, and it gives us the steps to validate and fix the problem. It looks like some configuration changes have removed the preferred route, forcing the traffic onto a backup path. Marvis even has a recommendation to fix it. Well, it's a relief that the router kept the network up.

So let's see what we can do about these SLEs. If we check out the latency view, we can see these failed tests have increased gradually over time. And if we want even more details, we can see the granular spike in latency. Overall, this aligns perfectly with what Marvis told us. So let's go ahead and look at what's going on with these excessive hops right here.

So since Friday's baseline, we can see 2 additional network hops. This new route is definitely the issue. So we're going to go back to Marvis action and see what we can do to fix this problem. Let me call over here. And there we have it, the missing prefix. Marvis knows exactly how to fix the issue and how to restore the prefix. So if we click here and look. I approve this recommended configuration change is a good idea. But you guys know what they say, trust but verify. So let's check the actions on those experienced twins one more time, just to be sure and everything is back to normal. The latency is restored, the networks is recovered, the application is performing as intended. And if I'm ready, I can even take one step closer to self-driving by letting Marvis do this automatically next time. And with that, we're back to enjoying our Saturday and watching the game with no one but us the wiser about the disaster that never happened.

And here I was, I thought that routing was hard.

Not anymore.

Not anymore. I see the power of AI native routing. It senses what's happening in real time. It reasons through the impact and recommends or even takes action before users feel the pain. This is a great example of how our Marvis AI engine is delivering impact across every domain. Katrina, thank you so much. Great to have you up here.

Thanks, Rami. Thanks, Vegas.

Okay. I want to now shift our focus to the data center because this is where the demands on AI are becoming very, very real. And in many ways, there is no greater test of a modern data center network than in media and entertainment. Few industries push infrastructure harder, massive amounts of content moving continuously across globally distributed product environments or production environments, real-time collaboration across teams and continents, ultra-high-resolution video workflows and production time lines where downtime is simply not an option. In that world, performance and resilience are everything.

So to talk more about that, please welcome Director of Global Networking at the Walt Disney Company, Ben Croy.

Nice to have you. Please have a seat. Can you introduce yourself maybe and tell us a little bit about your operating environment?

Sure. Thanks, Rami. It's great to be here. I'm Ben Croy. I lead Global Networking for the Walt Disney Company. My team owns network architecture, engineering and operations with a strong focus on media and production infrastructure. One of our largest production networks support studios like Marvel, Pixar and Lucas Film. At any one time, I have over 200 concurrent productions globally, and a major film could easily generate a petabyte of content. This data has to move quickly, securely between partners, creatives and departments globally.

So maybe give us an example of like a recent movie.

Okay. Sure for an animated feature like Zootopia 2, we may create the content in a single Burbank location and then regionalize it around the world to prepare day-in-date release in 35 or more languages. So the scale and the global movement of content is constant. And that's what our studio production network has to support.

It's interesting. How has the technology behind filmmaking evolved over the last, let's say, decade? And what does that meant for data centers that support your production environment?

For one thing, film is made digitally end-to-end now, driving a lot of change. A few things stand out. It's gone from terabytes to petabytes. Visual effect workloads have grown significantly with over 80% of the movies now relying on some type of EFX. Virtual production environments emerged as well on films like Mandalorian & Grogu. This is much more sophisticated than the green screen VFX processes of the past. And there's no physical fallback anymore. It's all data end-to-end.

So when you're supporting some of the largest and most complex media production like in the world, what do you need for your network infrastructure to ensure that productions are delivered always successfully?

Well, for us, it comes down to delivery and predictability for our large feature production we release globally at around the same time in every country. If we miss that window, there's real financial impact. So the network must be reliable, scalable and able to move very large data sets without becoming a bottleneck. We leverage HPE Mist platform for our campus and Apstra for our data center fabric. We found that these solutions give us more consistency across environments and a platform that allows us to scale.

Okay. So Mist and Apstra. What role do you think the network should play in the creative process? I mean, how visible should it be to the filmmakers, to the artists, to the production teams who depend on it every day?

Well, it's foundational. The network is a critical part of production, but it's ideally invisible. Our goal is simple. We want our filmmakers to focus on story, character, amazing visuals and sound, not to worry if the network will do what it needs to do.

Okay. So as you look to the future of content creation, the technology just keeps getting more and more powerful, but also keeps getting more complex, unfortunately. What's the challenge that you're really trying to solve for, let's say, over the next few years?

Two things, speed and simplicity.

That's it. Speed and simplicity.

That's it.

Okay, Ben. I'm proud to make the network invisible for you. It will be my mission. All right. Thank you, my friend. I appreciate you joining us.

Thank you very much.

Thank you. The pace of innovation in the data center space is absolutely extraordinary. Architectures are evolving rapidly. Workloads are advancing rapidly. Customers' expectations are changing rapidly. But guess what, fortunately, so are we. Our innovation engine is incredibly strong here, and we are moving aggressively to help customers build the AI data centers of the future. And yes, we are bringing self-driving operations to the data center as well.

So to dive into this, in more detail, I would like to bring up product lead for data center, Kyle.

Kyle, how are you?

Okay.

Great to have you with us. Tell us what's happening in the data center space at HPE?

Yes. Thanks, Rami, and you said it. What's really exciting is that we're bringing the power of AI native self-driving operations into the data center. Our solutions continuously collect rich real-time telemetry across the network from our switches, routers and firewalls. That telemetry flows directly into the Mist platform where Marvis AI turns it into deep visibility, operational insights and automated actions. And what makes us truly unique is that we bring design, deployment, assurance and operations together into a single life-cycle experience. The result is continuous visibility, automated operations and self-driving optimizations across the entire data center network. You want to see it in action?

Let's do it.

Excellent. With HPE networking, you get the proven self-driving capabilities you already know and trust applied to the data center, too. For example, the Marvis dashboard that Sunalini showed for wired and wireless actions also services key anomalies for the data center network. And it clearly explains how to fix them. In addition, service level expectations or SLEs, which are a key part of the Mist platform for optimizing user experiences are combined with our Knowledge Graph to continuously measure network health. We take this a step further with application awareness. We map application flows directly onto the data center network, allowing operators to instantly see the real impact of a switch or link failure. This moves application troubleshooting from reactive guesswork to intent-based decision-making.

And to minimize downtime altogether, Marvis AI-powered intelligence enables proactive maintenance with Marvis Minis running network tests that continuously validate network intent and services. Further, predictive analytics monitor system and optics behavior, tracking 30-plus metrics such as voltage, current, temperature, CRC errors and more. Machine learning models analyze these metrics to predict optic failures before they occur. And now we go even further with Agentic AI.

The Marvis AI system uses reasoning models and intelligent agents to solve problems the way a seasoned network engineer would. It correlates data from multiple sources, contextual data, switch telemetry, application flows, historical tech support cases and then it performs expert level reasoning, leveraging skills to rapidly identify the root cause and recommend next steps. But once took hours or days can now happen in minutes. And by demonstrating the logic process and reasoning used, the agents give operators insight into the solutions being recommended, building trust in the network and allowing it to act autonomously.

So I think the pattern is pretty clear by now. We are extending self-driving operations across everything, camps and branch, security, routing and data center networks as well. But of course, HPE does much more than networking, right? Inside the data center, we provide the networking, the compute, the storage and the virtualization and orchestration layer that brings it all together. So the obvious next question becomes, why just stop at the self-driving network? Why not extend self-driving operations through the entire data center?

Agreed. And we have been working hard to deliver a truly integrated data center infrastructure solution across HPE networking, compute, storage and hybrid cloud to enable faster deployments and streamlined operations. We integrated our management capabilities with OpsRamp, our hybrid cloud observability platform and compute ops management for our platform for managing and automating server infrastructure. These integrations deliver comprehensive observability, predictive assurance and proactive issue resolution across server, storage and now the networking domain.

Plus, we have now integrated with Morpheus, our management platform for virtualization in containers. We did this because many times, when a virtual machine gets provisioned, the virtual network could take hours or days to provision in the physical fabric. Meanwhile, the app team and the business are waiting by integrating data center operations with Morpheus, the network and server teams no longer work in silos.

These integrations really make a difference when it comes to streamlining data center operations. And this is the value that HPE like uniquely is able to deliver. So Kyle, can we see what that looks like in real life?

Yes. Let me show you. Here we are in Morpheus, a single pane of glass for managing your entire virtual infrastructure. VMs, cloud, clusters, workloads all here. First, we connect Morpheus to our managed data center fabric. That's the bridge between the virtual world and the physical network. Now we create 2 virtual networks, vnet 10 and vnet 11. We provision 2 VMs, a VM on vnet 10 and another VM on vnet 11. Both are live hosting critical workloads. Now Watch what happens. Without anyone touching the network, vnet 10 and vnet 11 were automatically created and pushed to the physical fabric. No ticket, no waiting, no manual configuration, no human error. And the proof, both VMs communicating perfectly across the HP managed fabric.

But here's where it gets interesting. What happens when a VM moves? Let's start a continuous ping between the 2 VMs. Now we migrate one of the VMs from server 9 to server 15. Watch the packet loss counter, 0, not a single dropped packet. The network and security policies follow the VM automatically and visibly instantly. We've automated networking and removed silos. This means that you get faster deployment in zero touch networking without manual errors. And when you move workloads for resiliency or server utilization reasons, the network and security policies follow automatically.

What you just showed us is obviously incredibly powerful. You just demonstrated is that infrastructure operating as a coordinated system, Kyle, where networking virtualization and cloud operations are part of a single intelligent workflow. No tickets, no manual networking changes, no operational lag between teams, just seamless automation from the VM all the way down to the physical infrastructure. And I love how when we integrate with the rest of HPE, customers get all the benefits of GreenLake, flexibility, cost savings and accelerated time to value, thanks to its industry-leading hybrid cloud platform. But Kyle, I know this integration goes beyond just that, right? I mean what more do you have for us?

That's right, Rami. Right here at Discover, we announced that HPE is expanding its AI data center solution to include our QFX switches managed by Apstra Data Center Director. This creates a full stack free integrated solution spanning compute, networking, storage, software and services, which accelerates AI data center deployment with assured interoperability, scale and performance.

So that's great news, obviously. But we all know that the automation and intelligence that comes from self-driving operations is kind of useless without a solid hardware foundation, which is why innovation in data center hardware has never been more important than let's face it, hardware is kind of hot again, right? So there has been a lot of new innovations recently to our AI data center product, tell me about them.

Absolutely. We continue to innovate in our data center. In addition to recent enhancements to the MX and PTX routing lines, we keep adding new QFX platforms for scale-out and scale-up networking. We recently introduced the industry's first Ethernet-based scale-up solution, the QFX5252 purpose-built for AMD Helios systems. Thank you. And you can see this product in the demo area right over there. Trust me, you can't miss it.

It's massive, like the size of fridge. In addition, we launched the QFX 5250, which was the first 100% liquid cooled switch using the Tomahawk 6 chipset. Just as we were the first OEM vendor to ship 800 gig, we did it again with 1.6T connectivity. And I'm excited to say this product is shipping now.

So let me just get this straight. If anybody wants to build a data center with 1.6 terabit connectivity, they really only have one option, right? It's us.

That's pretty damn cool. Congratulations to the team. Yes. All of this innovation across both data center software and hardware is definitely getting the market's attention.

Our solutions are positioned as a leader in the Gartner Data Center Networking Magic Quadrant and ranked #1 for enterprise build-out and #2 for AI Ethernet fabric in the Gartner Critical Capabilities Report, right? This recognition is based on decades of innovation.

So when organizations think about building data center networks for the AI era, I believe they can have tremendous confidence in what HPE networking is delivering.

Kyle, thanks so much for joining me here.

Appreciate it. Thank you, Rami.

Today, you heard a consistent message from every customer, every demo and every innovation we shared. The old way of operating networks has reached its limits. The scale is too large. The complexity is too high. The pace of change is too fast, and AI is accelerating all of it.

That is why the future belongs to networks that can think, that can adapt, that can optimize and that can protect themselves in real time, self-driving networks, not a futuristic idea, not a lab experiment, but as a practical necessity for operating modern infrastructure at scale. Now look, I know I'm not the first tech executive you've heard from talking about how their AI is better than everybody else's AI. And I'm certainly not going to be your last.

But I have deep conviction about what sets HPE networking apart from companies that are mostly showing you slideware. And it comes down to one key thing, efficacy.

Our self-driving network just works. It works at scale. It works under pressure and it works across our whole portfolio from the wired and wireless edge all the way to the data center, delivering real outcomes to customer environments every single day.

But honestly, the only way to truly believe it is to experience it yourself. So I would urge you to try it. And by the way, with HPE Financial Services, we can make it really easy for you to do just that, including a new network migration program to clear out old non-self-driving tech and reinvest in what is new. Because once you see the infrastructure that can continuously monitor itself, optimize itself, protect itself and get smarter every single day, you simply cannot unsee it.

At the beginning of this presentation, we talked about foundation. About what happens when the foundation underneath something is not built for the demand placed on top of it. AI is creating one of the biggest technology shifts any of us will experience in our careers.

Every company is being forced to rethink how they operate, innovate, secure their business and compete. And in moments like this, there are really only 2 choices. You can be disrupted by those who embrace the change faster or you can become the disruptor. If you want to be on the right side of this change, if you want to move faster than your competitors, if you want AI to become an advantage instead of a source of complexity and risk, -- you have to start with the right foundation, a foundation built to adapt and scale with whatever comes next. That foundation is IT. That foundation is the network. That foundation is the self-driving network. Thank you all so much.

Good afternoon, and welcome to the HPE Discover Investor Relations Summit. [Operator Instructions]

Please note, this event is being recorded. And just a moment, I will turn the conference over to Shannon Cross, Chief Strategy Officer.

If uncertainties materialize or if the estimates or assumptions prove incorrect, our results may differ perhaps materially from those expressed or implied by such forward-looking statements. HPE assumes no obligation to update such statements. Please find more information regarding forward-looking statements on our website at investors.hpe.com. So with that, let me welcome Antonio Neri, HPE President and CEO.

Good afternoon.

So Antonio, we all enjoyed your keynote today and Rami's talk during the networking general session. We've also had an opportunity to tour the show floor. It's clear HPE is leading the adoption of Agentic AI in the enterprise.

