Meta Platforms (Facebook) Aktie

Good afternoon. My name is Krista and I will be your conference operator today. At this time, I would like to welcome everyone to Meta's Third Quarter 2026 Earnings Conference Call. [Operator Instructions] And this call will be recorded. Thank you very much. Rob Bhavnani, Meta's Director of Investor Relations, you may begin.

Thank you. Good afternoon, and welcome to Meta Platform's Third Quarter 2026 Earnings Conference Call. Joining me today to discuss our results are Mark Zuckerberg, CEO; and Susan Lee, CFO. Our remarks today will include forward-looking statements, which are based on assumptions as of today. Actual results may differ materially as a result of various factors, including those set forth in today's earnings press release and in our quarterly report on Form 10-Q filed with the SEC.

We undertake no obligation to update any forward-looking statement. During this call, we will present both GAAP and certain non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in today's earnings press release. The earnings press release and an accompanying investor presentation are available on our website at investor.atmeta.com.

And now I'd like to turn the call over to Mark.

All right, everyone. Thanks for joining today. This has been a busy and exciting few months. Our family of apps is now used by 3.64 billion people each day, and we are seeing all-time highs and engagement on both Facebook and Instagram, driven by strong growth in video. Threads has now overtaken X in monthly actives and continues to grow quickly. WhatsApp is hitting all-time highs in messaging and is becoming an increasingly important platform for agents as well. In the last few months, we have had several major milestones towards our goal of delivering personal super intelligence to everyone.

We had one of our most successful launches ever with the Muse personal agent. We shipped Muse Wonder as our first frontier model, and we shipped several new glasses and gadgets with Muse integration, including coming soon to our new audio glasses. We've been discussing our vision for personal super intelligence for more than a year now. We want to help every person have an extremely capable personal agent that understands your goals and can help you achieve them and improve your life in whatever way you want whether that's improving your relationships, helping you save time, save money, build a business, create new things or anything else that you can dream of. Muse is the product form of this vision. So now we have the foundational technology in place to deliver this. There is still a lot of work ahead, but our initial position is very strong.

In just 1.5 months, more than 10 million people in the U.S. and Canada have set up used personal agents. Growth continues to be strong, and we have been #1 in the app store almost every day. Engagement and retention are also off to a strong start. About 25% of people who have set up Muse, use it on any given day. For our earliest adopters, the longer that they use news, the better the retention seems to get. So to put this in perspective, this is more than twice the level that threads started at or where we believe that other AI apps started at as well. The most common use case for Muse is shopping, with more than 1/4 of people who are using Muse engaging in this with an average purchase size of over $200. Plaid told us that they hadn't seen activity like they're seeing with their connector on any other service before. And in the first week, we had to work with Stripe to raise credit limits since the volumes were above what any of us expected.

Beyond shopping, the most common connectors are for personal productivity and connecting Muse to your other services like e-mail and calendar. Now this shows the importance of the Muse secure VM that we've built to provide state-of-the-art privacy and security and we are looking forward to following up soon with the Muse confidential VM, which will make it that even Meta cannot see your content. We've also seen that Muse Wonder improves quality, engagement and retention further which suggests that we continue to improve intelligence, Muse will perform better and better over time. Now that Muse has clear product market fit, we are entering a new phase and a more operational phase that is focused on scaling it. The goals for this next phase play to many of Meta's strengths. We need to do 3 things: scale the product, grow revenue and optimize compute costs.

We expect to grow Muse to serve billions of people. This is both our mission and our strategy. It's our mission in that we believe in putting the power of personal super intelligence in everyone's hands. We believe that, that is the path to ensuring a positive future for everyone with individual empowerment as the source of prosperity and balance of power is the foundation for safety. On a values level, we believe that delivering this capability widely is the best thing that we can be doing. This is also our strategy because we think that billions of Muses in a network at scale will be able to get more done for people and businesses by working together. So we'd like to create a large-scale use network as quickly as possible.

From a business perspective, the average revenue per American user for our existing Muse users, is -- sorry, for our existing family of apps users is several hundreds of dollars annually, and we expect that Muse will reach an even higher level over time due to the amount of commerce that we already see taking place. This will take some time, but we're starting to work on this now. Our business model will have a few components. For most people, the product should be free, and we will build commerce tools, so we get paid a fee based on the transactions that we help facilitate. We expect this will be paid for by businesses. For people who want the most advanced models or to use extra tokens, they subscribe directly and pay for business owners who want additional features who use Muse to grow their businesses or for companies that want people to use Muse for work, they will be able to subscribe directly and pay as well.

And for those who want us to help improve the personalization and quality of their content and ads across meta, they will be able to connect to those experiences as well. Just to clarify, the average revenue that I was talking about before is the family of apps revenue for our existing Muse users. On the compute side, we are building out the capacity to serve more than 1 billion people over the next few years. In the near term, Muse is very compute-intensive to run, but we expect to improve its cost efficiency by more than 10x in the next year, which will enable it to be profitable once the revenue lines are at sale. If compute efficiency continues improving quickly for another few years, which we think is highly likely, then Muse will be highly profitable over multiple years. In the near term, the strength of our core business enables us to scale the Muse network before then, and we believe that we are the best positioned company to pursue this strategy.

One other note on compute economics is that since training is a large fixed cost, now that we have leading models, we want to serve as much high-margin inference demand as possible. That is why we are building the Meta enterprise platform led by CJ Desai. Muse Wonder and even more advanced models that we're training can serve additional high-margin services beyond the muse network, including our API, Muse code and other enterprise services. A fundamental feature of our business is that we have many good uses for this compute. It is difficult to forecast the exact growth rates for the Muse network for our enterprise platform and for how we Muse Wonder and beyond will accelerate our core business. But through building up multiple ways to deploy these advanced models, we increase the chances that one or more of them significantly outperforms our expectations.

Beyond serving other businesses, we have not yet found the limit of how effectively we can apply compute to accelerating our own core business. We have implemented an eval with Muse Wonder to improve how effectively it can autonomously improve our recommendations and ad systems. And as our upcoming models get even more advanced, I expect that we're going to see an increasing rate of improvement in these systems. All right. Let's talk about the models now. Muse Wonder is state-of-the-art across many dimensions that matter for personal super intelligence. This validates the importance of training models ourselves to deliver the best experience. In addition to Muse, we have also seen the engagement per person in Meta AI has more than doubled since April when we started rolling out our new models. While Muse Wonder has been our most important release, I have been very pleased with our pace of execution here.

In the last few months, we've shipped Muse Spark 1.2 and 1.3 Muse code, 2 open weights models with Muse Glimmer and Muse Spark Open and also Muse Voice. Going forward, with the field advancing quickly, we plan to focus on both pushing the frontier of intelligence and distilling those models into much smaller and more efficient product models that will be optimized for personal agents. My view is that we will be able to build super intelligence through scaling. So we are methodically working our way up the ladder there. We are already working on larger models after Muse Wonder, and they're going to be very impressive and capable. We are also working on Muse smaller models that are coming out soon that will deliver Muse Wonder level intelligence more efficiently.

Moving on. Our work on personal super intelligence is now also the focus of reality labs, Muse is coming to mediglasses starting next month, and that will be an important step towards this. People already love many features of the glasses already, music, calling, capturing moments, hearing assistance and more. But over time, we expect muse and having your personal super intelligence with you throughout the day will be the most important part of the experience. Muse is also central to the business model for the glasses. Again, our mission and strategy are to serve everyone here. We believe in putting power of this technology in everyone's hands. So we have priced our devices not to make a large profit on the hardware, but to build a large software and super intelligence business on top of them over time. Glasses sales continue to grow rapidly. And now with Muse, we expect that paid subscriptions will increase and that this business will begin improving meaningfully from here.

I also shared at Connect that we are changing our approach to virtual reality to only ship new devices when we have major breakthroughs like we had with the Meta VR glasses this year, which, of course, are the first VR experience ever in a pair of glasses. I continue to believe that the path to delivering holographic experience of bringing together the physical and digital worlds and everyday glasses will come through VR first. But we will pace this aspect of our investment more to focus on delivering personal super intelligence first. So with these factors combined, we expect that the P&L of Reality Labs to improve from here with operating loss coming down to around $15 billion next year and trending towards profitability from there.

Before we conclude, I want to spend a moment on the shear velocity that we are seeing. An important part of our approach has always been to move fast so we can learn quickly and serve people better at scale. But recently, things have accelerated. In the last year, we shipped edits and it has already passed 50 million actives, new apps like Forum and seller were built by small teams and are scaling quickly to serve power users. Muse went from an idea on a whiteboard to a successful product in just about 8 months. And we're about to ship used churn, which will have gone from concept to a hardware device in people's pockets in less than a year. I am proud of what we are building. People have all kinds of different opinions about the future, but at Meta, we believe that the future is for everyone and that putting power in people's hands is how we get there. I think that this matters. There needs to be a company out there that represents these ideas. But this isn't just about ideas, it's also about execution. So I am grateful for everyone who is working so hard to deliver all of this to build and people around the world. And as always, thank you to everyone who is on this journey with us.

And now here is Susan.

Thanks, Mark, and good afternoon, everyone. Let's begin with our segment results. All comparisons are on a year-over-year basis, unless otherwise noted. Q3 total Family of Apps revenue was $64.2 billion, up 26%. Q3 Family of Apps ad revenue was [ $63.1 billion ], up 26% on both a reported and constant currency basis. In Q3, the total number of ad impressions served across our services increased 10%. Impression growth was healthy across all regions, driven by growth in engagement and users as well as ad load optimizations. The global average price per ad increased 14% year-over-year driven by ad performance gains and improvements in macro conditions relative to Q3 of last year. This was partially offset by strong impression growth, particularly from lower monetizing surfaces and regions. Family of Apps Other revenue was $1.1 billion and grew 59%, driven primarily by WhatsApp paid messaging and subscriptions revenue. Within our Realty Lab segment, Q3 revenue was $443 million, down 6% due to lower Quest headset sales, which were partially offset by continued strong growth in AI glasses revenue.

Moving now to our consolidated results. Q3 total revenue was $64.6 billion, up 26% on both a reported and constant currency basis. Q3 total expenses were $51.6 billion, up 68% compared to last year and included a $9.9 billion charge related to the August 2026 multistate attorneys general agreement. Year-over-year growth was primarily driven by legal-related costs, infrastructure costs and third-party AI token costs. The growth in infrastructure cost was driven by higher third-party cloud spend, depreciation and data center operating costs. We ended Q3 with over 66,000 employees, down 12% from Q2 following the May 2026 head count reduction.

Third quarter GAAP operating income was $13 billion, representing a 37% decline year-over-year and a 20% operating margin. Excluding the Q3 legal charges related to the aforementioned legal settlement, our third quarter operating income would have increased 11% year-over-year. Our tax rate for the quarter was 24% and above our outlook range, primarily driven by the nondeductible portion of legal expenses. Net income was $9.9 billion or $3.83 per share. Capital expenditures, including principal payments on finance leases were $34.9 billion, driven by investments in servers, data centers and network infrastructure. Free cash flow was negative $2.1 billion. We ended the quarter with $82.4 billion in cash and marketable securities and $83.7 billion in debt.

Turning now to the business performance. There are 2 primary factors that drive our revenue performance, our ability to deliver engaging experiences for our community and our effectiveness at monetizing that engagement over time. On the first, we continue to see significant gains from our content recommendation initiatives. On Instagram, global time spent per user reached an all-time high and grew double digits year-over-year in Q3 with strength across both reels and feed. On Facebook, Reels' time spent reached all-time highs, both globally and in the U.S. with both growing double digits year-over-year. Threads also continued to scale with time spent per user growing over 20% year-over-year. Engagement gains across our services continue to be driven by improvements we're making to the relevance, personalization and highliness of the content we recommend. Investments into our recommendations engine across retrieval and ranking have continued to yield positive results on Instagram and Facebook. For instance, on Instagram Reels, our Ray-Ban's launches in Q3 increased Reels time spend over driven in part by our upgraded model architecture that ranks sequences of Reels at a time instead of evaluating each reel independent as well as improvements we made to add wait or watch time signals.

On Facebook Feed and Reels, we completed the global launch of a new generative retrieval architecture, which inserted by LLM labels pieces of content with Symantec representations and generates potential recommendations helping our systems surface relevant new posts before they've accumulated significant engagement. This new architecture is now rolled out to roughly 1/4 of unconnected feed views and half of Reels' views and increased Reels watch time by 1%. Our recommendations work is also helping us deliver fresher and more personalized content. Nearly 45% of recommended Instagram Reels are now less than 1 day old, up more than 11 percentage points from Q2. On Facebook Reels, we doubled the amount of history our ranking model can consider, driving a 1% increase in Reels watch time in the U.S.

LLMs are also helping us understand content more deeply. We continue to process public Instagram posts each day through LLM based content understanding systems. These signals have improved the engagement with Instagram search results by 1% through the end of this quarter. This quarter, we also began moving our content understanding models onto our first-party muse model family. In off-line evaluations, these models are outperforming our prior production models and are reducing per unit content processing costs by more than 80% across Instagram Threads, Facebook and ads. Beyond improving quality, owning more of stack gives us greater flexibility to develop models around the signals that matter the most for personalization.

Turning to the second driver of our revenue performance, increasing monetization efficiency. We see significant opportunity to continue improving performance for the businesses who use our services. Within our ad systems, we are continuing to drive performance gains as we deploy more predictive models that understand users more deeply and deliver ad experiences better aligned with their unique preferences while also making our infrastructure more efficient. Last quarter, we introduced meta generative Recommender, which uses LLM to reason across ad content and user preferences to predict the best ad for each person.

In Q3, we launched this approach to generative retrieval on Facebook Feed, driving a nearly 2% increase in conversion across targeted segments. In addition, we expanded our GEM model for ads ranking to additional services -- surfaces, driving a 1.7% increase in conversions for off-site conversion campaigns and a 9.6% increase in app ad conversions. And we continue to scale our retrieval system efficiently. In Q3, improvements to Andromeda such as scaling model complexity and throughput increased conversions by 1.4% on Facebook. We're also leveraging AI to make it easier for businesses to manage campaigns develop creative and act on performance opportunities.

Meta AI business assistant is becoming a more proactive and capable partner for advertisers. We estimate that advertisers who adopted its recommendations drove almost 5% more conversions at the same cost per conversion and increased their advertising investment by 9% compared with similar non-adopters. We are rolling out additional capabilities, including creative insights, stronger business context and more agentic actions. These features will allow marketers to plan, create and optimize campaigns as well as scheduled tasks and alerts through a conversational experience with approvals and previews that keep them in control. I'll share one example of how Meta AI Business Assistant is empowering our advertisers, simply organic maple, a family-run maple syrup producer in Vermont used Meta AI business Assistant to improve its campaign messaging. After the business assistant recommended testing longer, more descriptive advertising copy, the business saw a 57% increase in purchases and a 36% reduction in cost per purchase. Its strongest revised ad delivered a 3x return on ad spend, a 52% improvement from the original.

Adoption of our generative AI ad creative tools also continues to scale. In Q3, nearly 10 million advertisers used at least one of these tools. Advertiser adoption of our video generation features grew 25% quarter-over-quarter, driven in part by ready-made videos that reduced generation time to just a few seconds from a single image upload. We're also expanding the opportunity at the intersection of creators and advertising. Partnership ads, which allow advertisers to run ads with creators, brands and other businesses continue to scale rapidly in Q3. Our partnership ads product grew over 97% year-over-year and surpassed $15 billion in annual revenue run rate.

In Q3, we also completed the global rollout of Creator Marketing Hub to all advertisers, bringing creator discovery content organization and partnership ad activation into one place with more than 10 million creators now discoverable. We are integrating Meta AI business assistant into the hub. So advertising conversationally search for creators identify content to use in partnership adds and get guidance across their creator marketing workflows. Outside of ads, we're also building out additional ways to monetize engagement across our ecosystem. In September, we announced a global expansion of Meta One, our unified subscription service for consumers, creators and business. Meta One is rolling out gradually with more than 50 features across Instagram, Facebook and WhatsApp, including Meta Business Agent and Meta AI and has surpassed 15 million subscriptions and trials to date. The core experience across our apps and Meta AI will remain free, while subscriptions offer more specialized capabilities and expanded AI usage.

And beyond monetizing and giving across our existing services, we are building new experiences and revenue streams around consumer agents. Mark highlighted the impressive early momentum with Muse. Users on many reasons to use Muse with the top 3 categories being shopping, personal productivity and creativity. We're encouraged by the breadth of use cases with shopping in particular being used by over 25% of Muse actives. While our near-term priority is to grow adoption and expand Muse's reach, we see multiple paths to monetization beginning with subscriptions. We also see a significant opportunity to extend Muse to small businesses. Based on a recent user survey, we estimate that at least 500 million self-identified small businesses use Meta's family of apps to reach customers and generate revenue. Muse could help these businesses acquire customers and operate more efficiently while helping people complete transactions. Over time, consumer and business agents could connect and transact directly creating opportunities around paid usage, commerce and business services.

Along the growth, we are focused on building sustainable unit economics. Muse's compute intensity means that expanding its user base will require significant investment, and we expect total serving costs to rise. At the same time, we expect to improve compute cost efficiency by more than [ 10 ] over the next year, driven by beer utilization of off-peak capacity, more efficient models and inference optimization. As revenue scales, these efficiency gains give us confidence in our ability to build a highly profitable business over time.

Moving now to our financial outlook. We expect fourth quarter 2026 total revenue to be in the range of $73 billion to $76 billion. Our guidance assumes foreign currency is an approximately 1% win to year-over-year total revenue growth based on current exchange rates. Turning to the expense and CapEx outlook. We are raising our expense outlook to incorporate the $9.9 billion charge related to legal proceedings recognized in Q3. We now expect full year 2026 total expenses to be in the range of $175 billion to $179 billion. Excluding the Q3 legal charge, we continue to expect 2026 operating income to be above 2025 operating income. At a segment level, we now expect Realty Labs losses to decline compared to 2025 to a range of $18 billion to $19 billion.

We anticipate 2026 capital expenditures, including principal payments on finance leases to be in the range of $135 billion to $145 billion, narrowed from our prior outlook of $130 billion to $145 billion. Absent any changes to our tax landscape, we expect our fourth quarter 2026 tax rate to be between 15% and 17%. Additionally, we continue to monitor active legal and regulatory matters that could significantly impact our business and financial results. For example, we continue to be scrutiny on youth-related issues in several markets and have a number of youth-related trials scheduled in the coming months in the United States. Looking beyond 2026, we are still developing our budget and capacity plans for next year, but we expect to continue investing aggressively to support both our core business and emerging AI initiatives. Our current capacity and what is scheduled to come online gives us the flexibility to allocate resources across model training, core business growth and the development and support of new products.

Our AI investments are already delivering substantial incremental advertising revenue through stronger engagement, better recommendations and more efficient monetization. Building on this progress, our MS investments are creating the technology and infrastructure needed to unlock significant new revenue opportunities and strengthen capabilities across the business. While the investments will take longer to generate returns than those in our core business, the breadth of monetization opportunities and internal efficiencies gives us confidence in their long-term potential. Looking ahead, although we are not providing a full year 2027 revenue outlook, the benefits our AI investments are delivering to the core business, combined with emerging opportunities for model advances and product launches, position us to sustain strong overall revenue growth throughout 2027.

Separately, we remain focused on improving Reality Labs profitability. As Mark mentioned, we expect losses to narrow next year to around $15 billion as we reallocate investment towards other priority areas, primarily AI. Finally, today, we announced plans to create a new class of nonvoting capital stock, which will be distributed to all existing shareholders through a stock dividend. For each existing share held stockholders will receive one share of the new nonvoting stock, effectively resulting in a 2-for-1 stock split. We expect trading to begin in late January 2027.

In closing, our business momentum continued in Q3 with strong execution across our core ads and engagement initiatives, and we're also excited by our progress towards bringing personal super intelligence to everyone.

With that, Krista, let's open up the call for questions.

[Operator Instructions] And your first question comes from the line of Brian Nowak with Morgan Stanley.

I have 2, one on Muse, one on enterprise. Maybe congrats on the success of Muse. But as you analyze the user behavior and the product, can you talk with us a little bit about any of the constraints to the slope of forward user growth that you're most focused on addressing, whether it's capabilities, model efficiency, capacity, et cetera, just so we understand what are the limiting factors to forward user growth that you want to address? And the second one is on enterprise. Mark, can you just walk us through your vision for the Meta enterprise product offering and the key value drivers you hope to provide for the enterprise adopting the service?

Yes, I think I can talk about both of those, and then Susan can jump in with anything else. All right. So right now for us, we are -- and one of the things I pointed out is we have more than 10 million people in the U.S. and Canada, who have set it up, but I think one of the biggest constraints so far is that because this is a compute-intensive service to run, we have not launched it globally yet. That's partially because of wanting to optimize it for each place, and there's some work around connectors for that. But it's partially that -- that we want to make sure that we do this efficiently. So I mean, there's pros and cons there. The pro, of course, is that just makes it even more impressive that you have 10 million people using it and signing up to use it in the first 6 weeks because those are all kind of Americans and Canadians in a small number of markets. So if you actually did roll it out more broadly, then you would expect that number to be significantly larger than what we're seeing today.

But then on the flip side, we actually have to go do the work to bring the efficiency in line to make that make sense. And so that's a big part of it that we need to do. We -- the compute efficiency work is going to be important on this also because we want to make sure that we've built enough compute to be able to serve what we think is going to be 1 billion-plus people if we can achieve certain compute efficiency targets, but we obviously have a lot of work to do there as well, and that will be -- that will be one of the limiting factors over the next year. We want to build out the business around it. So I talked about that. And I think doing that in tandem with bringing down the compute -- or increasingly compute efficiency is going to be important for getting to work from a business perspective, which I'm optimistic about, but I think that, that's one of the things that we need to do.

And then there's just continuing to increase the value in the product, which I think we will largely do by continuing to develop new models. That's why I mentioned that we launched used first on Muse Spark, and that's what is currently the default across the free tier. And we rolled out Muse Wonder and people can access that in the paid tier today, and we know that it performs a lot better. So we're working on getting an efficient version of Muse Wonder that we'll be able to roll out more broadly. But you can imagine just as we iterate forward in 2, 3 models into the future, when you have significantly higher intelligence and capability to solve many more problems for people and the ability to deliver that at a much more affordable price point, I think that's going to just unlock a lot more value to people. So we're very excited about that.

Transitioning to talking about the enterprise piece because I think these 2 pieces, there is some connection here in that the main enterprise offering in the near term is going to be the model API and Muse code, we -- business agents that we've been scaling for a while. And each of those 3 things is growing quite quickly. So we're excited about that. Over time, we would like to make it so that we have a whole kind of business network of Muse agents as well. And there's the business agent, which we've talked about on previous earnings calls for businesses to help out with sales and customer support. But what we see is a lot of -- a lot of people are actually using Muse today to help run especially small businesses. And we think that there's both an opportunity to help create an agent that people use to really enable them to automate and do a lot of the parts of spinning up a small business that are just kind of difficult and time consuming, but aren't the core thing that a person is trying to do is we think that's a big deal.

And then making a version of Muse that people can bring into different corporate environments and work there, we think it's going to be a big deal. And then beyond that, building what I refer to as the Muse network, which is making it set Muse as not just a kind of single player game where you have your Muse, but making it so that your Muse can work with your friends and your families and if you want to buy something, it's not just navigating to that business' website. It's talking to their agent. I think that getting that all to work well together is going to unlock tremendous value at scale. So I'm very excited about all of that. I think the enterprise use is going to be a big part of it, but it's also going to kind of have this very valuable boost in the utility for consumers, too. So I'm very excited about both of those.

Comments on enterprise. AI is creating huge amounts of disruption in enterprise today. And we don't -- from what we can see, we really believe that traditional enterprise companies have this locked down. And in fact, some of the leading companies in this space today didn't exist a few years ago. Those companies made large investments in model development, and we've done the same, and we have increasingly competitive models to bring to bear for the benefit of all of our users, including small businesses and including enterprises. And already today, we have 500 million businesses using our products. So we really believe that AI is going to reshape how businesses operate. We expect that businesses of all sizes are going to use AI agents across software development and productivity, sales, marketing, consumer customer service, et cetera. .

We have differentiated assets to serve this market, bringing together models, infrastructure, global distribution and long-standing relationships with businesses, and it's our hope that this will be able to be a new diversified recurring revenue stream for us that we'll be able to support multiple business models, including subscriptions and usage-based API and messaging and other revenue lines. So we expect that this will be an opportunity that will develop over multiple years, and we intend to scale our investment behind demonstrated customer demand and product market fit in this area.

Our next question comes from the line of Eric Sheridan with Goldman Sachs.

Maybe just the first one, building on Brian's question. When you think about the friction points around the genetics, how do you guys going to think about solving for friction that's rising where there are aggregators or suppliers that don't want agents to interact with their platforms or their services? And how you think about lowering that friction or creating the right balance between the existing services that consumers access today relative to how this might all be changed or altered by Agentic behavior over the medium to long term.

And the second one would be with the announcement of the new class of shares tonight with earnings. Any updated thoughts broadly about how you think about sources of capital and optimizing your capital structure to make sure you invest the proper amounts to capitalize on the opportunities that AI are presenting in the next couple of years?

I can answer the first one. So -- okay. So the question -- sorry, what was the first question there?

Friction points around Agentic Commerce.

Got it. All right. Yes. So the -- right now, what we're seeing is that overwhelmingly businesses are very excited about enabling this because they recognize that it's going to be a new opportunity for them not only to increase the sales and the quality of the experience for existing customers, but also potentially to bring in new customers. So now I mean across the whole business ecosystem, different people will kind of adopt this at different paces and how our strategy evolves on that to develop the ecosystem will depend on how that goes. But overall, I would say we've been very positively surprised by how quickly I think a huge amount of businesses have wanted to not only figure out how to make it so that Muse can use their services more effectively, but have reached out to us to integrate around it. So I view that as a very good sign. And obviously, not everyone wants to be an early adopter, that's fine. We'll work with folks on the terms that makes sense for them. But I do think that over time, this is going to be quite valuable for everyone to make this work well, and I think we're on a good trajectory on that so far.

Yes. I would add to that, and I will also answer your second question and just say, I think agents are going to be an important part of how reality evolves, and I think everyone is going to have to adapt to what that means, the new age of Agentic commerce. We obviously have had a lot of businesses we're very excited to partner with us. There are also businesses that are understandably focused on protecting customer relationships. And we expect the market, I think, ultimately to evolve away from agents attempting to navigate every digital experience like a human. And I think over the long run, this will move towards more permissioned structured interactions and that will include authenticated APIs, approved agent partnerships or direct communications between the consumer's personal agent and a business's own agent. And we're helping establish the standards needed to support that model.

Earlier this month, we previewed the personal agent protocol and standard across Muse and Meta business agents that we are developing with industry partners like Sierra, Walmart Stripe, Shopify, et cetera. The protocol is intended to define how a consumer's personal agent can interact securely and reliably with the business's agent and preserve appropriate controls for both the business and customers. So again, I think adoption will likely be uneven and will take time. And in the near term, we expect any experience -- many experiences will remain AI-assisted rather than fully autonomous. But as freshed integrations develop and agents demonstrate that they can generate incremental value and demand and operate safely. We expect that more and more businesses will be excited to participate.

Your second question was about the announcement of the new class of shares. So I should say, we created this new class of shares with the purpose of creating a publicly traded nonvoting stock usable for capital raising employee compensation, acquisitions, other general corporate purposes, while preserving the voting power of existing Class A and B shareholders. We believe this share class will maintain our founder-led structure. It will expand our funding tools and give us flexibility to invest behind the unprecedented momentum we're seeing across every part of our business, driven by AI. We don't have any specific financing plans to announce at the time, but we thought it was prudent to establish the appropriate capital structure framework to ensure that we have access to a diverse pool of funding sources.

Your next question comes from the line of Doug Anmuth with JPMorgan.

