Meta Platforms, Inc. second-quarter revenue exceeded expectations, surging 28% year-over-year, driven by a full-scale acceleration of AI technology across its core business and the unfolding blueprint for personal superintelligence. However, a weaker-than-expected third-quarter forecast and soaring AI capital expenditures dampened market sentiment.
On July 29, after the U.S. market close, Meta held its earnings conference call. CEO Mark Zuckerberg and CFO Susan Li maintained that the company’s AI investments are not only generating real returns but are also outlining a commercial landscape for "superintelligence" that will serve billions of people.
During the call, Zuckerberg dedicated significant time to a robust defense of the company's strategy. He argued that the current massive investment is crucial to seizing the window of opportunity in AI infrastructure, with returns expected to materialize through several avenues, including enhanced core advertising, enterprise services, API monetization, and the leasing of computing power.
Explaining the rationale for acquiring substantial computing resources, Zuckerberg noted that current capacity is far from sufficient to meet all demands. While the company has received numerous offers for its computing power, he emphasized: "Simply selling all of our compute capacity for short-term profit would be foolish. We believe the profit margins on selling intelligence will remain significantly higher than directly selling compute capacity."
He also stated that open-source models are currently less powerful than frontier models, and as a "full-stack technology company," Meta Platforms, Inc. must possess the capability to build its own models.
Initial AI Monetization Gains Traction, Advertising Business Leads the Industry
Investor concerns regarding revenue growth and AI monetization were strongly validated in the second quarter. Advertising revenue for the Family of Apps reached $59.4 billion, a 27% year-over-year increase. "Our advertising business reported faster year-over-year revenue growth in dollar terms than any other company's ad business," Zuckerberg stated on the call. "So, these AI investments are paying off."
The primary engine for this performance is the enhancement of recommendation systems and ad conversion rates through AI. Susan Li pointed out that, driven by large language model (LLM)-based predictions and sorting, the average price per ad globally rose by 12% year-over-year. Currently, nine million small businesses on Meta's platform use at least one of its AI ad creative tools. The company's AI-powered Advantage+ end-to-end solution continues to grow, with an annual revenue run rate exceeding $75 billion.
Zuckerberg Defends the Need for Massive Spending Under Pressure
Despite strong business performance, high infrastructure and computing investments remain a focal point of investor concern. Zuckerberg addressed market skepticism on the call, stating, "I understand it's a huge investment and a big bet. My personal bet is that those who invest in this will be rewarded and feel very good about it over time."
Explaining the need for vast computing power, he provided a supply-and-demand perspective: "Current computing power is far from meeting all demand." Facing external offers to buy Meta's computing capacity at a premium, Zuckerberg laid out the company's business logic of "not selling short-term compute but investing in long-term intelligence": "We've received numerous offers to buy our compute at significant premiums. But our thinking is that the profit margins on selling intelligence will remain significantly higher than directly selling compute... Simply selling all our compute for short-term profit would be foolish. When you have the opportunity to build intelligence on top of compute, it represents a multiple of the compute's value."
To support this seemingly bottomless demand for computing power, Meta has even broadened its financing channels. In addition to raising low-cost, long-term capital through bond issuance, the day before the call, Meta announced a strategic partnership with BlackRock to jointly develop a 1-gigawatt new data center in Texas.
Zuckerberg's Ultimate Ambition: Putting Superintelligence Directly in People's Hands
Beyond the core advertising business, personal and business AI agents form the most significant part of Meta's future performance guidance and long-term vision. When discussing the long-term AI vision, Zuckerberg used a powerful statement: "We are the only major tech company whose primary goal is to put superintelligence directly into people's hands. We are not centralizing superintelligence; we are focused on widespread distribution."
He made a bold prediction about the potential of personal agents: "If you look out five years, it's extremely unlikely that billions of people won't have a personal agent that knows your goals and works for you 24/7 to achieve them." Currently, Meta's business agents have been globally launched on WhatsApp and Messenger, with over one million businesses using them weekly. In the future, businesses would only pay Meta when a conversion is generated, potentially creating a new auction-based business model similar to the existing advertising system.
Addressing the debate between open-source and closed-source approaches, Zuckerberg made it clear that Meta Platforms, Inc. does not blindly follow one path. He believes that current open-source models are not as powerful as frontier models, and as a "full-stack technology company," Meta must have the capability to build its own models. "When you look at what many discerning customers and companies around the world want, they want to know they control their own destiny... I think open source will be important, but building models will also be a key part of it."
Full Translation of Meta's Q2 2026 Earnings Conference Call
Mark Zuckerberg, Founder, Chairman and CEO:
Alright. Hello everyone, thank you for joining today. Our community and business performed strongly this quarter, with 3.6 billion people using at least one of our apps daily. We reached several milestones. Instagram daily active users hit 2 billion, Threads surpassed 500 million monthly active users, making it the fastest-growing conversation app ever. Facebook has had over 2 billion daily active users for some time now.
WhatsApp just set a historic messaging record, peaking at 30 million messages per second during the World Cup final. Additionally, Kunal Shah just joined us as the new head of WhatsApp; he founded one of India's most important payment companies and will be a great addition to the team. We also released some powerful new models from Meta Super Intelligence Lab and launched new smart glasses. Overall, the scale and influence of our community across our apps are truly amazing, providing a powerful platform for us to deliver new innovations to billions of people.