We're leveraging our innovative networking, cloud and AI portfolio, helping customers move from AI experimentation to fully autonomous operations at scale.

During your keynote, you shared that we are deepening our work with NVIDIA with the next phase of AI factories and HPE Private Cloud AI, and we are excited about the opportunity to work with AMD on Helios. Rami also talked about HPE extending our leadership in self-driving networks as critical foundation for Agentic AI from campus and branch through data center.

I'm excited by what we've announced so far at Discover, and I hope everyone listening on the webcast as well as everyone here in the room can tune into Fidelma Russo's general session tomorrow, where she will discuss how HPE's innovation will continue to support customers on their AI journey. So I look forward to this Q&A session with the investors and analysts. But first -- and we'll take questions from the audience shortly.

But first, let me kick off with a couple of my own. So first, networking has become such an exciting part of our portfolio and our story since we combined with Juniper. How do today's announcement show the extension of our networking leadership in the AI era?

Well, first of all, good afternoon, and those who are tuning in on the webcast, thank you for joining us today. For those of you here in person, I hope you enjoy so far the event and the day. And I understand you did the tour on the show floor, which it takes 2 days to see everything.

But I hope you got a glimpse of the amazing portfolio that we have curated and built. And at this event, you can see the continuous innovation we continue to bring to the market to address the needs of both cloud and AI. But obviously, the biggest topic is AI. And in that context, we think about AI as a productivity tool that will change forever how we live and how we work.

But fundamentally, it's to power it. And at the core of that powering is the foundation, which sits on the network. We talked about the need to improve the cost per token or the first time to token. And fundamentally, every aspect of that stack needs to be productive.

And today, the network is a bottleneck. No question about it because we saw the tremendous advancements with the compute and accelerated computing.

But what we have done with networking and the portfolio that we built with the acquisition of Juniper is address the demands of AI and cloud from the edge of the network, which obviously is the on-ramp for many things, including going forward, the inferencing component of this all the way to the training side. And I thought what I covered this morning and what Rami covered, whatever just an hour ago, is a testimonial about how well this integration has done for us.

It is not just integrating 2 great companies and 2 great assets, which were very complementary with each other, but really scaling that integration with innovation perfectly tied to the inflection point, whether it's in AI, scale up, scale out and scale across. We have an amazing portfolio. And that's why we see the results we saw in the previous quarter in terms of orders, in terms of backlog, obviously, because of the supply availability and in terms of durability because the networks for AI demand is untouchable.

And for us, I think we are perfectly timed for that moment. And look, I believe in the Agentic enterprise is also using Agentic AI to make these solutions more autonomous and intelligent. And this concept of self-driving network is something that we started we were running a while back. But now it's live. It's available. I mean if you see some of the demos we put on the floor even on the routing side, it's a little bit scary to see how far you can self-optimize traffic across data center interconnect. And to give a sense, that 12,000 router, you can put together the entire population in New York and London together. And concurrently, 60 million people can broadcast a movie or watch a movie on that single rack.

So the performance and the ability to do it in an autonomous way it is just remarkable. That's just an example. But that's why we say we extend it to everything. And we already had it anyway in the campus and branch, and now we brought into Aruba side as we brought the switching to the Juniper side.

Yes. No, I think it's fantastic what we've done. It's a testament to the hard work, I think the team did in terms of the integration and the IP that sat in Juniper and in Aruba and now is being supported by the total [indiscernible].

When you think about it for a moment, right? So we closed the transaction on July 2. And by January 2, just exactly 5 months, we brought in 10,000 Juniper employees inside the company. We announced our strategy for networking. We announced the road map across the 4 key networking segments, campus and branch, data center switch and security and routing. And we integrated the sales force into one unified sales organization. And we announced all these products along the way. And of course, we are doing really well from a synergies point of view.

What comes next now is that vision to build the best networking business on the planet, and that includes also the back end of how we do business. But fundamentally, also the next chapter is also the synergy with the rest of the portfolio, particularly with the cloud portfolio, which we are integrating products, whether it's software in the virtualization stack or whether it's in the private cloud stack or whether it's with storage, which are sources of revenue and profit as we think about '27, '28, '29.

Well, that's a perfect segue to my next question, and then we'll open it up to the floor. So we've come out of a great quarter. How are you thinking about the durability of these results? And what gave you the confidence to provide the fiscal '27 framework that have called for double-digit growth for basically both revenue and EPS at the midpoint?

Well, I think at the core is structurally, the portfolio of our company has changed forever with the addition of Juniper, right? So the mix has changed dramatically.

I argue we're still undervalued in many ways from a peer multiple perspective against not just today and '27 guidance, but against the long-term potential, especially because the networking demand is very, very high.

So my view is that first is the mix of the business. Second is the demand that we see in the market, right? I mean we grew in security mid-teens all the way to the routing and 30%. And in between, you have upper 20% in the rest of the portfolio. And then we have an enormous backlog, obviously, that we need to clear.

But what gives Marie and I confidence to give 6 quarters of guidance because that's exactly what we did, second half and 2027. The pipeline is multiples of our backlog. Number two, our portfolio is being seek by customers and networking clearly is the driving force that's doing that.

Then obviously, the supply constraints in many ways, are driving demand because everybody wants to get in the line, in the queue to make sure they don't have to wait too long to get the supply. And let's be clear, that supply problem is not going to be solved anytime soon.

So it's like I make the analogy one time happened to me, I made a mistake. I went to the DMV, I took the ticket decision. I have to wait whatever, 40 people ahead of me. I left and then I come back the next day, shoot. Now I have to wait 60 people, right? So that's what we -- and then look, all the programmatic things we have done with Juniper synergies and catalyst, particularly catalyst in the way we work inside the company and the ability to improve our gross margin profile and operating margin profile to the OpEx.

It was a no-brainer for Marie and I to go out and give all that, which for us was the concept of durability because that's the key here, right? It's not one time.

And as I said in the earnings call, Q2 was not just a onetime event. It was a combination of many things we have done in many, many quarters. And in the AI scale, we have been very disciplined about what capital to deploy for what return.

That's it.

Great. Well, thank you, Antonio. So with that, let's open it up to questions from the audience that we have. We will have mic runners, so please wait until you have a microphone before you begin. And we are webcasting this session, so please state your name and company asking your question. And finally, can you just stick to one question. We will come back around as time permits.

We are under the earnings rules here.

Yes. And Wamsi, you have the mic.

Wamsi Mohan, Bank of America.

Nice to see all the integration progress you made. So we've heard a lot of exciting things about the networking portfolio.

When you look at your Q2 results, 10% growth. There were differences within the subsegments in there. You guided next year, 8% to 12%. And I'm thinking like why is that not -- why should that not be viewed as a very conservative guide just given the fact that we heard so many things here in the pipeline that are coming through in the next several quarters?

Yes, because you guys always look at revenue, we look at orders and the ability to convert to revenue. And that's the thing. Look, in many ways, it was a prudent guide from a pure revenue perspective, which is what drives profit and eventually free cash flow.

As I said, the supply availability will continue to be severely constrained into 2027. And to give a perspective, we already have the capacity allocated for 2026.

And what we do, the way it works every 90 days, we tell our suppliers how we want that capacity to be between, I don't know, server storage between this 64 gig to this 128 to 256 this speed, that speed. And so basically -- and I met one supplier here, which is a great partner, and I say, it's great what we're doing together, but I need more supply. And then he goes on and said, yes, we'll see what we can do, right? But it's really that the issue, Wamsi. And so we expect to exit 2026 with a higher backlog in many ways than we are today.

And we felt it was prudent to give that double digits based on what we believe we're going to be allocated in 2027. And that process is a process that's still going today because we have no firm final numbers because we are -- we have negotiated now the LTAs and those LTAs are not just 1 year.

Now they are multiyear commitments. But we need them to come back with the final, final, final. And so that's the reason why. But we expect in many segments of our business to continue to grow faster than that number on an orders perspective. And we're going to go back, but he [indiscernible].

Tim Long at Barclays. Sorry about your ticket, but given what the stock has done, you didn't have to go, you could have just paid the full price, I think.

The ticket for what.

The DMV. So I wanted to get back on networking, if I could, a 2-parter. You showed a lot of really good technology today. And in that AI piece, you raised the numbers a little bit last quarter. Two parts.

One, can you talk a little bit about leverage of the strength you guys have in server and storage? And have you started to see that at all impacting that line? And then second, some of these newer layers like scale up, you have the nice AMD deal and much more importance on scale across. Can you talk about how that might impact positively that AI data center line?

Yes, Tim. Fair to say that one of the areas that are growing the fastest is actually the scale across for sure. I think both the 10,000, 12,000 and the MX product are becoming key references for DCI and the on-ramp to the edge. Scale out, which is the QFX products continue to grow, and we have a number of marquee customers, which are adopting that.

That -- think about it in the case of NVIDIA, right, and the Ultimate II, that's the NVIDIA Grid announcements we made at GTC -- and there is a number of hyperscalers and, I'll call it, large service providers in new cloud that actually have adopted the QFX on top of the Spectrum-X to do the scale-out, right? And there is a number of reasons for that. They love the management control plane.

They love the AI operations that we built into that. They love the performance of the actual switch. And as you saw, we announced the 1.6 terabits first industry time to market with Broadcom on the Tomahawk 6. And even there, we offer the 2 distributions between Junos OS and SONiC-based OS. And both have AI embedded into it. Now with the Junos OS, you get more telemetry by definition. But the reality is that depending on the type of customer may pick one versus the other on how their environment work. In the case of Scale Up, in the NVIDIA case is NVIDIA. I mean that's a given because of the Spectrum-X, ConnectX and BlueField.

But in the case of Helios, which I hope you saw there is, I call it, a double fridge or, right, because it's an OCP design. That's going to be our Junos -- sorry, our SONiC OS with our 5252 QFX switch, which we are excited because as customers adopt alternatives to NVIDIA for certain training -- large training, this is a large environment, right, for training.

Then obviously, that's our networking in it. And then everything else will scale out and scale up will continue to be the same with Juniper. So -- and that obviously gets deeply connected with our compute, right? Because inside that Helios or NVL, there is an HPE server that comes with it, right? And so in the case of the AMD, there is a tightly integrated work between AMD, our server team and our networking team. It's not just GPUs from AMD with the network fabric. It's actually all 3 together.

Obviously, we have to see the fall when that comes out, right? And I expect a number of large marquee hyperscalers and/or service providers to adopt that. And so that will be clearly a tailwind for us as we go forward.

Yes, behind you and then Neil.

Asiya from Citi. There's investor perception that in this cycle where we have so many component constraints, HPE has done a really great job and specifically as it relates to networking chips, access to having silicon.

Maybe you can just help us understand like why do you think HPE is better positioned in this cycle to get access to components? And where could there be some upside to driving better component access as you go into fiscal '27?

Well, thank you for the question. I think it's important to remind ourselves our portfolio and what IP we own across the portfolio.

So first of all, on the routing side, we have 2 dedicated silicon road maps that we run. One is for the router MX, that's the TRIO silicon, our design, our silicon. And the other one is the PTX, which is our Express 5 silicon.

Basically, we do not use merchant silicon for any of our routers. So that's number one.

Number two, when you go to the campus and branch equation, our entire Aruba CX portfolio, maybe just a couple of products on the fringes, is our silicon. We designed that silicon for many, many generations now. And it was part of the original portfolio we won, which I reverse integrated into Aruba in 2015, which is our ProCurve business. And now that silicon is across the entire campus access layer and aggregation layer for the campus and branch. And what we announced today is that those switches now are also managed and available to the Juniper Mist, okay?

So in those 2 aspects of the portfolio, we have our silicon. We don't buy any one else silicon. When it comes down to security, you will see very quickly that, that campus switching silicon, I just talked about it, is converging with security. So unlike others that converge security at the software layer, we are converging security at the silicon layer. So the next generation of CX switches will be a converged silicon between networking and security.

That's a unique value proposition that is going to give us a huge advantage because that silicon is truly programmable. So all the algorithms are built in the silicon, so we can program that from our cloud control plane, whether it's Mist or whether it's Aruba Central.

And then in data center switches, we use, of course, Broadcom and Broadcom, we are now the largest OEM partner for Broadcom. So that's another reason why we have an advantage when it comes down to that. But look, there is constraints there in networking, too, and the networking constraints are mostly aligned to the same constraints you see in the market, which is memory.

Even though the memory footprint in a networking switch is much smaller, it's actually the older technologies, the DDR4, not even DDR5, which people have, of course, deemphasized. And so we are moving to the latest design in some of these switches. In fact, some of them, we may skip all the way to HBM. .

Louis Miscioscia, Daiwa Capital Markets. So looking beyond networking, I guess, for a moment, you had some very good questions on that. Can you differentiate the demand and the highest supply chain constraints for, let's say, GPU servers, AI GPU servers, CPUs, Vera, normal x86. Maybe leading the answer here a little bit.

A few years ago, I thought that maybe CPUs would see 10x the capacity needs of GPUs once you get to inference. It seems like you're starting to see that, but maybe you could help us with the differentiation.

Yes. On the GPU side, I would say there is not severe constraints, but the model works slightly different. Unless you're willing to invest way upfront and take a bet, you normally place orders POs based on orders.

You don't place and build a huge inventory. We learned that lesson early on in the cycle, right? Especially with the life cycle of these GPUs moving so quickly, it's -- maybe at the beginning, it was great to have some. And look, if I knew what I knew today, I will have more power -- both power and GPUs. But that was not the case.

So it's less about the constraints. It's more the lead times. And so if you build -- you're going to need to build a large AI training system, that's based on lead time based on where you place the POs. And obviously, generally tends to be a 3-way conversation between the customer who has sometimes significant relevance with NVIDIA and then just us, right?

So we work together on that prioritization. However, there are other constraints around that, power loops, cooling loops. This is you think chassis will be a problem. That's a problem. In networking transceivers, that's a challenge. So there is a number of things, peripheral things that goes around that. On the CPU side, there has been constraints.