I have two. First, I just wanted to follow up on the your last comment so just around as Muse evolves and more businesses adopt agents, just wanted to get your sense of how things kind of change and shift as you go to more permission-based and agent to agent transactions. And like what does that really unlock over time? And then secondly, just on kind of expenses and CapEx, thinking about '27. I know you're not really there yet. But if I go back to some previous comments, you've talked about maximizing compute in '26 and '27 and then having more flexibility in '28. A year ago, you talked about kind of notably larger dollar increase in '26 than '25, and then you've also used kind of growing operating income on an annual basis as well. So I'm just curious if there's any of those that you can kind of tie or to help us think about '27.

Doug, so on your first question about how the Agentic ecosystem evolves. I wish I could sort of predict that more perfectly. And I think it will take some time. And again, I think it will be a little uneven. But I definitely think that when I use Muse today, it's clear that there are a lot of parts of the experience that would be even better if Muse was able to interact more seamlessly with business agents on the other side. For example, when if Muse has to call business on my behalf, I think it would be easier for the business if that call was handled by business agents that were representing it. And I think also one of the things that is really distinctive about us and one of the things that makes the product experience so great is the degree of proactivity that exhibits on my behalf, right? I can give it a task and Muse is going to go execute against that task for me even as it encounter some obstacles or can't quite find the right thing and needs to chart a new path to helping me for the thing I want. And so I think the -- that sort of dimension of proactivity exhibited by either consumer agents or business agents, I think, are going to enable more transactions to happen or more activity to happen broadly that happens today where -- when things are when more actions are taken, primarily when like the latent demand expresses itself versus things where I can have Muse searching for things for me on an ongoing basis, giving me regular updates on things I'm looking for, sending me 10 new books months and agent should be doing similar things for businesses, and I think that creates sort of opportunities again for a new more proactive sort of framework for what agent to agent interactions look like over time.

On your second question, which is about 2027, we are quite early in developing our budget and capacity plans for next year. So we are, I would say, at the point where we are both investing -- sorry, evaluating a lot of investment proposals that have attractive return profiles over different time horizons. And we want to figure out how we can fund as many of those as possible within the construct of our overall budget and financial goals. But it's very dynamic and it's still pretty early. And so we don't have anything more specific or quantitative to share until we get to the Q4 call, which is typically when we share guidance for the next year.

Your next question comes from the line of Justin Post with Bank of America.

Great. One for Mark, one for Susan. Mark, you've got this powerful new model on your hands, and it seems like there's a lot you could do internally, besides externally. But specifically to the ad business, you rearchitected your systems around AI. How much room is there for improvement as you deploy that new model? How are you thinking about that? And then for Susan, you mentioned strong revenue growth next year. Just wondering if that's advertising or does that incorporate some incremental AI revenues hitting next year?

I'm happy to take both of those. Great. On your first question, I think our ads business has been benefiting for now many years from our ongoing investments in both kind of what we would have called traditional AI machine learning models. And then, of course, from the sort of more recent generation of models. I would say that there are 3 themes there that I think are worth outlining. The first is just the kind of traditional road map of performance improvements that we have in the pipeline based -- that is what I would describe more as business as usual machine learning work. That continues to be a robust and healthy pipeline. And as we are looking into 2027, again, we're not providing an outlook, but I would say that our investments and product improvements across both engagement and ads initiatives give us confidence that we have meaningful runway to grow our core business throughout 2027, that also gets to your second question also. But I should say, basically, I think the ranking and recommendations pipeline is healthy.

The second theme I would say is that we are excited about some of the new research bets that we have as we move into sort of a more LLM oriented AI infrastructure to help us more deeply reason about people's preferences and what they're interested in both the cross organic content and recommendations, and we expect that we'll be able to unlock -- we will be able to unlock benefits to our ranking recommendations engine from that research to. And then the third thing is that we expect that generative AI tools, which have already very valuable for advertisers will continue to grow in terms of the features that we make available to advertisers, primarily through the Advantage Plus suite, but features that make it easier for advertisers to quickly and seamlessly generate, whether it is creative or take an image and generate video or generate sound, all, I think, in service of making the ads increasingly personalized for the person that they're being served to. So I would say we feel like there is meaningful runway across all 3 of those things.

Your second question was about revenue growth next year. So as I mentioned, I think we're both feeling good about the budget investments that we expect we will make in both core ads and engagement initiatives next year. We're also building several new AI businesses, and we expect that those can begin contributing incremental revenue in 2027. But more importantly, can scale meaningfully over time. And those will include monetizing our AI products through subscriptions, transactions, advertising, developing enterprise offerings as we talked about in one of the earlier questions. So I would say while many of these opportunities remain early, and our near-term focus is really on building high-quality products and achieving product market fit and scaled product market fit. I think our track record of reaching billions of people and building businesses at scale makes us feel bad that AI is going to become a meaningful source of revenue diversification over time.

Your next question comes from the line of Mark Shmulik with Bernstein.

Mark, coming out of cash flow like we were ushering in the next era of the company and the business reorienting around kind of Muse at the center of everything. Is that a fair assessment of kind of how things are going internally? And if so, can you just talk a little bit more about how some of the business mandates and outcomes may have changed? And then Susan, it was interesting to kind of see the launch cadence where enterprise announcements kind of preceded some of the small business ones. And I appreciate your comments on some of what's going on there around the SMB ages opportunity. But can you just share more details on kind of the work required to serve and scale the offerings to SMBs?

Yes. I mean I can talk briefly about the first one. I mean, look, it was more than a year ago that I think I wrote the letter about personal Super Intelligence that's the North Star for the work that we're doing here. And it basically -- it fits with this line of work that we've done for more than 20 years at the company is focused on empowering individuals, giving people a voice, giving people the tools to connect in the ways that they want. And when you extend that to the work around AI now, we think that the most important thing that we can be building for this next generation of technology is giving -- is basically putting the power of this technology in everyone's hands, which is the personal super intelligence vision.

So we've been working on this for a while, and this is when we did the reboot around MSL and building the lab, a lot of it was this idea that we need to build models specifically to be good as personal agents, that we'd be able to build the best experience by owning and building and designing ourselves the end-to-end technology that includes the models and then the agents. And this has been this has kind of been in the works for a while. I mean, the vision on the glasses has also been this. I mean the -- we've wanted to build the next generation of hardware platforms so that we can deliver the best end-to-end experiences to people. And I think that, that's starting to come together as well, and I think we'll even more over the coming years. I mean these are -- these are long-term bets. I feel like this is what we're here to do is make long-term technology best and design the best experiences end to end that we can for billions of people around the world. And I think as long as we continue to do that, then over time the business takes care of it as long as we continue executing well, which we're really proud of the team for how we're doing on that.

Mark, your second question was around use for small business. So we launched -- we launched that, which you can think of as an expansion from Muse in late September and rather than a separate product, it's kind of a new -- a set of new skills and connectors inside the existing Muse app aimed at helping SMB owners run operations, find new customers that connect to businesses, for example, Instagram account analytics or Facebook pages or Meta ad accounts and a few clicks, along with dozens of third-party connectors that business is already run on like Shopify, QuickBooks, Canvas, et cetera. It already understand what businesses sell, it's brand voice, what customers are asking for and Muse can take actions, including analyze ad and content performance, draft campaigns, customer responses, flag issues, and we are actively listening to our SMB users to understand what their top use cases are.

We are seeing them use it for ads optimization for direct message organizing, use it to expand their sales footprint to keep their website current based on things like inventory and new products to deal with employee load balancing. So we're pretty closely following how businesses use news to make sure that we are being thoughtful about new features that we release in Muse to make it maximally useful and productive for SMBs. I'd also say we have unique touch points to, again, I mentioned in my opening remarks, 500 million self-identified small businesses. And we think that many of the investments that we make in consumer products are going to carry over to help power those businesses also. We have multiple excellent channels to surface those opportunities to small businesses globally. So this is a place where we feel like we have an existing extremely strong platform, and we're very excited about it. And we also believe the transition to AI has has broadly unlocked enterprise is a new opportunity for us as well.

Our next question comes from the line of Ross Sandler with Barclays.

Great. Mark, you touched on this earlier, but could you elaborate a little bit on maybe some high-level metrics around the user cost structure of us today? And then how you kind of see that cost are coming down as you go forward? And what are some of the things you're doing to bring that cost curve down?

I'm happy to answer that. And Mark, you should add anything that you feel like I'm missing. We are -- we don't have sort of specific numbers to share around the cost structure of Muse today. But of course, it is, as Mark said, pretty compute expensive and making sure that we are focused on compute efficiency is going to be one of the key vectors towards being able to both scale and use and grow it into the business that we want. I think there are a couple of dimensions of achieving that efficiency. I think some of the more near-term pieces include things like leveraging off-peak or spare computing capacity to run some sets of jobs using smaller models, there's a lot of work to be done across different inference efficiency initiatives, like all in service effectively of higher GPU utilization and other product efficiency work we can do also.

And I think over the longer term, I think this is a place where we expect that we're going to be able to, over time, as LLM scale, get more tasks done with fewer tokens and also rely on AI models themselves to help us improve the efficiency of the software that we use to run the model. So I think it's a place where we both feel like there is a lot more work on our efficiency road map that is going to help us bend the curve here. And we also expect that the sort of arc of how the of how the sort of AI-driven technological advances are going to apply here will also help us achieve larger savings over time.

Your next question comes from the line of Mark Mahaney with Evercore ISI.

Two questions. There's a bit of a playbook that there has been at Meta over the years when you surface new surfaces and you build up the user growth, engagement, then you layer in monetization. Do you think it's sort of a similar playbook here? Or does it have to be a little bit faster with Muse just given that the costs associated with it? Are materially greater arguably than launching those surfaces in the past? And then just a clarification on that commerce data point that you provided about 25% of users and the $200 or greater ASP. Are you including in commerce travel purchases? Or is that just a retail data point?

My guess, I think that we are going to work on monetization somewhat sooner on this than our normal playbook for a couple of reasons. One is just the size of the numbers that we're talking about in terms of both the costs and the opportunity, I think, make it kind of important to make progress on that in order to be able to scale it. The other is just that such a large part of what people are doing is commerce.

And to your question that does include travel to. I think we might consider travel a different case, but travel is actually one of the other top use cases for Muse is basically people planning that out and planning different kinds of things that they would get that they might get through our family of apps. So I think that this is going to be an important thing for us to work on to deliver the consumer value that people want. And in doing so, I think we will be able to build a model where we get a percent of the value that we are helping to create for people. So that's going to be -- it's going to be very important to kind of get the math on all of this to work out that way we can continue the compute build-out that we want to build in order to get this to reach everyone at scale.

And in addition to everything that Susan said in response to the last answer, I would also just add that, like we just see this trend that year-over-year, I just expect that we will be able to deliver the same intelligence that we have now in Muse Wonder or frankly, any of the current frontier models at a fraction of the cost in a smaller model a year from today. So to the extent that, that represents some kind of baseline level that people seem to -- that kind of is sufficient for product market fit for us, I think that we will just continue through kind of continued model development, make progress on that efficiency. The thing that is -- and of course, less clear is that I don't think in a year from now, most people are going to want to use the current model level of intelligence because that's also going to make progress.

So I think we're going to need to work on all of these things, both the revenue pieces of it and scaling the different aspects of that as well as progressing on the model development and all the efficiency in order to be able to both fund and support the compute build-out and reach billions of people, which, again, we want to get to that as soon as possible. Part of the reason why we're not just charging everyone upfront to use this is we want to establish this muse network, and we think that it's going to be more valuable for people at scale when there are a lot of people on businesses using Muse. So that way, they can collaborate and work together. I think that that's really a very important state to get to and one that we're quite excited about. And again, I don't think that there are many companies in the world that are well situated to be able to do that in the way that we are. I think we are very well situated to do that, and we plan to push on that quite aggressively.

And that does conclude our question-and-answer session. Mr. Bhavnani, I would like to turn the call back over to you.

Thank you for joining us today, and we look forward to speaking with you again soon.

And ladies and gentlemen, this does conclude today's conference call. Thank you for your participation, and you may now disconnect.

Good afternoon. My name is Christa, and I will be your conference operator today. At this time, I would like to welcome everyone to Meta's Second Quarter 2026 Earnings Conference Call. [Operator Instructions] And this call will be recorded. Thank you very much.

Chad Heaton, Meta's Vice President of Finance. You may begin.

Thank you. Good afternoon, and welcome to Meta Platform's Second Quarter 2026 Earnings Conference Call. Joining me today to discuss our results are Mark Zuckerberg, CEO; and Susan Li, CFO.

Our remarks today will include forward-looking statements, which are based on assumptions as of today. Actual results may differ materially as a result of various factors, including those set forth in today's earnings press release and in our quarterly report on Form 10-Q filed with the SEC. We undertake no obligation to update any forward-looking statement. During this call, we will present both GAAP and certain non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in today's earnings press release. The earnings press release and an accompanying investor presentation are available on our website at investor.atmeta.com.

And now I'd like to turn the call over to Mark.

All right. Hey, everyone. Thanks for joining today. We had a strong quarter for our community and business with 3.6 billion people using at least one of our apps each day. We reached several milestones. Instagram reached 2 billion daily actives. Threads crossed 500 million monthly actives, making it the fastest-growing conversation app ever. Facebook has reached more than 2 billion daily actives for a while now. WhatsApp just hit an all-time messaging record peaking at 30 million messages sent per second during the World Cup Final. And also, Kunal Shah just joined as our new Head of WhatsApp, he built one of India's most important payment companies and will be a great addition to the team. We also shipped some strong new models from MetaSuper Intelligence Labs and release new glasses.

Overall, the scale and reach of our community across our apps is pretty remarkable, and it gives us a strong platform for delivering new innovations to billions of people. The opportunity in front of us is massive. First, we are now at a point where our investments in AI are accelerating every major part of our core business. They're improving the experience for people using our apps, driving better performance for advertisers and helping our teams build new experiences and ship faster.

Second, we are developing new personal agents that will be the foundation for our next wave of products and revenue lines in the months and years ahead. And third, we see a large enterprise opportunity to sell to businesses, including APIs, business agents, potentially selling compute directly and other services that we're building for large customers. And I want to spend some time today laying out exactly how our investments are delivering results today and the opportunities that we see over time.

For accelerating our core business, we are seeing strong and promising results in several areas. In Instagram and Facebook, I'm very optimistic about our work to integrate large language models into our recommendation systems, LLMs add a first principles understanding of what the content is about and why it is compelling as well as a deeper understanding of what people are interested in and what their goals are when they're using our app. So this means that we can show more relevant and engaging content that better reflects people's goals and interests.

Our new Muse image and Muse video models will also dramatically expand the universe of content that people can discover across our platforms. There are already 2 large sets of content to draw from first from your friends and the people you follow and second from creators that you don't follow. But now there is going to be a whole new and nearly infinite universe of personalized content. This is going to make our services a lot more useful and engaging for people.

For ads, we are using LLMs to improve how our systems predict and rank the ads that we show. We've expanded the context that we can take into account around a person's organic and adds activity to determine adds relevance, driving significant increases in relevance and conversions on both Facebook and Instagram. And on a dollar basis, our ads business is reporting faster year-over-year revenue growth than any other company's reported ad business. So these AI investments are paying off.

We are also seeing a lot of demand for our new AI-powered creative tools. 9 million small businesses on our platforms are now using at least one of our AI ad creative tools, and we're rolling out new end-to-end creative solutions that help advertisers translate performance data into their creative decisions.

Muse image is going to supercharge this. The model can analyze images, improve its own work and produce better ad variations based on advertiser input. We're getting great feedback on this so far.

We also just launched Meta One, a new subscription offering that provides more tools and AI features across our apps. As demand grows, we're going to offer a variety of different tiers and pricing options there as well. And I'm also excited about how AI is helping our teams speed up product development. Earlier this year, we shipped Instagram Instance. We also just launched Forum, a stand-alone group's app and seller, a stand-alone marketplace app. And I expect it to become a lot easier to ship new apps. So we are planning to build out more ideas and use our recommendation systems to scale them to the people who will find them interesting as we've done with Threads.

It's been a little more than a year since we launched Meta Super Intelligence Labs and our trajectory is strong. In the last month, we shipped Muse Spark 1.1 and Muse Image. Since we rebuilt Meta AI and integrated Muse Spark, we have seen a 60% increase in the number of people interacting with the assistant each day, and that continues to grow quickly week-over-week. Muse Spark 1.1 is a strong agent encoding model that is very efficient and excels at computer use tool use and multimodal understanding. It's available through our new public API and we are ramping up distribution through partner channels and more coding agents over the coming weeks. We are also building out features to make it easier for enterprises to adopt to Muse Spark, and that is an area that we expect to continue focusing on.

One reason that we are so focused on making Muse Spark great at agentic capabilities is that we think that there's a very big opportunity to ship a few types of agents that are aligned with our mission and business. The first is personal agents. Soon, we will have agents that can work 24/7 on your behalf to help you achieve your goals and improve your life, your health, your relationships, your finances, whatever you want. The first domain that agents have really taken off in is coding, but engineers are more technical and willing to spend time making those agents work. So to build great personal agents, this needs to be a great consumer product that just works out of the box and is easy enough for billions of people to adopt and use. I'm very excited about this, and we're going to have more to share soon.

As we move towards a future where we're all interacting with multiple agents, again at WhatsApp and our other messaging services are going to become increasingly important. WhatsApp is already the leading surface where people engage with Meta AI and as we build out a platform for more agents across our messaging apps, we're going to innovate on how to deliver a private and secure AI experience. We launched Incognito mode this quarter on WhatsApp and the Meta AI app, allowing people to have private conversations with their assistant that even Meta can't see. We are planning to make strong privacy and security a fundamental part of the agents that we're building as well.

I'm also very excited about our progress with business agents. We made Meta business agents available globally this quarter on WhatsApp and Messenger, and there are already more than 1 million businesses using them to talk to their customers or complete sales every week. We're rolling business agents out on Instagram now, too. One interesting thing about having an agent talk to your customers every day is that it learns over time and can bring all of those insights back to you. So we're building more agentic capabilities to summarize all these conversations, digest what happened overnight and surface what customers are asking for.

And soon, it will go further including suggesting ways to grow your business, giving you competitive intelligence and real-time insights into what's working and what's not. And over time, we'd like to build this into a business-in-a-box service that can help you start and run a whole business using Meta's platforms.

In terms of how we will monetize these, we have a mix of subscriptions, volume-based pricing and I expect that we're going to continue to evolve more of these products to be like our ad systems where businesses only pay us when we achieve results for them. Over time, that will let us run an efficient auction over our compute similar to how we do that for advertisers today. As AI usage in our products and businesses continues to ramp, we continue to invest aggressively in infrastructure to meet the demand.

Yesterday, as part of our Meta compute effort, we announced a new strategic venture with BlackRock to develop a new 1 gigawatt data center in El Paso, Texas. Overall, we expect that a significant portion of our compute is going to go towards training our models, growing our core business and delivering personal agents and new products, but we also expect to grow a large business serving large customers as well. We've built our API. We're rolling out business agents. We're getting a lot of offers for compute at a significant premium over what we paid for it, and we have more coding and productivity tools on our road map as well, and we'll have more to share on all of this soon.

As we get closer to personal super intelligence, we are also going to need hardware that allows you to seamlessly interact with it. Glasses are the ideal form factor since they can be with you throughout the day and they can assist you without pulling you away from the moment. Our glasses remain one of the fastest-growing consumer electronics of all time, and we continue adding to the lineup. We just released our own line of Meta Glasses in collaboration with Essilor Luxottica including a style that we designed with Kylie Jenner. They're the first glasses to ship with Muse Spark out of the box so that they can understand what you're seeing and give even more helpful answers.

Early sales have been strong, exceeding our expectations, and we're going to have more to share on our glasses lineup at this year's Connect Conference on September 23, so I encourage you to tune in then.

Before I wrap, I want to mention that I just published an op-ed about why I'm so optimistic that we are building a positive future for everyone. At Meta, we have always built technology to put power in people's hands. So they can connect to the people they care about and shape the world in the ways that they want. It's why we've always focused on making our products affordable and accessible for everyone. And it's a strategy that has served both our community and our business well.

As we enter this next chapter, that same philosophy is guiding how we approach developing AI. We're the only major company building AI with the primary goal of putting super intelligence directly into people's hands rather than centralizing Super Intelligence, we are focused on distributing it widely and giving everyone the ability to direct it towards what matters to them. That's the way that society has always made progress. I think that these are the right values for building a positive AI future. And if we help build this, then I think that we will continue to build a very strong business as well.

So that's everything I wanted to cover upfront today. AI is improving our core business. It's making our apps more relevant and delivering better results for businesses. We are starting to deliver more novel products, and we'll have a lot more there soon as well. And we are investing aggressively because the potential is huge, and we know that there are many ways to deliver value here. As always, I'm grateful to all of you for being on this journey with us. And now here is Susan.

Thanks, Mark, and good afternoon, everyone. Let's begin with our segment results. All comparisons are on a year-over-year basis unless otherwise noted. Q2 total family of apps revenue was $60.4 billion, up 28% year-over-year. Q2 family of apps ad revenue was $59.4 billion, up 27% or 26% on a constant currency basis. In Q2, the total number of ad impressions served across our services increased 14%.

Impression growth was healthy across all regions, driven by growth in engagement and users as well as ad load optimizations. The global average price per ad increased 12% year-over-year, driven by ad performance gains, improvements in macro conditions relative to Q2 of last year and currency tailwinds. This was partially offset by strong impression growth, particularly from lower monetizing surfaces and regions.

For the first time, quarterly family of apps other revenue reached $1 billion and grew 73% year-over-year, driven primarily by WhatsApp paid messaging and subscriptions revenue. Within our Reality Labs segment, Q2 revenue was $431 million, up 16% year-over-year due to strong growth in AI Glasses revenue, partially offset by lower Quest headset sales.

Moving now to our consolidated results. Q2 total revenue was $60.8 billion, up 28% or 27% on a constant currency basis. Q2 total expenses were $42 billion, up 55% compared to last year and included $2.4 billion in charges related to legal proceedings and $1.2 billion in severance expenses in connection with the May 2026 headcount reduction. Year-over-year growth was primarily driven by increases in employee compensation, infrastructure costs, legal-related costs and third-party AI token costs. Excluding the previously mentioned severance expense, growth in employee compensation was driven by technical hires we've added over the past year, particularly AI talent. The growth in infrastructure cost was driven by higher depreciation, data center operating costs and third-party cloud spend.

We ended Q2 with over 75,000 employees, down 3% from Q1. This total includes approximately 8,000 employees impacted by the May 2026 headcount reduction. We expect the majority of impacted employees will no longer be captured in our head count by the end of Q3 2026.

Second quarter GAAP operating income was $18.8 billion, representing an 8% decline year-over-year and a 31% operating margin. Excluding the Q2 legal charges and severance expenses, our second quarter operating income would have increased 9% year-over-year. Our tax rate for the quarter was 16%. Net income was $15.8 billion or $6.18 per share. Capital expenditures, including principal payments on finance leases were $31.1 billion, driven by investments in servers, data centers and network infrastructure. Free cash flow was $784 million. We ended the quarter with $90.3 billion in cash and marketable securities and $83.7 billion in debt.

Turning now to the business performance. There are 2 primary factors that drive our revenue performance. Our ability to deliver engaging experiences for our community and our effectiveness at monetizing that engagement over time. On the first, we continue to see significant gains from our content recommendation initiatives.

On Instagram, global time spent this quarter grew double digits year-over-year this quarter, largely driven by improvements to our feed and reels recommendations.

On Facebook, video time spend increased 9% globally year-over-year and over 10% within the U.S. and Canada, where it was driven by ranking improvements. We are finding that LLMs are increasingly capable of delivering ranking and recommendations gains. First, they make our existing systems smarter by understanding what the content is actually about and generating better training data. Second, LLM powered agents are also helping with engineering development by evaluating content quality, detecting trends and testing ranking changes.

Earlier this year, we reached a milestone of every public reels and feed post on Instagram being automatically processed through an LLM and analyzed across dimensions from topics to tone and we're working towards including more services on Facebook as well. These signals can then be passed to downstream applications across ranking, recommendations and content policy enforcement, which is a key building block toward greater personalization.

This quarter, we also began using our Muse family of models to conduct content understanding across signals like video topic classification and summarization, and we've seen positive early results.

Finally, our recommendations are also becoming more personalized, surfacing more fresh content while giving people more direct control over what they see. On Reels, we shipped our largest single release ranking improvement to date combining faster inference with a new architecture that draws on deeper user history to improve predictions. This drove a 15 basis point increase in sessions on Instagram with particular strength in reshares and time spent, which are both strong indicators of better content to user matching. We are now bringing this to feed where early results look comparable.

We are also getting new content to people more quickly. Investments we've made in more real-time infrastructure and modeling improvements on new videos are allowing our largest ranking models to now identify high-quality new reels at creation. On InstruramFeed, over half of all recommended content is now less than 1 day old, more than double from a year ago.

We're also giving people more direct control of the content they see. Today, Instagram users can visit the Your Algo page, which lets users write natural language prompts to tune their recommendations. Similarly, on Facebook, we launched Shape Your Feed. Early results show over 80% retention among users who engage with it.

Looking forward, we're executing on our longer-term efforts to develop the next generation of our recommendation systems. This includes building foundation models that are designed to power organic content and adds recommendations simultaneously as well as developing LLM native recommender systems. We hit our first research milestone this half by continuously pretraining a large-scale model with recommendations data and observing healthy scaling laws in the process. We're encouraged by this milestone and expect continued progress in the second half of the year.

Turning to the second driver of our revenue performance, increasing monetization efficiency. The first part of this work is optimizing the level of ads within organic engagement. Here, we continue to enhance our systems to show ads of the optimal time and location. In Q2, we also expanded availability of ads on our newer services, including completing our global ads expansion on Threads.

On WhatsApp, we have introduced support for more types of ad destinations and advertiser performance goals and status and continue on track toward our global rollout.

Moving to the second part of increasing monetization efficiency, improving performance for the businesses who use our services. Within our ad systems, we're delivering performance gains as we deploy more complex and predictive models. This quarter, we introduced Meta Generative Recommender, a paradigm shift in how our ad system works. Rather than scoring every possible ad individually, we are now using LLM to reason about ad content and user preferences together and predict the best ad for each person. This makes our ad matching more intelligent and more precise, which compounds performance gains for advertisers.

We deployed the first generative model into our ads retrieval system and saw notable improvements in ads performance. Early pilots using LLM to better understand user preferences drove a 1% increase in app event conversions on Instagram.

In Q2, we also advanced our user understanding models to analyze ads and organic activity and simultaneously improve both user experience and advertiser performance. Combined with our GEM model for ads ranking and sequence learning, these advancements generated an 8.3% increase in ad clicks and a 15.7% uplift in conversions on Facebook.

We're also leveraging AI to empower businesses to more easily manage their campaigns, develop ad creative and engage with their customers. Our AI-powered Advantage+ end-to-end solutions continue to grow, reaching over $75 billion in annual revenue run rate this quarter. We're working to deepen adoption as advertisers who leverage multiple tools see compounding performance gains.

I'll share one example of how Advantage+ is making performance marketing meaningfully easier and more effective for SMBs. Underneat, an online apparel brand in India had been setting up each campaign manually across Facebook and Instagram. After adopting Advantage+ sales campaigns layered with Advantage+ audience placements and budget optimization, they saw a 13% incremental lift in purchases and a 16% increase in add-to-cart conversions.

Adoption of our GenAI ad creative tools continues to scale with over 9 million small businesses using at least one AI creative tool. Image generation, which now lets advertisers produce more creatives that scale from existing content, including a new ability to create images from video assets saw adoption more than double this quarter. We also introduced a new end-to-end creative solution that gives advertisers the AI infrastructure to translate real performance signals into their next creative decision while preserving brand identity and tone. We're building with agency integrations from day 1, so teams can diagnose, generate and scale high-performing creative without leaving their existing workflows.

Looking ahead with the rollout of used image, we expect to further expand advertisers' ability to generate high-quality on-brand creatives at scale. With Meta Business Agent, businesses are better able to serve their customers through our messaging apps by responding to inquiries, recommending products and handling support around the clock. Earlier this month, we also introduced the Meta Business Agent platform, which gives enterprises the infrastructure to build, customize and deploy their business agent at scale on WhatsApp. The platform provides larger businesses with enterprise-grade controls, guardrails and measurement built in so they can define rules and offer personalized experiences starting within the messaging apps that their customers already use.