The opportunities before us are immense. First, we are now at a stage where our AI investments are accelerating every major part of our core business. They are improving the experience people have using our apps, delivering better performance for advertisers, and helping our teams build new experiences and ship products faster. Second, we are developing new personal agents, which will form the foundation for the next wave of products and revenue lines in the coming months and years. Third, we see a large opportunity in selling to enterprises, including APIs, business agents, potentially direct compute sales, and other services we build for large customers. Today, I want to spend some time detailing how our investments are producing results and the long-term opportunities we see.
In accelerating the core business, we are seeing strong and promising results in several areas. On Instagram and Facebook, I am very optimistic about our work integrating large language models into our recommendation systems. LLMs add a first-principles understanding of what content is about and why it's engaging, as well as a deeper understanding of what people are interested in and what they aim to do when using our apps. This means we can show more relevant, engaging content that better reflects people's goals and interests. Our new Muse Image and Muse Video models will also vastly expand the range of content people can discover on our platforms. There are already two major data sources for content: first, content from your friends and people you follow; second, content from creators you don't follow.
But now there will be a whole new, almost infinite universe of personalized content, which will make our services more useful and engaging for people. For advertising, we are using LLMs to improve how our systems predict and rank the ads we show. We have expanded the range of context we can consider around a person's natural activity and ad campaigns to determine ad relevance, leading to significant improvements in relevance and conversion rates on Facebook and Instagram. Our advertising business reported faster year-over-year revenue growth in dollar terms than any other company's ad business.
So, these AI investments are paying off. We are also seeing strong demand for our new AI-powered creative tools. Nine million small businesses on our platform now use at least one of our AI ad creative tools, and we are rolling out new end-to-end creative solutions that help advertisers turn performance data into their creative decisions. Muse Image will further drive this. The model can analyze images, improve its own work, and generate better ad variants based on advertiser input. So far, we have received great feedback. We also just launched Meta One, a new subscription service offering more tools and AI features within our apps. As demand grows, we will offer a variety of different tiers and pricing options.
I am also excited about how AI is helping our teams accelerate product development. Earlier this year, we released Instagram Instance. We also just launched Forum, a standalone groups app, and Seller, a standalone marketplace app. I expect it will become much easier to launch new apps. Therefore, we plan to develop more ideas and use our recommendation systems to scale them to people who would find them interesting, similar to what we did with Threads. It has been over a year since the launch of Meta Super Intelligence Lab, and our trajectory is solid. Last month, we released Muse Spark 1.1 and Muse Image. Since we rebuilt Meta AI and integrated Muse Spark, the number of daily interactions with the assistant has increased by 60%, and it continues to grow rapidly week-over-week.
Muse Spark 1.1 is a powerful coding model, very efficient, and excellent at computer use, tool use, and multimodal understanding. It is available through our new public API, and we will increase distribution through partner channels and more coding agents in the coming weeks. We are also building features to make it easier for enterprises to adopt Muse Spark, an area we expect to continue focusing on. The reason we are so focused on making Muse Spark functionally excellent is that we see a very large opportunity in deploying several types of agents aligned with our mission and business. First are personal agents. Soon, we will have agents that can work for you 24/7, helping you achieve your goals and improve your life, health, relationships, finances, whatever you want.
The first area where agents are truly taking off is coding. But engineers are more technical and willing to spend time making these agents work. To build a great personal agent, it needs to be a great consumer product that works out of the box and is simple enough for billions of people to adopt and use. I am very excited about this, and we will have more information to share soon. As we move towards a future where we all interact with multiple agents, WhatsApp and our other messaging services will become increasingly important. WhatsApp is already the primary interface for people to interact with Meta AI, and as we build a platform for more agents within our messaging apps, we will innovate on how to provide private and secure AI experiences. This quarter, we launched Incognito Mode on WhatsApp and the Meta AI app, allowing people to have private conversations with the assistant that even Meta cannot see. We also plan to make strong privacy and security a fundamental component of the agents we are building.
I am also very excited about the progress of our business agents. This quarter, we globally launched Meta Business Agents on WhatsApp and Messenger, and over one million businesses are already using them weekly to talk to customers or complete sales. We are now also launching business agents on Instagram. Interestingly, having an agent talk to your customers daily, it learns over time and can bring all those insights back to you. Therefore, we are building more features to summarize all these conversations, digest what happened overnight, and present customer needs. Soon it will go further, including suggesting ways to grow your business, providing you with competitive intelligence, and real-time insights into what's working and what's not.
Over time, we want to build this into an "all-in-one business" service that can help you start and run your entire business using the Meta platform. Regarding how we will monetize these, we have a mix of subscriptions and volume-based pricing, and I expect we will continue to evolve more of these products so that, like our advertising system, businesses only pay when our help achieves results. Over time, this will allow us to run efficient auctions on our computing power, similar to what we do for advertisers today.
As the use of AI across our products and business continues to increase, we will continue to invest aggressively in infrastructure to meet demand. Yesterday, as part of our Meta Compute plan, we announced a new strategic partnership with BlackRock to develop a 1-gigawatt data center in El Paso, Texas. Overall, we expect a significant portion of our compute will be used for training our models, growing our core business, and delivering personal agents and new products. But we also expect to develop a large business serving large customers.
We have already built APIs. We are launching business agents. We have received numerous offers for compute at prices significantly higher than what we pay. We have more coding and productivity tools on our roadmap. We will share more details on that soon. As we approach personal superintelligence, we will also need hardware that allows you to interact with it seamlessly. Smart glasses are the ideal form factor because they can be with you all day and help you without taking you out of the moment. Our glasses are still one of the fastest-growing consumer electronics products ever, and we continue to expand the product line. We just launched our own collection developed in partnership with EssilorLuxottica, including a style designed with Kylie Jenner.