I think we have done a very good job navigating that. But look, you can have the CPU, but if you don't have the memory, it's a waste of time. So like I have the car, I have no wheels, okay? That's not very helpful, right? So you have to move as a system. But look, we have done a very good job in partnership with both AMD and Intel. Because of a long-term relationship. Their CPU is just early, right? We just introduced the product. And I think that product is going to do great in the inferencing space.

But when I think about between now and end of the decade, the vast majority of the demand will be in the inferencing, not in the training side. That's our view. And the question is, number one, where the inferencing will be done, what type of architecture you're going to deploy for that inferencing and it's not going to be one kind of unique architecture. Use cases and verticals will variate. And then ultimately will be more centralized or decentralized. These are all things we're going to learn as we go forward. But the ratios of CPUs to GPUs, I think it's going to variate based on the type of inferencing, whether it's 4 or 8, I don't know.

But look, having a server business that has that scale is important in that context. What I'm working with the team is that, okay, how the architecture really collapses because you don't want to build more layers and overhead in the architecture. I do believe there will be new emerging architectures between KV cache cores and networking fabrics coming together in a more efficient way because you have to solve for scale, cost, right? -- and eventually energy. So a lot of things will happen.

So that's why in my head, I always think about it, I have a server, I have a storage, I have a networking. No, I have network fabrics. I have cores, I have software, I have memory because storage is an extension of memory. And then how you bring it all together in a way that creates some differentiation. Yes, he was asking for a while.

Erik Woodring, Morgan Stanley. Antonio, can you maybe help us just to build on that question by contextualizing how your enterprise customers' compute needs are really changing in this environment with Agentic AI, with inferencing, more getting done on-premise. And the question maybe is, is this customers materially growing their server installed base?

Is this just refreshes to modern architecture, the sustainability behind that? Would just love to understand because we do seem like we're at an inflection point, but just the context behind that, that you're seeing would be really helpful.

Yes, sure. Look, first of all, I spend more than 50% of my time with customers because I always say the truth is in the coalface, when you talk to customers, right? They will tell you exactly what's going on. Fair to say maybe a year ago, things were a little bit slower. People are sitting on the fence is understanding how this is going to evolve.

But it's fair to say, at least in the last 6 months, there is a significant acceleration. But we're still early. That's the interesting part, right? So if you look at the ratio of compute for training versus inferencing, it's still maybe 70-30 right now. And at some point, that has to go the other way around. And then again, where it's done and what level of scale you need? Look, I'll give you an example of us as a company. We, as a company, are aggressively using AI everywhere we can.

We have 1,200-plus use cases in the life cycle. 250 are in production, meaning already deployed in production. Marie is not here, but Marie and finance have been super aggressive deploying AI everywhere she can. This is part of the modernization with Catalyst. And interesting, we are striking some unique partnerships where they use our AI factory for enterprise, take a private cloud AI to develop the agentic models that eventually get packaged and we can take it together to market.

That's an example of Deloitte with Zuora AI suite for CFOs. But when I talk to customers, and I was a couple of weeks ago in Chicago and in Europe, they are aggressively moving forward. What is interesting, people started with large language models, we call it frontier models.

And that's fine because it's very isolated, very contained and you can see the benefit of it. But I think the way you become an agentic enterprise is by actually bridging together, stitching together agents in a workflow by first digitizing, automating and then deploying AI on top of it.

So there is a little bit of process engineering going on, understanding value streams, then doing the hard work. But one of the barriers has been governance, data, preparing the data and all the regulatory goes around. That's why this morning on stage, I talked about everything we built in that AI factory. Unlike some of the competitors we have, they just resell just the hardware. We went on and build an entire software ecosystem inside our GreenLake Cloud that ultimately, the infrastructure that sits underneath is tightly couple for that, whether it's RAGing, whether it's small language model training or giving context to multi-modality and so forth.

Now when you say, okay, how many GPUs are you going to use? I can tell you, inside our company, 300 don't need more than that, right? And so what I always said is that if you have -- if you picture on a whiteboard 2 axis, you have the axis of training and the axis of inferencing, right, the question you ask.

Clearly, training is all GPUs for the most part. And the inferencing will be a mix of things. But then if you take a different view of that, which is service providers, model builders, hyperscalers, these are a very low number of customers, right, although it has been growing because everybody wants to participate in the build-out through some sort of financial engineering that's going on.

But let's say, 50 that are big enough to make a difference. Those 50 will consume millions of GPUs. So it's like here to here. And then you have the opposite, which is hundreds of thousands of customers who are going to consume the opposite, very low amount of GPUs. Now this transaction value is significantly lower than that transaction value.

So that's why you have to find the right balance. But ultimately, I argue it comes down to, honestly, working capital and our return on the working capital and obviously, the margins you can generate. And more software, more services embedded into it, the better it is because if you believe end of the year -- end of the decade, that's what it's going to be, you want to be ahead of that and now build an enormous revenue with huge amount of capital that eventually that's going to be a tough compare. So -- and in either case, we need when networking. It doesn't matter.

Kath Murphy from Goldman Sachs. So maybe extend on the conversation around the AI server opportunity, talking about sovereign in particular, realizing that sovereign use cases are not monolithic.

But can you share anything to think about the size of the opportunity and the maturity of the market and then where that falls on the training versus inference spectrum? And what in HPE's portfolio and your existing relationships is unique in kind of addressing that opportunity?

Yes. Sovereign is a unique customer segment because it's a combination of many things. You have what I call the traditional labs and government entities that they are making investments in order to deliver an AI cloud under the principal sovereignty. A lot driven by the geopolitical environment we live in, but also as acting some sort of service provider for the communities that obviously, the country or the regions are. But look, their ability to raise capital, it depends on the -- I mean, here in the United States is very easy. I mean, like how much you need and how much you're willing to pay. And you go to Europe, it's an ongoing fight, I will say.

They have a lot of ideas, but the ability to raise money is complicated. In fact, I was with one and he said, we are very excited. And I said, great. We're going to build a gigawatt factory fantastic. So how much money you have? -- simple, we have $4 billion. I can't get out of that for $4 billion, right?

This is literally -- the understanding of scale is so off, right? But then it goes through that process where basically they're trying to attract the private sector to bring the capital. And through regulation, through geopolitical inside their own is long. It's a long sales cycle. The bottom line is a very long sales cycle. But look, we built some AI clouds already. If you think about the U.K., the AI Bristol Cloud is a sovereign cloud. If you think about the European Union, the Lumi system is an AI system in Norway and is in service the European Union. So these are examples, and we are working with a number of them.

But I would say it's maybe 12 to 15 in total, right? And then what's going to happen, the sovereign AI clouds are going to be these new clouds that have the capital to become the driving force. So they're going to act as a sovereign AI cloud, although they are new clouds, but just happened, I'm going to build it in this country. I'm going to open with the same regulation so that you feel that we can serve you.

Obviously, Middle East was going strong until the conflict started and that put a significant slowdown to the process, although UAE is still going, I would say. So this is the challenge, right, that we see. But it's not just AI because all of them also need supercomputing. And supercomputing is very important because it's an incredible adjacency to AI. And if you take in the United States, all the national laboratories are supercomputing entities.

Oak Ridge, Argonne, Los Alamos, Livermore Lab, all are HPE supercomputers. Now we built those 3 years ago in some cases, they're all exascale systems. And now they are adding to it an AI system. And in the case of the Oak Ridge National Laboratory, that it was Frontier, the first exascale. Now we're going to what you saw a little bit of a [indiscernible] through of 2 cabinets, Mission mission.

And then next to that, you have Lux, which is the AI pure system. So that for us is an opportunity to grow in sovereign because we have the expertise and the trust, which is super important when it comes down to sovereignty.

And the other part is the networking.

Yes. And the networking, obviously, look, we're trying to be simplistic in the way we tell stories so that it become super, super technical. That's why I said, Rami, go do that later. But it's like when you build your house, you're not going to start putting trimming around the doors and all that until the plumbing the electricity.

To me, the network is exactly that. Without the plumbing the electricity and the plumbing being the network in this case, you're not going to finish the data center. And then obviously, you have the cooling that goes with it. But networking is going to be that core foundation because without a robust foundation, you can't deliver the rest.

Victor Santiago with Evercore ISI. Antonio, could you help us better appreciate how customers are navigating this higher-priced hardware environment we're in based on the conversations you're having with customers?

Just given the recent strength in traditional servers you guys saw where prices are multiples of where they were from prior year, how are customers managing their IT budget allocations? Are they extending life of existing assets or reallocating spend from other areas?

I mean, look, of course, they're not trying to extend what they can. However, the need to modernize their infrastructure to be able to adopt AI is stronger than ever.

Look, you need to find -- if you're going to host this on-premise, you need to find the space and the power. The demand for power are pretty significant. We -- when you look at the Helios Rack, there is a chance depending on how you deploy, I think, you may need 700 kilowatts power.

That's whatever, 5 feet, Y, whatever. But the reality is that they all need to modernize and save space and save cooling. We can show customers that we can take 7 Generation 10 servers of any vendor for that matter and reduce it to 1. So that's a 7:1 reduction just on space. And then we can save up to 65% energy and then increase the performance by a factor of x in terms of core and memory density and so forth.

So that's why we see the momentum we saw in what we call traditional servers, which was -- the orders were up triple digits year-over-year. Of course, a lot was also driven by the cost. So cost is clearly cause of concern, but it's not the reason why they stop it. So they have not stopped. And then we have a portfolio with HPE Financial Services that actually helps them because with HPFS, we can come in, accelerate depreciation of those assets, remove legacy infrastructure and then free up the capital for them to reinvest where it makes sense. And in many cases, they actually pivot to OpEx versus CapEx. And so it's not just, I don't know, free up $3 million, let me buy $3 million of CapEx.

They actually shift to OpEx because they believe, particularly with AI, is probably a prudent way to start small and then start growing from there on. So that's why our portfolio is not just technology, it's also the financing capability, but it's all delivered also to one unified control plane, which is our GreenLake Cloud because everything I've done now for 5 years plus, doesn't matter how you pay, you will be using our GreenLake Cloud to manage it. And that includes a subscription model to the software associated with that infrastructure.

So there's a number of ways. But look, we have not seen a slowdown because of the cost. If anything, we have seen an acceleration. And we believe that's going to continue to be the case because even in '27, that cost curve will be more stable, but it's going to stay very, very elevated. So do I wait 18 months, 2 years, who knows, right?

This is Michael Tsvetanov with Wells Fargo. With your new networking product announcements, which you talked a lot about today and obviously, the integration of Juniper going quite well, it seems you're very well positioned to attach your networking alongside the rapid server growth you're obviously seeing. So I'm curious if you can just speak to the attach rates that you sort of see today between your compute and networking and maybe if you would expect that to trend moving forward? And then just generally, I assume as you customers upgrade their compute portfolio, you kind of need to bring networking and storage with it. But kind of what's the timing lag between those 2 things? So if you can just expand on that, that would be really helpful.

Yes. Well, look, we are very early in the process in terms of integrating the data center switching with the rest of the portfolio. On the AI front, that's happening at the customer side because, again, in the case of NVIDIA, you sell the NVL 72 and then we win the footprint above on our own architecture for scale-out and eventually scale across.

So there is no really attach of anything. You need to win the footprint inside the data center in the AI space. When it goes to the traditional enterprise or cloud, that's where the data center switching attached to the rest of the portfolio is very important.

But the way you do it is not just here is the switch. You have to integrate the entire life cycle. So what we are doing and we already kind of done is the integration of the life cycle provisioning management for networking, which we call it Apstra with Morpheus enterprise and then eventually with the full rack for private cloud.

At that point, you basically slide the switch into the top of rack and you're kind of done. That's -- the hardware part is the easy part. It's like no different than any other vendor top of rack. The hardest part is the software integration, and the team have done a great job. That's one part.

The second part of this is the integration of the software-defined layer of networking with Morpheus and VM Essentials, which is our virtualization and container environment and that's done. And so we did that in record time.

So now in our Morpheus enterprise software, not only we provide orchestration and brokering with the public cloud and on-premises, but also we provide the full software-defined layer from network to compute, obviously, virtualization all the way to the storage layer. And then we integrated there also the OpsRamp for multi-cloud and multi-vendor observability. That will drive an attach rate, of course, of networking with compute. And then on the storage side, as we now have the 1.6 terabit, there is a lot of workloads that are perfectly tuned to use Ethernet as the fabric inside the storage. And so that's going to be a Juniper QFX switch into our Alletra MP portfolio where it makes sense.

But once we integrate the Astra software with our switches and then with our control plane, then everything else is flow. So this is why we expect that the revenue synergy of this will start '27 and then continue. And the private cloud footprint is the perfect instantiation because a server storage network and an infrastructure with a software-defined architecture with GreenLake, all in one tightly coupled package, whether it's virtualization or AI because it's the same thing.

The only thing changes is really what type of GPUs and storage you use underneath. And in that case, it will be a server with GPUs -- and in other cases, will be the storage file and block for -- file an object for unstructured data. I think Wamsi has a question.

Wamsi Mohan, Bank of America. Antonio, in this world of agentic AI, where there is more attach of traditional server storage, why should HPE not consider maybe selling to Tier 2 CSPs in this world where the margin structure of the entire portfolio can be higher relative to maybe the opportunity of selling just AI-based servers?

I don't think -- look, the traditional servers in the context of hyperscalers and large service provider, that already moved in the cloud space a long time ago. Like if you recall, in 2017, I decided to stop selling that because the margin rates were inexistent. So they go -- they have -- in many cases, they have their own design to begin with.

In many cases, they have their own silicon, if you think about AWS with Graviton and so forth. So there are elements of the hyperscalers where we participate and will continue to participate is more the edge of the hyperscaler where they use products like us as the on-ramp into the large data center. right? And that's more the distributor element of the hyperscaler as a point of entry to the large cloud.

But once you are inside the large cloud, it's not like I'm selling ProLiant DL products or any other products. In fact, none of us, including our competitors sell into that. They go straight to the CMs or ODMs who build that for them because they have a unique design based on the footprint, how they lay that cloud with CPUs. And I think for them is another CPU recipe that now just sort of inferencing for those centralized. Look, there is an opportunity, of course, we will do it. But I don't think that's going to be the biggest opportunity in my mind. We're going back there and then we'll come back.