Movida, one of Brazil's largest rental car companies with nearly 400 locations deployed a business agent on WhatsApp to handle the entire booking flow from vehicle selection and pricing to payment in a single conversation. Returning customers could complete a reservation in as few as 3 messages. In a 1-month period, Movida reported a 44% increase in daily bookings through WhatsApp when compared to the same period in the prior year and that 85% of conversations in the channel were resolved entirely by the AI agent without human assistance.

We're also building out additional ways to monetize our ecosystem. Two additional revenue streams are subscriptions and monetizing our competitive models through an API. Meta One is an evolution of our subscription portfolio to create more value for everyday users, businesses and creators so they get more features and AI tools to create connect and stand out. We are excited to bring this to more users and continue to build enhanced tools for our subscribers.

We also recently launched a high intelligence model API at a competitive price and are encouraged by the initial results. We recently made Muse Spark available on open router for U.S.-based developers, broadening its distribution and making it easier for developers to adopt the model. We soon expect to roll out the model API to more distribution channels, make it available in more countries and open it up for enterprises.

Our approach to building capacity is strongly influenced by several key elements. First, the broad environment for building infrastructure is dynamic and uncertain in both near-term and longer-term time horizons. The industry has underbuilt historically for the wave of AI adoption, making existing capacity, including our own extremely valuable. Longer term, the supply chains need to be built out to support the capacity that we anticipate we and others will need for AI-powered experiences. Second, we have high confidence in our ability to utilize capacity to scale and build on top of our existing experiences as well as continue to invest in foundational models that will create substantial new opportunities.

Consequently, our current plans are geared towards maximizing 2026 and 2027 capacity. When we have had incremental capacity in the past, it has proven extremely valuable in scaling experiences like wheels. And we are confident that this will be true in this time frame as well.

Longer term, it's harder to predict the exact usage scaling curves, but we believe that our distribution advantages will give us the opportunity to serve AI products that are valuable for everyone, both our 3.6 billion users and millions of businesses. This should be true regardless of whether our models are on the frontier, but we believe that being on the frontier will unlock new markets and opportunities for which we may need additional compute. Therefore, our longer-term capacity strategy aims to give us the flexibility to continue growing compute in 2028 and beyond by laying down data center and network foundations to accommodate future server decisions.

The long-lived nature of these assets inherently provides the flexibility that will make it possible to adjust our investment to the pace of AI adoption. In addition, we have been making strategic investments in areas like our internal custom silicon effort, which will provide long-term strategic flexibility and supply chain leverage. This will be helpful in driving better returns on those long-term investments.

Finally, we believe that overall industry capacity is going to remain tight for the foreseeable future. As we've said earlier, we strongly believe that the models, consumer experiences and enterprise offerings that we are building will be the best and highest ROI use of our infrastructure. Those enterprise offerings have the potential to take multiple forms, as Mark mentioned, agentic tools, our API or monetizing compute directly given outsized market demand. We expect that remaining nimble about these opportunities will help us fund our build-out more efficiently while preserving our strategic flexibility to have the compute when we need it and provide us multiple pathways to generate returns on expected -- on invested capital.

In funding these infrastructure investments, the strength of our balance sheet gives us the ability to attract capital from a wide range of markets to supplement the cash flow generated by our business. Our announcement with BlackRock yesterday is an example of the partnerships that we can structure to complement our approach to building infrastructure capacity.

Moving now to our financial outlook. We expect third quarter 2026 total revenue to be in the range of $61 billion to $64 billion. Our guidance assumes foreign currency is an approximately 1% headwind to year-over-year total revenue growth based on current exchange rates.

Turning to the expense and CapEx outlook. We are raising the lower end of our expense outlook to incorporate the $2.4 billion charge related to legal proceedings recognized in Q2. We now expect full year 2026 total expenses to be in the range of $165 billion to $169 billion. We continue to expect to deliver operating income this year that is above 2025 operating income. We anticipate 2026 capital expenditures, including principal payments on finance leases to be in the range of $130 billion to $145 billion, narrowed from our prior outlook of $125 billion to $145 billion.

Absent any changes to our tax landscape, we expect our tax rate for the remaining quarters of 2026 to be between 15% to 17%, an increase from our prior outlook of 13% to 16%.

Finally, we continue to monitor active legal and regulatory matters that could significantly impact our business and financial results. For example, we continue to see scrutiny on youth-related issues in several markets and have a number of youth-related trials scheduled for this year in the U.S., which may ultimately result in a material loss.

In closing, our business momentum continued in Q2 with strong execution across our core ads and engagement initiatives. We're also progressing in our efforts to bring personal super intelligence to everyone with exciting model releases, and we expect to build on that momentum over the course of this year with new products. With that, Christa, let's open up the call for questions.

[Operator Instructions] And your first question comes from the line of Brian Nowak with Morgan Stanley.

I have 2, one for Mark, one for Susan. Mark, I appreciate all the color on the big pipeline for new products across consumer and business agents, API tools and compute rental. There's a lot of opportunities here. My question is, as you look at these opportunities and the state of the current offerings, the compute capacity, which of them do you expect to be able to scale first sort of in '26 and '27 to sort of showcase quantifiable material ROIC for investors?

And then Susan, there -- there's been some public comments about capacity in '27, doubling capacity, and I appreciate your color about CapEx. Any early comments on '27 CapEx, even the philosophy around sources of upside or sources of downward pressure? Do you sort of think through different ways to finance this multiyear build and have '27 CapEx?

I can take the first one. So in terms of the different opportunities and how we think about the compute, overall, a substantial amount of the compute goes towards training models to be a leading lab. And I think that's an important investment, but then the rest of it goes towards a set of different products and revenue opportunities, which spans from optimizing and improving our core business to building new consumer products that we're releasing soon to the API, do the business agent's work, to the developer tools work on the road map that I alluded to and then also the opportunity to sell compute directly where -- I mean I mentioned that we have quite a number of offers at a meaningful premium over what we paid for the compute.

The question in thinking about this is we believe that there will continue to be a significantly higher margin on selling intelligence rather than selling compute directly. But we think that there is a big opportunity, obviously, to sell compute as well. So we're thinking through -- I think you asked what one area would likely scale the most. But I mean I'm actually quite optimistic that we're going to see meaningful growth in all of these areas. And I think we'll have more to share soon on a number of them.

Brian, on your second question, we aren't providing a specific outlook for 2027 CapEx at this time. Infrastructure planning remains highly dynamic. And even this year, there are a range of outcomes embedded in our outlook. I alluded in my main remarks to our focus on being -- gearing our current infrastructure plans towards maximizing capacity in 2026 and '27 and giving us the flexibility to continue to grow in '28 beyond and beyond, but also giving us the ability to make server decisions when we come to our being able to evaluate our actual needs in '28 and beyond.

So we're still working through what our capacities are going to be over the coming years. Generally, we believe near-term capacity is more valuable than long-term capacity, and it remains a very dynamic planning process.

Your next question comes from the line of Eric Sheridan with Goldman Sachs.

Maybe 2, if I can. When you frame up the enterprise opportunity, how much of that opportunity do you think is available to you today based on what you built out in terms of go-to-market strategy as extensions of the advertising and the marketing business, you have already versus go-to-market strategies that might have to be built to capitalize on the opportunity?

And then maybe, Susan, if I could just squeeze the second 1 in. When you think about sources of capital for the business for the next couple of years, you referenced the deal that was announced yesterday as an example of looking at ways to finance forward obligations. We continue to get questions from investors about how to think about the mix of debt and equity and sources of capital. Philosophically, how are you guys thinking about wanting to be ambitious on the spend, but then marrying that with the need for capital?

Sure, I can talk about the first part. I think for enterprise, there's going to be a combination of extending the current business, which is effectively selling to marketers and businesses that are like basically customer facing, trying to reach customers and sell to them directly. Obviously, that's the vast majority of the business across Facebook and Instagram. And we believe that there is an opportunity to extend this with business agents across messaging apps and other surfaces to continue to interact with customers. And just like the ad system, effectively, we will get paid when we deliver results for those businesses.

So we view this as an extension of the sales and the partnerships that we have with many millions of advertisers and hundreds of millions of small businesses that use our platforms. So that one, I think, should be quite a natural opportunity for us, and we're focused on delivering it in a way where we're not trying to kind of maximize the sale in the near term. We're trying to maximize the results for people and build out a robust auction and that's what we've seen has served the business well over time.

There are other enterprise customers who I think we're increasingly going to serve, too. We are building coding and developing an internal productivity tools, partially because we need to build them ourselves. And we need to make sure that we have tools that are tuned for ourselves. And now that we have those, we feel like there's a large opportunity to serve, whether that's small businesses or larger businesses, that is a somewhat different muscle than we have historically had. And we will share more soon on how we're planning to build that out, but I think that there is just a very large opportunity on this.

And the way that I think about this is there's obviously been a bunch of news about the compute side, but I think that the enterprise opportunity is kind of the sum of all of these different things. It's not just the selling compute, also the API services, the productivity services, the kind of business agents for other parts of the business beyond marketing are all parts of the overall offering. And I think that there's just a very, very large opportunity there.

So we're quite focused on that. That's going to be somewhat of a new muscle that we build as a company, but I think it's a very important one that we build. So that way we can make sure that we can maximize the opportunity ahead of us.

Eric, on your second question, in terms of sources of capital, this is something that we have been looking at thoughtfully as we think about financial planning for the future. Obviously, our strong operating cash flow certainly has put us in a position of strength as it pertains to funding our infrastructure build-out. But we've also been evolving our capital structure in recent years to include a greater mix of debt as we work to bring down our cost of capital and we have generally found it prudent to continue adding cost-efficient, long-duration sources of capital as we make investments in initiatives that themselves have long time horizons, especially AI infrastructure projects.

We've also broadened our aperture there to include partnerships, like the one we announced with BlackRock yesterday. And we'll continue to be thoughtful about evaluating the appropriate different sources of capital over time as we evaluate future projects.

Your next question comes from the line of Mark Shmulik with Bernstein.

Mark, everyone's got a story of kind of someone coming back from Silicon Valley, deep writing code, building agents with AI. And then they head home and they kind of tell their parents, they're using AI wrong. It's just kind of like a glorified search tool. Consumer behavior is always pretty difficult to predict, but kind of reading your update on AI for everyone. How do you think about whether consumer adoption can close this AI utility gap? Like are we on the cusp of something kind of breaking through? Or do we just need to be a bit more patient?

Well, I think that some things have already broken through. And I think one of the interesting things about AI compared to other technologies is that every year or so and the cycle may accelerate. But every year or so, you get new capabilities that create new possible product lines. So there's obviously the AI assistant market for consumers where that's what we're doing with Meta AI and then there's the other products that competitors have in that space.

And then in the last year, I think the coding agent market has really grown very quickly. And that's the first real agentic market. And I think that there's a number of reasons why coding is first, right? You have technical customers who are willing to do the work to make it work. Coding is an inherently digital and closed-loop activity. So it's sort of -- in some ways, it is somewhat of a kind of an easier system to train on.

But our bet is that -- or one of the bets here that we're making is that we think that consumer personal agents is going to end up being an extremely important and massive market. I think that it's extremely unlikely if you look out 5 years from now, for example, that whatever period of time you want, that you don't have billions of people with a personal agent that understands your goals and that is just working on your behalf, 24/7 to achieve your goals. And whatever the domain is that you care about, whether it's helping you with your health or your hobbies or your personal finances or your productivity in running your home better or improving and enhancing your relationships, helping with your career, helping -- just all these different things. This is like a very, very deep set of use cases.

But like you said, I think that building for consumers is a little bit different than building for developers. If you're building for consumers, especially if you're trying to build something that isn't used by millions of people, but it's used by billions of people, it needs to just work, right? It needs to be simple. And I think a lot of what we see with these agents today, a lot of the kind of proto agents is it takes a bunch of fiddling. You have to get into a terminal to get it set up. You get a use case going. It kind of seems magical, but then maybe it breaks down over time. And I think that the companies that can deliver the personal agents that just work, I think that this is going to be one of the next major opportunities in AI, and we are very excited about that in addition to what we're doing with Meta AI, in addition to what we're doing with business agents, in addition to all these other things, while also developing the models that we think are going to continue to advance new capabilities to create new product lines as well.

So this is why we're very optimistic. Now I get -- we're -- we're going to ship this at some point soon, and that's going to be very exciting. And we haven't done that yet. So this is -- there's kind of only so much that I can say on an earnings call about this. But I think, this is a very large opportunity and one that I think is almost inevitable that someone does. And I think it really plays to Meta's strengths as a company. We build consumer products that reach billions of people. We're great once we get something working, scaling it to a large number of people, and we're good at building the infrastructure to be able to support these intensive applications.

So I feel very good about this. I kind of understand that we need to deliver it for our community, and that's what we're very focused on.

Your next question comes from the line of Doug Anmuth with JPMorgan.

One for Susan, one for Mark. Susan, you talked about how LLMs are increasingly capable of delivering ranking and recommendation gains. Can you just talk more about the road map here and how far along you are in just leveraging better models and more compute?

And then, Mark, just in terms of the number of offers to monetize your compute externally, you also at the same time are purchasing capacity from a number of third parties. So just hoping you could help us understand some of the differences here? Is it just timing and stop gap issues? Or is it training versus inference, and leveraging the chips that are best suited for each?

Thanks, Doug. I'll take that first question about where we are on the recommendations road map. First of all, we certainly see further headroom to continue improving recommendations over the rest of the year and into 2027. And we expect that will help us drive additional gains on both engagement on Facebook and Instagram.

A couple of things I would highlight. First, we'll continue to make recommendations even more personalized and relevant to user interest. And that's in part by advancing our recommendation models and architectures to further capture user interest more precisely and respond faster to what people care about in a given moment.

Our AI investments are going to play a significant role in delivering on this vision, including the expansion of LLM based content understanding to develop a deeper understanding of post creators that people value, to capture user interests more precisely and respond more quickly to what they care about in the moment and using AI to surface high-quality, fresh and trending content and reduce the share of low-quality content. Second, we're continuing to improve our data infrastructure to allow our models to train on more data and leverage that data more effectively.

We're adding more detail to how we describe content that users have engaged with in the past and enriching past user interaction sequences with more granular content. That allows our models to more precisely learn which engagements are more or less valuable to users. We're continuing to scale up both the length of user interaction sequences we used during training as well as the complexity of our model architectures across Facebook and Instagram to take advantage of the larger data sets.

And then we've already made significant strides, leveraging LLM for content understanding. We're going to further incorporate them into both our recommendations and content policy enforcement stacks given their capability to more deeply understand content. And I would say, broadly, I think we're very optimistic about that body of work. We've also invested in using LLM based agentic approaches to transforming our recommendation system. And we grew the number of launches from our ranking agents this half. That also helps make our engineers more productive and that's another path that we're very excited about as well.

I can answer the second part of that. I mean, I think your question was about how we think about offers that we're getting to sell the compute, but then we're also buying the compute.

I mean, look, the high-level observation is that there's just nowhere near enough compute for all the demand. So that is why we see that like basically, we are getting a large number of offers for the compute that we have, but also we have a lot of internal uses that we think are going to be quite valuable.

Now in terms of running the business, obviously, a common trade-off that we need to make is around how much do you monetize something today versus develop future assets for the future? And I think that it's always a portfolio, right? It's not like you don't want to only do long-term things and you know like and not kind of prove the markets out that exist in the near term. But I also think it would be foolish to basically just sell all of the compute and take a short-term profit. But when you have the opportunity to build intelligence on top of it, which will be a kind of a multiple and that compounds the value of the compute on top of that.

So I think the answer is what we're doing, which is to basically use a lot of our capital to build out compute, having confidence that we have the ability to monetize the compute directly when that makes sense, but also knowing that we have quite a number of different use cases to monetize the intelligence on top of the compute, including the enterprise cases that we talked about and including some of the consumer and cases that we talked about, and including just the core business, which is not even necessarily new products that we haven't talked about, but just in terms of using that to be able to further add intelligence and improve the ranking and recommendations and ads in the core services.

So I think all of that is true, and it creates this dynamic where there is a lead time, where we're investing in building out these data centers now. They come online at some point in the future. You obviously are not getting value out of them until they're online. But that's -- we basically see a very large demand for all of this, and want to go maximize the opportunity to build out all of these different businesses.

Your next question comes from the line of Justin Post with Bank of America.

Mark, you've hired the senior leadership of your AI labs about a year ago. Could you just give us your thoughts on how the lab is performing? And do you think -- the Street will really see an uptick in product velocity as far as models or chips or other things coming forward? And then what kind of sustainable competitive advantages do you think the lab is building?

Yes. So I mean, I'm quite happy with the trajectory that we're on. We've released a few models that I think are quite impressive on our early scaling letter. And as I've said, we're in the process of scaling much larger and more advanced models and we're excited about those too.

I think that there's the intelligence aspect of this, and then there's the data aspect of it. And I think it is true that in any -- any kind of product category that you want to go into, there is going to be some kind of flywheel where you learn from the behaviors and how people in that community are using the product. So and kind of the better that you can do there, the more feedback you get and then that inevitably makes the products better. So I think this is a field where people understandably are obsessed with intelligence. But the data and the knowledge part of it to serve people well is really important. If you're talking about personal super intelligence, then obviously, having a very clear mental model of everything that is going on in the person's life and their goals is going to be a very important aspect of that.

And there are certain companies start with different advantages in different markets and different parts of that. But also, I think part of the reason for working on a number of these different things upfront is that, first of all, the technology is general. So when you build the intelligence out, it can apply to a number of different uses, but then you want to invest in building up the flywheel on them because I think that that's how you build the sustainable advantage over time.

But I think that we've definitely shown at Meta that when we have a product in a format that works, we -- I would go as far as to say that I think we're probably the best company in the world that's scaling those experiences to billions of people. So I feel quite good about the personal agent work. I feel very good about the business agent work where we kind of already have our base of advertisers and small businesses that we serve and the ability to scale that out and roll it out around those. I think that those are durable advantages as well as the data flywheels that we're building on them.

And then obviously, on the research side, you want to build a good culture. I mean this is just kind of the simple stuff, but it's -- I think that it's -- you just want to make sure that you're building the team in a way where you're managing things well and kind of consistently with low drama compounding over time. And that's sort of the goal that you aspire to. And I think if you can do that well, then that is hopefully may not be kind of a clear mathematical explanation of something, but I think that, that's kind of how business works. So yes, that's what we do.

Your next question comes from the line of Ross Sandler with Barclays.

Mark, sticking with the comments on the AI Lab. So Muse Spark 1.1 is pretty close to the Pareto Frontier, but it's at the lower end of the intelligence spectrum or lower cost then, I should say. And it sounds like from your last answer, you think it's important to compete at both the lower cost and the more expensive performance end. Could you just talk about that a little bit? And then your -- the leadership of your lab has also talked about going back into open source kind of where you were a couple of years ago. So how does that fit into the strategy and all this discussion around monetization of AI products and models?

Yes. So I mean, just to get into the kind of scientific process on this a little bit, basically, training large models at each stage of training a large model, you see novel behaviors. So basically, the way that you want to do this is you build up from training smaller models to building larger models. And the models that we've released so far are based on -- there are certain scale in climbing up the scaling ladder and we're continuing to scale larger models.

So I think for Muse Spark 1 and Muse Spark 1.1, I think that they are very impressive models for the scale of the model, the kind of stage of development of the lab. And I feel quite good about them. Now we want to have models that are more advanced as well, and that's why we're scaling larger models, and that's why we're building out a bunch of the research infrastructure around that. But at the same time, we also want to have good, more efficient models that will be a lot of what we serve to consumers at scale, right? If you're serving billions of people, you want the ability to have more advanced models for things that are very hard problems and you want the ability to serve the vast majority of prompts from simpler and more efficient models.

So I think both are going to be very important. The efficiency really matters, but I also think that we want to be able to solve the hardest problems for businesses and customers around the world. So we care about both of them.

On open source, I think we have always felt like open source was an important part of the ecosystem, and it's good for the world, and it creates its own feedback loops that are positive for us around getting the community invested in our infrastructure stack and our work and contributing improvements. But we've always basically said that we were going to do a mix of open and closed. And that continues to be true.

Now in ramping up Meta Super intelligence labs, in some ways, actually counterintuitively, it takes some more work to do open source models because you're not necessarily -- if you're doing something as a closed system that only -- you're only building for your own use cases, it can be a little more jagged. Whereas if you release it is open, and it's going to be used for a lot of things, you want to make it more well rounded. And I just wanted to make sure that the MSL team was uninhibited in building out the most intelligent models that we could. And we expect that we will get back to releasing some open-source models at some point soon.

But like we've always said, we're not dogmatic about this. We think open source is important. We want to contribute to that ecosystem. We plan to do a combination of open and closed models.

We have time for 1 more question, and that question comes from the line of Ken Gawrelski with Wells Fargo.

Two, if I may, please. I just want to maybe, Mark, just touch on the last point again, around open weight models. There's been a lot of discussion you've weighed in on the topic. Why or why not does that change Meta's view of developing closed proprietary frontier models. What -- is there an opportunity if open weight models proliferate that you don't have to develop your own frontier model. So that's question one.

And the second one, a clarification, if I may, for Susan. You noted that in your prepared remarks that you plan to maximize '26 and '27 capacity, is that a demand or a supply comment? Meaning are you suggesting that Meta plans for internal use of all the capacity built through '27? Or are you saying that you will evaluate '28 and beyond builds based on demand for Meta products and services?

I can take the open source question and let's see. So -- do we -- is basically the question is, do we think that because there are some open weight models that we -- that we can just rely on those. I mean, right now, the open-source models are not as strong as the frontier model. So no is the basic answer.

And then there's also just always the perpetual both policy debate and question around other companies' actions and whether that's actually a thing that we can -- that we -- that a company like Meta can rely on. And I think that, that's very tricky. So I think on both fronts, we believe we're going to be able to do better work, and we think that there's some risk in that reliance, I don't believe that, that is the right thing to do.

I think that we're a company that if you look at Meta from -- take a step back on this. So a lot of people view the surface layer of we build some social media apps and we have an ad business, we are really a full stack technology company. We built our own data centers, our own infrastructure, our own chips, our own low-level software. When we got started and like the -- like my background in engineering, like I wrote a lot of the systems code. But a lot of the reason why Facebook worked was because it actually -- it just worked, right? Like it literally worked when other social networks did not work fast and efficiently.

And I think we just have the ability to build things that can be more personalized, more optimized, more efficient. Some qualitative experiences are just not even possible for others to build because we go all the way down the stack. And it just seems to me, pretty clear that having kind of sovereign to you over building your own models is going to be an important part of that stack going forward, which is why it is important for Meta. But is also important -- is also why other people care about open source and why open source matters overall.

Because like other companies, if they -- even if they don't have the ability to build these models, do not want to have to just rely on a small number of closed labs. Like that there is a very important place in the world for there to be open source models. And to be clear, that doesn't take away from the API opportunity or any of the things that I'm talking about because someone still needs to run the models and run inference on them and be able to kind of run those models efficiently and have the compute to do that. PAUSE is going to continue to be a source of business advantage.

So I don't really think that these things are necessarily at odds. But when I look at like what a lot of discerning customers and companies are going to want around the world, they're going to know that -- they're going to want to know that they have control of their destiny and that they can trust the models that they're using and that their data is safe and that they're not sending it to competitors and all that. So I think open source is going to be important. But I think for us, also building the models is going to be a critical part of it.

It also -- for what it's worth, and I know that like the models all get evaluated on like a common set of evals and they get chocked up to like, okay, some models within like a couple of points of another model or whatever. But they all really do have different combinations of skills, kind of like people, right? It's like people have spikes in certain areas and have different personalities and are better and worse of different things. And if you're trying to build personal super intelligence for people or you're trying to build a business agent for small businesses, that needs to have potentially some different skills than what the other labs are tuning. And just like being able to do the full stack work on our -- on like Instagram recommendations on our ad system is how we've gotten the results there over time.

I believe that building the kind of full stack model is going to be a lot of the advantage over time and how we build personal super intelligence agents, business agents, all these different use cases for all of the different customers that we want to serve.

So in addition to the distribution that we have and the ability to reach all these people and scale products that work, I think this is a lot of the durable advantage is that you build something that is kind of specific to that use case and excellent in those use cases. And I mean, look, I get that this is a big investment and it's a big bet. We see the technology working. We're happy with the trajectory of the lab. I'm excited about the products that are coming, and we believe that this is going to be a big thing. So I mean, I get that this is sort of a big bet across the industry. My personal bet is that the people who invest in this are going to be rewarded and feel very good over time.

Ken, I'll just quickly answer your second question. When we refer to focusing on 2026 and '27 capacity, there are really two factors. One is we are today and expect to be in the sort of foreseeable future demand constraint, that really includes our core business, too, where there are -- we still have numerous ROI positive places that we would put compute toward if we had it. And the second, of course, is just the uncertainty over long-term constraints on building capacity, and we talked about some of the sort of the need to build out more of the supply chain earlier in my comments.

So I think beyond '27, when we look into '28 and further than that, the world is going to evolve a lot, our own internal demand will evolve. We'll have turned over a lot of cards by then. And so when we think about planning today for '28, we're really focusing on flexibility. That's just giving us kind of the ability to have land and power. But to really make the actual decisions about buying chips and other big-ticket items further in the future. So for now, I think we know we have a lot of good use cases for capacity in '26 and '27, and that's really what we're building towards.

Great. Thank you for joining us today, and we look forward to speaking with you again soon.

This concludes today's conference call. Thank you for your participation, and you may now disconnect.

Good afternoon. My name is Christa, and I will be your conference operator today. At this time, I would like to welcome everyone to Meta's First Quarter 2026 Earnings Conference Call. [Operator Instructions] And this call will be recorded. Thank you very much.

Kenneth Dorell, Meta's Director of Investor Relations. You may begin.

Thank you. Good afternoon, and welcome to Meta Platform's First Quarter 2026 Earnings Conference Call. Joining me today to discuss our results are Mark Zuckerberg, CEO; and Susan Li, CFO.

Our remarks today will include forward-looking statements, which are based on assumptions as of today. Actual results may differ materially as a result of various factors, including those set forth in today's earnings press release and in our annual report on Form 10-K filed with the SEC. We undertake no obligation to update any forward-looking statement.

During this call, we will present both GAAP and certain non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in today's earnings press release. The earnings press release and an accompanying investor presentation are available on our website at investor.atmeta.com.

And now I'd like to turn the call over to Mark.

All right. Hey, everyone. Thanks for joining today. We had a strong quarter for our community, our business and our progress towards AI. More than 3.5 billion people use at least one of our apps every day. We saw a small decrease in total family dailies due to Internet outages in Iran and blocks in Russia. But otherwise, trends across our apps are strong. Daily and monthly actives on Instagram and Facebook continue to grow with video driving all-time high engagement across both apps. WhatsApp continues to see strong momentum, too, including in the U.S. And Threads continues on its trajectory to be the leading app in its category.

Our biggest milestone so far this year has been the release of our Muse family of models and our first model MuSpark along with a significantly upgraded new version of Meta AI. This was the first release from Meta Super Intelligence Labs, and it shows that our work is on track to build a leading lab. Over the past 10 months, we have built the strongest research team in the industry and established the scientific and technical foundations to scale very advanced models. Spark is just one step on that scaling ladder, and we are already training even more advanced models.

But Spark has already made Meta AI, a world-class assistant that leads in several areas related to our vision of personal super intelligence, including visual understanding, health, shopping, social content, local, creating games and more. We're hearing very positive feedback on it so far. We've seen large increases in Meta AI use since releasing the updates, and the Meta AI app has consistently been near the top of the app stores as well.

Now that we have a strong model, we can develop more novel products as well. Since I first wrote about our vision for personal Super Intelligence last year, we've been focused on delivering personal and business agents to billions of people around the world. Our goal is not just to deliver Meta AI as an assistant, but to deliver agents that can understand your goals and then work day and night to help you achieve them. My view of AI is very different from many others in the industry. I hear a lot of people out there talk about how AI is going to replace people. Instead, I think that AI is going to amplify people's ability to do what you want, whether that's to improve your health, your learning, your relationships, your ability to achieve your personal career goals and more.