These are the first glasses to come out of the box with Muse Spark, so they can understand what you see and give more helpful answers. Early sales are strong and exceeding our expectations. We will share more about our glasses product line at our Connect conference on September 23rd, so I encourage you to tune in then. Before I conclude, I want to mention that I just published an op-ed outlining why I am so optimistic that we are building a positive future for everyone. At Meta, we have always been focused on building technology that puts power into people's hands, allowing them to connect with who they care about and shape the world the way they want.
This is why we have always focused on making our products affordable and easy to use for everyone, a strategy that has greatly benefited both our community and our business. As we enter the next chapter, the same philosophy guides our approach to AI. We are the only major tech company whose primary goal is to put superintelligence directly into people's hands. We are not centralizing superintelligence; we are focused on widespread distribution and empowering everyone to direct it towards what matters to them. This is how society has always made progress.
I believe these are the right values for building a positive AI future. If we help build this, then I think we will also continue to build a very strong business. That is all I wanted to preview 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 will have more progress on that front soon. We are investing aggressively because the potential is huge, and we know there are many ways to create value here. As always, I thank you for joining us on this journey. Over to Susan.
Susan Li, CFO:
Thank you, Mark, and good afternoon, everyone. Let's start with our segment results. Unless otherwise noted, all comparisons are year-over-year. Second quarter Family of Apps total revenue was $60.4 billion, up 28% year-over-year. Second quarter Family of Apps ad revenue was $59.4 billion, up 27% year-over-year, or 26% on a constant currency basis. In the second quarter, the total number of ad impressions served across our services increased 14%. Impression growth was solid across all regions, driven by user engagement and user growth, as well as optimization of ad load.
The global average price per ad increased 12% year-over-year, driven by improved ad performance, a more favorable macro environment compared to the second quarter of last year, and foreign exchange tailwinds. This was partially offset by strong impression growth, especially from regions and surfaces with lower monetization levels. Family of Apps other revenue reached $1 billion for the first time, up 73% year-over-year, primarily driven by WhatsApp paid messaging and subscription revenue.
In the Reality Labs segment, second quarter revenue was $431 million, up 16% year-over-year, driven by strong revenue growth from AI smart glasses, partially offset by lower Quest headset sales. Now to our consolidated results. Second quarter total revenue was $60.8 billion, up 28% year-over-year, or 27% on a constant currency basis. Second quarter total expenses were $42 billion, up 55% year-over-year, including $2.4 billion in legal-related costs and $1.2 billion in severance costs related to the May 2026 layoffs. The year-over-year increase was primarily due to higher employee compensation, infrastructure costs, legal-related costs, and third-party AI token costs. Excluding the severance costs mentioned, the growth in employee compensation was driven by an increase in technical hires, particularly AI talent, over the past year. The increase in infrastructure costs was due to higher depreciation, data center operating costs, and third-party cloud spending.
At the end of the second quarter, we had over 75,000 employees, down 3% from the first quarter. This total includes approximately 8,000 employees affected by the May 2026 layoffs. We expect most affected employees to no longer be included in our headcount by the end of the third quarter of 2026. Second quarter GAAP operating income was $18.8 billion, down 8% year-over-year, with an operating margin of 31%. Excluding the second quarter legal and severance costs, our second quarter operating income would have grown 9% year-over-year. The tax rate for the quarter was 16%. Net income was $15.8 billion, with earnings per share of $6.18. 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. At the end of the quarter, we had $90.3 billion in cash and marketable securities and $83.7 billion in debt. Now to business performance.
Two main factors drive our revenue performance: our ability to deliver engaging experiences for our community and our ability to effectively monetize that engagement over time. On the first point, we continue to see significant benefits from our content recommendation programs. On Instagram, global time spent grew double-digits year-over-year in the second quarter, primarily driven by improvements to Feed and Reels recommendations. On Facebook, global video watch time grew 9% year-over-year, and over 10% in the US and Canada, driven by ranking improvements.
We are finding LLMs increasingly capable of delivering ranking and recommendation gains. First, they make our existing systems smarter by understanding what content actually is and generating better training data. Second, LLM-powered agents also aid engineering development by evaluating content quality, detecting trends, and testing ranking changes. Earlier this year, we reached a milestone where every public Reels and Feed post on Instagram is automatically processed through an LLM, analyzed across multiple dimensions from topic to tone.
We are also working to bring more Facebook surfaces under this. These signals can then be passed downstream for ranking, recommendation, and content policy enforcement, which is a key building block for greater personalization. This quarter, we also began using our Muse model family for content understanding, including signals like video topic classification and summarization, and we are seeing positive initial results.
Finally, our recommendations are becoming more personalized, presenting more fresh content while giving people more direct control over what they see. On Reels, we released our single largest ranking improvement ever, combining faster inference with a new architecture that leverages deeper user history to improve predictions. This drove a 15 basis point increase in inbound sessions on Instagram, particularly strong in reshares and watch time, both strong indicators of improved content-user matching.
We are now bringing this to Feed, and initial results look promising. We are also bringing new content to people faster. Our investments in more real-time infrastructure and new video modeling improvements allow our largest ranking model to now identify high-quality new Reels at the time of creation. On Instagram Feed, over half of recommended content is now less than a day old, more than double what it was a year ago. We are also giving people more direct control over the content they see.
Today, Instagram users can access the Youralgo page, which allows users to write natural language prompts to adjust their recommendations. Similarly, on Facebook, we launched Shape Your Feed. Early results show over 80% retention for users who use the feature. Looking ahead, we are executing on long-term work to develop next-generation recommendation systems. This includes building foundation models designed to support both organic content and ad recommendations simultaneously, as well as developing LLM-native recommendation systems.