[indiscernible] With Loop Capital. Just to kind of double-click on some of the things you've already. talked about. In the last 6 months, you talked about the significant acceleration in demand. Could you elaborate a little bit more on how much of that you're seeing is due to inference on-prem and how you think HPE will kind of continue to outperform its peers if it is kind of in an on-prem world with inference?

Yes. Well, what we're seeing is that -- and it's a pattern that we see, right? Look, you have to see multiple data points over time is that when enterprise customers in particular verticals decide to go all in on the AI and they pick a mix of expert models. Once they attach that data to that model, they tend to want to have it under their control. And so yes, we are making things like private cloud with the security around punching an API out more secure.

But look, there are industries that even though you make that secure, they said, no, I'm going to have it on-prem, and I don't want anybody to touch it. So there are a number of verticals and use cases that will be definitely on-prem. And as they grow in deployment, the interest is going to grow. And that's why I said earlier, we believe that at least 80% of the demand by 2030 will be inference. And what we need to see is what's the mix between on-prem, off-prem, meaning large centralized environment.

But interesting enough, how many will be at the edge -- and we showcased, I think, on the floor, you have 4 products that they are designed specific for inferencing at the edge and large distributed enterprises. And you can see, look, warehouses of sorts, manufacturing floors of sorts, health care, right, where they need to process. I mean, look, there's a lot of regulation, they are HIPAA and everything else that putting around all the control is more expensive than just putting an inference server in their environment, be done with it.

I mean it's just math and physics. So we saw it. We believe we will continue to grow, but we need to see more about eventually how these architectures evolve and ultimately, what decisions to make.

He has a question, too, but if you can bring the mic and then we go back to Tim.

Just had a quick one for you, Antonio.

Erik Woodring, Morgan Stanley. Quantum. So we saw quantum computer out on the show floor. Some of your competitors that have talked about quantum benefit significantly when we think about valuation multiples. How relevant is Quantum to HPE?

Well, I will not comment on valuations of sorts because sometimes, look, you can go back to other valuations, right, of other company. I'm speaking from a practical point of view, and obviously, as an engineer, I used to be at least one.

I'm bringing a pragmatic view. Look, quantum is awesome for many things. But the scalability of quantum is not there. To put it in perspective, [indiscernible] there probably has, I don't know, 75 qubit is not on because you can't do it here. It needs to be in a vacuum location.

But if you really, really, really stretch it, we may be, as an industry, 1,000 qubits. And we have multiple technologies being tested for qubit creating the qubit. So that's a fight we don't participate. And that's why we are qubit agnostic. So now to do something very, very useful with quantum, you need at least 10,000 qubits.

So we are far, far away from 10,000 qubits. However, our approach was work, we can accelerate quantum by focusing on 3 things: Number one is the ecosystem. Number two is the network. Networking, again, comes to play, and I will explain why. And number three is the environment to develop quantum applications. And so on the ecosystem, you saw a number of vendors with us. Number two is the network. So what if you use the same principle of traditional computing, applying scale-out architectures to quantum.

So normally, if you want to get to 10,000 qubit, you have to scale up. Remember, HPE had and still has today Superdomes. Superdome is a scale-up system that today can scale to 64 terabytes of memory. you can put a lot of -- you can put an entire database there, right? And it was great at the time and still great for many applications for that.

But to get 10,000 qubits on that is going to take time. What about if we take 1,000 qubits, scale it out, that requires networking. And now with Juniper and other assets we have inside the company, including silicon photonics and other things, we can go enable that. And then number three, probably the most important one, which is how we develop quantum applications today, not when we have 10,000 qubits. And that's where we use traditional compute environments in this scale-out model to allow them to develop applications for those who want to build it -- and that's why the connectivity between the quantum system to traditional system like HPC or supercomputers are very important.

So you can start simulating this thing. The question is, will people develop scale-out applications in quantum or wait for a scale-up? That's the question, and that's a technology decision people would have to make. So that's my view, is the pragmatic reality. And the reality is, look, I'm old enough to have seen every inflection point in IT, mainframes to PC clients, to the Internet to mobile cloud to now AI.

All the other stuff still alive. Many of you work in banks here, you have a mainframe guaranteed, right? -- you still have applications that were developed 20 years ago. So -- and quantum will be another one. I think quantum will be great for cryptography and all the things. But ultimately, I think as a form of an accelerator to traditional computing.

We already connected a supercomputer to quantum computing. And so what happens is that the supercomputer is doing all this work. And they said, for this specific task, let me give it to the quantum. They do it faster for that thing and give the answer back so that the outcome of the whole thing goes out faster. That's how we think about it. Tim?

Sorry, Tim Long of Barclays.

Antonio, I wanted to go back to the kind of the core enterprise business. Obviously, servers, you went through some of the reasons why they're so strong. Just curious your take on why not just HPE, but the industry seems to be seeing less storage growth in the enterprise than compute server growth. So why do you think that is? And do you think at some point, there will be a catch-up? Or any perspectives there, that would be great.

I think there are a few reasons for that, Tim. First of all, on the training side, you can't think of storage the way you normally think about storage, right? I think about a large amount of boxes with traditional compute and a lot of SSDs that the fast object gets laid on top of it as a software-defined layer.

But that's in the training side of the equation. We go to enterprise, look, you already have data all over the place. And what we did with the team is giving an intelligent context to that data through the MCP aspect of this. But I expect data to grow once the inferencing continues to grow. So my view is that, that data will grow as the inferencing grows, not because of what we're doing today.

So if that inferencing kicks off, then it will go faster. And then the question is how it shows up. I don't think it's a traditional storage only. I think this memory of -- concept of memory and K-V cache attached straight to the fabric and the cores is going to be -- let me -- I don't think it will be a storage array only. Lou?

Quick question, Lou, and then we're going to wrap up.

Okay. I think it's pretty quick. Lou Miscioscia at Daiwa. Just obviously, you talked about the biggest point is really the supply chain. What about power and data center space in the U.S.? Will that be a constraint?

Power and data center space.

Yes, both individually in '26 into '27.

Yes. I think Louis, it's fair to say we are behind. Look, I'm sure you do the analysis and all of you will probably have different numbers to begin with. That's a given. But how many gigawatts have been announced? I don't know, 150.

We think by 2030, there will be 250 gigawatts of data center somewhat announced. I don't know how much of that 100% will be in production. But the biggest limitation today is power and cooling. And in the United States, we don't have space problem. You may have a community that doesn't like to have a data center in the back of their house.

But look, I think as I reflect the back history, this is also an opportunity to innovate entire -- around the entire ecosystem. There will be no new sources of power. I think it will be independent grid to power this data center. I don't think the vast majority of the energy will be connected to the main grid. And that also forces new innovation in gas turbines. I mean you saw Siemens Energy has using AI as a way to create next generation of gas turbines, obviously, nuclear power and small reactors of sorts.

We talk about putting AI data center in space. Look, we already have -- funny enough, HPE already has a very small AI data center in space. It's going around the earth, 256 miles above our head every day of the week. That's called Spaceborne 2 and the Astronauts are using that to do research on the international space station a small scale. And then here, probably by the end of the year, October, November, Artemis 3 is going to take the first rover on the moon, that compute module and the network to connect back to us is Hewlett Packard Enterprise, and that's Astrobotic.

So the first flip rover on the moon will be powered by Hewlett Packard Enterprise because before you show up there as a human, you need to have some infrastructure, right? And Hawthorne is going to control that rover through that connection. And so -- but yes, I expect Lou that -- and that's why United States has a huge advantage, put aside the politics, right? Getting into this faster, including regulations compared to other geographies of the world, I don't think anybody can really match us.

So we had a suggestion that the next investor event we have is going up to the International Space Station.

Yes. Good luck with that. I have -- I want to say 0 liability with that. I can take you to one of the space mission controls where actually we are going to power that as well. So the mission control where the rover is going to be controlled is also Hewlett Packard Enterprise. So it's a fun thing to do, but actually you learn a lot.

Yes. That's great. Well, thank you, everyone, for joining us here in the room and on the webcast, and thank you to Antonio.

Yes. Thank you for everyone that logged in through the webcast. I appreciate it. Thank you for spending the time with us in the next couple of days.

Welcome to Day 2 of Bank of America's Global Tech Conference. I'm Wamsi Mohan. I cover IT hardware here for the bank. Delighted to welcome HP Enterprise today to the stage. We have Shannon Cross, who's Chief Strategy Officer. A lot of you probably know Shannon Cross from her prior roles as well. So Shannon, welcome. Pleasure to have you here.

Thank you. Very excited to be here. It's a great time to be talking about the company.

Yes, absolutely. I mean this has been a pretty incredible earnings season from a hardware standpoint and you guys really knocked it out of the park. The question that we get a lot is how sustainable is this? And you expressed confidence by giving an outlook for '27 as well. So I would love to -- for you to frame that a little bit for everyone.

Sure. I think when we see what we -- obviously, we're very pleased with what we were able to report for the quarter, how we're looking at the growth that we expect, I mean, in '26, we took up our EPS target by 40%. And then we did provide a financial framework for '27 that I think underscores our belief that what we're seeing is durable and sustainable. What we've been getting in terms of questions clearly on the sustainability and durability side is far more on the server versus networking. I think networking good story, people understand it. We're -- and I'll touch on that a bit in the future.

So right now, I'll focus on server. I think I've been around this industry for a long time. And clearly, I think we're seeing the creation of a new TAM, the creation of a situation where servers are seen as a really critical important component of company's agentic AI journey. In terms of the data points and the reasons why we think this is going to be sustainable, when we talk to our customers, they're clearly at the early days of their agentic AI journey. They see that they want their data to be more on-prem. They want to do more of the compute on-prem, there -- we're seeing demand at the higher end of our platforms. So more memory, more compute power.

Clearly, there's a reason when you've seen ASPs increased as much as they have over the last year, that customers are still saying, hey, we want to buy, we're going to start new programs, we're willing to pay the higher prices because we clearly see that there's an ROI in whatever they're putting out there that justifies the purchase of the hardware. And I think that's something that's different than we've seen in the past. We're also seeing customers who when we had contracts that were not particularly advantageous for us because, remember, we used to have 90-day quote validity. We pulled that back in. We went back to our customers and we said, "Hey, we need to renegotiate. We need to look at this." They weren't all happy. I mean, I'm sure we would have been super happy in this situation. But they've seen what's gone on with the memory provider.

They've seen what the stocks have done. They've seen what obviously, some of the bonus payouts and things that are being talked about right now. And so they realized this isn't a server situation. This is an industry situation. So we were able to come to terms with a lot of those customers in mutually beneficial arrangements. And so I think that plays into it. I would say that we're seeing and we talked about this quarter, we're seeing triple-digit order growth in the quarter. So even in light of the higher ASPs, people are buying. We are -- we have the highest backlog that we've ever had, a record backlog. Our pipeline is multiples of our backlog. We -- the conversations we're having, clearly, customers are looking at what they're going to need. It's not -- this isn't just -- I think we've talked about for a while this idea of a data center refresh. I'm sure that plays in somewhat.

But I do think that they're fundamentally seeing that the server is driving a higher ROI. I mean I kind of think about it when we look at the company, like our company and we talk about it, we have cost of workforce, cost of our consultants. And then we, I think, are going to have sort of cost of whatever you call the agents within the server ultimately. And so again, it's just -- it's become so much more a part of the way that people are doing business. Now I know there's questions about pull-ins, we are seeing minimal pull-ins. Are there some? I'm sure there are, and we have heard of some cases.

But in general, we believe this is just sustainable demand. And we do track. We look for cancellations. We look for signs of double ordering. We're monitoring the situation very closely, but everything seems strong. And again, we're seeing now where the orders that we're placing are at ASPs that are substantially higher than earlier in '26. And so when you figure you leave '27 at that higher ASP, that also provides some underlying support.

Yes. No, that's a great framework to start with. Maybe just to remind a little bit right, like what would you say specifically changed over the last 90 days from customer conversations. Is that sort of an acknowledgment that this inflationary environment is just going to stay longer? Is it agentic? Like what is it that's kind of driving this elevated level of interest but now is going to sustain itself?

Well, I think -- and we all -- we think in 90-day increments, right? But I think a lot of the building blocks for what we're seeing now have been in place over the last 6 months or so. You could even go back to November when we were the first to raise prices and customers continue to buy, and so we continue to see sustained demand. I will say we are hearing more about agentic AI. I would say if you think about our PCAI business, when we first launched it, we had small, medium, large, extra large, and then we realized people wanted developer sizes so more like the extra small and we're seeing those extra small purchasers now moving up the stack and buying multiples of PCAI and also just buying bigger ones.

And so to me, that means they're finding agentic AI use cases that they're able to deploy. Internally, we're -- we have a multitude of use cases that we have and we've been looking at it very closely in terms of we've got grassroots, how do we put the enterprise framework on top of it. And part of my team is working on that. So I've been talking to a lot of our other Chief Strategy Officers or transformation officers about what their journey is. I think we're all sort of in the same camp, and we're all moving forward, we see significant opportunity. And again, underlying that is the server, the data center. I don't want to sell networking short. I think there's a significant opportunity for networking as well, and we're super happy to have Juniper in the fold. So I think that's part of it.

And just in general, I think the idea that customers continue to come and like I said, triple-digit order growth in the quarter, no signs of slowdown. And so I think all of that is kind of playing into the opportunity. I mean I do think one of the questions we've been getting at this meeting is -- or in the conferences we've been at is basically where is the money coming from? I think that's a legitimate question over time. But I think that right now and certainly for the foreseeable future, technology is what's going to enable this incredible improvement that we should see in productivity, and we're right at the center of that.

Yes. Yes. And anecdotally, like we've heard about basically parts of whatever was allocated within corporate, within finance, within sales and marketing, within all these other functions of HR. Money is coming from all of those to kind of support a lower cost structure and higher productivity essentially for an organization. So kind of aligns with the increased use of agentic AI and enterprise. Maybe, Shannon, you can talk about how, as a company, you're thinking about the inflationary environment and how you're managing that. I mean I think there's a lot of concern in the market that for particularly hardware OEMs, it's going to be a really difficult time with the rate and pace of DRAM and NAND price increases.

So how are you managing that? And what's kind of your strategy on a go-forward basis? Are you doing anything with LTAs, are you doing anything in terms of procuring how much supply do you have access to? Because I think supply, I think you said it was a gating factor in growth.