My view is that human progress has always been driven by people pursuing their individual aspirations. And I believe that this will continue to be true in the future. People will be more important in the future, not less. Meta believes in empowering individuals. And those are the kinds of products that we're going to build, and I believe that they're going to be some of the most important and valuable products of all time.

We are building a personal agent focused on helping people achieve the diverse goals in their lives. We're also building a business agent focused on helping entrepreneurs and businesses across the world, use our tools and others to grow their efforts, reach new customers and serve existing customers better. These agents will work together to form an ecosystem. And whether you use our personal or business agents to achieve your goals, I believe that the future will see a massive increase in entrepreneurship from people creating new things that they've always wanted to exist, but previously didn't have the tools to bring into the world. We're already testing an early version of business AIs and weekly conversations have grown 10x since the start of this year.

We're also working on using Spark in our upcoming models to improve our recommendation systems and core business in Facebook, Instagram and ads. Right now, our apps primarily help people accomplish 3 important goals: connecting with people, learning about the world and entertainment. But we've always wanted our apps to understand more of people's goals so we can help improve their lives in all the ways that they want. These new AI models will let us understand this in more detail. So instead of just looking at statistical patterns of what types of people engage with what content, for the first time in Meta's history, we're going to be able to develop a first principles understanding of what you care about and what each piece of content in our system is about -- is that way we can show you more useful things for what you're trying to accomplish. And we'll also be able to create personalized content specifically for people to help you achieve your goals as well.

Since our recommendation systems are operating at such a large scale, we'll phase in this new research and technology over time. But the trend over the last few years seems clear that we are seeing an increasing return on the amount that we can improve engagement for people and value for advertisers. This encourages us to continue investing heavily in what we expect will provide increasing value over the coming years as well.

On that note, we are increasing our infrastructure CapEx forecast for this year. Most of that is due to higher component costs, particularly memory pricing, but every sign that we're seeing in our own work and across the industry gives us confidence in this investment. That said, we are very focused on increasing the efficiency of our investments, and as part of that, we are rolling out more than 1 gigawatt of our own custom silicon that we're developing with Broadcom, as well as significant amount of AMD chips to complement the new NVIDIA systems that we're rolling out as well.

One of the primary goals of our Meta compute initiative is to lead the industry in efficiency of building compute, and we expect that will be a strategic advantage over time. Talking about building physical goods at scale, our AI glasses continue to perform well with the number of people using them, daily tripling year-over-year. This continues to be one of the fastest-growing categories of consumer electronics ever. We released Ray-Ban Meta optics this quarter designed for all day wear rather than primarily of sunglasses. And building on our release of Oakley last year, we have some exciting new partnerships and styles that I think are going to have the potential to reach even more people coming later this year.

All of our glasses are designed to easily update to use our newest AI models and features. I'm also really excited to see the glasses evolve from being able to answer questions to being able to be a personal agent that's with you all day long, helping you remember things and achieve your goals.

Beyond glasses, I am excited for more of our Metaverse efforts to be powered by the AI models we're training as well. We remain the biggest investors in the VR space across the industry, but we are focused on making our VR business sustainable as we invest more in other areas like AI and glasses.

Before wrapping, I want to talk for a moment about how AI is transforming our work. We are seeing more and more examples where one or two people are building something in a week that would have previously taken dozens of people months. And I want to make sure that Meta is the best place in the world for these types of people to come and make an impact.

We're building the next evolution of our company around these people. And there's a lot that we can do to enable this, building the best infrastructure for creating and delivering products at scale, streamlining our teams so they aren't bigger than they need to be. recognizing and rewarding the people who are having outsized impacts and setting ourselves up to try many more ideas and take on many new projects in the future. Of course, we will continue pushing to increase our efficiency as well. But overall, I think the future is about building many more higher-quality things than we've ever built before.

All right. That is what I wanted to cover today. We are living through a historic technological transformation. We are among the few companies positioned to shape the future, and we are on track to do that. I'm looking forward to delivering personal super intelligence to billions of people. And as always, I am grateful for the hard work of our teams and to all of you for being on this journey with us.

Thanks, Mark, and good afternoon, everyone. Let's begin with our segment results. All comparisons are on a year-over-year basis, unless otherwise noted. We estimate 3.56 billion people used at least one of our family of apps on a daily basis in March, which declined slightly from December due to Internet disruptions in Iran and a restriction on access to WhatsApp in Russia. Absent these impacts, growth in family daily active people would have been positive quarter-over-quarter. .

Q1 total family of apps revenue was $55.9 billion, up 33% year-over-year. Q1 family of apps ad revenue was $55 billion, up 33% or 29% on a constant currency basis. In Q1, the total number of ad impressions served across our services increased 19%. Impression growth was healthy across all regions, driven primarily by growth in engagement and users as well as ad load optimizations. The global average price per ad increased 12% year-over-year in Q1, with broad-based growth as we benefited from ad performance improvements, better macro conditions versus Q1 of last year, and currency tailwinds in international regions. This was partially offset by strong impression growth, including from lower monetizing regions.

Family of Apps Other revenue was $885 million, up 74% driven primarily by WhatsApp paid messaging and subscriptions revenue. Within our Realty lab segment, Q1 revenue was $402 million, down 2% year-over-year due to lower Quest headset sales, which were partially offset by continued strong growth in AI glasses revenue.

Moving now to our consolidated results. Q1 total revenue was $56.3 billion, up 33% or 29% on a constant currency basis. Q1 total expenses were $33.4 billion, up 35% compared to last year. Year-over-year growth was driven mainly by infrastructure costs and employee compensation. The growth in infrastructure costs was due to higher depreciation, data center operating costs and third-party cloud spend. The growth in employee compensation was driven by technical hires we've added over the past year, particularly AI talent.

We ended Q1 with over 77,900 employees, down 1% from Q4 as the impact of headcount optimization efforts in certain functions was partially offset by hiring in priority areas of monetization and infrastructure. First quarter operating income was $22.9 billion, representing a 41% operating margin. Q1 interest and other income was negative $1.1 billion, driven by unrealized losses on our equity investments.

Our tax rate for the quarter was negative 23%, which was favorably impacted by a tax benefit of $8.03 billion. This benefit partially relieves the $15.93 billion noncash tax charge we recorded in the third quarter of 2025, which reflects updated guidance from the U.S. Treasury issued in February 2026 regarding the tax treatment of previously capitalized R&D expenditures in the United States. Absent the tax benefit, our Q1 tax rate would have been 14%.

Net income was $26.8 billion or $10.44 per share. Absent the tax benefit, our net income and EPS would have been $18.7 billion and $7.31, respectively.

Capital expenditures, including principal payments on finance leases were $19.8 billion, driven by investments in servers, data centers and network infrastructure.

Free cash flow was $12.4 billion. We ended the quarter with $81.2 billion in cash and marketable securities and $58.7 billion in debt.

Turning now to the business performance. There are two primary factors that drive our revenue performance, our ability to deliver engaging experiences for our community and our effectiveness at monetizing that engagement over time. On the first, we're continuing to see significant gains from our content recommendation initiatives. On Instagram, the ranking improvements that we made in Q1 drove a 10% lift in real time spent. On Facebook, total video time increased more than 8% globally in Q1, the largest quarter-over-quarter gain in 4 years.

Within the U.S. and Canada, ranking improvements we made drove a 9% increase in video watch time on Facebook in Q1. These games are benefiting from advances we're making across the full stack. Starting with data, we doubled the length of user interaction sequences we use for training on Instagram in Q1 and increase the richness of how each user interaction is described, enabling our systems to develop a deeper understanding of user interests.

Within our models, we've significantly increased the speed with which our ranking models index new posts, which is enabling us to recommend them sooner after they are published. We're also applying more advanced content understanding techniques, which is enabling us to quickly identify posts that may be interesting to someone even if they haven't engaged with a lot of similar content. These and other improvements have enabled us to increase the diversity and recency of recommended content with same-day posts now representing more than 30% of recommended reels on both Instagram and Facebook more than double the levels 1 year ago.

We're also using AI to unlock more inventory by auto translating and dubbing videos into a viewer's local language, enabling us to recommend a more diverse set of content. Over 0.5 billion users on each of Facebook and Instagram are now watching AI translated videos weekly.

Looking forward, we're making several investments we expect will deliver more valuable recommendations. This year, we will continue scaling up our models in several dimensions, including their size and complexity, while incorporating LLM to deepen content understanding across our platform. This will enable us to better match people to a wider variety of content aligned to their interests. At the same time, we are executing on our longer-term efforts to develop the next generation of our recommendation systems. This includes building foundation models that power organic content and ads recommendations as well as developing LLM based recommender systems.

Our focus this year is validating the model architectures and techniques in these domains before we scale them out in future years. Aside from our recommendations work, we are focused on deploying the models from Meta super intelligence labs to enable a new set of product experiences. We're seeing encouraging results within Meta AI since we began powering responses with the first model from MSL, Muhspark.

In tests we ran leading up to the launch, we saw meaningful engagement gains that accelerated week-over-week with each new iteration of the model. We're seeing similar games within Meta AI following the broad rollout of our new model with double-digit percent increases in Meta AI sessions per user.

Mu Spark is now powering Meta AI in direct chat threads across our family of apps as well as the stand-alone MetaAI app and website, giving billions of people globally access to our latest model. Overall, we're very encouraged by the momentum within our research and product road map and look forward to sharing more detail on what we're building over the course of the year.

Turning to the second driver of our revenue performance, increasing monetization efficiency. The first part of this work is optimizing the level of ads within organic engagement. Here, we continue to enhance our systems to show ads at the optimal time and location. In Q1, we also expanded availability of ads on our newer services, including bringing ads on Threads to people in more markets. On WhatsApp, we're making good progress with the rollout of ads and status with hundreds of millions of people now viewing them daily.

Moving to the second part of increasing monetization efficiency, improving performance for the businesses who use our services. To do so, we're deploying AI more deeply across each layer of our systems and tools. Within our ad systems, we're delivering performance gains as we deploy more complex and predictive models. In Q1, enhancements we made to Lattice's modeling and learning techniques, along with advances in our GEM model architecture, drove a more than 6% increase in conversion rate for landing page view ads. In addition, we've been investing in more performing inference models for 1 more serving ads. In the second half of last year, we began rolling out our new adaptive ranking model, which is an LLM scale adds recommender model that we use for inference. This model improves our inference ROI by routing requests to more compute-intensive inference models when it determines there is a higher probability of conversion.

In Q1, we expanded coverage of our adaptive ranking model to support off-site conversions, which drove a 1.6% increase in conversion rates across the major surfaces on Facebook and Instagram.

We're also leveraging AI to make it easier for businesses to manage their customers, develop ad creative and engage with customers. The Meta AI business assistant has now been fully rolled out to all eligible advertisers on supported meta buying services, providing personalized recommendations to advertisers, resolving account issues, and servicing campaign insights to help optimize results. Performance has been strong since we began testing the assistant in Q4 with common account issues being resolved at a 20% higher rate. This week, we're also introducing meta ads AI connectors in open beta, providing advertisers the ability to connect their meta ad account directly to an AI agent.

We've always supported advertisers both on our platform and through tools like the marketing API. And now we're extending that to AI. So businesses and agencies can analyze and optimize campaigns with the tools they're already using.

Usage of our ad creative tools is also scaling with more than 8 million advertisers using at least one of our Gen AI ad creative tools and particularly strong adoption among small- and medium-sized advertisers. These tools are benefiting performance as well with advertisers using our video generation feature seeing more than 3% higher conversion rates in tests. We're also seeing good traction in using AI to facilitate customer engagement.

In Q1, we expanded business AIs on WhatsApp to SMBs across Latin America and Indonesia as well as on Messenger in Asia Pacific. We now have more than 10 million conversations each week being facilitated through business AIs, up from 1 million at the start of the year. We'll further expand access to more countries this quarter while adding more capabilities to the AIs.

We also continue to invest in the value optimization suite, which helps advertisers maximize their return on ad spend by prioritizing the highest value conversions rather than optimizing solely for the most conversions at the lowest cost. Adoption by businesses has been strong following performance improvements we've made over the past year with the annual revenue run rate of our value optimization suite now over $20 billion, more than doubling year-over-year.

Last, I want to touch on our commerce efforts. People discover products on our platforms through ads and organic posts with brands increasingly turning to creators to promote their products. This is contributing to rapid growth in our partnership's ads product with its revenue run rate more than doubling year-over-year in Q1 to $10 billion.

To support the product discovery and purchasing happening through creators, we're expanding our solutions beyond ads. Last month, we rolled out our affiliate partnerships offering on Facebook to more test partners, so creators can tag products from participating retailers on their posts and earn a commission when someone makes a purchase. We have also started testing similar experiences on Instagram. We see a real opportunity to help people more easily discover and buy products within our services, particularly as we incorporate AI deeply across our platforms.

Next, I would like to discuss our approach to capital allocation. Compute is becoming increasingly important as it determines the quality of services we can provide including powering more capable models and delivering innovative new products. It is also becoming more critical to how we work at a -- as we are entering a world where employees are managing agents to help them generate new ideas, run experiments, execute tasks and build products.

We are investing aggressively to meet our infrastructure needs, and ensure we maximize our strategic flexibility over the coming years. This includes substantially expanding our own data center footprint and striking deals throughout the supply chain to secure necessary components for future capacity.

We're also signing cloud deals that will come online over the course of this year and 2027, allowing us to scale more quickly. These multiyear cloud deals and our infrastructure purchase agreements drove a $107 billion step-up in our contractual commitments this quarter.

Our investments will support our training needs for future models and most importantly, provide us the inference capacity necessary to deliver personal and business agents to billions of people around the world, along with several other AI product experiences we're developing.

As we grow our infrastructure spend, we remain committed to operating efficiently, and we recently shared internally that we plan to reduce the size of our employee base in May. We believe a leaner operating model will allow us to move more quickly while also helping to offset the substantial investments we're making.

Moving to our financial outlook. We expect second quarter 2026 total revenue to be in the range of $58 billion to $61 billion. Our guidance assumes foreign currency is an approximately 2% tailwind to year-over-year total revenue growth based on current exchange rates.

Turning to the expense and CapEx outlooks. We expect full year 2026 total expenses to be in the range of $162 billion to $169 billion, unchanged from our prior outlook. We continue to expect to deliver operating income this year that is above 2025 operating income.

We anticipate 2026 capital expenditures including principal payments on finance leases to be in the range of $125 billion to $145 billion, increased from our prior range of $115 billion to $135 billion. This reflects our expectations for higher component pricing this year and to a lesser extent, additional data center costs to support future year capacity. Absent any changes to our tax landscape, we expect our tax rate for the remaining quarters of 2026 to be between 13% and 16%.

Finally, we continue to monitor active legal and regulatory matters, including headwinds in the EU and the U.S. that could significantly impact our business and financial results. For example, we continue to see scrutiny on youth-related issues and have additional trials scheduled for this year in the U.S., which may ultimately result in a material loss.

In closing, Q1 was a solid start to the year with strong execution across our core ads and engagement initiatives. We're also making exciting progress on our AI research and product efforts and expect to build that momentum over the course of this year.

With that, Crystal, let's open up the call for questions.

[Operator Instructions] And your first question comes from Brian Nowak with Morgan Stanley.

I wanted to ask you just about the level of investment you're making in sort of the signposts you're watching to ensure you're going to generate ROIC and all these investments - us and the other products. So if you could just sort of let us know some of the key factors you're watching over the next 12 to 24 months, whether it's Meta AI, Muse advances, chloralgorithm, what are you sort of watching foremost just to make sure that you're on the right path to generating healthy ROIC on all this CapEx and infrastructure spend?

That's a very technical question for basically where -- the things that we're watching are to make sure that we're on track building leading models and leading products. The formula for our company has always been build experiences that can get to billions of people and focus on monetizing them once you get to scale. That's -- we're seeing a little bit of that here where basically we invest in advance to build leading models, and we convert that into leading products. And then we think that these are going to be some of the most important products that get built over the next decade. So I think just like anything else that we've done over time, the basic milestones that I look at are around, first, technically, are we delivering the quality to enable a great product; then second, when you have the product, how is it scaling; and then third, you look at the monetization and then you drive up the efficiency of it towards increasing profitability.

I don't -- I mean like I don't think we have a very precise plan for exactly how each product is going to scale month-over-month or anything like that. But I think we have a sense of the shape of where these things need to be. And I think if you look at the usage of these and the quality of the products and the quality of the models that are out there and the use that other frontier models are getting and the trajectory of that, I'm quite comfortable that a, the lab that we're building is on track to be a leading lab in the world. I think MuSpark was a very high-quality model. It powers Meta AI, which I think is now a world-class assistant. We have an ability to be able to grow that and have a large amount of engagement. And over the coming quarters, we're just going to be tracking how do our next set of training runs go. How do our products scale how excited are we about the products in the pipeline, where right now, we're very excited. And then we'll also ramp up monetization over that period of time as well. So those are the set of things that I look at. I think for the kind of specific financial questions, I think Susan can jump in if there's anything more to add.

Your next question comes from the line of Mark Shmulik with Bernstein.

Mark, I guess now that we've got MuSpark kind of out there launched -- how are you thinking about the team's focus here kind of divided on to further model training runs and kind of further specialization in that personal intelligence goal versus product launches and kind of shipping more product out the door. And Susan, I guess, kind of as a follow-up to Brian's question, I know it's too early to discuss 2027 CapEx. But we've had peers mention tonight a potential significant step-up. Any way to think about dimensionalizing kind of how we think about some of the returns or traction this year and how it might affect the 2027 spend?

I mean I think the road map from the team is -- has been pretty consistent. So we have the research team, which is focused on scaling increasingly intelligent models with capabilities for the specific things that we're focused on, which are business and personal agents. So we're -- we just released our first model, and I talked about in my comments how we're climbing the scaling ladder towards greater capabilities and scale for the models. That work continues. We have our next set of more advanced models in training now. And that is -- that work will, I think, just continue. I mean that's a loop. I don't think you were going to be done with that anytime soon. We're going to have teams that are just consistently focused on training more intelligent and more capable models and the way that we want.

Then we have our product team, and that team is now really unlocked to be able to build things on top of our models because we now have a very strong model. So before this, we have been prototyping a bunch of things using other different models, whether it was our previous older models or kind of using the APIs from other companies. And now we're unlocked to be able to go build things and get them to scale on top of our own models. So I think you will see that over some period of time. I tried in my opening remarks to give a bit of a sense of where we're going, but I think that more of the details of that will become clear over the coming months. And I think that these are just both loops that we'll iterate on. We'll keep on iterating on the intelligence. We'll keep on working on building new products and scaling the products. And then as we get to product market fit, we're also going to increasingly focus on building the businesses around them and decreasing the costs. And this is kind of how we've done everything over the last 20 years of running the company, and that is basically the plan.

Mark, on your second question, we aren't providing a specific outlook for 2027 CapEx. And we are, frankly, undergoing a very dynamic planning process ourselves as we're working through what our capacity needs will be over the coming years. Our experience so far has been that we have continued to underestimate our compute needs even as we have been ramping capacity significantly as the advances in AI have continued and our teams continue to identify compelling new projects and initiatives. And now to, there are very compelling internal use cases. So our expectation is that compute will become even more central to the business going forward. And it will be critical to determining the quality of the models we develop, the types of products we can introduce, how productive we can be as an organization. So we're going to continue building out our infrastructure with flexibility in mind. And if we end up not needing as much as we anticipate, we can choose to bring it online more slowly or reduce our spending in future years as we grow into the capacity that we're building now.

Your next question comes from the line of Eric Sheridan with Goldman Sachs.

Maybe if I can build out on one of the topics that was discussed in the prepared remarks. But just the opportunity set that sits in front of the company with respect to putting agentic compute in front of both consumers and enterprises. You've long been associated with sort of the consumer landscape. And I am curious about how you're thinking about extensions of the media engagement parts of your business model and the commerce part of the business model to become more agentic over time. But what do you see also as the opportunity set that sits in front of you across SMEs and enterprises where historically, you maybe haven't had as much product velocity?

Thanks, Eric. So I would say in the near term, obviously, the sort of biggest focuses are some of the areas that you mentioned around deepening sort of engagement, obviously, with our existing community and user base, making ad experiences meaningfully more personalized, more engaging, more valuable, helping SMBs find and engage with customers across our platform. Those are some of the, I think, most intuitive and adjacent opportunities to the business that we have today. And then, of course, as we are able to build out more agentic capabilities enabling agents to help people be more productive, but also agents for businesses and enabling, frankly, those agents to interact with each other and build what we hope will be a thriving commerce ecosystem on our platform.

So I would say some of these are a little bit further out, especially in that latter category of things. Again, the focus is on building personal super intelligence, building a consumer agent that can work for you and help you get things done. That right now is a consumer experience that we're focused on, but we think there will be clear monetization opportunities over time. You can imagine commission structures or a premium offering. And on the business side, we're seeing a large opportunity, of course, around agents and scaling our business AI initiatives. I think I mentioned earlier in my remarks that there are over 10 million weekly conversations between people and business AIs on our messaging platforms. That's up from $1 million at the start of the year, and we're going to continue expanding globally in Q2. And business AIs today are currently free for most businesses on our messaging apps. But as we make more progress, we expect that we will also work towards establishing a longer-term monetization model. And we'll also consider other services services that we can offer to businesses in the future, but we don't have anything more to share today.

Your next question comes from the line of Youssef Squali with Truist Securities.

Maybe one for Mark and one for Susan. Mark, ray-Ban, Oakley AI glasses continue to perform really well for you guys, but Essilor-Luxottica owns and manages a lot more brands. What are the gating factors to see the launch of additional classes under these other brands this year? And what would be a successful year for you as you look back at 2026, maybe in terms of units sold?

And then Susan, on that 10% risk, how much of that is due to efficiencies for maybe AI implementation versus just the need to stay fit? And as you look at your employee needs over time, how do you see that growing maybe relative to your overall top line growth?

I can go ahead and take both of those. I might answer your second question first. And I'm just trying to make sure I got all of the parts of the question. So in terms of what the sort of kind of the optimal size of the company, I think, over time, we don't really know the optimal size of the company will be in the future. I think there's a lot of change right now with AI capabilities advancing rapidly. We're very focused on leveraging AI tools to substantially increase our productivity, and we're seeing that reflected in the accelerating output from our engineers. And we're generally approaching -- we're approaching this with a bias towards wanting to use these tools to build even more products and services than we would have before. At the same time, we're making very significant investments in infrastructure, and we are very focused on continuing to operate efficiently. So I think we will be continuously evaluating how we're structured just to make sure we're best set up to deliver against our priorities over the coming years. So that is, I think, your second question.

The first question was about the AI glasses. We have -- we're continuing to see strong growth in, obviously, the AI glasses sales over the course of Q1. Demand for the expanded portfolio lineup has generally been quite strong, and we're seeing sales shift now from the prior generation of Ray-Ban metas to the latest generation, which I think speaks to the value of the improved features like extended battery life and higher features like higher resolution video capture. So we're pretty excited about the progress we've made with glasses. We see strong interest now in the Meta Ray-Ban displays with the Meta neurlbands. So that's an encouraging sign that there is consumer appetite for display glasses, which is kind of the next generation of how this product evolves. And yes, so I think this is an area that we will continue to be -- that we continue to be excited about and are investing in.

Your next question comes from the line of Justin Post with Bank of America.

Mark, it took about 10 months to get you Spark out. I think it's a pretty good pace. Just help us understand what kind of unlock that is for some of the new products you're developing? And how is the product cadence going to be over the next 9 months on either consumer or business enterprise products built on top of that model?

I mean the field is moving pretty quickly. So I mean, I'm very happy that we're -- I think the lab that has gone the fastest from standing up the lab to having a very kind of widely accepted a strong model. So that's good. I think that is a very significant validation of the effort that the team is working well together, that the infrastructure is working, that that the effort is on track. And I think that, that's basically the main thing that we've learned over the last quarter that I would take away is like where and we started what is this pretty big bet, and it's on track for our plan.

In terms of what exactly the cadence is going to be? It's tough for me to say both because I don't really want to share competitively sensitive information and because I think some of the stuff we are more focused on quality than hitting a specific date. I mean, on the research side, this is research, right? We are trying novel things. You don't exactly know when they're going to land. And on the product side, I think we care a lot about just having. Let me put it this way. There's a lot of agents out there, right, that people are building for different things. And there aren't that many that I would want to give to my mother. And I think getting to like that quality bar is something that I care about more than hitting a specific week for launching or something like that.

So -- but with that said, I mean we're in a zone here where the teams don't check in with me like once a quarter, like we make meaningful progress like day over day. I think that's part of the fun of developing in this world is that people can make very rapid progress. small groups of people and teams can make very rapid progress. So I think we're going to see a lot of innovation. The timing of this call is it's good in some ways because the MuSpark release, I think, was was positive. The Meta AI first released, I think, is positive. I think that, that shows that we're on track. I'm trying to kind of paint a picture of the very high-level direction that we're going in, but I think that the picture is going to come into focus a lot more over the subsequent quarters.

Your next question comes from the line of Ross Sandler with Barclays.

Yes. Mark, just sort of related to that last answer, but there's a lot of new consumer applications kind of cropping up everything from like an open claw to something a little bit more consumer friendly than you would build for your mom, like you said, with like Pak or Dreamer, which you recently acquired. So how are these new ideas, I guess, changing your view around the direction that core Meta AI or Dream or kind of your overall agentic strategy needs to go? And then the second part of it would be, do you think the lab will stay in this consumer lane? Or do you think you need -- or you want to go down the route that others are going down with code writing and like the recursive self-improvement loop and in that direction kind of in parallel just thoughts on that.

Yes. So look, on the Open Claude and other agents, I think that they give you a very exciting glimpse of what types of things should be possible. Now they're pretty rough systems today. And to set up Open claw you need to like install a computer locally and then get into a terminal and configure a bunch of things that, again, like there's -- maybe there's hundreds of thousands of people or small numbers of millions of people who can do that. But what we're talking about is delivering personal super intelligence for billions of people around the world.

So how do you make a version of that experience that is a lot more polished and dialed and easy and that has all the infrastructure basically done for people already and that just works. And that's kind of what we're focused on, on the consumer side. and I'm really excited about that. I think if you had something like that, that worked quite a bit better than those systems and was easy enough that people could just get then I think you go from having something that hundreds of thousands or millions of people are going to use to something that is going to be addressable to billions of people. And that has been our primary focus from day 1 of the lab is being able to deliver something like that as a product, and I think it's just going to be very exciting. By the way, the same thing is true for businesses, right? I mean, there's the personal version of this -- but there's also a lot of people's goals are they want to create things, right? They want to create websites. They want to create products. They want to grow their products.

These are all things that good agents are going to be able to help people do, which I think is partially why this is so exciting. And in my opening comments, I talked about how today we can handle a few goals for people, they're big goals, right? We can help people stay connected with people they care about, learn about the world. These are big things that people care about. But they're not the only things that people care about.

And one of the things that I would love for our products to be able to do is just understand people's goals specifically and then be able to just go work on them for them, and check back in and whenever you have questions that you need to answer it. So whether those are personal goals or you're trying to create a business or do work. I think that this is like -- this is stuff that I think literally every person in the world is going to want some version of it. And also, I think it is something that scales where the more you want to get out of it, I think people are going to also be willing to pay a lot of money to have premium or high compute versions of it.

So I think that this is like it's a very exciting area. But I think what you all should be waiting to see is like whether we can build the version that really like just works and how effective we are at converting people who are using our products into being hundreds of millions and then billions of people using this stuff. And then over time, how can we effectively convert that into something that's increasingly profitable by monetizing it and getting the costs down. So I think that that's the road map of what we need to do.

You asked about whether we're primarily focused on consumers or also recursive self-improvement. I think that we've talked about two main goals for the the team. I mean one is this kind of agents version vision of what we're doing. The other is that self-improvement is really important because you can't build a leading AI product if you don't have leading models. So -- and you're not going to have leading models in the future if your models can't improve themselves, right? So you're getting to a point where today, the models are still able to learn from people -- and then I think at some point, the models will have to improve themselves. And that's how the growth is going to -- an improvement in the models is going to happen. And if you don't -- if we don't have an ability to do that, then we or anyone else, I think the companies that don't do that are not going to be leading labs, then they're not going to produce leading products. So I think at that like that is a table stakes thing that we are focused on.