We reached our first research milestone in the first half of this year, continuously pre-training a large-scale model on recommendation data and observing healthy scaling laws in the process. We are encouraged by this milestone and expect continued progress in the second half of the year. Turning to the second driver of our revenue performance, improving monetization efficiency. The first part of this work is optimizing ad levels within organic engagement. Here, we continue to enhance our systems to show ads at the optimal time and place. In the second quarter, we also expanded ad availability on new surfaces, including completing the global ad expansion on Threads. On WhatsApp, we introduced support for more ad objectives and ad business goal and status, and continue to make progress on the global rollout. Turning to the second part of improving monetization efficiency, which is improving performance for businesses using our services.
Within our ad system, we are achieving performance gains as we deploy more sophisticated and predictive models. This quarter, we launched Meta Generative Recommender, a paradigm shift in how our ad system works. Instead of scoring every possible ad individually, we now use an LLM to jointly reason about ad content and user preferences and predict the best ad for each person. This makes our ad matching smarter and more precise, leading to compounding performance improvements for advertisers.
We deployed our first generative model into our ad retrieval system and saw significant improvements in ad performance. Early pilots using LLMs to better understand user preferences resulted in a 1% lift in in-app event conversion rates on Instagram. In the second quarter, we also advanced our user understanding model to analyze ad and organic activity and simultaneously improve user experience and advertiser performance.
Combined with our GEM model for ad ranking and sequence learning, these advancements drove an 8.3% lift in ad clicks on Facebook and a 15.7% lift in conversion rates. We are also leveraging AI to make it easier for businesses to manage their campaigns, develop ad creative, and interact with customers. Our AI-powered Advantage+ end-to-end solutions continue to grow, with an annual revenue run rate exceeding $75 billion. We are working to deepen adoption, as advertisers using multiple tools see compounding performance improvements.
I will share an example of how Advantage+ significantly simplifies and improves the efficiency of performance marketing for SMBs. (Inaudible), an Indian online apparel brand, was manually setting up Facebook and Instagram ads for every campaign. After adopting Advantage+ sales campaigns and layering on Advantage+ audience, placement, and budget optimization, they saw a 13% lift in purchases and a 16% improvement in ad-to-cart conversion rates. Adoption of our Gen AI ad creative tools continues to expand, with over nine million small businesses using at least one AI creative tool.
The image generation feature now allows advertisers to produce more creative assets at scale from existing content, including a new feature to create images from video assets, which saw adoption more than double this quarter. We also launched a new end-to-end creative solution, providing advertisers with AI infrastructure to turn real-time performance signals into their next creative decision while maintaining brand identity and tone. We built integrations with agencies from day one, so teams can diagnose, generate, and scale high-performing creatives without leaving their existing workflows. Looking ahead, with the launch of Muse Image, we expect to further enhance advertisers' ability to generate high-quality, on-brand creatives at scale.
With Meta Business Agents, businesses can better serve customers through our messaging apps, responding to inquiries, recommending products, and handling support around the clock. Earlier this month, we also launched the Meta Business Agent Platform, which provides enterprises with the infrastructure to build, customize, and deploy their business agents at scale on WhatsApp. The platform offers built-in enterprise-grade controls, guardrails, and measurement tools for large enterprises, so they can define rules and deliver personalized experiences within the messaging apps customers already use. Movita, one of the largest car rental companies in Brazil with nearly 400 locations, deployed a business agent on WhatsApp that handles the entire booking process from vehicle selection and pricing to payment in a single conversation.
Return customers can complete a booking in just three messages. Within a month, Movita reported a 44% year-over-year increase in daily bookings through WhatsApp, and 85% of conversations in the channel were fully resolved by the AI agent without human assistance. We are also building other ways to monetize our ecosystem. Two other revenue streams are subscriptions and monetizing our competitive models through APIs. Meta One is an evolution of our subscription product portfolio, designed to create more value for everyday users, businesses, and creators, giving them access to more features and AI tools to create, connect, and stand out. We are excited to bring this to more users and continue building enhanced tools for our subscribers. We also recently launched a high-intelligence model API at a competitive price and are encouraged by initial results. We recently made Muse Spark available to US developers, expanding its distribution and making it easier for developers to adopt the model.
We expect to roll out the model API to more distribution channels soon, make it available in more countries, and open it up to enterprises. Our approach to building capacity is strongly influenced by several key factors. First, the broad environment for building infrastructure is dynamic and uncertain in both the short and long term. The industry has historically under-built capacity for this wave of AI adoption, making existing capacity, including our own, extremely valuable. In the long term, the supply chain needs to be built up to support the capacity that we and others anticipate is needed for AI experiences. Second, we have high confidence in our ability to use capacity to scale and build our existing experiences and to continue investing in foundation models that will create significant new opportunities. Therefore, our current plans are aimed at maximizing capacity in 2026 and 2027. When we have had incremental capacity in the past, it has proven very valuable in scaling experiences, and we believe this holds true for this time frame.
In the longer term, the exact growth curve for usage is harder to predict, but we believe our distribution advantage will give us the opportunity to provide valuable AI products to everyone, whether our 3.6 billion users or millions of businesses. This should hold true regardless of whether our models are at the frontier, but we believe being at the frontier will unlock new markets and opportunities that may require additional compute. Therefore, our long-term capacity strategy aims to give us the flexibility to continue growing compute capacity in 2028 and beyond by laying the foundation for data centers and networks to accommodate future server decisions. The long-lived nature of these assets naturally provides flexibility, allowing us to adjust our investments to the pace of AI adoption.