Yes. I think -- I mean, one thing to keep in mind is HPE has been in this business. I mean, we'll go back to Compaq. We've been in it for decades. And so we have very, very long close relationships with all of the component suppliers. We've had LTAs for a long time. We've had various different agreements. So are people leaning a little more heavily into them? Sure, but it's not like this is a significant change in the way we do business with our partners. And I do think that our long-term relationships play in here. I also think the memory providers want to have diversity of customers. And so one of the areas when I talk to the Head of Supply Chain, they're definitely very cognizant that they want to keep us healthy and our competitors healthy, frankly. It's an interesting kind of position.

So I think from that perspective, we're benefiting. In terms of managing the margin side of it, we were the first to raise prices back in November, everybody else followed. We've continued to raise prices. Everybody -- the entire industry changed sort of the quote validity period. And then, again, the conversations, and I think this was -- it's a testament to the relationships and the trust our customers have in us. When we went back to them and said, "Hey, that quote we gave you 90 days ago, we can't honor it because it's -- the prices have moved so much." They worked with us, and we worked with them. And I think that goes back to very tight, long-standing relationships. When I got to the company, I think that was one of the things that I found very positive relative to maybe how I thought about the way the world works is that our customers are sort of the who's who of, I don't know, everywhere, like all industries, we're global.

It's just we're kind of there. And they do look to HPE as a trusted resource. They respect us. They take our advice. They love our technology. And so I think all of that kind of plays into the ability for us to have the conversation with customers and then have the positive outcomes that we have seen over the last few months. So all of that kind of comes in, it's just -- it's talking to them -- talking to our supply chain all the time, talking to all of our partners and working with them. And it's -- I think it's been a -- it's a tough -- I mean it's a tough environment. Don't get me wrong. And as Antonio said on the call, if we had more supply, you would see upside because demand is not an issue. It's supply. But I think we're managing through it as good as could be expected.

Yes. No, I mean that's actually impressive to see the margin results that you guys are delivering in the face of this. Maybe sticking with servers for a minute more before we switch to networking. I do want to ask about AI servers and storage because it feels as though that the AI servers you guys have been selective about the type of deals that you do, and not maybe participating in the super competitive end of that spectrum, whereas on the storage side, I would love to also get your perspective if the server strength is translating and if there is more yet to come on the storage side? And how do you see that group playing out?

I think when it comes to AI server, I think we've been pretty upfront with -- if you go back 2 years ago, we've announced that we're going to buy Juniper. It's a $14 billion acquisition, largest company has done, and we levered up to do it. And so then as you sort of move through the AI server journey, we wanted to focus on making sure that we maintain investment grade that we paid down the debt that we were very prudent with the way that we approach working capital and that we also focused on where we thought we had the right truly to win and to add extra because when you add extra, that drives more margin. And so from our perspective, and I think it's played out, we focused on sovereign and enterprise and we do -- I mean, we bid for hyperscale deals. We will play in there where it makes sense and it fits our framework.

But so far, I think our strategy has definitely been working. And clearly, from a Juniper perspective or from a capital -- from a balance sheet perspective, if you look at it right now, we're now going to be at 2x net leverage by the end of '26, which is a year ahead of our plan. So super happy about how that's working out. And so I do think there are areas that you should remember, we'll get to networking. But I think don't disregard the fact that when we think about the hyperscale opportunity and the large model builder opportunity, networking is going to play a significant role. We've come from a very small position in what we call networks for AI just a few quarters ago to now talking about having $2 billion -- over $2 billion in cumulative orders by the end of '26 so we may not lean super heavily into the lower margin opportunities for hyperscalers and AI server, but we're certainly looking at what we can do from a networking perspective.

And we're also interested -- we work very closely with NVIDIA, but we're also very interested in what we can do with Helios because on the AMD side, we actually have the Juniper technology that's been designed into the system for scale up, and then we have our server technology. And so that, I think, may be an entree for us to be maybe a little more aggressive on the hyperscale server side as well.

Okay. And what about storage, are you seeing any pull in?

I think it's still early for storage. We're happy with what we're seeing in terms of Alletra MP demand. We had triple-digit growth there for -- I can't remember what number consecutive quarter. I mean, we're doing well in that space. But I think it's still coming. And what I really do like though, is the quality because I mean we're not the biggest storage provider out there. I mean I don't think it's a surprise to anybody. But when I talk to the customers, and I talked to the leadership in that business, it's really the technology that's pulling customers along.

They're very happy with the idea that they can have one platform that has file block and object on it. We are seeing both and this is really important storage because another thing I didn't quite realize when I was on your side of the table was just how sticky storage is, it's really hard to rip and replace. And what we're seeing here is not just a refresh of our installed base, but also significant new logos coming to the platform. So it's a smaller part, but it also just like networking has a really nice margin profile.

Yes. Yes. So maybe now switching to networking, right? So clearly, you're now showing some better traction with the Juniper acquisition. And if we just think back to SAM versus now? I mean, you obviously -- I think people felt like your guidance at SAM might have been conservative feels like it is conservative, right? So what are some of the opportunities that you think were not properly calibrated by the market from a Juniper standpoint. Let's start with the revenue side and we will talk about...

Well, I think it's fair because we have consistently raised our revenue expectation through the year. So I'm not sure it was just market. It was how we were looking at things because I do think -- look, this is the largest acquisition the company had done. We had to bring together sales forces, we had to bring together products, which we still are. We had to bring customers along on the journey with us. And so I think we approached guidance for '26 from a networking perspective very prudently, clearly too conservatively given what we're seeing. But I also think that things have played in that are better than we expected. So clearly, campus and branch is doing well. Wi-Fi 7 refresh. There's significant demand there. Our orders were high 20% this quarter. So we're seeing positive moves there.

I think from where there's really excitement and I think probably we were too conservative as you think about networks for AI and the opportunity in data center scale out, scale across what we're seeing in terms of the routing opportunities is strong. But again, in the $2 billion, it's both data center networking as well as data center interconnect. So there's a combination there. And I think the -- whether Juniper could have gotten there themselves, I think having HPE for the scale and the scope, some of the introductions has really helped. And I mean we're at the table with a lot of the large logos that you guys would all be very excited about because they do want alternatives.

And we also -- we're time to market -- first to market with Tomahawk 6 in terms of direct liquid cooling and we're 100% direct liquid cool there. And that's something we started investing in months ago. We started investing in the Helios stack many months ago before it was really maybe something that everybody else wanted to lean in on. So I think that it's a testament to the Juniper and Aruba combined management team that they had the foresight to put the money and the dollars in those places. And it's -- we've made the right bets, I think, and they're coming through.

But if you went back to October, a lot of that was just in very infancy stage. So we weren't necessarily sure where that was going to play out. But I think overall, the reaction has been extremely positive. The combination of the sales force, I mean, that's always kind of a risk, right? You have one salesperson at one account, and then you got the HPE and the Juniper and you put them together, you figure out who's the best, either you keep both or you probably don't, but all of that can cause disruptions. And I think we've managed it exceptionally well. And that is one thing, too, I would say, while we didn't want the extra time, and obviously, some of it was pencils down as we were dealing with the DOJ and all of that. I do think that this integration has been extremely well planned out, gone through -- there are issues. I mean everything is going to have an issue, but I think it's been very smooth relative to what could have happened. And I think that's a testament to the leadership that we have.

Yes. No. It's -- I mean your execution has been obviously very strong here. And it's a complicated deal. It's a very big deal and complicated portfolio emerging, technology emerging. So great job on that. I guess, if I -- you mentioned like data center interconnect. What's the sort of IP that you're able to bring to the table on DCI and how do you think about that opportunity?

Well, we have our own silicon. And I think maybe that's something that people don't necessarily appreciate as much, whether it's parts of campus and branch, whether it's routing, obviously, we work with Broadcom in other areas, too. But HPE has and Juniper has their own silicon. And I think it does create a competitive differentiator that again, I probably underappreciated it. So I'm guessing it's maybe not necessarily as well understood out there. So that's part of it. I think AIOps and some of the technology we have, obviously, the Mist platform is extremely popular and provides customers with significant savings and better quality of service and just their experience.

And some of that, we're able to bring over just put in, and I think that's pretty key. And Juniper has been doing this for a very long time. I mean, they helped to build the original one of the Internet for those of you who -- I remember back then, but -- some people that investors get younger every day or every year. But yes. No, I think it's just -- it's the quality of the IP too. And I think that's something where maybe it hasn't had the opportunity to shine as much as both the market and the acquisition and the support we've given them have allowed them to do.

When you think about the scale-up opportunity, is there room for also scale out within sort of within the AMD system that you refer to?

Well, I mean we'll provide the technology there as well. But I think the key here is this is the first opportunity for us to do scale up, right? And I mean we had -- we had Slingshot. We had some -- again, that goes back to the Cray technology for supercomputing. But here, I mean this is with an AI server rack that we can do scale-up. And then, of course, we could do top of rack and scale-out as well. That would be -- that's obviously more on the Broadcom side.

Yes. Maybe pivoting a little bit to margins. Your networking margin guide is pretty strong. I think when you look at considering where pre-acquisition Juniper margins were where HPE's margins are so really, really strong performance. What's driving some of that? I know you've been leading Catalyst. There's been a whole bunch of other things that have been happening at HPE. So how should we think about the trajectory and the step-up here in margins?

I think there's -- part of this is, obviously, you have the synergies coming together and we're at or ahead of our synergy targets. So that's possibly like what we're seeing, what we thought was going to happen is coming through. So I think that's been a big positive in terms of bringing the companies together. I think there's been a focus, I want to make sure everybody realizes we do appreciate that we spent $14 billion on Juniper from an HPE perspective, so we're not going to starve the company. We have continued to invest and that's seen in what we've talked about with Helios and what we're seeing in terms of scale across and the silicon investments and everything we're doing there. So we do continue to invest while we're getting these margins out.

I think gross margins, there's been some benefit from that perspective, too, in terms of bringing the supply chains together and all of that. And then scale, frankly. I mean, you get scale leverage as we're growing revenues. And we're talking about around 10% revenue growth for Juniper for this year, and we guided to 8% to 12% for next year. So within that, there's opportunity. And then the Catalyst program, we're bringing in, I think, OpEx discipline, which is key. And we do -- we're not double counting. We are very careful that we have the synergies and we have, obviously, what we're doing Catalyst. But I think overall, from a company perspective, we're thinking about how can we do processes better? How can we then layer on AI because you don't want to just throw out like, "Hey, go use Copilot or hey, go do this." You have to actually change how people do work. And I think that's a pretty key part of it. But yes, no, I think there's -- and I would give Marie Myers, huge credit, our CFO. She is exceptionally focused on making sure every single dollar we spend drives value, has an ROI.

And I'm not saying it wasn't part of the core fabric of the company before, but it is absolutely something that we talk about, focus on, track very carefully. And I think that's a mind shift in the company that I think is very beneficial because everything I do at the company, I'm constantly focused on, again, given my history, given shareholder dollars. Is there a return? Are we doing -- when you care about your employees, you care about your customers, absolutely. But we definitely need to think about what can flow through to the bottom line. And I think you're seeing that in terms of the EPS guides that we've given for both '26 and '27.

And on the '26, '27, particularly '27, are you baking in continued price increases and inflationary environment as you think through because it doesn't -- it feels like it's going to be that way, but I'm curious how you'll talk about it.

I think what we've said is we expect to continue inflated environment. We're not getting any specifics on units versus pricing. But we expect to see continued pressure in terms of the supply that's out there as you look at next year. Clearly, DRAM, it's over 60% of the BOM now. I'm not sure exactly where it's at. So that's an area to watch when you think about what you would expect for pricing. There are other inflationary parts, but that's clearly the biggest focus. And I think that's on the server side. I mean, in terms of networking, there's a bit less pressure there. But clearly, from a supply chain perspective, given the significant ramp in demand that we're seeing, you have to kind of manage that through. I mean it's not the same, and I think there's been a lot of learnings, but think back to some of the challenges that everybody had in AI servers when everything was ramping and there were just shortages of various random pieces. I think we're working very carefully to try to alleviate any of those concerns before they really become an issue.

Yes. Now you also completed your H3C divestiture, and that's been great. I think those -- if you rewind 2 years ago, and people had no idea if this money would ever come your way or not and obviously...

Yes, team did a great job.

Yes. And you guys have executed on that, so kudos to you. Now that you have that, as you think about maintaining a certain level of leverage and then capital return, how are you thinking about that?

Well, I don't think there's been a change to how we're thinking about capital return, except that we brought it forward a year. So we had said that we -- by the end of '27, we get to 2x net leverage and then kick in our 75% return of cash to shareholders. We are now bringing that forward to the end of '26. So you can expect that we'll be out buying back stock more aggressively starting in '27 as well as maintaining our dividend. And when we announced our dividend increase. I think it was SAM, I can't. Anyway, whenever we did that, we talked about the fact that we'd like to -- our goal, obviously, it's a Board decision. Our goal is to be a dividend growth company so that's how we think about that as well.

Okay. Well, we're coming up on time only a couple of minutes left. So Shannon, any message you want to give investors on how to think about HPE. I mean, I'll say that when the stock was at $20, we thought it was super undervalued, it was one of the stocks that told people like it was something that could double within my portfolio, like the stock that would be a potential double and you're already 3x. So now, I guess, it still seems that the revisions are outpacing any expectations that people might have had. So would love to put that in context and your message to investors.

Sure. I think -- I'm super excited. I joined the company 2 years ago. I saw the vision that Antonio had about what we could create. I think the Juniper acquisition has been a home run and will continue to benefit us. And it's beyond just thinking about what Juniper was when we bought the company. It's where Juniper and the networking industry can go over the next several years. And the key position that networking will play as sort of the orchestration layer of the data center and what we're doing in AI data center. I think on the server side, we are sort of in a new renaissance for server again, I've been around a long time. I think this is beyond just your typical cycle. I do think that servers and the value of servers are changing, and I think that, that's been reflected in what customers are doing.