Now does that make us a developer tools company? Not necessarily. I mean, I'm not against having an API or coding tools or anything like that. But it's not our primary focus. But I actually think people conflate coding with self-improvement more than they should. Coding is one ingredient for the model self improving. It's not the only thing. And we are focused on all of the parts that are going to be necessary for self-improvement in service of the personal super intelligence vision that we have for people and businesses.

Your next question comes from the line of Ron Josey with Citigroup.

Mark, maybe a quick follow-up to a prior question around personal agency and business agents. And with Sparks now live and more models in development, do you look at the personal agent opportunity, which we talked about earlier on the in the call, more of a short-term, medium-term, long-term goal, I'm sure it's a never-ending goal, but when we see a product, is the question short or medium term? And then Susan, I think the ranking recommendation model improvements are are very impressive to see, given the size and scale of both Instagram and Facebook. Could you help us understand just how doubling the length of these interaction sequences can drive greater usage. There's a thesis out there that maybe some of the rating recommendation improvements are long in the 2. So it seems if there's a lot more room to go. So any help there would be helpful.

I mean I think that the agents work, there's going to be short-term versions of it, but then I think that there's going to be massive upside for delivering more intelligence and more capabilities in the models. And you're kind of seeing this across the industry. Each month, each generation of models, they just have more capabilities and can do more things and people absorb it. and are able to get more superpowers and it's awesome. It's like the most exciting time in the industry. So I think of the agents as the product vehicle for delivering that capability to people. And we certainly -- I think this year is going to be a key period for establishing that as the vehicle for how people are going to use this, but then the model improvement, I think, is going to be something that's going to go on for a very long time. So there's a lot to do here in both the short, medium and long term.

And then on your second question, which I think is about the ranking and recommendations improvements that we talked about in our -- that I talked about in my earlier remarks, I think they're -- first of all, there is still a lot of room to continue improving recommendations over the rest of the year, and we expect we'll be able to do that to drive additional engagement on both Facebook and Instagram. A couple of the things. First, we're going to continue to improve our data infrastructure that's going to allow our models to train on more data. And we're adding more detail to how we describe the content that users have engaged with in the past and scaling up the complexity of our model architecture to take advantage of those larger data sets like using even longer histories of content interactions, and that should all be in service of improving the overall quality of recommendations.

We also are focused on making the recommendations even more personalized and more relevant to any given users interest. There's work we're doing to redesign our content retrieval system to show more content that matches the full range of a user's interest and to tailor the diversity of the topics we recommend to the broadness of someone's interest. So someone with particularly concentrated interest might see relatively more of that content while people with a broader set of interest might see kind of a greater range in the topics that we show them. And then finally, we're continuing to make improvements to our sort of LLM based two neuro algorithm features that allow users to provide more granular natural language feedback on what they want to see more of or less of in their feed. So the sort of the kind of the sequence length, which is the thing that you called out is 1 of really many improvements we made in Q1, and there is a big road map of further improvements going forward.

Your next question comes from the line of Doug Anmuth with JPMorgan.

Mark, how do you think about the step up as you go from leveraging smaller models in the ad business to use Spark and future large language models going forward, where are some of the key unlocks across engagement and monetization? And then on Manus, can you just talk at all about the strategic importance and the role in developing genetic products for Meta and then just current status around the tech and the deal.

I'll take that question. On Manus, we're still working through the details. So we don't have an update right now. On your first question, which is about sort of the -- going from leveraging smaller ads businesses -- smaller models in the ads business to kind of the ads sort of models growing. There's already some work underway, and I think I alluded to some of this in my earlier remarks, even kind of in the current landscape of the ads road map, where we're basically trying to advance the architecture here to allow sort of -- to allow us to leverage the abilities of larger models. Historically, we haven't used larger model architectures like GEM for inference. -- because their size and complexity would make them too cost prohibitive. And the way we drive performance from those models is by using them to transfer knowledge to smaller, more lightweight models that are used at run time.

The inference models are bound by strict latency requirements since they need to find the right ad within milliseconds, and that has, again, historically prevented us from meaningfully sizing up -- scaling up their size and complexity. But in the second half of last year, we introduced a new adaptive ranking model, which enables us to leverage LLM scale model complexity of 1 trillion parameters, and we made advances in the model architecture and codesign the system with the underlying silicon, so it maintains the sub-second speed that is required to serve ads at scale.

We also developed an approach that intelligently routes request more compute-intensive inference models if it determines that there is a higher probability of conversion and that lets us drive both better performance and increased inference ROI. So there's a lot of work being done there before we even sort of incorporate more of the LLM work into our underlying ads ranking models.

We have time for one more question, Ken Gawrelski with Wells Fargo.

Two, if I may. First, if I -- you talked on the Mu Spark launch. You've talked about two categories or two verticals. You talked about health and wellness and shopping. Can I dive a little bit -- ask you to dive a little deeper into the latter on the shopping and commerce side. And maybe if you could -- were there any learnings and the 2021, '22 phase where you push deeper into commerce on Instagram and on Facebook. Any learnings from that period that you might apply? Is there an opportunity for a next-gen marketplace-type business in commerce?

And then the second, please, Maybe, Susan, can you talk a little bit about -- based on your model improvements and the content recommendations, where -- how much visibility do you think you have to kind of the growth trajectory on the core business? You continue to grow at basically double the pace of the industry despite being a very large share of the industry. Could you just talk about a little bit about your visibility into that continued performance?

Yes. So I might give you a somewhat loftier answer to the question. You're asking about shopping. I think it's sort of an interesting example of the way in which the work that we're doing is different than what I think others are doing out there. These products, they -- AI agents get better when you fully optimize the stack. That's why we believe that we need to be a company that builds frontier models in addition to building the agents. And then in order to do that, you, of course, need to build your infrastructure in order to be able to do that well. So we're undertaking this large investment to be able to do that top to bottom. And I think a lot of the way to think about the investment that we're making is a bet that the individual things that people care about and that people are going to be more important in the future. And that's sort of like -- and I think it should be a pretty obvious thing to say. But I think so much of the rhetoric around AI in the industry is around like a company trying to build some kind of centralized thing that like does all the productive work in society in some way or something like that. And that just is very different from how we see the world.

Like our vision for the future is one where society makes progress by individuals pursuing their own aspirations. And some people care about big grand things like curing diseases. And a lot of people care about personal things like finding the right for my daughter. And I just think that we want to -- we're going to build things that help deliver this vision for personal agents for people. And I think that part of the lane and what is interesting and differentiated about what we're doing is that that's just so different from how I hear everyone else talking about the work that we're doing. So even though I think some of these ideas, they seem like they should be so obvious. I actually think that our approach of trying to empower individuals and building consumer things is just in the details extremely different from what others are doing.

And shopping might be one kind of specific example that I think is going to have interest in commercial implications. And I think people Consumers are going to like it. But I don't hear any other labs out there talking about how they're building an AI that's really good at shopping. And I think that the reason for that is like not because shopping is the most important thing by itself, but because like empowering people to do the things that matter in their lives, whether that's local or understanding social context, or shopping or personal health things or understanding what's going on around them visually, which is going to be really important on the glasses. These are all elements of the personal super intelligence vision. I think like a lot of this, and when you're thinking about kind of the investment in Meta over time, I think you should think about it as coming down to these set of values around what do we want AI to do in society. And if what you want it to do is empower individuals and build a world where the AI is in service to individuals goals, then that is what we are going to build, and I think it's going to be incredibly valuable.

Gosh, I almost wish we could end on that answer, but I will answer the second question, which I think kind of is 2 versions. One is a version of like what's the revenue outlook? And obviously, we gave the Q2 guide, which embeds, I think, both a range of kind of macro outcomes, but also the work that we've -- the ongoing work that we have to continue improving both the sort of usage and engagement on our family of apps and then our ability to continue making the ads better and more performance.

I think the second question is maybe more of a -- the second version of that question is more of a higher-level question about kind of the overall trajectory of the road map here. And one of the things I will say, having been working on this for a very long time, I'm always really impressed by the team's ability to continue to advance the state of the art here. And our planning process now is, I think, really fine-tune around this. So I've mentioned on a couple of calls, the budgeting process in which we run a very sort of ROI-based process to make sure that we are funding all of the ads initiatives that we think will drive growth in future years. And that's something that is both quite dialed in. And I think that our ability to measure the impact of that has been pretty robust, and it's been a very important driver of our ads revenue growth, and that continues to be a process that again, we ran in this past budget and -- as far as we can -- as we have line of sight, we feel good about the investment opportunities ahead of us.

Great. Thank you, everyone, for joining us today. We look forward to speaking with you again soon.

This concludes today's conference call. Thank you for joining, and you may now disconnect.

All right. Good morning, everyone. Welcome to our next fireside chat here at the Morgan Stanley 2026 TMT Conference. We are thrilled today to welcome the CFO of Meta, Susan Li. Susan, welcome back.

Thank you so much for having me.

It's always good to see you in our conversations we have about the industry, the company and everything exciting going on in the overall ecosystem. So...

A very sort of even keeled humdrum period in which to be presiding over one of the most conservative planning cycles that we've been through as an industry.

Clarity is high. We all know ROIC. It all makes sense. Exactly next question. Let me start with the important disclosures, including the personal holdings disclosures and Morgan Stanley disclosures appear on the Morgan Stanley public website at www.morganstanley.com/researchdisclosures. They are also available at the registration desk. Some of the statements made today by Meta may be considered forward-looking. These statements involve a number of risks and uncertainties that cause actual results to differ materially. Any forward-looking statements made today by the company are based on assumptions as of today, and Meta undertakes no obligation to update them. Please refer to Meta's Form 10-K with the SEC for a discussion of the factors that may impact actual results. One day I'll have an agent that's just going to do that.

I know. Or are you the person reading really fast on the pharma ads?

Yes, that's right, exactly. Yes.

Okay. So there is a lot that's been changing. Let's -- I want to sort of reflect on the last year external investor conversations and the perception of what Meta is doing versus where we are now. A year ago, we were sitting here talking about drivers of multiyear durable growth, ROIC on the core platform, all of these call options like Meta AI, Meta was going to build. And the market said, Meta is the AI winner, and that's it. One year later, as you know from the discussions, they're different now. There are more questions about ROIC. There's more questions about the positioning of the company versus other tech peers. Maybe let me start with, as you look at it internally, what has changed over the last year about how you think about the strategy of the overall core platform and those call options that we used to talk about 1 year ago?

Yes. Well, it's funny because when you look back -- and actually, it's a very sort of a very apropos timing that you're asking this question, right now is our performance review season. So I've just delivered a lot of performance reviews. And when you look back 1 year ago, it seems almost clean. When you look back at what the questions we were just debating a year ago, the things we were -- the biggest challenges on the horizon and you look now what has transpired over the last 12 months, right? And more importantly, what you think will happen over the next 12 months.

And when we look back at the last 12 months, okay, so we take stock of what's happened. The core business continues to perform, I think, extremely well. We have felt very good about our sort of -- our ability to run what has been now for many years, I think, an increasingly robust and measurement-driven process around how to evaluate and fund investments in the core business. That's both across the sort of organic content side, that's also across the ads ranking and recommendation side. I've been at the company for, gosh, 18 years now, and I have worked on ads the entire time. And when I -- sometimes I think back to when we were launching ads on mobile and then that was when we just began and then kind of the ad load ramp from nothing to 12.5%, that was like the first 4 years. And every half, we meet with the ad teams, and they come with kind of the list of here are the different ads initiatives we have. These are the front-end things. These are the ranking recommendation things. These are the kind of -- we have an internal metric called IREV, which is basically how we measure the performance of ads. And here's the list and here's what they add up to.

And it is, I think, one of the maybe modern wonders of the world that we have continued to generate basically half after half a list of improvements that continue to generate IREV gains every half and those continue to compound on each other. And so I described the monetization a bit in more detail, but that's true on the organic side, too. And I would say the core business is very healthy.

On the GenAI side or the AI side, we've done a lot of rebuilding in the last 12 months. We have built the MSL team and assembled, I think, an incredible leading basically cohort of talent in this space of AI researchers, but also AI leaders, product leaders, et cetera, to come together to complement the existing talent that we had. And that team, I've spent a lot of time with them. Mark has obviously spent much, much more time with them. They've come together really well. They are hard at work producing both the foundation models and also thinking about and building the product experiences that we're going to need. And as we said earlier, we expect the first models that come out of that team to be good, and we expect over the sort of course of the remainder of the year and next year for us to -- we hope to be part of pushing the frontier too.

We've also -- the way we think about capacity, and I assume this is something that has been echoed by all of my peers sitting in this chair just continues to evolve. But we thought there was going to be enough capacity 36 months ago, 24 months ago, that continues to change. And we are continuing to build for what we both know that we need today based on what we need in the core, based on what we need to train, but also what we plan to need for inference, inference across a lot of dimensions, both because of the customer, consumer experiences we want to build, also inference for like personal -- not personal, sorry, productivity -- internal, that's what I mean, internal productivity use cases. And so that's a place where we have, frankly, been playing some catch-up through the year. We are still playing catch-up. We are doing a lot to grow our O&O infrastructure footprint. But as it turns out, data centers are a long lead time project and a lot of the stuff we're doing today won't come online until '27 or later. And so we've also started taking down some cloud capacity.

So really, I think when I think about the last 12 months, core business, very healthy, very excited about the ongoing opportunities there. It also lets us fund the work in AI from a position of strength and confidence. And then on the AI side, we have been all hands on deck, assembling both the talent and the capacity, as I said. And I would tell you, in the way that only he can, I think, Mark is just like an incredible -- an incredible leader who responds with tremendous sort of focus to the problem at hand. And whether it is over the course of the last summer, if you had seen Mark, he was like the recruiter in chief, identifying and bringing folks on board, really helping the team come together. When we were looking at capacity and the fact that we didn't have enough data center capacity, frankly, to put servers in for the anticipated needs.

Mark is the person pushing us to be more creative about data center infrastructure. And I think we've talked a little bit about some of the projects we have. We have some of the finest tents in the world. It turns out you can get tents that are rated to stand for 25 years and withstand tornadoes and all these things to get capacity up faster. So I think that we've learned a lot of lessons over the last 12 months, but I think we are also -- we are never a company that is -- that's not going to respond to the challenge at hand and kind of with the most focus and energy and attention we can bring to bear.

It's a good preamble for me to sort of dig into a little more. Let me go back to that IREV internal metric you talked about of all of the improvements you've been making to the algorithms on organic content, the algorithms on ad rank and the advertising content. I think my team counted 20 changes you listed in the fourth quarter alone. So yes, you've been IREVing. Can you give us a little more quantification about some of the products that are really driving better engagement, driving better conversion? Give us a little quantification of those. And how have all those sort of changed your visibility and confidence about revenue growth to come throughout '26 and even in '27?

Yes. So we have now for -- we run a budgeting process every year. And when we do, teams come to us across the -- especially the ranking and recommendation teams have now a very buttoned-up process, both on the organic content side as well as on the ad monetization side. They run a lot of experiments to identify what they think are the highest confidence experiments, and they're able to measure returns on those experiments, frankly, over both a 1-year and a multiyear basis, so we can kind of look at this over a few years. And I would say, roughly, those things fall into the bucket of -- on the organic content side, how do we basically continue to grow the sort of interestingness and relevance of what we're showing users. And I think in Q4, for example, some of the product ranking work that we did on Facebook resulted in like a 7% lift in organic content views. That basically -- that was the product launch that drove the highest revenue impact in the last 2 years.

And we have a sort of -- we have a healthy pipeline of work ahead of us to basically to continue making the content more relevant through a couple of things. One is just scaling up the amount of data we can use that lets us increase sort of the history of content interactions, makes the overall corpus of data available to the recommendation engine larger. The second thing is we're really focused now on -- in the same way that we talked in the past couple of quarters, the way we are really trying to redistribute ad loads so that what we care about is right now, are you in a position where you're interested in engaging with an ad where you want to buy something where you're in a period of commercial intent. In a similar way on the organic side, we want to make the content recommendations most relevant and adaptive to the way you are engaging with content right now. Like what are you looking at right now and what's most interesting to you now?

And we're also investing in using LLMs to deepen our content understanding. They are -- as the models continue to become smarter and the sort of understanding and reasoning capabilities become better. Using LLMs to kind of help us understand content helps with recommendations in part because the traditional recommendation engine relies a lot on engagement signals and then you need a lot of engagement to happen to get the engagement signals, but LLMs can reason in real time about whether this is a piece of content that would likely be interesting to you based on what we know. So that's on the sort of organic content side.

On the ads ranking and recommendation side, we continue to do a lot of work across all of the -- we've talked a lot about our models, Andromeda, which is on the retrieval side, Lattice, which is on the model consolidation side, GEM, which is on the ranking side, how do we continue again to scale up models and make sure -- in a kind of a similar vein that the ads that we are showing you are the most likely to be relevant to you at this particular moment in time and that you want to engage with the ads in real time to the maximum degree that we can.

We're also trying to -- that's kind of the organic and ads bucket. We're also trying to do work in part because we're compute constrained on compute efficiency, how do we use the compute we have today for the highest impact. One of the launches we had on Instagram last quarter grew, I think it was like a 3% conversion lift on Instagram by applying compute to the highest impact sort of ads problems. And so we have a lot of work in all of those pipelines. And that I would generally describe as like work that gets funded through a very ROI-driven process. That's not even the new more research stuff be like, hey, we've got to line up some big bets because we want over the course of the next years to have other things in the quiver. Maybe we have slightly less solid understanding of exactly how that will look today.

That's some of the foundation -- the new foundation model work we're doing. We -- I think there was an announcement earlier this year where we're merging some of the research efforts across ads and across the discovery engine teams. The idea is to build sort of a unified foundation model and also doing work to, frankly, build some of our model architecture on top of LLMs and then fine-tune them with engagement data. That's sort of newer research efforts that we hope will pay off over kind of the longer run.

But all this is to say, I feel, again, just very good about the pipeline of initiatives we have had, that we have. And the thing that I think now at some point I was up here talking about this, it used to worry me, and it still does, to be clear, I'm just like engineered that way, that if you added up all these initiatives, sure, you could measure the return on each one because of that individual experiment, but you didn't know where on the slope of the curve you were. And so maybe actually, if you add up these 20 things, then you need to discount them by 80% because like the slope becomes much steeper. That has not turned out to be the case.

These -- the work that we have done has turned out to be more additive than we expected. And there is a virtuous cycle that you get into with advertisers, right? Like you make the ads perform better. That, in turn, drives costs down for advertisers. That in turn drives their budgets on us up. And then on the platform up. And then hopefully, that's good for their business, and that's a like long-term virtuous flywheel because now it's good for their business. They have more money to spend in the next cycle around doing this with us. That's really hard to measure, right? That's a multi-month, sometimes multiyear process. It's hard for us to measure that very directly. We have, in fact, done our best and run experiments that have been, I think, not statistically significant is probably the way that -- because you're trying to measure something that's so diffuse. But from everything we can observe, that appears to be happening on the platform. And our goal every day is to be the best place advertisers can come and spend their money relative to anywhere else.

You have a lot going on. A lot of pipeline. The one area you mentioned that I want to dig into because I think there is a little bit of a misperception externally about what you're doing now with LLMs on the core versus how you think about using LLMs in the future the core. So you mentioned it a little bit, but maybe just remind us how are you using LLMs now on the core? What does it look like 2 years from now? And what do you think that could do to signal and engagement things?

Yes. So today, LLMs are not a big part of kind of the work in core ranking and recommendations. It's not to say we don't use them at all. There are places we use them more than others, Threads, because Threads is text-based is a place that we're a little bit further ahead in terms of using LLMs to help with the ranking recommendations work on Threads. We are investing in using LLMs to help understand content today for the purposes of, again, better predicting whether the content will be relevant to you. But we are not, by and large, using LLM architecture to do ranking and recommendations work yet.

And that is, I think, something that is a little bit again of a longer-term research effort. We don't know exactly what that will look like, but we think it's worth investing in. And we hope that there will be very meaningful gains when we're able to do that successfully in the future.

And it'll take a lot of capacity, which requires a lot of CapEx, which is a hot button topic. I'm sure you get a lot of questions about. So you talked about the pipeline and new products to come. You have Mark on the last public call referring to the current systems as primitive to what they will look like over time with a lot of new projects to come both on and off the core. What types of analysis are you doing as the CFO when you're thinking through this is the right amount of CapEx to spend. How do you arrive at these numbers? How are you sort of putting math to it just to ensure that there's going to be enough revenue in a reasonable amount of time to deliver ROIC for the shareholders?

Yes. So the process is -- the process is kind of -- there are 2 parts of the process. One is, again, on that core work. And there, I think I alluded to this, we have a sort of a pretty robust budgeting process at this point where we kind of take the -- again, the expected inputs that is both across headcount needs and capacity needs, right? Those are kind of the inputs and then we look at, okay, well, this is what we expect the 1-year return to be. This is what we expect the 4-year return to be. Does this make sense? And that is a process we've run now for many years, I think, quite successfully. And that's how we build kind of what the capacity needs for the core are.

On the newer things, right, we are, I would say, at an earlier stage in the process. And so this is a little bit more -- if the prior process I described was very science, this is a little more art. And by that, what I mean is the teams that are working on basically AI training today, they have the most immediately sort of clearly defined buttoned-up road map for how much capacity, let's say, they think they need to train models for the next 12, 24 months. And then we need to be -- so that's still like -- that's kind of like a demand road map from the teams that they have more certainty into. The part I think that is the most challenging for us to have certainty around is inference needs because that's both -- you have to predict meaningfully into the future because of the lead time and getting capacity.

And we don't know exactly yet how the product experiences are going to take shape, which are the ones that will have the most scale, how will we use inference capacity internally, where will it drive the most productivity. These are all sort of questions that we are like living every day and trying to make our best predictions for what will be true in 12 months, 24 months, 60 months, right? And so what we want to do there is we have scenarios that we are looking at there, and we want to make sure that those scenarios pencil out next to what we also think the returns from these models, but more importantly, the experiences that we build on top of the models can look like. And so that is a little bit more of a -- again, it's like a little bit of a -- we're fitting together a picture of, okay, this is how much inference we need. And if this is how much we think that we're going to be able to grow the core business because of, again, using AI to make content and ads vastly more personalized to you, vastly more interactive.

I think this is just -- this area in particular, I think I worry almost that we will underestimate it because when we put things out, they scale so quickly if they're part of our existing FOA experiences, which then immediately billions of people are -- will have access to. And so I think that's a place where it's -- that's probably a place where it's easy for us to underestimate what our inference needs could be. And then you're thinking about new things again that don't exist at scale, and that's a totally different kind -- what should the trajectory of business agents out in the world look like? We certainly think the opportunity makes a lot of sense. We expect that in the not-distant future, every business is going to have at least one, if not many AI agents in the way that they have websites and e-mails and customer service and all of those things today, they're going to have agents that handle some number of these things on their behalf.

But it's a -- but because that doesn't exist at scale today, it's like a little harder to know what the trajectory should look like and how we want to roll that out and make sure we're doing it thoughtfully and doing it well. So we're really trying to make predictions over all these different spaces and think about the returns from each of those things also and make sure, again, that the math works out. And that's an imprecise. That's not like, okay, in 2026, the ROI is this in 2027, the ROI is this and so on, which pains me to be clear. I really wish that, that were the world we live in, but it's not. And we have to be willing to sort of make temporal bets, and that's a big part of what we have to do in an intelligent and thoughtful way.

Okay. As a Morgan Stanley analyst, we love our scenario analysis. So it sounds as you're doing scenarios on the ROIC.

Yes. That's right.

And one of those that involves the scenarios around sort of the frontier models and how to think about the super intelligence efforts and the different models that Mark just talked about. So a few questions. First, Mark talked about having some models to show us in coming months. Anything you want to talk about today on new models or not yet?

That I am going to leave to Mark and the AI folks to unveil at the right time.

No avocados [indiscernible], fine. Then let me ask you more conceptually then. There's been a lot of discussion about open source versus closed models. What is the company's latest philosophy on the importance of open source? And how do you think about the main monetization nodes of an open source model 3 years from now?

Yes. So I mean, first of all, I think our approach to open source has been one -- I mean, there's nuance to it, right? And that's been true historically. We don't open source everything we do. And I think we believe a lot, obviously, in open source as kind of a driver of innovation and standardization. And standardization brings efficiency, all things I love. But as the models become more and more capable, I think each model is kind of going to require its own thoughtful decision-making and discussion about whether to open source.

So now for -- in terms of, sorry, where we get returns from the models, it is really, we believe, going to be from the consumer experiences that we build. And I alluded to this a little bit in my earlier answer across the family of apps. How do we make the content that you engage with just like better, right? And what is possible today is different from what was possible -- what we thought possible 5 years ago, 10 years ago, 15 years ago for a lot of reasons, right? Part of it is the underlying technology infrastructure has gotten better. So when we first started, people were using this on their desktops, eventually, they moved to mobile. As mobile infrastructure got better, then the content moved from being more text-based to now visual and then there were a lot of photos being shared. And then we had things like photo tagging that made photo sharing even more delightful and you're kind of in this like great cycle of leveraging innovation from a lot of different areas and better infrastructure to make kind of the content experiences more engaging.

Now we have seen over the last couple of years, I think a big shift towards short-form video as being one of the sort of most engaging forms of content, again, enabled by a lot of things, including the infrastructure that lets people see short-form video and the ability to kind of again, rank and figure out what to show you. I think that, that is going to continue evolving on all those dimensions, like what if you could interact with the video? What if you could watch a video and say, "Oh, I wonder what would happen next if X," right? And then that's able to adapt for you, right? So I think AI kind of -- I don't know exactly what the word will be, whether it's AI-assisted content or maybe AI-generated content, I think has the potential to -- like I watch like a lot of videos with my kids about science because we're nerds. And like I wish like at the end of the video, there are a lot of great like YouTube videos for kids about like elements.

But often at the end, you're like, I have 4 more questions. My kid has more questions about this topic. It'd be great if they could just ask and then the next part of the video happens, right? Instead of I'm like, okay, sorry, let me go search for what this carbon allotrope is. And then -- but it would be better if you could just interact with it and it could give you what you're looking for. So I actually think the kind of intuitive extension of making content really interactive is something I'm, a, very excited about, and I think it's just like an obviously large adjacency to what we do today. And ads, I think, just to not belabor the point, are an extension of that. You get the individualized ad for you, I get the individualized ad for me from the same advertiser.

The advertiser doesn't even -- as part of our ongoing continuous now multi-decade effort to make advertising as streamlined as possible for the advertiser where they just come tell us how much they want to spend on something and we deliver it, this is like the next step in that journey for them. I think that the opportunity that comes from those 2 things, neither of which requires us to launch a brand-new business off the ground that doesn't exist today is already extremely large. And I'm very excited about both of them, and they require 0 leaps of the imagination, I think, to understand why they would be big businesses.

Setting aside those things, there are obviously more -- I mean, basically new AI experiences, business agents being one of them. Again, I think this is going to be something that will be very commonplace in a few years, even though today, we're not quite -- there's a funny -- there's a hotel that I was trying to book in Orlando. It turns out when you have kids, all the kid events are in Orlando for some reason, home of the greatest convention centers in the world. And I called this hotel and I thought I was -- like it seems very organic.

You called them.

Yes. I called them and I thought I was talking to someone. It took me about 4 minutes to realize I was not talking to a human. Like I was going through what was turning out to be a really confusing phone tree. But in the very beginning, I thought I was talking to a person for quite a long time. That experience should be way better over time, right? We bought Manus. We are excited to -- Manus already obviously exists as a very promising business, and we are excited to scale Manus and grow it and kind of have the notion of multipurpose AI, I think, is going to continue becoming more valuable for folks, and I think there will be a market there. And then I think there are things that, again, don't exist at scale today from like from a product perspective. But certainly, I think as we have those AI experiences that are built out, whether it's in Manus, whether it's in some future version of Meta AI, I think it will be I think monetizing those consumer experiences is not the hard part. I think it is growing the consumer experiences that is the hard part.