Additionally, we have been making strategic investments in areas like in-house custom chips, which will provide long-term strategic flexibility and supply chain leverage. This will help achieve better returns on these long-term investments. Finally, we believe overall industry capacity will remain tight for the foreseeable future. As we have said before, we firmly believe that the models, consumer experiences, and enterprise products we are building will be the best and highest-return use of our infrastructure. These enterprise products have the potential to take many forms, as Mark mentioned, including tools, our API, or direct monetization of compute, given the significant market demand. We expect to remain flexible on these opportunities, which will help us have the compute we need when we need it while maintaining strategic flexibility and providing us with multiple avenues to generate investment capital returns that exceed our expectations, thereby more effectively funding our buildout.
In funding these infrastructure investments, the strength of our balance sheet allows us to attract capital from a wide range of markets to supplement the cash flow generated by our business. Our announcement yesterday with BlackRock is an example of the partnerships we can build to complement our approach to building infrastructure capacity. Now to our financial outlook. We expect third quarter 2026 total revenue to be between $61 billion and $64 billion. Our guidance assumes that, based on current exchange rates, foreign currency will be a roughly 1% headwind to total revenue year-over-year growth. Regarding expenses and outlook. We are raising the lower end of our expense outlook to incorporate the $2.4 billion in legal-related costs recognized in the second quarter.
We now expect full-year 2026 total expenses to be between $165 billion and $169 billion. We continue to expect operating income this year to be higher than operating income in 2025. We expect 2026 capital expenditures, including principal payments on finance leases, to be between $130 billion and $145 billion, narrowing from our previous outlook of $125 billion to $145 billion. Assuming no changes to our tax environment, we expect our tax rate for the remaining quarters of 2026 to be between 15% and 17%, up from our previous outlook of 13% to 16%.
Finally, we continue to monitor active legal and regulatory matters that could have a material impact on our business and financial results. For example, we continue to see scrutiny related to teen issues in several markets, and there are several teen-related trials in the US this year that could ultimately result in significant losses. In closing, our business momentum continued in the second quarter, with strong execution on core advertising and engagement initiatives. We are also advancing our efforts to bring personal superintelligence to everyone, launching exciting models and expecting to continue this momentum with new products for the rest of the year. With that, Krista, let's begin the Q&A session.
Q&A Session
Operator:
Thank you. We will now begin the question-and-answer session. (Operator instructions) Your first question comes from Brian Nowak of Morgan Stanley. Please go ahead.
Brian Nowak, Analyst:
Thank you for taking my questions. I have two, one for Mark and one for Susan. Mark, thank you for the rich information on the pipeline of new products like consumer and business agents, API tools, and compute leasing. There are many opportunities here. My question is, as you look at these opportunities and the current state of products and compute capacity, which ones do you expect to scale first in '26 and '27 to demonstrate quantifiable material returns on invested capital to investors?
Then Susan, regarding some public comments about '27 capacity and doubling capacity, thank you for the color on CapEx. Any early commentary on '27 CapEx? Even just thoughts on upside sources or downside pressure to help us think about the different ways to finance this multi-year buildout.
Mark Zuckerberg, Founder, Chairman and CEO:
I can answer the first question. Regarding the different opportunities and how we think about compute, overall, a significant portion of our compute is used for training. As a leading lab training models, I think this is an important investment. But the rest is used for a range of different product and revenue opportunities, which spans from optimizing and improving our core business, to new consumer products we are about to launch, to APIs, business agent work, the developer tools work on our roadmap, and the opportunity to directly sell compute, where I mentioned we have received numerous offers at prices significantly higher than what we pay.
The key point in thinking about this is that we believe the profit margins on selling intelligence will remain significantly higher than directly selling compute. But we also think there is a significant opportunity in selling compute. So, we are thinking about—I think you asked which area might scale the most, but actually, I am quite optimistic that all of these areas will achieve meaningful growth. I think we will share more information on several of them soon.
Susan Li, CFO:
Brian, on your second question, we are not currently providing specific guidance for 2027 capital expenditures. Infrastructure planning remains highly dynamic, and even this year, our outlook includes a range of potential outcomes. I mentioned in my prepared remarks that a key focus of our current infrastructure plan is to maximize capacity for 2026 and 2027, and to give us the flexibility to continue growing in '28 and beyond, while also allowing us to make server decisions as we assess actual demand for '28 and beyond.
So we are still working on our capacity needs for the coming years. In general, we think near-term capacity is more valuable than long-term capacity, and this remains a very dynamic planning process.
Operator:
Your next question comes from Eric Sheridan of Goldman Sachs. Please go ahead.
Eric Sheridan, Analyst:
Thank you for taking the questions. If I could, I have two. When you frame the enterprise opportunity, how much of it is immediately available based on your established go-to-market strategy as an extension of the existing ad and marketing business, versus how much will require building a new go-to-market strategy to capture? Then, Susan, if I could sneak in a second one. As you think about funding the business over the next few years, you mentioned the transaction announced yesterday as an example of exploring ways to finance future obligations. We get constant questions from investors on how to think about the mix of debt and equity and sources of capital. Philosophically, how do you balance your spending ambition with your need for capital? Thank you very much.
Mark Zuckerberg, Founder, Chairman and CEO:
Alright. I can take the first part. For enterprise, I think it will be a combination of an extension of existing business and building new business. The existing business is really about selling to marketers and businesses that are essentially customer-facing, trying to reach customers and sell directly to them. Obviously, this represents the vast majority of the Facebook and Instagram business. We believe there is an opportunity to use business agents to continue engaging with customers through messaging apps and other services, and, like the ad system, effectively get paid when we deliver results for businesses.