And then I also would say underlying all of this is a level of operational discipline that we placed in the company that I think, again, the Catalyst program is becoming more sort of the fabric of how we run the company. And so I think that also provides us with support from a profit perspective. And then overall, we've got the cash flow, and I think that's something that's going to be really important. We feel very comfortable with how we're managing cash. And we're -- now that we're kind of through paying down Juniper, we're going to be returning it to shareholders.

And I think that will be helpful as well. So I'm super excited. I don't think we're done. I think there's a ton of opportunities as we look forward. And I think we're just starting this journey. And I think we're in an incredibly good position. As Antonio says, we've got hybrid cloud, server and networking, and that's kind of the core of where people are going to be investing going forward.

Amazing. Well, with that, we'll have to wrap. Thank you so much and really appreciate you being over here. Thank you.

Thank you, Wamsi.

Perfect. Good morning. Good afternoon. Good morning, everyone. Good afternoon, I guess, to be on the webcast on the East Coast. Really delighted to have with us Shannon Cross, the Chief Strategy Officer, Head of Investor Relations and AI deployments at HP Enterprise. I'm going to add things to your title, Shannon.

Thank you. I have many hats.

You have many hats. Perfect. Well, really appreciate the time and being here. We don't have a lot of time, but you folks had some fairly, fairly impressive numbers last night. I think the part that surprised me was the free cash flow numbers that were out there for 2 years ago, got pulled into this year. I think same with EPS and profits.

Maybe spend a couple of minutes just recapping the earnings, you guys -- what you folks are seeing from a demand perspective, and then I'm going to dig into a couple of questions on service and networking.

Yes. And thank you very much for having me here. Hello to everyone. When I think about the quarter, we were very pleased, as you might imagine, both with what we were able to achieve as well as our guide. I think there were several areas that we've been working on for a long time that came to fruition in terms of our plans. Beyond that, we're seeing strong demand from our customers. And so overall, just came together, I think, very well in terms of this quarter and then looking forward.

Specifically, if you take the various categories, I think starting with networking, we did about 10% revenue growth in networking in the quarter. We're guiding to around 10% for the year, and our longer-term outlook is 8% to 12%, which we gave a fiscal '27 framework basically. And so within networking, we're seeing very strong demand. Our order growth is quite high. Campus & Branch, Routing and Enterprise data centers were all in the 20% plus range.

We see a significant opportunity in our networks for AI, which we took up -- we have a cumulative year-end '26 target. We took that up to $2 billion. And really, what we're seeing there are hyperscalers, enterprise, people who are very excited about what we bring to the table with the Juniper technology, what we can do with Tomahawk 6.

One of the things I'm really excited about is that we've been able to take our R&D from the two companies and kind of combine them. The Tomahawk 6 QFX platform that we've announced uses direct liquid cooling. We're leveraging the direct cooling technology we have from Cray there. So overall, I think it's a testament to the integration of the company.

And then if you go into Routing, clearly, there is significant opportunity for us on what is scale out effectively combining data centers. And that with the PTX platform has been really well received, and we're very happy with the logos we're talking to around that. Security was up 18%. So that was positive. And then we had the scale across with the QFX platform.

So overall, things, I think, in networking are going well. And the Juniper integration is at to ahead of schedule. And so when I first came to the company, we just announced Juniper, this company had never done an integration of this scale. And I will tell you the planning that went into it. The extended period of time we had to do planning helped probably a little bit, but we could have done without it. But in general, we're really pleased with how everything is coming together. So that's networking.

I think on the Server side, traditional Server, it's really interesting. I truly believe there's a fundamental inflection point going on, and there's a new TAM being developed. with regard to Agentic AI.

I mean we're seeing customers with ASPs that are significantly higher than they were 4 quarters ago, coming in with new use cases. They're buying up in terms of compute power and memory requirements on the device. And so to us, this feels very sustainable, durable, and it's a change in the way people look, I think, at their server infrastructure.

And then on the AI server business, we continue to play in the areas that we think we have the most opportunity, both from a margin and a cash flow or working capital perspective, which is enterprise and sovereign. We will look at hyperscale deals where they make sense, but I think we continue to address that market very prudently.

And then finally, I would just say, from a cash flow perspective, we're really excited about the cash that we were able to generate in the quarter, $915 million. It was up $1.8 billion year-over-year. So it was a significant improvement from that perspective. And I see our working capital management and, frankly, just our operating profit really flowing through to free cash flow.

And we did take up our net leverage target in terms of when we expect to hit or 2x to the end of this year. We had been at the end of '27. And as we've told everyone, our plan is to return 75% of free cash flow to shareholders through dividends and share repurchase as soon as we get to the 2x net leverage target, which means we'll start that early in '27.

Perfect. Thank you for that. One of the things I got a lot last night, and I'm sure you got this a fair amount, too, which is just trying to understand the durability of the growth you're focusing, especially on the traditional server side of the market, right? I think revenues were up like 33%. But orders, I think, were up triple digits.

I mean the question is like, is this just a big pull-in? Or how durable is it? Just anything you can talk on that front would be helpful.

So we -- as you might imagine, for those of you who don't know, I used to be a sell-side analyst. And so I push very hard with the team on things like pull-in and double ordering and all the things that I've heard about 20-some-odd years on Wall Street, right?

So we do a ton of analysis on what we're seeing in terms of the order book, what we're seeing in terms of our pipeline, conversations with customers. At this time, we're seeing minimal pull-ins. I wouldn't say we're not seeing any because I think that would be kind of silly. There certainly are probably a few. But generally speaking, what we see is sustained demand.

And the triple-digit orders were very positive. Underlying that, there was some AUP increase, but there was also very strong unit orders. And so when we look at what people are using it for, there's Agentic AI use cases, obviously. There's traditional data center refresh that's going on because you have effectively the ability to replace several older generation servers with one, and that saves on space, cooling, power, et cetera. So that's also helping.

And we're seeing customers -- and I've talked to a number of our EC leaders, our executive committee leaders who deal with customers all the time. And they're hearing from people about literally new use cases and also repatriation of data on-prem.

Again, I covered the space for a long time, and the cloud came and everybody said things would repatriate potentially at some point. And we do have hybrid cloud, which I think my company was one of the leaders in and is one of the leaders in. But we are absolutely, I think, starting to see customers say, because of regulatory, because of compliance, because of data latency, data security; we're going to bring things on-prem.

And then the other thing that I think is very interesting and it's early days, and we're candidly doing it internally as well; is looking at the tokenomics of what's going on in terms of all of this Agentic AI and inferencing in that. And how do you do it most cost effectively. And if you can manage to know sort of what you need on-prem, you can use open source models, you can use much more targeted models, you can use our PC AI offering, you can use traditional servers.

And I think that is something that probably is a bit underappreciated in terms of customers right now where they're at, understanding what their token costs are because we are in that journey as well. And then what's the best way to optimize their IT spend around that.

So I think that's another area where you'll probably see some more shift on-prem. I mean it's a great time to be in the hardware space. I would say that's a long time coming in making that statement, but it's very exciting.

And I think the other thing that I would just say is the combination of HPE server and HPE networking is something that I think is going to be really interesting over time, whether it's the Helios stack that we're doing with AMD or it's what we can do in the data center cross-selling because remember, networking has a better margin profile. I think the opportunity to continue to go up into the right is pretty significant.

Maybe just on networking, and we'll go back to servers as well. But on the networking side, right, one that was folks like 21% operating margins, give or take, last quarter. But the guide for '26 and then '27 actually has a really good step-up in networking margins getting back to this mid- to high 20% range, I believe. or at least improving from the levels.

What is driving this margin expansion in networking? Is it the Juniper integration? Is it revenue synergies?

Well, I think in general, we're very pleased with what we're seeing in terms of the operating margin. Keep in mind, we have -- we're having a very strong quarter and a very strong year. And so along with others in the industry, we are seeing an increase in variable compensation. So that is something that you don't think will necessarily repeat in the coming year depending on how the company does. So that is an area where we would expect to see some improvement as you look at '27. Now it's going to stay elevated in '26.

Beyond that, you do have the Juniper synergies. And again, those are at ahead of schedule, very pleased with how that's going. And then I think from a -- there are some areas where we're actually seeing some Catalyst savings as well. So that plays in.

And then just continuing the revenue growth, I think, is going to be really important as you look to '27. And we've said 8% to 12% for networking because you'll get scale leverage off of that. And that should really help us as we think about the overall operating profit.

But I think the business is being run very well. It's -- I would say the management team there is very prudent in terms of where they spend money. We definitely have a lot of oversight, which I think is good. I didn't cover Juniper, but I've heard stories about they had a lot of things that they wanted to invest in.

And I think we're -- the positive here is they're able to leverage our scale and our scope and our supply chain and our sales force and everything else that we kind of brought to the table, and it gives them the capacity to really invest where it drives incremental revenue.

Are you seeing signs of revenue synergies where you're able to bring Juniper along in deals as you sell traditional servers, AI servers? And that are you seeing net revenue synergies unlocked at all from that asset?

It's early days, I would say, but there are anecdotal stories and areas where we have been able to -- either HPE has brought in Juniper or Juniper has brought in HPE. So I think that the sales forces are learning how to work together and how to drive that.

I do think that there's -- one of the products that I'm very excited about, again, very early, we'll have to see how it all plays out is the Helios rack stack from -- it's the AI stack from AMD. And Juniper is actually designed in on a scale-up basis. And so we have Juniper trays in there. We'll have HPE's server technology in there.

I think it's going to be a really interesting use case. And again, the hyperscalers have all committed to silicon diversity. So over time, we'll see how that works. But I think we have a really good shot there.

Got it. If I zoom out a bit, servers are doing very well for you guys. Your peers are kind of seeing the same exact growth rates, if not better in some cases. Demand is fairly robust for a lot of companies. But at the same time, I think we look at IT budgets, like these don't seem to be growing triple digits or high double digits.

That would be nice.

That would be nice. Where do you think the money comes from for all this stuff from an IT budget perspective? Or where enterprises getting the money to spend on all this stuff?

I don't know. I think that there's probably a mix shift away from lesser important areas, and they're probably sweating some technology that they don't need quite as significantly or at least at present. I do think it's interesting that servers used to be what was sweated. And so apparently, that's not the case right now, which is really nice.

We're still seeing strong demand on our Storage side. I mean our Electro MP orders were up triple digits. Our revenue was up triple digits. And I think that's a testament to, I would argue, leading technology that's positioned toward AI or that's positioned toward next-generation data centers is an area where investors continue to -- or not investors, sorry, where customers continue to invest.

I do wonder -- I mean, obviously, there's labor savings. I mean you see it, you hear it across the Valley certainly in terms of what people are doing to be able to invest more aggressively in AI.

And beyond that, there's -- the one thing that we look at, and maybe this is a fair way of thinking about it, we look at AI internally because I help to lead the AI strategy of the company. And we look at it as, yes, fine, a cost savings optimization, but so much more of it is literally let's change the process, let's try to improve both for the employees as well as the customers, the experience. And we're seeing benefit there.

And I think the other thing I try to emphasize at the company with Catalyst is it's not just a cost-cutting exercise. It is literally a way that we can reallocate resources and really drive better outcomes for our customers and, frankly, for our shareholders over time. And it's funny because when I was on sell side, people say, "Oh, we're going to reallocate resources." I'm like, "Oh, okay, that makes sense."

It's like it's really hard. I mean this is not something where you just snap your fingers and things move around, people's lives are affected in that. And there's a significant amount of training, and they call it management of change where you have to go in and actually make sure everybody is sort of on the journey with you.

But I think that, that's really where you're going to see increasing investment. And again, a lot of that is -- goes back to hardware and what sort of underlies it. Obviously, software and other things play in there, too. But I think we're in a good position.

The other topic -- other dynamic on this is really -- you obviously have a very robust set of numbers that you've guided for the next couple of quarters, next 18 months really. How do you think about component availability? And how much of the memory or other things you have spoken for already versus you still have to go out and procure?

Well, we have long-term agreements. I mean we've been in this business for a very long time going back to Compaq. And I mean, I'm sure some of the components probably go back to oscillators, if you go back 80 years. But I think the key here is what we have guided to is what we see line of sight to do. Now we have a significant backlog, and we have a really big pipeline. Our pipeline is multiples of our backlog sort of across the board.

And this is one of those quarters where everything looked good. It was the -- I mean, pretty much everything was really solid. And so when you think about that and you think about our component environment, if we can procure more components, it will drive upside. But we're very comfortable with the numbers that we put out and the growth rates and the expectations that we have for the company.

Again, we're at 8% to 12% for both networking as well as Cloud & AI revenue next year. Our operating margin guidance is 12% to 16% operating margin and our EPS growth is 12% to 16%, yes, 12% to 16% EPS growth. Sorry, it's been -- it was a very late flight last night. So at the end of the day, we are very pleased with what we're seeing.

And it's incumbent upon us to execute. And that's what I keep telling everybody in the company. We -- I will tell you, we have really good assets. I've said that since the day I walked in 2 years ago. We just need to be consistent and execute. And I think we've gotten on the path where we can do that. And it's also helping that the underlying market is very strong.

So -- but I do think we are in a unique -- with networking, with Hybrid Cloud, with Server, with our services capabilities, with our financing capabilities; we are the only ones that can bring all of that to table. And we haven't even talked about GreenLake, which is the platform on which people are able to actually purchase and utilize our technology from sort of a -- on a capacity basis.

And what that provides, especially now when you think about the cost of a server and what servers have done, I mean, prices are up significantly, and you could have effectively sticker shock. But if you're able to buy it on a ratable basis as you need it, I think that makes it a lot more palatable. And again, it allows customers to do things that are a cloud-like experience on-prem, which certainly in the Agentic space right now with the concerns around data, I think, is very attractive.

On the AI system side, right, I think you talked about a $5.9 billion backlog that you're sitting at right now with AI systems, you always talk about "Hey, the focus is enterprise and sovereign." You said this earlier, "We look at hyperscalers that makes sense, but we don't want to be aggressive there."

Does that start to change as maybe other companies back away from it, like a Supermicro, for example, where the profits get more attractive, where it makes sense to engage with new clouds? Or is it really the focus on the enterprise and sovereign?

I mean, look, I think we're happy to work with anyone, and we will evaluate deals, and we bid on things. I mean, certainly, where it makes sense. I would say customers -- some customers are starting to look for alternative suppliers.