I actually -- you brought Meta at the very end. That's why I want to go because you didn't say Meta AI and then now you brought it up. So over the last 2 years, there's been a lot of ebbs and flows that sort of investors' confidence in Meta AI's positioning. I would say a couple of years ago, there was a lot more anticipation Meta AI could be a leading agent to compete against Gemini or GPT. I remember sitting up here in the past talking about booking your trip to Orlando through Meta AI. Now there is this external perception that Meta AI is behind, falling behind and not going to be able to catch up to the other players. What's your reaction to that? And how do you think about sort of the pace of product innovation on Meta AI throughout 2026?

Yes. Well, Meta AI has over 1 billion people who use it despite not being on a state-of-the-art foundation model. So while -- like I think I am totally clear-eyed in assessing where we are today, I am meaningfully more sort of optimistic than that framing, I think, about where this could go for a lot of reasons. One is, again, just the kind of scale of distribution. I think one of the playbooks we really have dialed in as a company is when you have a good product with consumer market fit, how do you leverage the distribution network we have to put it in front of a lot of people. But you don't want to do that until it's like a very valuable experience.

The second thing is I think that our ability to personalize -- to personalize your conversational AI agent to you is going to be second to none. I think both based on just your deep history of interacting with the platform already and our ability to understand that information and sort of use it to make sure we're building a good experience for you. So I feel -- I think that when we have a frontier model, I feel quite confident that the combination of that, the combination of the distribution graph, the network effects, the fact that there are a lot of very natural places to have Meta AI interact with you. I can have -- I can be in a WhatsApp conversation with my friends, and we can be talking about going to dinner and it can just -- an AI agent in the chat can just book the restaurant, right? Or we're planning a ski trip and an AI agent can like tell us where there's snow, evidently nowhere. But if there is snow, like where should we go and when and like what kind of gear do we need.

Like I think that -- I think kind of how -- I think the ways in which the family of apps as it exists today, I think, are a great scaffold for AI experiences to fit very neatly within them, I think, is -- are very intuitive. I'm excited for that, and I'm excited to see how that unfolds.

Okay. We're excited to see the product evolve. The other area that I want to sort of touch on before we get to the end is custom silicon. This is another part that sort of has been a multiyear evolving strategy, and it sort of has expanded. So maybe remind us where are you using your own custom silicon now? What have been the early learnings from that from an efficiency perspective? And how do you think about the next couple of years of expanding the use of custom silicon as opposed to using third-party chips?

Yes. I'll actually expand the question a little bit. Custom silicon is one of, I think, many -- I mean it is one and a very important part of an overall like how do we bring the cost of compute down strategy over time. And obviously, chips are the sort of most expensive piece of that. And again, apparently to date, the shortest-lived version because new chips come out and you want to leverage the better performance that you can get, but obviously, who knows over the longer term. And so for us, because we have so many -- the scale of our silicon needs across AI training, across anticipated AI inference, across the core ranking and recommendation work, across CPUs for keeping the site running in kind of the like bread and butter of running the family of apps.

We are -- we're really focused on basically making sure that we are getting the optimal chip for each workload and each -- and that combination is still at a scale that like lets us do this in a sort of cost-efficient way, if that makes sense. And so you've probably seen a number of announcements come out between us and some of the different chip providers. And that's all in service of that effort, which is like based on what we know today and our current needs, what do we think is the best chip to use for each of these use cases and some of them are totally off the shelf. Some of them are somewhat customized, some of them are very customized. And can we negotiate like what -- the volumes we need at what we think are attractive prices for kind of for each of those chips.

Custom silicon is a big part of that, obviously. Some of our workloads really are very customized to us. The sort of ranking and recommendations workloads have been where we have started, and that's the place where we have rolled out custom silicon at the most scale. But we expect and are hopeful that we are going to expand that over time, including eventually to training AI models. So that is obviously later in the road map. But we're feeling both -- I think we're feeling quite optimistic about the way those chips are performing today because, again, they really let us optimize for performance per dollar and like the total cost of ownership of the chips we need for each use case.

Great. Let me ask you one more. We talk a lot about the GenAI or the GPU opportunities and everything is sort of the growth in the pipeline. What is sort of the most underappreciated challenge or the factor that keeps you up at night when you're sort of thinking about making sure you execute on the right factors in this whole GenAI era in the next couple of years?

I think there are 2 dimensions to it. On the product side, I think I alluded to this earlier, but I think -- again, and I think this is so natural, but when we, including the industry at large, talk about kind of what the future sort of products and experiences are going to be. And clearly, there have been some great new products and experiences that have been built. We tend to think about new things. And I -- again, I think we underestimate how big the sort of taking like AI technology and making things that exist better, I think, is just going to be a -- that is going to be a massive opportunity. And I think I want to make sure that we are resourcing against that appropriately.

I also think AI tools are changing the way we work. And I think clearly, if you were starting a company today, you would use a lot of AI tools very differently, and you might set up a lot of your workflows very differently and you might set up your teams very differently. And we don't want to -- we -- like as a company that has now existed for over 20 years, we don't want to find ourselves behind companies that are being born today and that are AI native from like the very day of inception.

And so making sure that we stay on top of how would -- what would a team that is solving this problem like at a start-up, how would they think about solving it, right? And of course, we can't just perfectly do that because we have an inherently very large code base that already exists, and we've got lots of processes and things that exist for good reason. But I do think making sure that we're asking ourselves that question is very important so we don't get left behind.

And what we have found, I think I mentioned this on the last call that like AI tools are making our developers more productive. They are making our most effective -- our developers who are most effective at using them much more productive. They -- I think we -- I think 80% was sort of the stat we shared in terms of increase in coding productivity. And so -- and then what does that mean for -- well, how should we think about like how you use these tools to not only increase your own productivity, but also make it easier for you to work with other people, what should teams look like in that future? How do we -- we've got senior executives of the company who are like using AI tools to build their own agents and stitch together data from different sources and things that you used to have to ask, you sent an e-mail into like a team and would wait several hours for multiple different data sources to get stitched together for you, like people are getting that information much faster now, and that enables them to make decisions more quickly and to make more decisions, right?

And like the long-run goal of this is to do more things, right, and to do more things and build more experiences. And so making sure that we for a company as kind of at the size and scale that we are that we don't work any less efficiently than companies that are AI native from the start. That is, I think -- that's something that I think about a lot and want to make sure that we are as well set up to compete as any of them.

Great. Susan, thank you very much. We're excited to see all the efficiency and you guys do more and more things in the years to come.

Thank you so much.

Good afternoon. My name is Christa, and I will be your conference operator today. At this time, I would like to welcome everyone to the Meta Fourth Quarter and Full Year 2025 Earnings Conference Call. [Operator Instructions] This call will be recorded. Thank you very much.

Kenneth Dorell, Meta's Director of Investor Relations. You may begin.

Thank you. Good afternoon, and welcome to Meta Platform's Fourth Quarter and Full Year 2025 Earnings Conference Call. Joining me today to discuss our results are Mark Zuckerberg, CEO; and Susan Li, CFO.

Our remarks today will include forward-looking statements, which are based on assumptions as of today. Actual results may differ materially as a result of various factors, including those set forth in today's earnings press release and in our quarterly report on Form 10-Q filed with the SEC. We undertake no obligation to update any forward-looking statements.

During this call, we will present both GAAP and certain non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in today's earnings press release and an accompanying investor presentation are available on our website at investor.atmeta.com.

And now I'd like to turn the call over to Mark.

All right, everyone. Thanks for joining us. We ended 2025 strong with more than 3.5 billion people now using at least one of our apps every day. That includes more than 2 billion daily actives each on Facebook and WhatsApp and just shy of that on Instagram. Our business also performed very well, thanks to record-breaking holiday demand and AI-driven performance gains.

We are now seeing a major AI acceleration. I expect 2026 to be a year where this wave accelerates even further on several fronts. We're starting to see agents really work. This will unlock the ability to build completely new products and transform how we work.

In '25, we rebuilt the foundations of our AI program. Over the coming months, we're going to start shipping our new models and products. I expect our first models will be good, but more importantly, we'll show the rapid trajectory that we're on. And then I expect us to steadily push the frontier over the course of the year as we continue to release new models.

I'm very excited about the products that we're building. Our vision is building personal super intelligence. We're starting to see the promise of AI that understands our personal context, including our history, our interests, our content and our relationships. A lot of what makes agents valuable is the unique context that they can see. And we believe that Meta will be able to provide a uniquely personal experience.

We're also working on merging LLMs with the recommendation systems that power Facebook, Instagram, Threads and our ad system. Our world-class recommendation systems are already driving meaningful growth across our apps and ads business, but we think that the current systems are primitive compared to what will be possible soon. Today, our systems help people stay in touch with friends, understand the world and find interesting and entertaining content. But soon, we'll be able to understand people's unique personal goals, and tailor feeds to show each person content that helps them improve their lives in the ways that they want.

This also has implications for commerce. Our ads today help businesses find just the right very specific people who are interested in their products. New Agentic shopping tools will allow people to find just the right very specific set of products from the businesses in our catalog. We're focused on making these experiences work across both our feeds and across business messaging, significantly increasing the capabilities of WhatsApp over time.

New kinds of content will soon be possible as well. People want to express themselves and experience the world in the most immersive and interactive way as possible. We started with text and then moved to photos when we got phones with cameras and then moved to video when mobile networks got fast enough. Soon, we'll see an explosion of new media formats that are more immersive and interactive and only possible because of advances in AI.

Our feeds will become more interactive overall. Today, our apps feel like algorithms that recommend content. Soon, you'll open our apps, and you'll have an AI that understands you and also happens to be able to show you great content or even generate great personalized content for you.

Glasses are the ultimate incarnation of this vision. They're going to be able to see what you see, hear what you hear, talk to you and help you as you go about your day and even show you information or generate custom UI right there in your vision. Sales of our glasses more than tripled last year, and we think that they're some of the fastest-growing consumer electronics and history. Billions of people wear glasses or contacts for vision correction and I think that we're in a moment similar to when smartphones arrived, and it was clearly only a matter of time until all those flip phones became smartphones. It's hard to imagine a world in several years where most glasses that people wear aren't AI glasses.

For Reality Labs, we are directing most of our investment towards glasses and wearables going forward while focusing on making Horizon a massive success on mobile and making VR a profitable ecosystem over the coming years. I expect Reality Labs losses this year to be similar to last year, and this will likely be the peak as we start to gradually reduce our losses going forward while continuing to execute on our vision.

As we plan for the future, we will continue to invest very significantly in infrastructure to train leading models and deliver personal super intelligence to billions of people and businesses around the world. I recently announced Meta Compute with the belief that being the most efficient at how we engineer, invest and partner to build our infrastructure will become a strategic advantage. Dina Powell McCormick also joined us as President and Vice Chairman, and she will lead our efforts to partner with governments, sovereigns and strategic capital partners to expand our long-term capacity, including ensuring positive economic impact in the communities that we operate in around the world.

An important part of Meta Compute will be making long-term investments in silicon and energy. We will continue working with key partners while advancing our own silicon program. We're architecting our systems that we can be flexible in the systems that we use, and we expect the cost per gigawatt to decrease significantly over time through optimizing both our technology and supply chain.

The last thing that I want to mention is that I think that 2026 is going to be the year that AI starts to dramatically change the way that we work. As we navigate this, our North Star is building the best place for individuals to make a massive impact. So to do this, we're investing in AI native tooling so individuals at Meta can get more done, we're elevating individual contributors and flattening teams. We're starting to see projects that used to require big teams now be accomplished by a single, very talented person. I want to make sure that as many of these very talented people, as possible, choose Meta as the place that they can make the greatest impact to deliver personalized products to billions of people around the world. And if we do this, then I think that we're going to get a lot more done. And I think it's going to be a lot more fun.

All right. That's everything I wanted to cover. This is going to be a big year for delivering personal super intelligence, accelerating our business, building infrastructure for the future and shaping how our company will work going forward. As always, I am grateful for all of the hard work of our teams and to all of you for being on this journey with us. And now here is Susan.

Thanks, Mark, and good afternoon, everyone. Let's begin with our segment results. All comparisons are on a year-over-year basis unless otherwise noted. Our community across the Family of Apps continues to grow, and we estimate more than 3.5 billion people used at least one of our Family of Apps on a daily basis in December. Q4 total Family of Apps revenue was $58.9 billion, up 25% year-over-year. Q4 Family of Apps ad revenue was $58.1 billion, up 24% or 23% on a constant currency basis. In Q4, the total number of ad impressions served across our services increased 18%. Impression growth was healthy across all regions, driven primarily by engagement and user growth and, to a lesser degree, ad load optimizations. The average price per ad increased 6% year-over-year, benefiting from increased advertiser demand, largely driven by improved ad performance. Family of Apps other revenue was $801 million, up 54%, driven by WhatsApp paid messaging revenue growth as well as Meta Verified subscriptions.

Within our Reality Labs segment, Q4 revenue was $955 million, down 12% year-over-year. As we noted on the last call, the year-over-year decline in Reality Labs revenue is due to us lapping the introduction of Quest 3S in Q4 of 2024 as well as retail partners procuring Quest headsets during the third quarter of 2025 to prepare for the holiday season, which was recorded as revenue in Q3.

Moving now to our consolidated results. Q4 total revenue was $59.9 billion, up 24% or 23% on a constant currency basis. Q4 total expenses were $35.1 billion, up 40% compared to last year. Year-over-year growth was driven primarily by employee compensation expenses, legal expenses and infrastructure costs.

Growth in employee compensation expenses reflects the technical hires we've added this year, particularly AI talent. Legal expense growth was due to both lapping legal accrual reversals in Q4 of '24 and charges recorded in Q4 '25. Infrastructure expense growth was driven by higher depreciation, cloud spend and other operating expenses. We ended Q4 with over 78,800 employees, up 6% year-over-year. driven by hiring in priority areas of monetization, infrastructure, Meta Superintelligence Labs as well as regulation and compliance.

Fourth quarter operating income was $24.7 billion, representing a 41% operating margin. Q4 interest and other income was $609 million, driven primarily by unrealized gains on our equity investments. Our tax rate for the quarter was 10%, slightly lower than our outlook of 12% to 15% due to the settlement of matters with tax authorities. Net income was $22.8 billion or $8.88 per share. Capital expenditures, including principal payments on finance leases were $22.1 billion, driven by investments in data centers, servers and network infrastructure. Free cash flow was $14.1 billion. We ended the quarter with $81.6 billion in cash and marketable securities, and $58.7 billion in debt.

Turning now to the business performance. There are two primary factors that drive our revenue performance: our ability to deliver engaging experiences for our community and our effectiveness at monetizing that engagement over time. On the first, we're continuing to drive incremental engagement from ranking and product improvements. Instagram Reels had another strong quarter with watch time up more than 30% year-over-year in the U.S. Engagement is benefiting from several optimizations we made to improve the quality of recommendations including simplifying our ranking architecture to enable more efficient model scaling. This unlocks the ability for our systems to consider longer interaction histories to better identify a person's interests.

On Facebook, video time continued to grow double digits year-over-year in the U.S., and we're seeing strong results from our ranking and product efforts on both feed and video surfaces. The optimizations we made in Q4 drove a 7% lift in views of organic feed and video posts on Facebook, resulting in the largest quarterly revenue impact from Facebook product launches in the past two years.

We're continuing to increase the freshness and originality of content recommendations as well. On Facebook, our systems are surfacing over 25% more reels published that day than the prior quarter. On Instagram, we grew the prevalence of original content in the U.S. by 10 percentage points in Q4 with 75% of recommendations now coming from original posts. Threads is also seeing strong momentum again, benefiting from recommendation improvements. The optimizations we made in Q4 drove a 20% lift in threads time spent.

Turning to 2026. We see a lot of opportunity to drive additional gains. This includes scaling the complexity and amount of training data we use in our models while continuing to make our systems more responsive to people's real-time interest. We're also focused on incorporating LLMs to understand content more deeply across our platform, which will enable more personalized recommendations.

Another big area of investment this year is developing the next generation of our recommendation systems. We have several big bets on this front, including building new model architectures from the ground up that will work on top of LLMs, leveraging the world knowledge and reasoning capabilities of an LLM to better infer people's interests.

Beyond improvements to our recommendation systems, we expect to use the models developed by Meta Superintelligence Labs to deliver compelling and differentiated AI products. One area we're already seeing promise is with AI dubbing of videos into local languages. We are now supporting 9 different languages with hundreds of millions of people watching AI translated videos every day. This is already driving incremental time spent on Instagram, and we plan to launch support for more languages over the course of this year.

We are also seeing strong traction with our media creation tools. Nearly 10% of the reels people view each day are now created in our Edits app, almost tripling from last quarter. Within Meta AI, the number of daily actives generating media tripled year-over-year in Q4. This year, we expect to advance the capabilities of our underlying media generation models and ship new features to further enhance the product experience.

Another area we're focused on for Meta AI is personalization. We're seeing in our early testing that personalized responses drive higher levels of engagement, and we expect to significantly advance the personalization of Meta AI this year. This dovetails with our investments in content understanding, which will enable our systems to develop a deeper understanding of each person's interests and preferences while also identifying the most relevant content across our platform to pull into responses.

Turning to the second driver of our revenue performance: increasing monetization efficiency. The first part of this work is optimizing the level of ads within organic engagement. Here, our focus remains on tuning our systems to identify the right time and place to deliver ads. In some cases, this enables us to grow the overall level of ad load while preserving the user experience. However, an increasingly important part of this work is finding opportunities to drive incremental conversions within the same overall level of ad load by determining when a person is more interested in seeing an ad. In fact, in the second half of 2025, our initiatives on Facebook to redistribute ads across users and sessions delivered a nearly 4x larger revenue impact than Facebook ad load increases.

We also continue to make progress on bringing ads to our newer services. Within threads, we're beginning to expand ads to all remaining countries this month, including the U.K., European Union and Brazil. On WhatsApp, we expect to complete the rollout of ads in status throughout the year with the level of ads remaining low in the near term while we follow our standard approach of optimizing ad formats and performance before ramping inventory.

Moving to the second part of increasing monetization efficiency: improving performance for the businesses who use our tools. We're seeing very strong results from the ad performance investments we made throughout 2025 with year-over-year conversion growth accelerating through the fourth quarter. We expect the set of investments we're making in 2026 will enable us to drive further gains as we continue to integrate AI across all layers of the marketing and customer engagement funnel.

The first area is our ad system where we're continuing to scale the complexity and size of our models to better select which ads to ship. In Q4, we doubled the number of GPUs we used to train our GEM model for ads ranking. We also adopted a new sequence learning model architecture, which is capable of using longer sequences of user behavior and processing much richer information about each piece of content. The GEM and sequence learning improvements together grow a 3.5% lift in ad clicks on Facebook and a more than 1% gain in conversions on Instagram in Q4. This new sequence learning architecture is significantly more efficient than our prior architectures which should enable us to further scale up the data, complexity and compute we use in our future ranking models to deliver performance gains.

As we scale up our foundational ads models like GEM, we are also developing more advanced models to use downstream of them at run time for ads inference. In Q4, we launched a new run time model across Instagram Feed stories and reels, resulting in a 3% increase in conversion rates in Q4.

We continue to progress on our model unification efforts under Lattice as well. After seeing strong success with the consolidation of Facebook feed and video models in the first half of 2025. In Q4, we consolidated models for Facebook Stories and other services into the overall Facebook model. This, along with a series of back-end improvements drove a 12% increase in ads quality. And in 2026, we expect to consolidate more models than we had in the prior two years as we continue to evolve our systems towards running a smaller number of highly capable models.

Moving to the next area: ads products. We continue investing in ways to help businesses leverage AI to reduce the friction of setting up and optimizing an ad campaign. In Q4, we started testing our Meta AI business assistant with advertisers which helps with tasks like campaign optimization and account support. In the coming months, we'll make it available to more advertisers, so each business has an AI assistant they can chat with that remembers their businesses' goals and provides personalized recommendations on how to improve performance.

Another area we're deploying AI to improve performance is ad creative. The combined revenue run rate of video generation tools hit $10 billion in Q4, with quarter-over-quarter growth outpacing the increase in overall ads revenue by nearly 3x. We are also seeing very good results from our incremental attribution feature, which optimizes for incremental conversions in real time. Our latest model rollout in Q4 is driving a 24% increase in incremental conversions versus our standard attribution model, and this product has already achieved a multibillion-dollar annual run rate just 7 months since launching.

The last area of our monetization [ work ] I'll cover is business messaging, where we're seeing strong momentum across our portfolio of solutions. Click to message ads revenue growth accelerated in Q4 with the U.S. up more than 50% year-over-year, driven by strong adoption of our website to message ads which direct people to a business's website for more information before choosing to launch a chat. Paid messaging within WhatsApp continues to scale as well, crossing a $2 billion annual run rate in Q4. Finally, we're seeing good early traction with our business AIs in Mexico and the Philippines, with over 1 million weekly conversations between people and business AI is now happening on our messaging platforms. This year, we will expand availability of our business AIs to more markets, while also extending their capabilities so they not only answer questions on topics like product availability, but can help people get things done right within WhatsApp.

We speak a lot about how AI is improving our products, but I'd like to take a moment to give an update on how it's changing the way we work. Mark mentioned our focus on making Meta a place where individuals can have significant impact. A big focus of this is to enable the adoption and advancement of our AI coding tools where we're seeing strong momentum. Since the beginning of 2025, we've seen a 30% increase in output per engineer with the majority of that growth coming from the adoption of agenetic coding, which saw a big jump in Q4. We're seeing even stronger gains with power users of AI coding tools, whose output has increased 80% year-over-year. We expect this growth to accelerate through the next half.

Next, I would like to discuss our approach to capital allocation. We have significant opportunities to improve our core business in 2026. We plan to continue to prioritize investing in the business to support these opportunities while also positioning us for an entirely new and exciting product cycle over the coming years, powered by our AI models.

Procuring sufficient infrastructure capacity is central to these initiatives, and we're working to meet our silicon needs by deploying a variety of chips that optimally support each of our different workloads. To that end, in Q4, we extended our Andromeda ads retrievable engine, so it can now run on NVIDIA, AMD and MTIA. This, along with model innovations, enabled us to nearly triple Andromeda's compute efficiency. In Q1, we will extend our MTIA program to support our core ranking and recommendation training workloads in addition to the inference workloads it currently runs.

More broadly, as we invest in infrastructure to meet our business needs, we continue to prioritize maintaining long-term flexibility so we can adapt to how the market develops. We're doing so in several ways, including changing how we develop data center sites, establishing strategic partnerships, contracting cloud capacity and establishing new ownership structures for some of our large data center sites. We have a strong net cash balance and expect our business will continue to generate sufficient cash to fund our infrastructure investments in 2026, which is reflected in our expectations. Nonetheless, we will continue to look for opportunities to periodically supplement our strong operating cash flow with prudent amounts of cost-efficient external financing, which may lead us to eventually maintain a positive net debt balance.

Moving to our financial outlook. We expect our first quarter 2026 total revenue to be in the range of $53.5 billion to $56.5 billion. Our guidance assumes foreign currency is an approximately 4% tailwind to year-over-year total revenue growth based on current exchange rates.

Turning to the expense and CapEx outlook. We expect full year 2026 total expenses to be in the range of $162 billion to $169 billion. The majority of expense growth will be driven by infrastructure costs which includes third-party cloud spend, higher depreciation and higher infrastructure operating expenses. The second largest contributor to total expense growth is employee compensation driven by investments in technical talent. This includes 2026 hires to support our priority areas, particularly AI as well as a full year of expenses from 2025 hires. At a segment level, we expect expense growth to be driven by the Family of Apps with Reality Labs' operating losses remaining similar to 2025 levels. We anticipate 2026 capital expenditures, including principal payments on finance leases to be in the range of $115 billion to $135 billion, with year-over-year growth driven by increased investment to support our Meta Superintelligence Labs efforts and core business.

Despite the meaningful step-up in infrastructure investment, in 2026, we expect to deliver operating income that is above 2025 operating income. Absent any changes to our tax landscape, we expect our full year 2026 tax rate to be 13% to 16%.

Finally, we recently aligned with the European Commission on further changes to our less personalized ads offering, which we will begin rolling out this quarter. However, we continue to monitor legal and regulatory headwinds in the EU and the U.S. that could significantly impact our business and financial results. For example, we continue to see scrutiny on youth-related issues and have a number of trials scheduled for this year in the U.S., which may ultimately result in a material loss.

In closing, 2025 was another strong year for our company. The investments we've made to improve our business are continuing to drive strong growth, and we have an exciting road map this year to deliver new experiences and services for our global community. As always, thank you to our teams for their hard work and commitment to our mission.

With that, Christa, let's open up the call for questions.

[Operator Instructions] Your first question comes from the line of Brian Nowak with Morgan Stanley.

I have one for Mark, one for Susan. Mark, one is a long-term question. As you think about ramping all this investment and the personal intelligence opportunity, the Meta compute opportunity. Can you walk us through a little bit how you think about the largest revenue or ROIC long-term opportunities you're trying to unlock with those over the next, call it, 3, 5, 10 years through all the investment?

And then Susan, a little more near term, more like '26. I think the guide is the fastest growth you've had in almost 5 years. I know you have a lot of improvements on recommendations and monetization efficiency. But can you just sort of help us a little bit understand 2 or 3 of the biggest drivers of this inflection you're seeing on revenue in '26?

Yes. I guess I can start with the first one. Although I have to say upfront that I think my answers to a lot of your questions on this particular call, maybe somewhat unfulfilling because we're in this interesting period where we've been rebuilding our AI effort, and we're 6 months into that, and I'm happy with how it's going. But we are going to be rolling out our initial set of models and products and businesses around that over the coming months. And I will have a lot more to share on all of those fronts at that point. So I'm happy to offer kind of a high-level view of some of the stuff, but I apologize in advance that not much of this is going to be particularly detailed, but it will be exciting as we roll it out.

I think the theme on the business, I mean, this is -- I don't think I'm going to break any new ground here, but there are several major business opportunities that we're focused on. I think that -- one is just going to be improving the core products and accelerating the current business. I talked about that in terms of the connecting of the recommendation systems and the LLMs, which I think will both improve the quality of the organic experience and of advertising.

We're going to see the generation of media improve the quality of content, which, coupled with the improvements in the recommendation systems, we expect to generally accelerate the quality and effectiveness of the core business, both for people who use it organically and for businesses. So I think that will have a compounding effect.

And then there's going to be several, many, I think, new business opportunities that come up. I mean, we have been working on Meta AI for a while. I think you're starting to see some of the way that products like that get monetized across the industry when we get that to a scale and depth that we want. We think that there are going to be opportunities, both in terms of subscriptions and advertising and all of the different things that you see on that.

And I mean, yes, I think -- there's a number of things on shopping and commerce that I'm quite excited about that I alluded to in the comments upfront. And as the models launch and we demonstrate some of the capabilities, both in the first set of models and over the year, I think the models are going to be a lot better, too. We'll be able to have different products paired with those that I think will facilitate different businesses for businesses who use us and our platforms as well as direct-to-consumer businesses.

I guess it's probably also worth flagging because I don't think either of us mentioned that the Manus acquisition in the upfront comments, I mean that is going to -- is a good example of -- you have a significant number of businesses that already pay a subscription to basically use their tool to accelerate their business results and integrating that kind of thing into our ads and business managers, so that way we can just offer more integrated solutions for the many, many millions of businesses that use and rely on our platforms is going to be really powerful, both for accelerating their results using the existing products that we have and I think adding new lines as well. So a somewhat high-level answer and the -- I think the picture will become clearer. And I think more exciting if we do our jobs well over the course of the year.

Brian, on your second question, there's obviously a range of outcomes captured in the Q1 '26 revenue outlook. It overall reflects our expectation for a strong quarter of growth. The range embeds an outlook for accelerated growth, and that's really underpinned by the strong demand that we saw through the end of Q4 and continuing into the start of 2026.

Now I will say we also expect foreign currency to be a 4-point benefit to year-over-year growth. So that is a 3-point larger tailwind than it was in Q4 '25 as we lap the strengthening of the U.S. dollar a year ago. But overall, we see that advertisers are responding to a performance improvements that we made. They're driving strong conversion growth. We've made a lot of these investments over the course of 2025, including advances to our ads ranking and delivery systems, the more effective redistribution of ad load, new features and ad products like Advantage+, better measurement and just a lot of great work that has helped to drive the continued performance of our ads.

Your next question comes from the line of Eric Sheridan with Goldman Sachs.