So, we see this as a natural extension of our existing sales and partnership relationships with millions of advertisers and hundreds of millions of small businesses that use our platform. So, I think this should be a fairly natural opportunity for us, and we are focused on delivering it not by trying to maximize sales in the short term, but by maximizing outcomes for users and building a robust auction mechanism, which we see as serving the business well over the long term. There are other enterprise customers, and I think we will increasingly serve them as well.
We are building coding and internal productivity tools, partly because we need to build them for ourselves. We need to ensure we have the right tools tuned for our own use. Now that we have them, we think there is a huge opportunity to serve others, whether small businesses or large enterprises. This is somewhat different from what we have historically been good at, and we will share more information on how we will build this soon.
But I think the opportunity here is very large. My perspective is that, while there is a lot of news around compute, the enterprise opportunity is the sum of all these different things. It's not just selling compute; it's also API services, productivity services, business agents for the non-marketing parts of a business—all part of the overall product.
I think there is a very, very large opportunity there. So we are very focused on it. This will be a new capability the company needs to build, but I think it's a very important capability so that we can maximize the opportunity in front of us.
Susan Li, CFO:
Eric, on your second question about funding sources, this is something we are thinking about carefully as we do our financial planning for the future. Obviously, our strong operating cash flow gives us an advantage in funding our infrastructure buildout. But we have also been developing our capital structure in recent years, including increasing the proportion of debt, to try to lower our cost of capital.
We find it prudent to continue adding cost-effective, long-term sources of capital as we invest in initiatives with long-term horizons, especially AI infrastructure projects. We have also broadened our financing channels, including through partnerships like the one we announced yesterday with BlackRock. We will continue to prudently evaluate different funding sources as we assess projects in the future.
Operator:
Your next question comes from Mark Shmulik of Bernstein. Please go ahead.
Mark Shmulik, Analyst:
Yes, thank you for taking my questions. Mark, everyone has a story about someone coming back from Silicon Valley, deep coding, building an agent with AI, and then they tell their parents they are using AI wrong, that it's like an advanced search tool. Consumer behavior is always hard to predict, but reading your op-ed on "AI for everyone," how do you see consumer adoption bridging this AI utility gap? Are we on the verge of a breakthrough, or do we just need to be more patient? Thank you.
Mark Zuckerberg, Founder, Chairman and CEO:
Well, I think some things have already broken through. One interesting thing about AI compared to other technologies is that roughly every couple of years (and this cycle might be accelerating), you get new capabilities that create new possible product lines. So clearly there is a market for consumer-facing AI assistants, which is what we are doing with Meta AI, and what competitors are doing in that space. Then in the past year, I think the market for coding agents has grown very fast, and it's the first real agent market. I think coding emerged first for a few reasons, right? You have technical customers willing to spend time making it work. Coding is inherently a digital, closed-loop activity. So in some ways, it's an easier system to train. But our bet—or one of the bets we are making—is that we believe consumer personal agents will ultimately be an extremely important and large market.
I think if you look out five years, whatever timeframe you want, the likelihood that billions of people won't have a personal agent that knows your goals and works for you 24/7 to achieve them, in whatever area you care about—whether it's helping you improve your health, hobbies, personal finances, making your home life more efficient, or improving and enhancing your relationships, helping your career, helping with all these different things—is extremely low.
This is a very, very deep set of use cases. But as you say, I think building for consumers is slightly different from building for developers. If you build for consumers, especially if you are trying to build something not for millions of people, but for billions, it needs to just work, right? It needs to be simple. I think the problem with many of the agents and prototype agents we see today is that they require a lot of debugging, you need to go into a terminal to set them up.
You start a use case, it seems a bit magical, but it might break down over time. I think the company that can deliver a personal agent that works "out of the box" will be the next major opportunity in AI. In addition to our work on Meta AI, business agents, and all these other things, we are very excited about this, and we are also developing what we believe will continue to advance new capabilities to create new product lines. So that's why we are very optimistic. Now, we will launch this product at some point in the near future, and it will be very exciting.
We haven't done it yet. So, I can only say so much about it on an earnings call. But I think this is a very large opportunity, and it's almost inevitable that someone will do it. I think this really plays to Meta's strengths as a company—we build consumer products that reach billions of people. Once we get something working, we are very good at scaling it to a large user base, and we are good at building the infrastructure to support these intensive applications. So I feel very good about this. I understand we need to deliver it to our community, and that's what we are very focused on.
Operator:
Your next question comes from Doug Anmuth of JPMorgan. Please go ahead.
Douglas Anmuth, Analyst:
Great. Thank you for taking the questions. One for Susan, one for Mark. Susan, you talked about LLMs becoming increasingly capable of delivering ranking and recommendation gains. Can you elaborate a bit more on the roadmap here and how far along we are in leveraging better models and more compute? Then Mark, on the number of external offers to monetize your compute, you are also buying capacity from multiple third parties. So hoping you can help us understand some of the differences? Is it just a matter of timing and transition, or is it about training versus inference, and using the best chip for each task? Thanks.
Susan Li, CFO:
Thanks, Doug. Let me start with the question about the roadmap for recommendations. First, we certainly see room for further improvement in recommendations for the remainder of this year and into 2027. We expect this to help us drive further engagement gains on Facebook and Instagram. I want to highlight a few points. First, we will continue to make recommendations more personalized and relevant to user interests, partly by advancing our recommendation models and architectures to capture user interests more precisely and respond more quickly to what users care about at any given moment. Our AI investments will play a major role in achieving this vision, including expanding LLM-based content understanding to better understand the posts and creators that users value, capturing user interests more precisely, responding more quickly to what users care about in the moment, and using AI to surface high-quality, fresh, and trending content while reducing the share of low-quality content.
