Meta Platforms (META) Q1 FY26 earnings call: guidance and Q&A

What signposts is Meta watching over the next 12-24 months to ensure a healthy ROIC on its significant AI CapEx and infrastructure spend?

Mark Zuckerberg stated the focus is on building leading models and products that can scale to billions of people, then monetizing them. Key milestones include technical quality of models, product scaling, and then monetization efficiency. He expressed confidence in the lab's progress and the quality of Muse Spark and Meta AI.

“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.”

asked by Brian Nowak · answered by Mark Zuckerberg

How is the team balancing further model training and specialization versus product launches? Also, any insights into 2027 CapEx given potential significant step-ups mentioned by peers?

Mark Zuckerberg explained a consistent roadmap: research teams focus on scaling intelligent models for agents, while product teams build on these strong models. Susan Li stated no specific 2027 CapEx outlook, noting Meta has consistently underestimated compute needs due to rapid AI advances and internal use cases. They will continue building infrastructure with flexibility.

“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.”

asked by Mark Shmulik · answered by Susan Li

How does Meta see the opportunity for agentic compute extending its media/commerce engagement, and what is the opportunity for SMEs and enterprises?

Susan Li highlighted near-term focuses on deepening engagement, personalizing ads, and helping SMBs. Longer-term, the vision is for personal and business agents to interact, forming a thriving commerce ecosystem. She noted 10 million weekly conversations with business AIs on messaging platforms, up from 1 million, and expects monetization models to evolve.

“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.”

asked by Eric Sheridan · answered by Susan Li

What are the gating factors for launching more AI glasses under other brands, and what would define a successful year for units sold? Also, how much of the 10% RIF is due to AI efficiencies versus fitness, and how will employee needs grow relative to top-line?

Susan Li stated strong growth in AI glasses sales, with demand shifting to the latest generation due to improved features. She did not provide unit targets but expressed excitement for display glasses. On headcount, she noted the company is leveraging AI to increase productivity and continuously evaluating structure, without specifying a direct link to the 10% RIF or future growth relative to top-line.

“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.”

asked by Youssef Squali · answered by Susan Li

How does Muse Spark unlock new products, and what is the expected product cadence over the next 9 months for consumer or business enterprise products?

Mark Zuckerberg emphasized that Muse Spark validates the team's effort and infrastructure. He stated that the field is moving quickly, and while he couldn't share competitively sensitive information or specific dates, the focus is on quality over speed. He expects a lot of innovation and that the picture of the roadmap will become clearer in subsequent quarters.

“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.”

asked by Justin Post · answered by Mark Zuckerberg

How do new consumer applications influence Meta's agentic strategy? Will the lab stay consumer-focused or pursue recursive self-improvement like others?

Mark Zuckerberg highlighted Meta's focus on making AI experiences polished and accessible for billions, unlike rougher systems. He sees massive upside in delivering more intelligent models through agents. He confirmed the lab's focus on both agents and self-improvement, stating that leading models require self-improvement capabilities, but emphasized that Meta's primary focus is empowering individuals and building consumer products.

“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.”

asked by Ross Sandler · answered by Mark Zuckerberg

Is the personal agent opportunity a short, medium, or long-term goal? And how do ranking and recommendation model improvements drive greater usage, given the scale of Facebook and Instagram?

Mark Zuckerberg stated personal agents have short-term versions but massive long-term upside as models improve. Susan Li explained that ranking improvements are ongoing, driven by better data infrastructure, more complex model architectures, and personalized recommendations. She noted that doubling sequence length is one of many improvements, with a big roadmap ahead.

“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.”

asked by Ronald Josey · answered by Susan Li

What are the key unlocks from leveraging LLMs like Muse Spark in the ad business? And what is the strategic importance and status of the Manus deal?

Susan Li explained that LLM-scale models like GEM are being integrated into ad systems through adaptive ranking models, which intelligently route requests to more compute-intensive models for higher conversion probability, improving performance and inference ROI. She stated that Meta is still working through the details on Manus and has no update.

“We also developed an approach that intelligently routes requests 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 increase inference ROI.”

asked by Douglas Anmuth · answered by Susan Li