The other thing you have to keep in mind, though, is there's a finite amount of components. So if we're sitting on a bucket of components, we have a backlog, we have a very strong order book, and we have existing customers and loyal customers and maybe some new customers that we want, you have to really do -- it's one of the things that I think I'm seeing right now is a lot of account strategy about where you put the precious components you have and how you drive the best shareholder return for what you do with those components.

So make sure they go to the highest margin or you make sure they go to the customers that have the most potential long term. I mean you have to really think about some of the stuff strategically. And I think that's an area where we're working on and we've become more nimble. And I would just throw in, I think that COVID and the tariffs actually helped probably not just our company, but other companies become far more nimble in the way they address the market.

Perfect. The red light blinking over there. So I think I'm up on time. But maybe I'll turn it back to you any closing comments, anything we did not touch on that you want to flag our way?

No, I think we're very pleased with the fact that we're 2 years ahead of our plan that we put out. I think it's a testament to the hard work that the company has done and frankly, the assets that we have amassed, whether it's on the networking side or what we developed in AI server and then leveraging, frankly, our traditional Server business.

So I think we're just getting started. I see tremendous opportunity for this company over the next several years. And whether it's catalyst on the cost and the optimization side or it's everything we can do from a revenue growth standpoint, I really do think we've reached an inflection point.

Perfect. I look forward to the next fiscal '28 targets now.

Good day, and welcome to the fiscal 2026 Second Quarter Hewlett Packard Enterprise Earnings Conference Call. [Operator Instructions] Please note, this event is being recorded. I would now like to turn the conference over to Paul Glaser, Head of Investor Relations. Please go ahead, sir.

Good afternoon. I'm Paul Glaser, Head of Investor Relations for Hewlett Packard Enterprise. I would like to welcome you to our fiscal 2026 Second Quarter earnings conference call with Antonio Neri, HPE's President and Chief Executive Officer; and Marie Myers, HPE's Chief Financial Officer.

Before handing the call to Antonio, let me remind you that this call is being webcast. A replay of the webcast will be available shortly after the call concludes. We have posted the press release and the slide presentation accompanying the release on our HPE Investor Relations web page. Elements of the financial information referenced on this call are forward-looking and are based on our best view of our business and the external factors affecting us as we see them today.

HPE assumes no obligation and does not intend to update any such forward-looking statements. We also note that the financial information discussed on this call reflects estimates based on information available at this time and could differ materially from the amounts ultimately reported in HPE's quarterly report on Form 10-Q for the fiscal quarter ended April 30, 2026.

Figures used in verbal remarks are rounded for ease of discussion. For more detailed information, please see the earnings materials as well as disclaimers relating to forward-looking statements that involve risks, uncertainties and assumptions. Please refer to HPE's filings with the SEC for a more detailed discussion of these risks. For financial information that we are showing on a non-GAAP basis, we have provided reconciliations to the comparable GAAP information on our website.

Please refer to the tables and slide presentation accompanying today's earnings release on our website for details. Throughout this conference call, all revenue growth rates, unless noted otherwise, are presented on a year-over-year basis. Unless otherwise noted, all financial metrics and growth rates discussed today are non-GAAP and EPS refers to non-GAAP diluted net earnings per share. Certain financial information featured in the presentation today has been normalized to include Juniper Networks results as of the beginning of HPE's fiscal 2025. Antonio and Marie will reference our earnings presentation and their prepared comments.

With that, let me turn it over to Antonio.

Thank you, Paul. Good afternoon, everyone. HPE delivered an exceptional quarter with record-breaking results, disciplined execution and clear proof that our strategy is working. We made excellent progress in our Juniper integration and in our catalyst initiative with both running ahead of schedule.

Revenue in the quarter reached $10.7 billion, up 40%. Non-GAAP earnings per share of $0.79 increased 108%, significantly above the high end of our outlook. We generated $915 million in free cash flow, an improvement of $1.8 billion, driven by strong cash from operations and improved cash conversion cycle performance. Demand was even stronger than revenue growth. Orders more than doubled, significantly outpacing revenue, resulting in a record company backlog.

Customer investments in agentic AI and AI inferencing accelerated. We also saw broad-based demand strength across the portfolio driven by ongoing investment in compute infrastructure modernization, unstructured storage data growth and private cloud adoption for AI. Last year, at the security analyst meeting in New York, we laid out our strategy and fiscal 2028 financial commitments. Based on our strong first half 2026 results, our record backlog and our visibility into the second half demand, we now expect to deliver $3.40 in non-GAAP earnings per share at the midpoint and at least $3.5 billion in free cash flow in fiscal 2026.

That is 2 years ahead of our committed long-term plan. Marie will provide more detail on our third quarter and full year fiscal 2026 outlook as well as our fiscal 2027 framework which is grounded on durable customer demand and the profitability of both business segments.

Now let me turn to our business segment highlights, starting with networking. I am particularly pleased with the progress we are making on the Juniper integration. We are ahead of our integration milestones and synergies commitments, and the unified portfolio and sales force are already strengthening our market position and growth momentum. Our combined networking portfolio and vision for self-driving networks is resonating with customers and that enthusiasm, together with strong go-to-market execution is reflected in our results.

Networking delivered revenue of $2.7 billion, up double digits on a normalized basis with orders growing significantly faster than revenue. We saw increased demand in Campus & Branch network for AI and security. I am pleased with the strong demand we saw from enterprise customers for our networking portfolio. Campus & Branch orders reached a new record high, growing in the upper 20% range on a normalized basis.

We won multimillion dollar deals across multiple verticals, including retail, automotive, government and technology. WiFi 7 access points sales increased more than 7x reflecting a clear shift toward network modernization. HPE for the 20th time in a row was named a leader in the Gartner Magic Quadrant for enterprise wire and wireless LAN infrastructure. We believe this independent industry analyst validation reinforces how far ahead we are in enterprise networking beyond even incumbents.

Customers trust us with their most critical network and infrastructure decisions as they expand their digital initiatives and AI investments. One customer taking advantage of the power of our combined Campus & Branch networking portfolio is Lowe's. With over 1,750 stores across North America, Lowe's chose HPE to deliver the network foundation for a major technology transformation to support its digital on-ramp and AI-enabled operations.

The solution is built on our HPE Mist AI platform for wire and wireless networking infrastructure alongside our HPE EdgeConnect SD-WAN solution. Last month, we reached a milestone with the launch of new autonomous agents powered by agentic AI for optimizing networking performance. The self-driving network is no longer a concept, it is a reality.

The U.K. Ministry of Justice is an early adopter example. It was able to reduce the number of incidents seen by its network operations center by approximately 75% after deploying a suite of solutions that included our new HPE self-driving network capabilities. In enterprise data center switching, orders increased nearly 20% on a normalized basis. Our data center switching pipeline remains strong. Cross portfolio product integration and sales across server, storage and networking are driving deeper customer engagement and larger deals.

Security orders grew in the mid-teens on a normalized basis. We continue to make the network our first line of defense that responds to threats in real time. This quarter, we launched the HPE Juniper SRX400 series, bringing carrier-grade firewall protection to the branch for large distributed environments. We see significant runway as more customers consolidate networking and security with a single vendor forcing convergence all the way to the silicon layer of the stack where HPE will have further differentiation.

In our service provider customer segment, revenue increased double digits on a normalized basis. We're routing orders growing significantly faster than revenue. Routing orders increased nearly 30% on a normalized basis, driven by data center interconnect deployments in large cloud service providers. customers are choosing HPE because we help them scale in every dimension, scale up by increasing the performance and density of individual platforms for the most demanding AI workloads; scale out by expanding the network fabric to connect thousands of GPUs and accelerators within a single data center. And scale across by extending high-bandwidth interconnectivity between data centers, across campuses and into the wide area network.

So AI services can run wherever they are needed. HPE is developing a scale-up Ethernet switch and software designed specifically for the AMD Helios AI rack-scale architecture, which we expect will be introduced in the fall. In scale out, we lead with our AI-driven QFX switching fabric. HPE is the first OEM to productize a Tomahawk 6 based 100% liquid cooled switch with industry-leading performance and power efficient for AI infrastructure. In addition, our leading fabric management [ NII ops ] capabilities reduce congestion, latency as well as operational complexity.

We expect our road map to extend this leadership through co-packaged optics resulting in lower overall power consumption. Scale across. The Juniper PTX Series delivers 800 gigabit density with exceptional power efficiency, leveraging our distinct Express silicon and simplified AI-native automation. Because of our leading innovation and market momentum in networks for AI, we are raising our cumulative fiscal 2026 networks for AI order target to at least $2 billion.

We are laser focused on building the best networking business in the industry. Our priorities are clear: extend AI-driven automation across the portfolio, help customers scale modern AI infrastructure with secure, high-performance networking and lead in the convergence of networking and security. We have the team, the capabilities and the momentum to convert the opportunity in this segment into durable shareholder value.

In our cloud and AI business segment, we executed with strong discipline across all business lines. Revenue was $7.7 billion, up 23%, driven by exceptional traditional server orders and very strong demand in AI systems Alletra MP storage, private cloud and GreenLake software and services. Traditional server orders increased triple digits as customers continue to modernize their compute infrastructure and invest in AI inferencing. We are working very closely with our silicon and memory partners to continue to secure supply, which we factor into our new fiscal 2026 guide.

We're also engaging customers and channel partners on lead times and configuration options to help them plan effectively. We saw strong demand in AI training throughout the quarter. We booked $1.8 billion in new AI systems orders, bringing cumulative AI systems bookings to $16.4 billion. We entered Q3 with $5.9 billion in backlog, primarily composed of enterprise and sovereign orders. We are seeing a broad pattern across industries. Enterprises want the flexibility of choosing multiple AI models with the governance and control of on-premises. We will continue to manage AI systems opportunities with a focus on profitable growth and prudent working capital management. Storage had an outstanding quarter. Alletra MP storage orders increased triple digits, the sixth consecutive quarter of strong growth. Several weeks ago, we expanded the platform with new file storage in Agentic AIOps capabilities.

This extends Alletra MP into the growing unstructured data market. Our HPE Morpheus Enterprise and HPE VM Essential software offerings continue to build momentum. Revenue grew sequentially for the fourth consecutive quarter. VM Essentials customer count increased 43% in the first half with another rise in net new logos.

Private cloud AI orders increased again this quarter with a growing base of new customer wins. We recently launched our second-generation PCAI offering designed for enterprise AI inferencing and cloud-gapped sovereign environments, which position us for continued growth. More broadly, we are embedding agentic AI capabilities across our storage and data protection portfolio to help customers automate AI data pipelines and operations.

We continue to add new cloud and AI agentic services to our GreenLake cloud platform, acquiring new customers and increasing the net retention rates for our GreenLake services business, which remains near 110%. We exited Q2 with approximately 50,000 customers operating their IT in our GreenLake cloud, managing more than 6.7 million systems, up from 5.3 million a year ago. One win that brings the power of the full HPE portfolio together is the Dallas Cowboys, the most valuable sports franchise in the world. They came to HPE with a clear objective, modernize their infrastructure, simplify operations and build the right secure foundation for AI.

We delivered a comprehensive solution anchored on our HPE GreenLake private cloud offering, spanning ProLiant servers, Alletra MP storage, and HPE Morpheus Enterprise. The Cowboys are also adopting HPE VM essentials as their preferred virtualization layer. This is a strong example of the value customers can unlock when they chose HPE as an end-to-end technology partner. Lastly, HPE Financial Services delivered another outstanding quarter with record return on equity.

Financial Services deepens customer relationships supports our GreenLake cloud adoption and remains a meaningful competitive advantage as customers ramp their investment in AI. Before I close, I want to highlight 2 important upcoming events. In 2 weeks, we are hosting HPE Discover in Las Vegas. We will share updates on our networking cloud and AI strategies, including major product announcements, along with a live Q&A for investors and analysts. I hope to see you there.

Then later this fall, we will host a dedicated network in Investor Day. In closing, HPE delivered an exceptional quarter. Our results demonstrate that our strategy continues to pay off. We now expect to significantly exceed our original fiscal 2028 non-GAAP earnings per share target and generate at least $3.5 billion in free cash flow in fiscal 2026, 2 years ahead the plan. The market trends driving our performance remains strong and well aligned to our strategy.

We expect demand strength to continue into fiscal 2027 and beyond, which will accelerate durable shareholder value as we continue to scale profitably. We are executing with strong discipline, creating meaningful value from the Juniper acquisition and strengthening our position at the intersection of networking, cloud and AI. With the combined strength of HPE and Juniper, we have the portfolio, the talent and the go-to-market scale to lead in the market. I want to thank our team members for their focus and strong execution.

With that, let me turn it to Marie to take you through the financial results and our 2026 and 2027 outlook. Marie?

Thank you, Antonio, and good afternoon, everyone. I'm pleased with our outstanding second quarter results. We exceeded our commitments, delivering record revenue and EPS, driven by disciplined execution and a strong demand environment. Our large backlog, favorable industry tailwinds and improved demand visibility support a higher growth outlook. In addition, we are achieving catalyst cost savings and Juniper synergies ahead of schedule. .

As a result, we are increasing our fiscal '26 EPS outlook by over 40%. I will address the drivers behind the strong EPS and free cash flow outlook shortly. But first, let's take a look at our Q2 performance. Revenue of $10.7 billion was above the high end of our guidance range, led by traditional server as customers accelerated investments in agentic AI and AI inferencing, and by networking where we saw broad-based growth across the portfolio. Sequentially, revenue grew 15%, reflecting higher average selling prices within our server business, driven by ongoing DRAM and NAND inflationary costs and supply constraints.

We continue to work with our partners to secure long-term agreements while executing the pricing actions we discussed last quarter. Gross margin improved to 36.9% driven by mix as we shape demand to higher-margin products. Catalyst savings and Juniper-related synergies also contributed to improvement on a year-over-year basis. Operating profit was $1.4 billion, above our expectations, representing a 13.3% operating margin. As the company scales and we continue to capture accelerated catalyst cost savings and Juniper synergies, we expect operating profit growth to continue to outpace our top line.