Second, we continue to improve our data infrastructure to allow our models to train on more data and use that data more effectively. We have added more detail in describing the content users have interacted with in the past and enriched past user interaction sequences with more granular content. This allows our models to learn more precisely which interactions are more or less valuable to users. We continue to expand the length of user interaction sequences used in training, as well as the complexity of our model architectures on Facebook and Instagram, to leverage larger datasets.
Then, we have already made significant progress in using LLMs for content understanding, and given their ability to understand content more deeply, we will integrate them further into our recommendation and content policy enforcement stacks. And, I would say, overall, we are very optimistic about this set of work.
We are also investing in using LLM-based agent approaches to revamp our recommendation systems, and the number of releases from our ranking agents has increased in the first half of this year. This has also helped improve our engineers' productivity, which is another path we are very excited about.
Mark Zuckerberg, Founder, Chairman and CEO:
Alright. I can answer the second part. I think your question is about how we view the offers we receive to sell compute, while at the same time we are buying compute. I mean, the high-level observation is that there simply isn't enough compute to meet all demand. That's why we see, basically, we have received numerous offers for the compute we have, but we also have many internal uses that we consider very valuable.
Now, in running the business, obviously a common trade-off we need to make is how much you monetize it today versus developing future assets for the future. I think it's always a mix, right? You don't want to only do long-term things, and you don't want to only do short-term things and not validate the near-term market. But I also think it would be foolish to simply sell all your compute for short-term profit.
But when you have the opportunity to build intelligence on top of it, that is a multiple, and it compounds the value of the compute underneath it. So I think the answer is exactly what we are doing, which is basically using our significant capital to build compute, being confident in our ability to directly monetize compute when it makes sense, while also knowing we have many different use cases to monetize the intelligence on top of the compute.
Including the enterprise use cases we talked about, the consumer use cases we talked about, and the core business, which isn't even necessarily a new product we haven't talked about, but using it to further increase intelligence and improve ranking, recommendations, and ads in our core services. So I think all of these are real. This creates a dynamic where we are investing now to build these data centers, which will come online at some point in the future.
Obviously, you don't get value from them until they come online. But we basically see the demand for all of these as very large and want to maximize the opportunity to build all these different businesses.
Operator:
Your next question comes from Justin Post of Bank of America. Please go ahead.
Justin Post, Analyst:
Great. Thanks. Mark, you hired a senior leader for the AI lab about a year ago. Can you talk about how the lab is performing? Do you think Wall Street will actually see an uptick in product velocity in terms of models, chips, or other things? And then, what sustainable competitive advantage do you think the lab is building? Thanks.
Mark Zuckerberg, Founder, Chairman and CEO:
Yes. So, I mean, I am quite satisfied with our trajectory so far. We have released some impressive models in our early scaling ladder. As I said, we are scaling larger, more advanced models, and we are excited about that too. I think there is an intelligence aspect and a data aspect. I think in any product category you want to enter, there is a kind of flywheel effect where you can learn from the behavior of people using the product in that community. So, the better you do on that front, the more feedback you get, and then this inevitably makes the product better. So I think this is an area where people are understandably obsessed with intelligence, but the data and the knowledge of serving users are really important parts.
If you are talking about personal superintelligence, then obviously, having a very clear mental model of everything happening in a person's life and their goals will be a very important aspect of it. Certain companies have different advantages in different markets and different parts of this field. But I think part of the reason for working on these different things early is that, first, the technology is general.
So when you build intelligence, it can be applied to many different uses. Then you want to invest in building flywheels for those uses, because I think that's how you build a sustainable advantage over time. But I think we have shown at Meta that when we have a product in an effective form, I would say, we are probably the best company in the world at scaling those experiences to billions of people.
So I feel good about the personal agent work. I feel very good about the business agent work, we already have our advertiser base and the small businesses we serve, and our ability to scale and promote them. I think these are also enduring advantages, along with the data flywheels we build on top of them. Then, obviously, on the research side, you want to build a good culture. I mean, it's just basic things, but I feel like you just need to make sure you are building the team in a well-managed, consistent, stable, low-drama, compounding way, which is what you aim for. I think if you can do that, then maybe it's not a clear mathematical explanation, but I think that's how business works. So, yes, that's what we are doing.
Operator:
Your next question comes from Ross Sandler of Barclays. Please go ahead.
Ross Sandler, Analyst:
Yes. Hey, Mark. Continuing on the AI lab comments. Muse Spark 1.1 is very close to the frontier, but it's at the low end of the intelligence spectrum, I should say the low-cost end. From your last answer, it sounds like you think competing at both the low-cost end and the more expensive high-performance end is important. Can you talk a bit about that? And then, your lab leadership has also talked about returning to open source, returning to where you were a few years ago. So how does this fit into the strategy and all this discussion around monetizing AI products and models? Thank you very much.
Mark Zuckerberg, Founder, Chairman and CEO:
Yes. So, let me delve a bit into the science of this. Basically, at each stage of training a large model, you see new behaviors. So basically, what you want to do is build up from training smaller models to building larger models. The models we have released so far are based on—at a certain scale on the scaling ladder, we are continuing to scale larger models.
So I think for Muse Spark 1 and Muse Spark 1.1, for the scale of that model, given the development stage of the lab, they are very impressive models, and I feel good about them. You also want to have more advanced models, which is why we are scaling larger models, and why we are building a lot of research infrastructure around this. But at the same time, we also want to have good, more efficient models, which will be many of the services we deliver to consumers at scale.
