Key Takeaways
- • Mark Zuckerberg said on the Stripe-cofounder-hosted podcast Cheeky Pint that Meta greenlit an AI buildout of "like $100 billion, $200 billion" before there was proof it would work, with Meta guiding to as much as $145 billion in capital expenditures this year.
- • Meta is pursuing a "full-stack" strategy, building its own data centers, chips, and low-level infrastructure, a approach Zuckerberg links to years of constraints imposed by Apple and Google on Facebook's mobile apps.
- • Zuckerberg said a one-gigawatt AI data center costs $50 billion to $60 billion to build and roughly $15 billion a year to operate, and Meta is weighing a cloud business to sell AI computing capacity.
- • Meta's Muse personal agent could disrupt its ad model, which generated about 98% of revenue last quarter; Zuckerberg has proposed subscriptions and small commerce fees as alternatives but has not directly addressed impact on ad impressions.
- • Bank of America analyst Justin Post named rising AI spending, not agents, as the top risk to Meta.
Summary
Zuckerberg describes the effort as a "full-stack" operation in which Meta builds its own data centers, chips, and low-level infrastructure. He links this to nearly two decades of friction with Apple and Google, which control the dominant smartphone platforms. Facebook's iPhone app launched in July 2008 and its Android app in September 2009. Meta later sought workarounds, including pushing Android updates outside the Google Play store in 2013, which prompted Google to ban the practice. Apple's 2021 change allowing iPhone users to block app tracking was a major blow; Meta later said it would cost about $10 billion in 2022 alone. In a 2024 post, Zuckerberg wrote that "building our services constrained by what Apple will let us build on their platforms" was a formative experience.
Unlike Amazon, Microsoft, and Google, Meta does not operate a cloud business renting data centers to outside customers, so it bears those costs itself. Zuckerberg said a one-gigawatt AI data center costs $50 billion to $60 billion to build and roughly $15 billion a year to operate, and that Meta's Hyperion campus in Louisiana is designed to scale to five gigawatts. Bloomberg reported in July that Meta is building a cloud business to sell AI computing power and is weighing selling raw capacity, and Zuckerberg later confirmed cloud plans on the earnings call. On the podcast he described the compute as a hedge against failure.
Meta's ambitions extend to agentic commerce. Zuckerberg and Wang want Muse Wonder and Muse Wonder Full to power Muse, Meta's personal AI agent. Advertising accounted for about 98% of Meta's revenue last quarter, and eMarketer projects Meta's worldwide ad revenue will reach $243.46 billion this year, surpassing Google's for the first time. Collison questioned whether a business that "only succeeds by overcharging consumers" would want agents excluded, while Zuckerberg argued that businesses seeking to block agents will lose out and that Meta is receiving calls from businesses seeking deeper integration. Zuckerberg named subscriptions as "going to be a big one," with Muse free to use and paid plans at $20 and $100 a month, and floated taking "a small cut" of commerce savings. On the July earnings call he described business agents working "just like the ad system."
Zuckerberg has not directly addressed whether agents will reduce the ad impressions that generate nearly all of Meta's revenue. Wall Street is not currently pressing the point; Bank of America analyst Justin Post wrote last week that Meta is "back on offense with AI," identifying rising AI spending as the top risk.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.