EPS was $0.79, well above the high end of our guidance. GAAP EPS was $0.44. We delivered Q2 free cash flow of $915 million, fueled by strong operating profit. Now let's turn to our segment results. Networking delivered another solid quarter. Revenue of $2.7 billion was up 10% on a normalized basis as growth accelerated. Our backlog continues to grow, given elevated demand and supply constraints and this is reflected in the greater than 40% sequential growth we saw in our purchase commitments.

We continue to see strong demand for our networks for AI portfolio and now expect cumulative orders to reach at least $2 billion by fiscal year-end '26. Within our product categories, Campus & Branch, normalized revenue growth accelerated to 10%, again by large deals across multiple industries. Security growth inflected positively to 18%, benefiting from improved backlog conversion and solid in-quarter demand.

Data center networking and routing grew 6% and 9%, respectively, reflecting robust networks for AI demand. We are optimistic about the demand momentum we are seeing based on our growing pipeline. Across customer verticals, service provider revenue grew 13% and enterprise grew 9% on a normalized basis. Our AI-native self-driving network solution is clearly resonating as customers prioritize AI use cases and simplify their network operations.

Network operating margin of 21.6% was in line with guidance, reflecting improved operating leverage as Juniper synergies continue to ramp. The sequential decline in margin reflected 2 factors. First, Q1 benefited from certain onetime items. And second, Q2 absorbed higher variable compensation expense. We remain focused on disciplined execution, operational efficiencies and synergy realization to improve profitability and expand margins in the second half and beyond.

Moving to cloud and AI. We delivered revenue of $7.7 billion, up 23% as strong order activity and pass-through of higher cost of new orders, a traditional server and storage drove the upside, partially offset by supply constraints and timing of AI server shipments. Financial Services continues to perform well. Scale benefits drove operating profit of nearly $1 billion, up 48% sequentially and triple digits year-over-year, pushing operating margin to 12.4%, up 220 basis points sequentially and server revenue increased 33% as ASP growth and additional server more than offset supply-constrained unit volumes.

Demand remained broad-based, as orders more than doubled year-over-year and increased strong double digits sequentially. We see accelerating demand in high memory configured service, targeted and agentic AI workloads, supporting our expectation of sustainable growth. AI systems orders of $1.8 billion were more balanced and broad-based this quarter. Demand is expanding beyond AI server factories into broader AI workloads like orchestration, data movement and agentic AI. Service Provider orders exceeded the combined total of its prior 4 quarters, underscoring the inherently lumpy nature of our large-scale AI deals.

Our backlog increased nearly 20% sequentially to a new high, and our pipeline remains multiples of our backlog. We continue to expect AI revenue to improve in the back half of the year now peaking in Q4. Storage revenue grew 2%, driven by strong orders, the ongoing mix shift towards high-value owned IP and disciplined pricing execution. Alletra MP customer migration momentum accelerated sequentially, driving triple-digit year-over-year growth in both orders and revenue. Continued demand strength in private cloud and our expanding backlog are driving improved revenue visibility.

Lastly, financial Services revenue was up 6% and generated an all-time high in return on equity exceeding 30%. Turning to our catalyst initiatives and Juniper synergies. I'm pleased that we are running ahead of plan as we work on a range of programs to reduce cost of sales and OpEx across our business. As a result of these programs, at quarter end, we reported an employee base of just over 65,000, the lowest level at which we have operated as a combined company and reflecting an over 9% decline since both programs began.

Within Juniper synergies, we continue to focus on the 4 pillars we laid out at SAM. Phase 1 of the integration, which we completed in January, focused on reducing overlapping corporate functions and optimizing sales and service organizations. As our Mist and Aruba portfolios converge, we expect to optimize our R&D spend. In addition, we plan to continue to leverage overall HPE scale to improve commodity prices and consolidate our vendor footprint to drive savings through supply chain integration.

Finally, regarding customer support, we intend to leverage scale and digital capabilities inherited from Juniper to further improve efficiency and the customer experience. We expect to exceed our annual target of $200 million by the end of fiscal year '26. I'm pleased with our progress on Catalyst, and we are ahead of plan. Workforce transformation continues to drive the majority of our savings and Gen AI-enabled process simplification now represents nearly 20% of our fiscal '26 initiative savings. We are leaning into GenAI to increase productivity and reduce costs across the organization, including customer support, HR and marketing.

Our teams are driving greater automation, redefining work management and reducing costs. We are also rationalizing our global lab footprint by more than 2/3 and reducing our contractor base and supply chain customer service by over 90% through targeted consolidation. Taken together, we are building a leaner, more efficient organization and delivering meaningful benefit to our operating margin.

Turning to free cash flow. We delivered operating cash flow of $1.4 billion. Free cash flow totaled $915 million in Q2, bringing our first half fiscal '26 total to $1.6 billion, about 75% above our prior comparable period, high reported in fiscal '21. Our cash conversion cycle improved by 2 days from Q1, driven primarily by increase in days payable due to higher purchases to support future shipments. This was offset by an increase in days of inventory due to higher inventory in anticipation of second half AI service shipments.

Days receivable increased by 5 days from the prior quarter due to strong revenue performance towards the end of the quarter. Inventory ended the quarter at $9 billion, up year-over-year and sequentially, supporting second half AI installations and targeted commodity purchases. We remain committed to our capital allocation strategy. During Q2, we returned $343 million to shareholders, including $189 million in common dividends and $154 million via share repurchases. We refinanced $2 billion of debt, received gross proceeds of approximately $1.4 billion after closing our previously announced H3C transactions last month and used cash on hand to retire our term loan.

We expect the net impact will reduce annual net interest expense by approximately $75 million. Importantly, we improved our pro forma net leverage ratio to 2.3x at quarter end, down from 2.6x last quarter. Turning to guidance. We are taking up our outlook on the back of Q2 results and greater visibility into the second half demand environment.

Starting with Q3, we expect total revenue will be between $11.5 billion and $12.1 billion, driven by strong demand. For Networking, we expect revenue to grow 73% to 78% year-over-year and on a reported basis or approaching 10% on a normalized basis. We expect revenue performance and synergy realization to help offset the impact of inflationary component costs, driving an operating margin rate in line with our full year guidance.

In cloud and AI, we expect revenue to grow with the high teens, reflecting demand durability, elevated pricing and improved AI systems revenue. We expect operating margins to be in the low to mid-teens. On a consolidated basis, we expect Q3 total operating expense to increase sequentially, supporting seasonal marketing expense and networking R&D investments.

We expect our operating margin rate to be up on a sequential basis, driven by improved operating leverage. Consequently, we expect EPS between $0.88 and $0.93 and GAAP EPS between $0.84 and $0.89. For fiscal year '26, we are raising our EPS outlook range to $3.35 to $3.45. We are also raising our GAAP EPS range to $2.42 to $2.52. We are making the following updates to our outlook.

We are raising our full year consolidated revenue growth to 29% to 33% on a reported basis or high teens on a normalized basis. We are also raising our full year consolidated operating profit growth outlook to 80% to 85% on a reported basis. For cloud and AI, we expect server demand and pricing to remain durable driving sustainable revenue growth. Consequently, we are raising our full year cloud and AI revenue growth to the low 20% range from our prior mid- to high single-digit range, driven by higher ASPs in our traditional server business and improved AI systems revenue.

We are also raising our operating margin rate outlook to low to mid-teens. We are raising our full year networking revenue growth, 72% to 75% on a reported basis or approaching 10% on a normalized basis, reflecting accelerated business performance as our integration efforts take hold. We are lowering our OI&E outlook to a range of $420 million to $460 million, reflecting lower net interest expense expectations.

Lastly, we are increasing our free cash flow outlook to at least $3.5 billion, up from our prior outlook of at least $2 billion. We are confident in our new fiscal '26 outlook as we see continued order momentum in the business thus far in Q3. Based on the durability of demand we are seeing in our results, we are providing an initial framework for fiscal '27. We see sustained secular tailwinds driving consolidated revenue growth of 8% to 12% with a similar range for both of our networking and cloud and AI segments.

Our outlook assumes that acceleration in AI systems revenue growth. With the improved operating margins of 12% to 16% for the company and expect to see a year-over-year reduction in operating expense. We forecast net working margin in the mid- to high 20% range, driven by scale mix and synergies with cloud and AI operating margin in the range of 10% to 15%, depending on the mix of AI business and the pace of Catalyst savings. We expect revenue growth and operating leverage to deliver EPS growth of 12% to 16% and free cash flow of at least $4.5 billion. Our outlook is expected to enable faster debt pay down.

As a result, we now expect to reach out 2x net leverage goal by the end of fiscal year '26, one year ahead of schedule. Once we reach our leverage target, we expect to return at least 75% of free cash flow to our shareholders via dividends and share repurchases. To close, Q2 was an outstanding quarter for HPE. We scale the business, expanded margins and generated significant free cash flow. We raised our outlook and are building a stronger, more profitable HPE.

I am confident in our ability to create long-term value for our shareholders. With that, I'll turn the call back to the operator to begin the Q&A.

[Operator Instructions] The first question will come from Asiya Merchant with Citi.

Marie and Antonio, I guess folks are kind of talking about enterprise budgets, just given the price inflation that you guys are seeing through and being passed through, where do you see enterprise budget still sustaining themselves? And obviously, your guide here now into fiscal year '27. Many people are concerned that there is some kind of like demand cliff that you could see even past the more near-term outlooks and growth forecast that you're seeing. What gives you this confidence to now provide fiscal year '27 guide if you could help us understand, I see it's between networking as well as cloud and AI. What gives you the confidence that you're seeing in being able to provide an early outlook into '27? That would be great.

Well, thank you, Asiya, for the question. I think it's multiple factors. Factor number one and probably the most important one is the durability of the demand based on what my conversations with customers and the large, large pipeline, which remains multiples of the current backlog. And when you look at that demand and the pipeline is driven by the use cases, we see with the deployment of AI or the build-out of new data centers for AI, obviously, and then the modernization taking place in enterprise.

And so it's a combination of multiple things that ultimately give us the confidence to not only provide the new guide for '26, but an early view of '27. So when you think about our results in the first half, and the backlog we have and the supply that we have on hand and what is coming, that solidifies the '26. And then in '27, the momentum we have in networking is outstanding across all customer segments and as well product segments. We talked about some of the demand that we see today in Campus & Branch, upper 20% and cloud networking, which is in the 30% and so forth. And customers, when you think about budgets, obviously, they are challenged because of the price increases we have seen driven by the cost of commodity.

But I can tell you, we have not seen any pull-in. We don't see a cliff. And in many ways, I think customers are prioritizing getting access to technology now faster than ever before because nobody wants to be left behind when it comes down to deploying. I give an example in our own company. We have 1,200 use cases in AI. Marie, just next to me here, is one of the early adopters and I will say, aggressive adopter but we have more than 250 use cases, mostly agentic AI, which has been already deployed. And now we see this across the entire spectrum. I was last week in Chicago. I met with a number of customers and partners, and they see this.

And when you go through that motion, then AI inference is growing. And so we expect that the AI inference is going to be an accelerator of our demand as we go forward. And therefore, points to be durable in our demand and our ability to convert that.

The next question will come from Wamsi Mohan with Bank of America.

Really impressive set of results here and guide. Antonio, can you give us maybe some rough mix of the opportunity that you see. You mentioned scale up, scale out, scale across, especially as you look into fiscal '27, how do you see that evolve? And if I could, Marie, the growth in free cash flow is well in excess of EPS for '27, hoping you might be able to share some color on the drivers of that.

Well, thank you, Wamsi. I think it's fairly balanced across the 4 product segments, Wamsi. I mean let's start with Campus & Branch, which has been and is the lion's share of our networking mix. I think the self-driving network vision and now the execution of it is absolutely resonated with customers. We announced our road map last December at the HPE Discover in Barcelona. You will see new announcements here in 2 weeks. Our ability to support Aruba switches with Mist and cross-pollinating 2 platforms. But ultimately, that AI-driven experience is resonating, and I mentioned one of the customers, as an example, taking advantage of that.

When I think about data center networking, we grew 20% in enterprise. That obviously is driving synergies with the rest of the portfolio because now we have a full conversation with customers across server, storage and networking. Now we are introducing in the fall, the new switch with Helios stack reference architecture. That's the first time to market tomahawk 6, 1.6 terabits. That will be a tailwind as 2027 start adopting the footprint in large service providers.

And then obviously, scale across, the PTX platform is, I will say, the reference when it comes down for data center interconnect and that PTX10000 and 12000 is resonating to drive data center interconnect. So is fairly balanced, I will say, and we are early, early in the process. Obviously, you have to win the reference architecture and kind of the discussion with the customers at that level.

But then it's going to be synergies across the rest of the portfolio with compute. So I feel very good about the momentum and kudos to the team who has executed flawlessly. And when you think about an integration of this scale done as so fast because if you put it in perspective, we integrate the R&D teams, we announced the road map. We didn't miss a beat on network and innovation. We integrated the sales force in January, and we're ahead of the integration milestones of synergy.

So I think this is a reference for how to do large acquisitions in the market. So Marie, what do you want to...

Wamsi, I'll answer your question on the strong free cash flow guide to '27. And it's pretty simple. It's really just based around the expectations that we have of the operating profit growth. And obviously, that supports higher profitability, and that's translating, frankly, into cash flow. And just one other thing just to bear in mind our '26 baseline, actually, has the charges associated with the Juniper synergies, which, obviously, you're not going to see that level of magnitude repeated in '27. So that's also one of the benefits that's playing as well into the free cash flow guide. Just to close, as I said in my prepared remarks, that will get us to 2x leverage by the end of '26. And I think I mentioned that we will pull in our share repo now into early '27 as well. So I just want to sort of close with that comment.

Please limit questions to 1 per analyst.

The next question will come from Amit Daryanani with Evercore.

I can't ask a multipart question after Paul just called us out on that, I guess, now. So maybe I'll stick to one. And Antonio, this may sound like a bit of a silly question. But given these numbers are so strong, especially when I look at the fiscal '27 guide, can you just talk about what's the bigger gating factor to growth as you go forward? Is it customer demand? Or is it more component availability? And I'm really trying to understand whether the outlook reflects the level of demand you're actually seeing today? Or is there actually additional demand that could be served if component supply and availability becomes a bit more easier. Just love to understand just on the component side, what's going on.