But if you are serving billions of people, you want the ability to use more advanced models for very difficult problems, and you also want the ability to use simpler, more efficient models for the vast majority of prompts. So I think both are very important. Efficiency is certainly important, but I also think we want to be able to solve the most difficult problems for businesses and customers around the world.
So we care about both. Regarding open source, we have always believed that open source is an important part of the ecosystem, beneficial for the world, and it creates a positive feedback loop for us, where the community invests in our infrastructure stack, our work, and contributes improvements. But we have always said we will take a hybrid open-source and closed-source approach. This remains true. Now, in scaling Meta Super Intelligence Lab, it is actually somewhat counterintuitive that doing open-source models requires more work, because if you build a closed system just for your own use cases, it can be more rough. Whereas if you release it as open source, it will be used for many things, and you want it to be more comprehensive.
I just want to ensure that the MSL team is not constrained in building the smartest models we can achieve, and we expect to resume releasing some open-source models in the near future. But as we have always said, we are not dogmatic about this. We think open source is important. We want to contribute to that ecosystem. We plan to combine open-source and closed-source models.
Operator:
We have time for one final question, from Ken Gawrelski of Wells Fargo. Please go ahead.
Ken Gawrelski, Analyst:
Thanks. If I may, I have two questions. First, Mark, I wanted to touch again on your point about open-weight models. There is a lot of discussion on this topic, and you have commented on it. Why or why not? Would this change Meta's view on developing closed, proprietary frontier models? If open-weight models proliferate, is there an opportunity that Meta doesn't need to develop its own frontier models? That's the first question.
Second, if I may, a clarification for Susan. In your prepared remarks, you mentioned plans to maximize capacity for '26 and '27. Is that a demand or supply comment? Meaning, are you implying that Meta plans to internally use all the capacity it builds before '27? Or will you assess further buildout in '28 and beyond based on Meta's product and service demand? Thanks.
Mark Zuckerberg, Founder, Chairman and CEO:
I can answer the open-source question. Let me see, is the essence of the question whether, because we have some open-weight models, we can rely on those? I mean, currently, open-source models are not as powerful as frontier models. So the basic answer is no. Then, there is also the long-term policy debate and question about the actions of other companies, and whether this is something a company like Meta can rely on.
I think that is very tricky. So I think on both fronts, we believe we can do better, and we think relying on that is risky. I don't think that's the right approach. I think we are a company—if you look at Meta as a whole, many people see the surface layer, that we build some social media apps, we have an ad business, but we are actually a full-stack technology company.
We built our own data centers, infrastructure, chips, underlying software. When we started, because of my engineering background, I wrote a lot of systems code. A large part of Facebook's success was because it actually worked, right? When other social networks couldn't run fast and efficiently, it did. I think we are fully capable of building something more personalized, optimized, and efficient.
Certain qualitative experiences, others can't even build, because we have always been deep in the entire tech stack. In my view, having the ability to build our own models will obviously be a crucial part of the future tech stack, which is why this is important for Meta, and why others care about open source, and why open source overall is important, because other companies, even if they don't have the ability to build these models, don't want to rely solely on a few closed labs. I know open-source models have a very important place in the world. To be clear, this does not affect the API opportunity or anything else I talked about, because someone still needs to run the models, do the inference, and running them efficiently will continue to be a source of business advantage. So I don't think these are necessarily contradictory.
But when I look at what many discerning customers and companies around the world want, they want to know they control their own destiny, they can trust the models they use, their data is safe and not sent to competitors, etc. So I think open source will be important, but I think for us, building models will also be a key part of it.
By the way, I know all models are evaluated on a common set of benchmarks and then reduced to, okay, a certain model is within a few points of another model or something, but they do have different skill sets, just like people, right? Like people have strengths in certain areas, have different personalities, and perform differently on different things. If you are trying to build personal superintelligence for people, or business agents for small businesses, that might require some skills that are different from what other labs tune for. Just as being able to do full-stack work on our Instagram recommendations and ad systems is how we achieved these results over time, I believe building full-stack models will be a major advantage in how we build personal superintelligence agents, business agents, and all these different use cases for all the different customers we want to serve.
Therefore, in addition to the distribution advantage we have and the ability to reach all these people and scale effective products, I think this is a lot of enduring advantage: you build something specific to that use case and excel at it. I mean, I know it's a huge investment and a big bet. We see the technology working. We are satisfied with the lab's trajectory. I am excited about the products we are about to launch, and we believe this will be a big deal.
So, I mean, I know this whole industry is a big bet. My personal bet is that those who invest in this will be rewarded and feel very good about it over time.
Susan Li, CFO:
Ken, I will briefly answer your second question. When we mention focusing on '26 and '27 capacity, there are actually two factors. One is that we are in a demand-constrained situation now and for the foreseeable future, which actually includes our core business; if we had compute, we still have many places with positive ROI to deploy it.
Second, of course, just the uncertainty around the constraints of building long-term capacity. We talked about the need to further build out the supply chain in our previous comments. So I think after '27, when we look towards '28 and beyond, the world will be very different. Our own internal demand will evolve. A lot will be settled by then. Therefore, when we think about planning for '28 today, we are really focused on flexibility. It's just about giving us the ability to have the land and power, but the actual decisions about buying chips and other commodities will be made much later. So for now, I think we know there are many good use cases that need capacity in '26 and '27, and that's what we are really building for.
Ken Gawrelski, Analyst:
Great.
Mark Zuckerberg, Founder, Chairman and CEO:
Thank you all for joining our call today, and we look forward to speaking with you again soon.
Comments