Maybe two, if I could. In prior periods, you've talked about being capacity constrained internally and not having enough compute to sort of achieve the goals you have on the platform on a product standpoint. I want to know if we can get any update on currently how you think about your own internal needs for compute against that road map?

And the second part of the question would be, as we continue to see the ads business sort of scale, especially in terms of dollar growth year-on-year. Have we yet seen the full first order effects of scaling the business against applying more compute to it? Or how should investors think about the direction of relationship between applying more compute and rate of change in terms of outcomes on the monetization side?

On your first question, we do continue to be capacity constrained. Our teams have done a great job ramping up our infrastructure through the course of 2025. But demands for compute resources across the company have increased even faster than our supply. So we expect over the course of 2026 to have significantly more capacity this year as we add cloud. But we'll likely still be constrained through much of 2026 until additional capacity from our own facilities comes online later in the year.

With that said, I think we have done a good job internally mitigating the impact of compute constraints on our business. I expect that will continue to be the case in 2026. We're continuing to focus on increasing our infrastructure efficiency in several ways, including by optimizing workloads, improving infrastructure utilization, diversifying our chip supply and just investing in efficiency improvements as part of our core technology development efforts in areas like content and ads ranking. So that was your first question.

The second question about how the ads business scales. I think we don't -- I don't have an extremely precise answer to this question. What I'd say is one of the ways that we are working to drive ads performance improvements is by improving our larger scale models along with our lighter weight ones that we use for ads and for instance, run time. We don't typically use our larger model architectures like GEM for inference because their size and complexity would make it too cost prohibitive. So the way that we drive performance from those models is by using them to transfer knowledge to smaller lightweight models used at run time. But I would say that we think that there is room for our larger models to benefit from having more compute. And I think as we scale up the compute available to those models, and the foundational models in different areas that power the different stages of ads ranking and recommendation, we expect that we will see gains coming from that.

Your next question comes from the line of Mark Shmulik with Bernstein.

Two if I may. Mark, kind of with your comments that you kind of expect to see some meaningful changes and how work and things are done this year. I guess would you be surprised just kind of by the end of the year, we've yet to see meaningful progress and adoption on some of the newer products and initiatives that you're launching? Or should we just be a bit more patient on the time line here?

And then, Susan, kind of with the guidance provided on [ OI ] still expected to grow kind of faster this year than last year? Let's say in a few months, we realize we need more investment and resources to continue to go after the AI opportunity, but perhaps macro might be a bit weaker. How hard of a line is there in terms of this high of investment levels to core performance?

I think the first question was asking about kind of when do I expect the product impact to be. I mean we're going to roll out new products over the course of the year. I think the important thing is we're not just launching one thing, and we're building a lot of things. I think they're -- like AI is going to enable a lot of new experiences. I outlined thematically a bunch of these in the upfront comments around personal AI around LLMs combining with the recommendation systems. I think that's a somewhat longer-term research project that I think will yield dividends over a long period of time, but we're already definitely seeing optimizations of the recommendation systems as we're including more of the AI research improvements and advances into that. The content is going to improve. There are going to be new formats. There are going to be improvements on the glasses. There are all these different things as well as several things that we think are new that we're going to try that are not just extensions of the current things that we're doing.

So yes, I mean, I would expect that we'll roll these out over the course of the year and that sometimes it takes a few iterations for things to really hit and reach the kind of product market fit that you need. But I think we have enough time, hopefully to -- we're starting off early enough in the year that I would expect that we'll see some successes by the end of the year on this as well as on the work side, what we were talking about is.

I think it's very hard for anyone exactly to predict what the shape of how organizations working is going to feel. But it I just think the fact that agents are really starting to work now is quite profound. And I think it is going to allow -- we're already starting to see the people who adopt them are just being significantly more productive. And there's a big delta between the people who do it and do it well and the people who don't. And I think that's going to just be a very profound dynamic for, I think, across the whole sector and probably the whole economy going forward in terms of the productivity and efficiency with which we can run these companies, which I think -- my hope is that we can use that to just get a lot more done than we were able to before. And I'm most focused on making sure that Meta is a great company to have a big impact, where you'll be able to use these kind of agentic tools anywhere, but you will only be able to come and ship things to billions of people if you join a company like Meta and there aren't that many companies like Meta. So I think if we make it so that we can harness these kind of tools, then I think that we should, over some period of time, start to see a real acceleration in the amount of output that we could have.

Now how to predict exactly the time frame for adopting that is somewhat hard, right? I'm not going to predict a specific quarter or something like that. But the trend seems like unmistakably like this is going to happen. And that, to me, assuming that is very exciting. And like I said in my comments upfront, also, look, honestly, kind of fun, right? I think it just makes it more fun to be able to build a lot of things. And that's what we're here to do.

Mark, on your second question, I want to make sure and clarify something. So I think in the question, you had said that operating income growth in '26 would be higher than '25. And I want to make sure my comments were super clear. In 2026, we expect to deliver operating income above 2025 operating income. So this is comparing absolute dollars, not year-over-year growth.

So to give some context on that, we are going into 2026 with strong revenue growth at the start. Of course, we are just a few weeks in set against a healthy macro backdrop. So obviously, hard to extrapolate the current trends to the full year, and there are many moving variables in the current landscape. We're really taking advantage of the current business strength to reinvest a lot of the revenue into what we see as very attractive investment opportunities in AI infrastructure and talent. It's hard to assess what all of those investment opportunities will be over the course of the year as we continue to work through our capacity options. And of course, it remains a very competitive hiring market. But we'd like to invest aggressively where we can. We continue to use our framework that we shared at this point several years ago of growing consolidated operating profit over time to guide those investments and based on where our plans are rolling up today, again, in '26, we expect to deliver more operating income than we did in 2025.

Your next question comes from the line of Doug Anmuth with JPMorgan.

One for Mark and one for Susan. Mark, could you just provide more detail on the progress of the MSL team several months in? And more on your view on the path to a frontier model this year?

And then, Susan, I know you expect to grow operating income in '26. Do you also expect to have positive free cash flow? Just how should we think about the current and any future JVs for data center and compute build-out?

I'm not sure I have anything else to add on the current progress on this. I mean I -- that's why I said upfront that I think this is somewhat of an unfulfilling time to be answering some of these questions.

We're about 6 months into building MSL. I'm very pleased with the quality of the team. I think we have the most talent-dense research effort in the industry and some of the early indicators look positive. But look, I think that this is going to -- is a long-term effort, right? We're not here to do this to ship like one model or one product. We're doing a lot of models over time and a lot of different products. And I want to make sure that the work can speak for itself and also that we all internalize that this is a journey that we're on. And the first set of things that we put out, I think, are going to be more about showing the trajectory that we're on rather than being a single moment in time. So yes, I'm quite optimistic, but don't have anything else particularly concrete to share.

Doug, on the first part of your question, we are making very significant investments in infrastructure capacity this year to support our AI efforts. And we believe we're in a strong position to support them with the cash generation of the -- of our business this year. And at the same time, we'll continue to explore different paths as we build out our infrastructure capacity that help us provide that help provide us the long-term flexibility and option value that we look for as we support our future capacity needs against the backdrop of a very wide range of possible capacity demand over the years to come.

So we don't have anything additional to announce at this point. We are looking at all of the different opportunities to stand up capacity across kind of the different time frames that we need them.

Your next question comes from the line of Justin Post with Bank of America.

A couple, maybe one for Mark and one for Susan. It just seems like you're going to have a tremendous amount of capacity. How do you think about expanding your opportunities beyond ads, things like subscriptions or licensing cloud models? Just with all the interesting things you're building, I don't expect any product announcements, but can you do things beyond ads?

And then for Susan, it's really interesting to see the acceleration even [ ex FX in ] advertising. I'm just wondering if you're seeing a general acceleration in e-commerce activity, where do you think the dollars are coming from? And is the entire Internet ecosystem accelerating? I'm just wondering your thoughts on that.

So yes, we are focused on things beyond ads, I think the numbers make it so that for the next couple of years, ads are going to be, by far, the most important driver of growth in our business. So that's why, as we're working on this, we have a balance of new things that we're trying to do, while also investing very heavily and making sure that all of the work that we're doing in AI improves both the quality and business performance of the core apps and businesses that we run there. But yes, I mean, we'll have more to share on that. But I mean all these things, even if they scale very quickly are going to take some time to be meaningful at the scale of what the ads business is. And while we're doing that, we're just very focused on also delivering more value to businesses and more quality in the apps that we run ads in.

Justin, on your second question, we saw healthy year-over-year growth across all verticals in Q4 with the exception of politics as we lapped the U.S. presidential election last year. The online commerce vertical was the largest contributor to year-over-year growth. That was followed by professional services and technology. So in online commerce, year-over-year growth was strong. It was actually relatively consistent with Q3 levels and that was broad-based across advertiser regions and sizes. In general, we saw that the demand leading up to the holiday shopping period that's sustained through Cyber 5 and into the end of the year was very healthy for us. Professional services in this category, we saw a strong broad-based growth of nice contributions from lead generation ads due to product improvements we've made, including from Advantage+ lead campaigns that we fully rolled out at the start of Q4 and the tech vertical continues to be strong for us, too, again, broad-based across advertiser regions and sizes. So in general, I would say it was very healthy, broadly growing growth.

Your next question comes from the line of Ross Sandler with Barclays.

Mark, you mentioned bringing Horizon Worlds into mobile. We haven't heard much from the Horizon Worlds squad on these calls. So interesting that that's making it in. It seems like the combo of AI and what you guys have built with Horizon might open up the door to a bunch of new potential areas in gaming or new forms of kind of communication. So could you just elaborate on what the plan is there?

Yes. So let me talk about the basic theme here. One core idea that I've talked about on some of these calls over the years is that people always want to express themselves and experience the world in whatever the richest format is that they can. So I talked about the upfront today. It's when we started a lot of this was text, right? That was the kind of the best we could do. Then we all got phones, then cameras, like a lot of this medium became visual but with photos, we went through a period where the mobile networks were kind of weak and every time you wanted to watch a video, it would buffer. And once that got worked out, now the majority of the content is video.

And one of the core ideas that we have had for a while is that, that is not the end of the line, right? Video will continue to be here for a long time. It's going to continue growing, it's not going anywhere, just like photos and text in many ways, continue to grow even as the market continues to grow beyond that. But I don't think that video is the ultimate kind of final format. I just -- I think that this is going to get -- we're going to get more formats that are more interactive and immersive and you're going to get them in your feeds. So you can imagine this, I mean there's obviously a lot of details to fill in on this, but you can imagine people being able to, easily through a prompt, create a world or create a game and be able to share that with people who they care about and you see it in your feet and you can jump right into it and you can engage in it. And there are 3D versions of that, and there are 2D versions of that and Horizon, I think, fits very well with the kind of immersive 3D version of that. But there's definitely a version of the future where any video that you see, you can like tap on and jump into it and like engage and like and be kind of like experience in it in a more meaningful way. And I think that the investments that we've done in both a lot of the virtual reality software and Horizon as well as a number of other areas around the company are actually going to pair well with these AI advances to be able to bring some of those experiences to hundreds of millions and billions of people through mobile.

So anyway, that's a thing that I'm quite excited about, but it's just sort of one flavor of a theme that I think is going to be very interesting. I think there are going to be lots of different types of interactive and immersive content that become possible. And I think Horizon is going to be one very interesting example that I'm quite excited to see how this unfolds.

Your next question comes from the line of Ron Josey with Citi.

I wanted to drill down, maybe, Susan, on your comments around ranking recommendation model changes. Clearly, lots of tailwinds here given the results from GEM, Andromeda, Lattice, consolidation of models, et cetera. So can you help us understand a little bit more just about the road map and where we stand within ranking recommendation model changes. There's a thesis out there that maybe where -- there's a limiting factor, maybe we're waiting on newer models, but any insights there would be very helpful as we think about the next -- as the future going forward.

Yes. Thanks for the question, Ron. We have -- I just -- I was sort of sorting out if your question was more specific to ads or if it was more specific to kind of the engagement side, but I'll try to talk a little bit about both. So on the sort of core engagement piece, we launched several ranking improvements in Q4 on Facebook and Instagram that drove incremental engagement. And there isn't really one single launch that is driving most of the gains. It's multiple optimizations to our recommendation systems that are helping us make more accurate predictions about what will be interesting to each person. And I talked a little bit about some of these -- the specific instantiations on both Facebook and on Instagram. And we see a lot of headroom to improve recommendations in 2026 which we expect will drive additional engagement growth on both apps.

First, we plan to continue scaling up our models and increase the amount of data we use, including a longer history of content interactions to further improve the overall quality of recommendations. We're also going to start validating the use of ad signals and organic content recommendations as we continue to work towards having a more shared platform for organic and ads recommendations over time.

Second, we're going to continue to make recommendations even more adaptive to what a person is engaging with during their session. So the recommendations we surface are more relevant to what they're interested in at that moment.

And finally, we will work on more deeply incorporating LLMs into our existing recommendation systems, given their capability to more deeply understand content. And so this will, I think, in particular, be useful for content that has been more recently posted since there's less engagement data to base recommendations off of.

On the ad side, again, we have -- we've talked about a lot of the sort of model work in the ads world across Andromeda and Lattice and GEM. I'll touch maybe specifically on GEM, in Q4, we extended GEM to cover Facebook reels. Now it covers all major surfaces across Facebook and Instagram. We also doubled the size of the GPU cluster we used to train it. And in 2026, we're expecting to meaningfully scale up GEM training to an even larger cluster, increasing the complexity of the model, expanding the data that we trained it on, leveraging new sequence, learning architecture that we had begun deploying in Q4. And we're also going to further improve how we transfer the learnings from our GEM foundation models to the runtime models that we're using.

So there are -- there's a lot more headroom, I think, across many, many components of the stack. This is the first time we have found a recommendation model architecture that can scale with similar efficiency as LLMs. And we're hoping that this will unlock the ability for us to significantly scale up the size of our ranking models while preserving an attractive ROI.

Our next question will come from the line of Ken Gawrelski with Wells Fargo.

Two, if I may, please. First, for Mark, how critical is it for Meta to have a leading general purpose model or is there a sufficient capability in a model that really excels at specific use cases, maybe similar to what you see at Anthropic in coding today? We'd love if you could opine on that.

And then second, maybe I just want to push again maybe on this last question, Susan. On the visibility you have, you talked about the improvements you're making in '26 on the models, the fine-tuning of the core, both in engagement and ad relevance. Could you talk about -- are you seeing any signs of diminishing returns to those investments? And do you think -- do you have visibility beyond '26 into further opportunities there?

I think the question was around how important is it for us to have a general model. The way that I think about Meta is we're like a deep technology company. Some people think about us as we build these apps and experiences, but the thing that allows us to build all these things is that we build and control the underlying technology that allows us to integrate and design the experiences that we want and not just be constrained to what others in the ecosystem are building or allow us to build. So I think that this is a really fundamental thing where my guess is that Frontier AI for many reasons, some competitive, some safety oriented are not going to always be available through an API to everyone. So I think like it's very important, I think, to be able to have the capability to build the experiences that you want if you want to be one of the major companies in the world that helps to shape the future of these products. So that I think is -- it's going to be, I think, important from a business perspective. And I think it's just important from like a creative and mission perspective to be able to actually design and build the experiences that we believe that we should be building for people. But yes, I mean I think it's quite important. Otherwise, we wouldn't be so focused on this. We're clearly extremely focused on this.

On your second question, interestingly, a year ago on this call, I think I talked about the set of investments we were making in 2025. as part of our 2025 budgeting process across our ads performance and organic engagement initiatives. And those investments have generally paid off, and we feel really good about kind of the -- the process we ran in terms of using projected ROI to stack rank investments, make sure that we had a robust measurement system funded things that were positive ROI and then tracking how they performed over the course of the year. And we are -- we've just finished running our 2026 budgeting process, and we have funded a similar set of investments, which we expect will enable us to continue delivering strong revenue growth in 2026.

Having said that, I expect both full year reported and constant currency revenue growth to be below the levels in Q1 for a few reasons. First, we would expect that currency tailwinds will dissipate later in the year based on current rates. Second, we'll be lapping stronger periods of growth later in the year that benefited from our 2025 ad performance investments and the strong macro landscape. And finally, we expect there could be some headwinds from our introduction of the revised less personalized ads offering in the EU that begins rolling out later in Q1.

But again, similar to '25, we feel good about the process by which we identified investment opportunities with attractive ROIs and funded them as part of our budget to support key initiatives across our ranking and recommendation systems and to increase the capacity efficiency of our models, all of which are key to driving growth for us.

Your next question comes from the line of Mark Mahaney with Evercore.

Okay. Two questions, please. Meta AI. Any update on what you're seeing there in terms of engagement and usage? And do you think you're just starting to be able to apply improvements to that specific functionality? And then just real quickly on share repurchases, Susan, I don't think you bought any stock back in the quarter. It's been a while, maybe a year since you haven't bought anything back, you talked about capital allocation a little bit into the year. It didn't sound like you're going to be buying back stock anytime soon, but just do you want to clarify that?

Yes. I'm happy to take both of those. So Meta AI, the quick update there is it's now available in over 200 markets. the largest daily active user markets for Meta AI align with our app -- we're aligned with where our apps are also very popular though the apps people engage most with Meta AI differ in some places, it's primarily WhatsApp driven, for example, India or Indonesia, in the U.S., Facebook is a stronger driver of engagement. And in general, we see a lot of opportunity to make it easier for people to accomplish the tasks that they already come to our services for every day. And if we do that well, then the way people use our products will continue to expand. So we're focused on making Meta AI the most personalized assistant while tapping into the vast amount of information, trends, content from our platform to offer differentiated insights and we think we have a very strong track record in building highly personalized experiences, and we're bringing that into Meta AI so that we can tailor responses to each person's personal interest and preferences.

On your second question, which is about share repurchase. Share repurchase levels will vary from time to time for a lot of reasons, including whether we believe there are areas that have a greater near-term need for capital. Right now, we think the highest order priority for the company is investing our resources to position ourselves as a leader in AI. And so that is really the -- that's kind of the first order of use of capital, but we'll continue to be opportunistic and evaluate repurchases versus other uses of cash.

Great. I think we will wrap it here. Thank you, everyone, for joining us today. We look forward to speaking with you again soon.

This concludes today's conference call. Thank you for your participation, and you may now disconnect.

Good afternoon. My name is Christa, and I will be your conference operator today. At this time, I would like to welcome everyone to the Meta third quarter earnings conference call. [Operator Instructions]

This call will be recorded. Thank you very much. Kenneth Dorel, Meta's Director of Investor Relations. You may begin.

Thank you. Good afternoon, and welcome to Meta's Third Quarter 2025 Earnings Conference Call. Joining me today are Mark Zuckerberg, CEO; and Susan Lee, CFO. Our remarks today will include forward-looking statements, which are based on assumptions as of today. Actual results may differ materially as a result of various factors, including those set forth in today's earnings press release and in our quarterly report on Form 10-Q filed with the SEC. .

We undertake no obligation to update any forward-looking statement. During this call, we will present both GAAP and certain non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in today's earnings press release. The earnings press release and an accompanying investor presentation are available on our website at investor.atmeta.com. And now I'd like to turn the call over to Mark.

All right. Thanks, Ken. Thanks, everyone, for joining today. We had another strong quarter with 3.5 billion people using at least one of our apps every day. Instagram had a major milestone with 3 billion monthly actives, and we're seeing good momentum across our other apps as well, including Threads which recently passed 150 million daily actives and remains on track to become the leader in its category.

I am very focused on establishing Meta as the leading frontier AI lab. Building personal super intelligence for everyone and delivering the app experiences and computing devices that will improve the lives of billions of people around the world. Our approach of advancing open source AI means that when Meta innovates, everyone benefits. Meta Super Intelligence Labs is off to a strong start. I think that we've already built the lab with the highest talent density in the industry. We're heads down developing our next generation of models and products and I'm looking forward to sharing more on that front over the coming months.

We're also building what we expect to be an industry-leading amount of compute. Now there's a range of time lines for when people think that we're going to get super intelligence. Some people think that we'll get there in a few years. Others think it will be 5, 7 years or longer. I think that it's the right strategy to aggressively frontload building capacity so that way we're prepared for the most optimistic cases. That way, if super intelligence arrives sooner, we will be ideally positioned for a generational paradigm shift in many large opportunities.

If it takes longer, then we'll use the extra compute to accelerate our core business which continues to be able to profitably use much more compute than we've been able to throw at it. And we're seeing very high demand for additional compute, both internally and externally. And in the worst case, we were just slow building new infrastructure for some period while we grow into what we build. The upside is extremely high for both our existing apps and new products and businesses that are becoming possible to build. Across Facebook, Instagram and Threads, our AI recommendation systems are delivering higher quality and more relevant content, which led to 5% more time spent on Facebook in Q3 and 10% on Threads.

Video is a particular bright spot with video time spent on Instagram up more than 30% since last year. And as video continues to grow across our apps, Reels now has an annual run rate of over $50 billion. Improvements in our recommendation systems will also become even more leveraged as the volume of AI-created content grows. Social media has gone through 2 eras so far. First was when all content was from friends, family and accounts that you followed directly. The second was when we added all of the creator content. Now as AI makes it easier to create and remix content, we're going to add yet another huge corpus of content on top of those. Recommendation systems that understand all this content more deeply and can show you the right content to help you achieve your goals are going to be increasingly valuable. Our ads business continues to perform very well, largely due to improvements in our AI ranking systems as well.

This quarter, we saw meaningful advances from unifying different models into simpler, more general models, which drive both better performance and efficiency. And now the annual run rate going through our completely end-to-end AI-powered ad tools has passed $60 billion. And one way that I think about our company overall is that there are 3 giant transformers that run Facebook, Instagram and ads recommendations. We have a very strong pipeline of lots of ways to improve these models by incorporating new AI advances and capabilities.

And at the same time, we were also working on combining these 3 major AI systems into a single unified AI system that will effectively run our family of apps and business using increasing intelligence to improve the trillions of recommendations that we'll make for people every day. I'm also very excited about the new products that we're going to be able to build. More than 1 billion monthly actives already used Meta AI -- and we see usage increase as we improve our underlying models. I'm very excited to get a frontier model into Meta AI and I think that the opportunity there is very large.

The same goes for our business AI. Every day, people have more than 1 billion active threads with business accounts across our messaging platforms, ranging from product questions to customer support. Our business AIs will enable tens of millions of businesses to scale these conversations and improve their sales at low cost and the better our models get, the better this is going to work for all businesses. This quarter, we also launched Vibes which is the next generation of our AI creation tools and content experiences. Retention is looking good so far. And its usage keeps growing quickly week over week. I'm looking forward to ramping up the growth of Vibes over the coming months.

More broadly, I think that Vibes is an example of a new content type enabled by AI, and I think that there are more opportunities to build many more novel types of content ahead as well. As our new models become ready, I'm looking forward to starting to show everyone some of the new kinds of products that we're working on. At Connect, we announced our 2025 line of AI glasses, and the response so far has been great. The new Ray-Ban Meta glasses and Oakley MetaVanguards are both selling well. As people love the improved battery life, camera resolution, new AI capabilities and the great design. And there's our new Meta Ray-Ban display glasses, our first glasses with a high-resolution display and the metaneural band to interact with them.

They sold out in almost every store within 48 hours with demo slots fully booked through the end of next month. So we're going to have to invest in increasing manufacturing and selling more of those. This is an area where we are clearly leading and have a huge opportunity ahead. Taking a step back, if we deliver even a fraction of the opportunity ahead for our existing apps and the new experiences that are possible, then I think that the next few years will be the most exciting period in our history. We've got a lot to do. But we're making real progress, delivering strong business results, building the talent density and infrastructure needed for the next era and leading the way on AI devices that will define the next computing platform.

I am proud of how our teams are rising to the challenge, and I'm grateful for their dedication, hard work and creativity. As always, thank you all for being a part of this journey with us. And now here is Susan.

Thanks, Mark, and good afternoon, everyone. Let's begin with our segment results. All comparisons are on a year-over-year basis, unless otherwise noted. Our community across the family of apps continues to grow, and we estimate more than 3.5 billion people used at least 1 of our family of apps on a daily basis in September. Q3 total family of apps revenue was $50.8 billion, up 26% year-over-year. Q3 family of apps ad revenue was $50.1 billion, up 26% or 25% on a constant currency basis. In Q3, the total number of ad impressions served across our services increased 14%. Impression growth was healthy across all regions driven by engagement and user growth, particularly on video services.

The average price per ad increased 10% year-over-year, benefiting from increased advertiser demand, largely driven by improved ad performance. This was partially offset by impression growth, particularly from lower monetizing regions and services. Family of apps other revenue was $690 million, up 59%, driven by WhatsApp paid messaging revenue growth as well as meta verified subscriptions. Within our Reality Lab segment, Q3 revenue was $470 million, up 74% year-over-year.

The significant year-over-year growth in Q3 was partly due to retail partners stocking up on Quest headsets ahead of the holiday season. We did not have a similar benefit in the third quarter of last year since our Quest 3S headset launched in the fourth quarter of 2024. Aside from this, strong AI glasses revenue also contributed to revenue growth in Q3.

Moving now to our consolidated results. Q3 total revenue was $51.2 billion, up 26% or 25% on a constant currency basis. Q3 total expenses were $30.7 billion, up 32% compared to last year. Year-over-year expense growth accelerated 20 percentage points from Q2 due primarily to 3 factors: First, legal-related expense growth was higher than in Q2 due to charges we recorded in the third quarter as well as us lapping a period of accrual reversals in the third quarter a year ago.

Second, employee compensation growth accelerated, driven by technical hires, particularly AI talent. Finally, growth in infrastructure costs accelerated due to increased infrastructure operating costs associated with our expanded data center fleet, depreciation on our incremental CapEx spend and third-party cloud spend.

We ended Q3 with over 78,400 employees, up 8% year-over-year, driven by hiring in priority areas of monetization, infrastructure, reality labs, MetaSuper intelligence labs as well as regulation and compliance. Third quarter operating income was $20.5 billion, representing a 40% operating margin. Q3 interest and other income was $1.1 billion driven primarily by unrealized gains on our marketable equity securities.

Our tax rate for the quarter was 87%, which was unfavorably impacted by a onetime noncash reduction in deferred tax assets that we no longer anticipate using under new U.S. tax law. Our tax rate would have been 14%, excluding this charge. Although the transition to the new U.S. tax law resulted in an accounting charge in the third quarter, we continue to expect we will recognize significant cash tax savings for the remainder of the current year and future years under the new law, and this quarter's charge reflects the total expected impact from the transition to the new U.S. tax law.

Net income was $2.7 billion or $1.05 per share. Excluding the onetime tax charge, our net income and EPS would have been $18.6 billion and $7.25 per share, respectively. Capital expenditures, including principal payments on finance leases were $19.4 billion. driven by investments in servers, data centers and network infrastructure. Free cash flow was $10.6 billion. We repurchased $3.2 billion of our Class A common stock and paid $1.3 billion in dividends to shareholders.

We ended the quarter with $44.4 billion in cash and marketable securities and $28.8 billion in debt. Turning now to the business outlook. There are 2 primary factors that drive our revenue performance. Our ability to deliver engaging experiences for our community and our effectiveness at monetizing that engagement over time. On the first, daily actives continue to grow year-over-year across Facebook, Instagram and WhatsApp. We're continuing to see improvements to our products and recommendations drive incremental engagement with year-over-year growth in global time spend accelerating on both Facebook and Instagram in Q3.

In the U.S., overall time spent on Facebook and Instagram grew double digits year-over-year, driven by continued video strength as well as healthy growth in non-video time on Facebook. The engagement gains continue to be driven by product work and ongoing improvements to our recommendation systems as we optimize our model architectures, implement advanced modeling techniques and integrate more signals about people's interests. We also continue to focus on increasing the freshness of recommended content. On Facebook, our systems are now surfacing twice as many wheels published that day than at the start of the year.

Looking to 2026, we expect to advance our recommendation systems across several dimensions. On Instagram, one focus is evolving our systems to surface content across a broader set of topics that cater to the diverse interest of each person. This follows a similar approach we've implemented on Facebook that has driven good results. We also expect to make significant progress on our longer-term ranking innovations in 2026. We're seeing promising new results from our research efforts to create foundational ranking models and expect the new model innovations we're developing as part of this will enable us to significantly scale up the amount of data and compute we use to train our recommendation models in 2026, yielding more relevant recommendations.

Another large focus next year is leveraging LLM to improve content understanding. We expect this is going to enable our systems to more precisely label the keywords and topics within videos and posts, which will allow our systems to both develop deeper intuition about a person's interest and retrieve the content that matches them.

Finally, we're making good progress with Meta AI and Threads. The number of people using Meta AI across our family of apps continues to grow, and we're increasingly leveraging first-party content into Meta AI results. with the majority of Meta AI's responses to Facebook Deep Dive queries in the U.S. now showing related reels. We're also seeing a lot of traction with media generation. People have created over 20 billion images using our products. And since launching Vibes within Meta AI in September, we have seen media generation in the app increased more than tenfold. On Threads, we see strong growth in both daily actives and the depth of engagement as we continue to improve recommendations.

The ranking optimizations we made in Q3 alone drove a 10% increase in time spent on threats. We also continue to ship new features, including launching direct messaging in Q3, so anyone on threads can now message one another within the app. Now to the second driver of our revenue performance. increasing monetization efficiency. The first part of this work is optimizing the level of ads within organic engagement. We continue to refine ad supply across each of our major surfaces within Facebook and Instarem. -- to better deliver ads at the time and place they are most relevant to people.

Longer term, we have exciting ad supply opportunities on both threads and WhatsApp status. Ads are now running globally in feed on threads and we're following our typical monetization playbook of optimizing the ads formats and performance before we ramp supply. Within WhatsApp status, we're continuing to gradually introduce ads and expect to complete the rollout next year. The second part of increasing monetization efficiency is improving marketing performance. Advancing our ad systems remains a critical aspect of this work. and we are driving performance gains through ongoing improvements in our larger scale ads ranking models. For example, we continue to broaden the adoption of Lattice, our unified model architecture. In Q3, we rolled out Lattice to app ads, which drove a nearly 3% gain in conversions for that objective. Since introducing Lattice back in 2023, along with other back-end improvements, -- we have now cut the number of ads ranking and recommendation models by approximately 100 as we consolidated smaller and more specialized models into larger ones that use the Lattice architecture to generalize learnings across surfaces and objectives. We continue to observe performance improvements as we combine models and expect to drive additional gains as we consolidate another 200 models over the coming years into a smaller number of highly capable models.

In addition to advancing our foundational ads models, we're innovating on our run time models we use downstream of them for ads inference. For example, we began piloting a new run time ads ranking model in Q3 that leverages more compute and data than our prior models to select more relevant ads. In testing, we've seen this new model drive a more than 2% lift in conversions on Instagram. We also significantly improved performance of Andromeda in Q3 and by combining models across retrieval and early-stage ranking into a single model, driving a 14% increase in ads quality on Facebook Surfaces. Within our ads products, we're seeing continued momentum with Advantage Plus. In Q3, we completed the rollout of our streamlined campaign creation flow for Advantages lead campaigns. So now advertisers running sales app or lead campaigns have end-to-end automation turned on from the beginning. -- allowing our systems to look across our platform to optimize performance by automatically choosing criteria like who to show the ads to and where to show them. The annual run rate of revenue running through our end-to-end automated solutions has now reached $60 billion following the implementation of the new streamlined creation flow, as we continue to see more advertisers leverage the performance benefits of our solutions.

Within our Advantage Plus Creative Suite, -- the number of advertisers using at least 1 of our video generation features was up 20% versus the prior quarter as adoption of image animation and video expansion continues to scale. We've also added more generative AI features to make it easier for advertisers to optimize their ad creatives and drive increased performance.

We introduced AI generated music. So advertisers can have music generated for their ad that aligns with the tone and message of the creative. Finally, business messaging remains a significant opportunity for us. We're seeing strong growth across our portfolio of solutions, including with click to WhatsApp ads, which grew revenue 60% year-over-year in Q3. We're also making good progress on our business AI efforts. -- where we've been focused on building a turnkey AI that helps businesses generate leads and drive sales.

We've been opening access in recent months to more businesses within our initial test markets, the Philippines and Mexico. And I've seen strong usage with millions of conversations between people and business AIs taking place since July. This month, we expanded availability within WhatsApp and Messenger to all eligible businesses in Mexico and the Philippines, respectively. In the U.S., we're also starting to roll out the ability for merchants to add their business AIs to their website so we can support the full sale funnel from ad to purchase.

Next, I would like to discuss our approach to capital allocation. Our primary focus is deploying capital to support the company's highest order priorities including developing leading AI products models and business solutions. As we make significant investments in infrastructure to support this work, we are focused on preserving maximum long-term flexibility to ensure we can meet our future capacity needs while also being able to respond to how the market develops in the years ahead. We're doing so in several ways. -- including staging data center sites so we can spring up capacity quickly in future years as we need it as well as establishing strategic partnerships that give us option value for future compute needs. The strong financial position and cash generation of our business enable us to make these investments while also accessing additional pools of cost-efficient capital.

Moving to our financial outlook. We expect fourth quarter 2025 total revenue to be in the range of $56 million to $59 billion. Our guidance assumes foreign currency is an approximately 1% tailwind to year-over-year total revenue growth based on current exchange rates. Our outlook reflects an expectation for continued strong ad revenue growth partially offset by lower year-over-year Reality Labs revenue in Q4. The anticipated reduction in Reality Labs revenue is due to us lapping the introduction of Quest 3 in Q4 of last year. as well as retail partners procuring Quest headsets during Q3 of this year to prepare for the holiday season, which were recorded as revenue in the third quarter.

Turning to the expense and CapEx outlooks. I'll first start with 2025 before providing some commentary on our planning for 2026. We expect full year 2025 total expenses to be in the range of $116 million to $118 billion, updated from our prior outlook of $114 million to $118 billion and reflecting a growth rate of 22% to 24% year-over-year. We currently expect 2025 capital expenditures, including principal payments on finance leases to be in the range of $70 million to $72 billion, increased from our prior outlook of $66 billion to $72 billion.

On to tax. Absent any changes to our tax landscape we expect our fourth quarter 2025 tax rate to be 12% to 15%. Turning now to 2026. We are at an exciting point for our company. where we have continued runway to improve our core services today as well as the opportunity to build new AI-powered experiences and services that will transform how people engage with our products in the future. We expect the set of investments we're making within our ads and organic engagement initiatives next year will enable us to continue to deliver strong revenue growth in 2026. while our progress on AI models and products will position us to capitalize on new revenue opportunities in the years to come. A central requirement to realizing these opportunities is infrastructure capacity.

As we have begun to plan for next year, it's become clear that our compute needs have continued to expand meaningfully, including versus our own expectations last quarter. We are still working through our capacity plans for next year. but we expect to invest aggressively to meet these needs, both by building our own infrastructure and contracting with third-party cloud providers. We anticipate this will provide further upward pressure on our CapEx and expense plans next year. As a result, our current expectation is that CapEx dollar growth will be notably larger in 2026 than 2025. We also anticipate total expenses will grow at a significantly faster percentage rate a than 2025, with growth primarily driven by infrastructure costs, including incremental cloud expenses and depreciation.

Employee compensation costs will be the second largest contributor to growth. as we recognize a full year of compensation for employees hired throughout 2025, particularly AI talent and add technical talent in priority areas. Finally, we continue to monitor active legal and regulatory matters, including the increasing headwinds in the EU and the U.S. that could significantly impact our business and financial results. For example, in the EU, we continue to engage constructively with the European Commission on our less personalized ads offering. However, we cannot rule out the commission imposing further changes to that offering that could have a significant negative impact on our European revenue as early as this quarter. In the U.S., a number of youth-related trials are scheduled for 2026 and may ultimately result in a material loss. In closing, this was another good quarter for our business. We have an exciting set of opportunities to continue improving our core business while delivering innovative new experiences and services for the people and businesses using our products in the years to come. With that, Krista, let's open up the call for questions.

[Operator Instructions] And your first question comes from the line of Brian Nowak with Morgan Stanley.

I have 2 for Susan. The first one, Susan. So the pipeline for for core improvements to come in '26 with models and ad ranking models and more types of compute seems very exciting and the infrastructure build team sizable behind that. So can you help us a little understand some of the early quantifiable signals you're seeing on AB tests from some of these improvements to come that sort of make you most excited and give you confidence you're going to get ROIC from all this CapEx? That's the first one. .

Second one is a little faster. How large is the Reality Labs revenue headwind in the 4Q guidance?

Thanks, Brian, for the question. I think your first question had a couple of parts to it. So I'm going to try to disaggregate those parts, and let me know if this addresses what you're getting to. I will say that the growth in 2026 CapEx relative to 2025 comes from growth in each of the core areas core AI as well as non-AI spend. So all of those areas are growing, but the MSL AI needs are growing the most.

In terms of the core AI pipeline, I think we talked about last year when we were going into the 2025 budget process, we had a road map of resource investments across both head count and compute that we thought would pay off in 2026. And it's really a very broad range of sort of different ads ranking and performance efforts. And we're continuing to see that those have paid off through the course of the year. There is a long list of specific efforts, but 1 of the measures that we look at to monitor this is how are we driving ad performance, how are conversions growing? -- conversions is a complex metric for us because advertisers optimize for so many different conversions on different values. But when we control for that and look at value-added conversion rates, we're seeing very strong year-over-year growth and conversion -- weighted conversions continue to grow faster than impressions.

We also talked about some of the new model architecture over the course of the year and the degree to which the new model architecture is enabling us also to take advantage of having more data and more compute to drive ads performance. So we expect that, that's going to be a continued story in 2026. -- we are, in fact, at the beginning of our 2026 budgeting process now, and we see a similar list of revenue investments. that we're excited to be able to invest in. And so we think that, that's going to be a big part of our ability to continue to drive strong revenue performance throughout the year.

On your second question, which is the Reality Labs revenue headwind. I don't think we have quantified the exact size of that. We expect that Q4 Reality Labs revenue will be lower than last year for a couple of reasons that I alluded to, the biggest factor is we're lapping the introduction of Quest 3S in Q4 of last year, and we don't have a new headset in the market this year. We also recorded all of our holiday-related Quest 3S sales in Q4 '24, since the headset was launched in October 24. This year, we're recognizing some of those Quest 3S sales in Q3 as retail partners have procured Quest headsets in advance of the holiday season. We're still expecting significant year-over-year growth in AI Glasses revenue in Q4 as we benefit from strong demand for the recent products that we've introduced, but that is more than offset by the headwinds to the [indiscernible].

Your next question comes from the line of Doug Anmuth with JPMorgan.

I appreciate the strategy to front-load capacity for super intelligence. Can you just talk about your thought process and kind of triangulating the CapEx dollar growth and the significantly faster expense growth next year with core growth in the business and then the impact on earnings and free cash flow? And do you have targets that we should be thinking about for cash on hand or net cash overall? .

Thanks, Doug. We're, right now, I would say, in the process of relatively early, actually still in the process of putting together our budget for 2020. And it is on the capacity side, a particularly dynamic process. We're certainly seeing that we wish we had more capacity today than we do. We would be able to put it towards good use certain not only with the MSL team appreciate having more capacity, but we'd be able to put it towards good and ROI-positive use in the core business as well. So we're really trying to plan ahead not only to ensure that we have the capacity we need in 2026, but also to give ourselves the sort of flexibility and option value to have the capacity that we think we could need in '27 and '28. So that said, there are lots of moving pieces in the budget. It's not baked yet. It's still sort of in the process of coming together. We don't have specific targets to share -- but we do feel like our strategic priority is really making sure that we have the compute that we need to be well positioned to succeed at AI, and that's sort of the foremost priority as we're putting together the budget.

Yes. I mean, I'll add a few thoughts on this, too, although I mean, as Susan said, we're still working through the actual budget, and I think we'll typically have more to share on that early next year. But -- to date, we keep on seeing this pattern where we build some amount of infrastructure to what we think is an aggressive assumption. And then we keep on having more demand to be able to use more compute, especially in the core business in ways that we think would be quite profitable, then we end up having compute for.

So I think that, that suggests that being able to make a significantly larger investment here is very likely to be a profitable thing over some period because if the primary use of it is going to be to accelerate the AI research and the new AI work that we're doing and how that relates to both the core business and new products. But any compute that we don't need for that we feel pretty good that we're going to be able to absorb a very large amount of that to just convert into more intelligence and better recommendations in our family of apps and ads in a profitable way. Now I mean, it's, of course, possible to overshoot that, right? And if we do, I mean, this is what I mentioned in my comments, then we see that there's just a lot of demand for other new things that would build internally, externally, like almost every week, people come to us from outside the company asking us to stand up an API service or ask if we have different compute that they could get from us and we haven't done that yet. But obviously, if you got to a point where you ever built, you could have that as an option.

And then the kind of the very worst case would be that we effectively have just prebuilt for a couple of years, in which case, of course, there would be some loss and depreciation, but we'd grow into that and use it over time. So my view on this is that rather than continuing to be constrained on CapEx and feeling in the core business like we have significant investments that we could make that we're not able to make that would be profitable, but the right thing to do is to try to accelerate this to make sure that we have the compute that we need, both for the AI research and new things that we're doing and to try to get to a different state on our compute stance on the core business. So that's kind of how I'm thinking about that overall.

Of course, there's a lot of operational constraints too on what one can build, right? So -- so we're basically trying to work through this all, and I think we'll have more to share in the coming months and over the course of next year. But I can think there's just a huge, huge amount of opportunities ahead here.

Your next question comes from the line of Eric Sheridan with Goldman Sachs.

Mark, I wanted to reflect on some of your comments with respect to scaling towards super intelligence and bringing it back to consumer AI, maybe reflect a little bit on the signals you've gotten on the way consumers across family of apps interact with Meta AI today? And how you think about scaling and exiting models from the super intelligence effort might change the utility and behavior around Meta AI in the years ahead. .

Yes. I mean A lot of people use Meta AI today. I mean, as I said in my comments upfront, there's more than 1 billion people who use it on a monthly basis. And what we see is that as we improve the quality of the model, primarily for post-training Lama at this point. We are -- we continue to see improvements in usage. So our view is that when we get the new models that we're building in in there and get like truly frontier models with novel capabilities that you don't have in other places, then I think that this is just a massive latent opportunity, right?

We know -- I mean I would guess that Meda I think has the best track record of any company out there of taking a new product that people love and getting it to billions of people in terms of usage. So I think that the ability to plug in leading models is going to -- I would predict lead to a very large amount of use of these things over the coming years. So I'm very excited about that in terms of new products. It's not just Meta AI is an assistant. I think that there are going to be all kinds of new products around different content formats, and we're starting to see that with video and content creation, but I think there's going to be a lot more like that, that I'm quite excited about. And then there are the business versions of all these 2, like business AI. And then that's, of course, one part of the story is the new things that will be possible to build. And then the other part is how more intelligent models are just going to improve the core business. And improve the recommendations that we make across the family of apps and improve the recommendations and advertising. And I think there's just a -- as we've shown, there's sort of this very large amount of headroom and the opportunity there keeps growing as we as we are improving and optimizing the AI there. And I think that, that really shows no sign of being near the end. I think that there's quite a bit more to do there. And like I said in response to the last question, we are sort of perennially operating the family of apps and ads business and a compute starved state at this point, which is, on the one hand, sort of an odd thing to say, given the compute that we built up. But we really are taking a lot of the resources and using them to advance future things that we're doing, and we think that there's a lot more compute that we could put towards these that would just unlock a huge amount of opportunity in the the core business as well.

Your next question comes from the line of Mark Shmulik with Bernstein.

Susan, as you think about the visibility into kind of the runway next year of continued add performance and engagement improvements, how do you think about kind of the scale of those improvements versus kind of the progress we've seen over the last 2 years? And then, Mark, as you think about kind of the timing of some of these newer efforts coming out of super intelligence labs, is us anchoring to kind of an updated frontier model launch sometime next year like the right way for us to think about it? Or should we be looking at kind of progress from new products you're excited to see ship like wise? .

Thanks, Mark. So on the sort of adds improvement side, some of the innovations that we have been launching actually involve sort of improving our larger scale models. So we don't use our larger model architectures like for inference because their size and complexity would make it too cost prohibitive. The way that we drive performance from those models is by using them to transfer knowledge to smaller lightweight models that are used at run time. And then in addition to the foundation model work, we are working on advancing our inference models by developing new techniques and architectures that then allow us to scale up compute and complexity in an ROI-positive way.

So in general, we obviously, have a very large base of advertisers. There's a lot of demand liquidity in the system and even small-scale improvements that we are able to make in terms of driving basis point improvements in the performance of ads or single-digit increases in conversions relative to impressions in a given quarter off of a large base. I mean that we're really able to continue to grow the absolute dollars of revenue growth in a pretty meaningful way.

Your next question comes from the line of Justin Post with Bank of America.

Justin, just give us one second. I think there was a second Mark's question that we just want to get to on MSL.

Yes. I mean, I'll keep it quick. I mean I don't think we have any specific timing to announce certainly on the models or products, but I expect that you will see both. We expect to build novel models and novel products, and I'm excited to share more when we have it.

So Mark, you mentioned the prior 2 constant cycles, and obviously, you've been able to generate very attractive margins on them. As we get into the AI cycle, obviously, some concerns on the investment. But -- can you talk a little bit about how you're thinking about tools that could be coming out for users? I know there's some new competition -- and then secondly, how do you think about margins in this content cycle? Any reason to think they would be different versus prior cycles.

I think it's too early to really understand what the margins are going to be for the new products that we build. I mean, I think certainly, every each product has somewhat different characteristics. And I think we'll kind of understand how that goes over time. I mean, my general goal is to build a business that maximizes value for the people who use our products and maximizes profitability, not margin. So I think we'll kind of just try to build the best things that we can and try to deliver the most value that we can for most people.

Your next question comes from the line of Ross Sandler with Barclays.

Great. And Mark, some of the goals for competing AI labs are around achieving AGI or these other milestones that are kind of like out there and a little esoteric how are you setting up your new team in terms of achieving those types of goals versus products that can generate revenue from meta kind of right out of the gate -- and is the goal that you had articulated to us previously around giving billions of people kind of a personal AI to use still the direction of travel that you see? Or is there other things like kind of the vibe or Sora angle that you think are potentially important. How should we think about like the overall direction .

Sure. So the way that I think about this is that the research is going to enable new technological capabilities to exist. And then those capabilities can get built into all kinds of different products. So the ability to reason more intelligently is, for example, very important across a large number of things. It would be useful for an assistant. It will also be useful in business AI. It will also be useful in the AI agent that we're building to help advertisers figure out what their campaigns are going to be. It will also have implications for eventually how we do ranking and recommendations of people's feeds and make different decisions there. That's just one example.

I mean certainly, the capability to be able to produce very high-quality good video is going to be useful for giving people new creative tools. it will help increase the amount of content inventory that can be shown in Instagram and Facebook and therefore, should enable an increase in engagement there. it should help advertisers be able to create creative that will help us monetize better. So you can just go kind of down the list of capabilities that you'd expect. And I think each one will enable a bunch of different things. And I think the art of product development here is looking at the list of technology capabilities and figuring out what new products are going to be useful and prioritizing those.

But fundamentally, I would sort of expect this exponential curve in new technology capabilities that are going to become available. And the other thing that I expect is that I think being the best in a given area will drive great returns rather than -- this is not like a check-the-box exercise of like, okay, we can generate some kind of content and someone else can. I think that like the company that is the best at each of these capabilities, I think, will get a large amount of the potential value for doing that. So there are lots of different capabilities to build I'm not sure that any one company is going to be the best at all of them. I doubt that's going to be the case. But a lot of what we're trying to do is not like not kind of do some things that others have done. We're really trying to build novel capabilities. And I'm keeping this high level because I'm not -- I don't want to necessarily from a competitive or strategic perspective, get into what we're prioritizing. But that hopefully gives you a sense of how we're thinking about what we're doing.

We want to be able to kind of build novel things, build them into a lot of our products and then have the compute to scale them to billions of people. And we think that that's going to both show up in terms of new products being possible in new businesses and very significant improvements to the current business, too.

Your next question comes from the line of Mark Mahaney with Evercore ISI.

Just a ask just a question on Meta AI and both product and a monetization path. So when you look at it, what you've seen that's most encouraging to you in terms of the adoption and and the use of Med AI? And then when you think about -- I know you generally like to roll out and then deepen engagement and then later think about monetization -- like where do you think you are on that path now? Is it clear to you what the monetization options are for Meta AI.

I mean, I think the most promising thing that we're seeing is, one, that we were able to build something that a large number of people use, and I think that's valuable. And then secondly, that as we -- there is a clear correlation as we improve the models in ways that we think make them better, that people use them more. So that shows that we have a runway to basically be able to improve engagement and turn this into a product that's leading over time. .

In terms of where we are on this, and we basically just did this huge effort to boot up Meta super Intelligence Labs and build what I am very proud of is, I think, the highest talent density lab in the industry at this point. There are a lot of really great researchers and infrastructure folks and data folks who are now a part of the effort who are focused on training the next generation of work and doing some really novel work. And when that is ready, I think that we will be able to plug that into a number of the products that we're building, and I think that, that will be very exciting. But I think that that's really the next thing that we're looking at. And then from there, I think that these models will also improve monetization in all of the different ways that we've talked about so far in terms of improving engagement, improving advertising, helping advertisers engage.

I mean there's 1 opportunity that we just usually talk about it on these calls, but hasn't come up as much here is just the ability to make it so that advertisers are increasingly just going to be able to give us a business objective and give us a credit card or bank account and like have the AI system basically figure out everything else that's necessary, including generating video or different types of creative that might resonate with different people that are personalized in different ways, finding who the right customers are. All of these -- all of the capabilities that we're building, I think, go towards improving all of these different things. So I'm quite optimistic about that.

Your next question comes from the line of Ronald Josey with Citi.

And this maybe dovetails perfectly off Mark, what you just talked about. And we heard a lot about -- in automation here, I think, reaching a $60 billion ARR wanted to hear about -- if you can talk to us more just about adoption rates amongst the advertisers? And then maybe bigger picture, as you incorporate ranking recommendation changes like in drama or Gimsorlattice, just talk to us how this automation is driving, call it, a higher ROI for advertisers overall, as we bring it all together. .

Yes. So we've been we've been sort of laying the continued brick-by-brick build of Advantage plus and extending the set of objectives that it applies to over time. And so we completed the global rollout of the streamlined campaign creation flow for Advantages lead campaigns. So now advertisers who are running sales Apple campaigns have end-to-end automation turned on from the beginning -- and like the kind of application of the streamlined campaign creation flow for other objectives, this generally allows advertisers to optimize and automate several aspects of the campaign setup process at once. That includes things like audience selection where to show the ad, how the budget gets paced and distributed across ad sets to drive the most efficient outcomes.

And we see that Advantage Plus continues to drive performance gains, advertisers who run lead campaigns using Advantages are seeing a 14% lower cost per lead on average than those who are not. And I would say that we think there is still a lot of opportunity generally to grow adoption of Advantage Pllus. A lot of advertisers only use our end-to-end automated solutions for a portion of their campaigns so we can grow share there. And to capture that opportunity, we're focused on driving continued performance improvements and addressing some of the key use cases that we still need in order to grow adoption.

We're also working to broaden adoption among advertisers who use 1 of our single step automated solutions. For example, advertisers who might only use a piece of it like Advantage Plus audiences by helping them understand the benefits of using more than one automated solution at the same time. So I would say Advantages is sort of an ongoing platform by which we both continue to expand the feature set that is available in Advantages and then expand the extensibility or the coverage of that feature set to sort of the broader set of advertisers. I think Mark mentioned that the annual revenue run rate now for advertisers who are using these automated options is $60 billion. And again, we see that there is room to continue growing that.

Your next question comes from the line of Youssef Squali with Truist Securities.

Great. Mark, on wearables, in particular, do you think you'll be able to sell enough hardware to recoup your investment? Or is that dependent on maybe creating new avenues for revenue from things like advertising services and commerce through that new computing platform? And if so, what are kind of the gating factors there? And then Susan, how do you see the on-balance sheet versus off-balance sheet financing of your AI initiatives? You've recently struck out for the Louisiana data center. Is that part of the CapEx guide for '26? And if it's not, how significant will that way of funding for Meta going forward? And basically, would that slow down your CapEx growth past 2026.

I can talk about wearables, and then Susan can jump in on the other part. So there are a few pieces here. One is that the the work on Ray-Ban Meta and the Oakley Meta product is going very well. I think, yes, I mean, at some point, if these continue going as well as it has been, then I think it will be a very profitable investment. I think that there's some revenue that we get from basically selling the devices and then some that will come from additional services from the AI on top of it. So I think that there's a big opportunity. Certainly, the investment here is not just to kind of build just the device. It's also to build these services on top. Right now, a lot of people get the devices for a range of things that don't even include the AI even though they like the AI. But I think over time, the AI is going to become the main thing that people are using them for and I think that that's going to end up having a big business opportunity by itself.

But as products like the Ray-Ban Meta and Oakley metas are growing, we're also going to keep on investing in things like the more full field of view, product form of of the Orion prototype that we showed at Connect last year. So those things are obviously earlier in their curve towards getting to being a sustaining business. And our general view is that we want to build these out to reach many hundreds of millions or billions of people and that's the point at which we think that this is going to be just an extremely profitable business.

Youssef, to your second question, so the JV that we announced with [indiscernible] all is sort of an example of finding a solution that enabled us to partner with external capital providers to codevelop data centers in a way that gives us long-term optionality in supporting our future capacity needs just given both the magnitude, but also uncertainty of what the capacity outlook in future years looks like.

In terms of how that is recognized as CapEx, our prior CapEx reflected a portion of the data center build cost prior to the joint venture being established. Going forward, the construction cost of the data center will not be recorded in CapEx as the data center is constructed, we will contribute 20% of the remaining construction costs required, which is in line with our ownership stake, and those will be recorded as other investing cash flows.

Your last question comes from the line of Ken Gawrelski with Wells Fargo.

Just one for me, please. Mark, as you think about with the -- hopefully, a leading frontier model next year in hand, could you talk about where you think the value will accrue in this evolving ecosystem? Will it be with the platforms? Or do you think that this will be mostly -- the value will accrue to the scaled first-party applications.

I guess I'm not exactly sure what you mean by platform versus application in this context. But I mean I think that -- I mean, I think there's just a lot of value to create with AI overall. So I mean, clearly, you're seeing the people who are making the hardware NVIDIA is doing an amazing job, right, I think extremely well-deserved success. The cloud partners and companies are making -- are doing very well. I think that, that will likely continue. I think there's a huge opportunity there. And -- but if you look at it today, the companies that are building apps, I mean, a lot of the apps are still relatively small. And I think that that's obviously going to be a huge opportunity. .

And I think what we've seen overall is basically people take like individual technology advances and build them into products that then build either communities or other kinds of network effects and then end up being very sustaining businesses. And I think what we haven't really seen as much in the history of the technology industry is the rate of new capability is being introduced because around each of these capabilities, you can build many new products that I think each will turn into interesting businesses.

So Yes. I don't know. I mean I'm generally pretty optimistic about there being a very large opportunity. But in terms of new things to build, I think being able to build them and then scale them to billions of people is a huge muscle that Meta has developed, and I think we do very well. And I certainly think that, that's going to deliver a huge amount of value, both in the core business for all the ways that we talked about, how it's going to improve recommendations and the quality of the services as well as unifying the models together. And so that way, when these systems are deciding what to show they can just pull from a wider pool. And that we've -- these are things that we've just seen over the 20-plus years of running the company that they just deliver consistent wins that we're going to keep on being able to make the systems more general and smarter and make better recommendations for people and have a larger pool of inventory. And that is all going to be great.

And then there's going to be a lot of new things that I think we're going to be able to take and scale to billions of people over time and build new businesses, whether that's advertising or commerce supported or people paying for it or different kinds of things. So Yes, it's -- I think it's pretty early, but I think we're seeing the returns in the core business. That's giving us a lot of confidence that we should be investing a lot more, and we want to make sure that we're not under-investing.

Great. Thank you, everyone, for joining us today. We look forward to speaking with you again soon.

This concludes today's conference call. Thank you for your participation, and you may now disconnect.