Meta Pivots On AI Under The Cover Of A Superb Quarter
Better-than-expected results helped Mark Zuckerberg navigate what might have been a tricky transition from an open-source model to “personal superintelligence.”

On Wednesday, Mark Zuckerberg addressed every constituency that mattered. In a video posted to Reels, and in a 616-word written manifesto, he first talked directly to his users about his desire to build a new “personal superintelligence” that will help people rather than take their jobs.
This is distinct from others in the industry who believe superintelligence should be directed centrally towards automating all valuable work, and then humanity will live on a dole of its output.
Then he spoke to his new AI superintelligence team, and those who might be interested in joining it, by stressing Meta Platforms Inc.’s strengths in the AI race:
We have the resources and the expertise to build the massive infrastructure required, and the capability and will to deliver new technology to billions of people across our products.
And later, after the closing bell, he gave investors everything they wanted to hear and more. Revenue rose 22% in the second quarter compared with the period a year earlier, and profit was up 36% — both ahead of analysts’ estimates. For the current quarter, Meta forecast revenue of as much as $50.5 billion, again ahead of estimates. Astoundingly, the company is still expanding its user base, up 6.4% from the quarter a year earlier: 3.48 billion people now use Meta-owned apps daily. Shares jumped as much as 12% in after-hours trading.
As with previous quarters, the strong performance of Meta’s advertising business has been enough to convince investors to look the other way when it comes to the less pleasant parts of its results. Costs are increasing sharply as Meta, like its peers, invests in more data center infrastructure. The company has stood apart from the pack in its willingness to shovel obscene amounts of money in the direction of AI talent. Costs will keep growing in 2026, Meta said.
And no Meta check-in would be complete without checking in with my favorite number: Losses from Meta’s Reality Labs unit, the “metaverse” division, were $4.53 billion for the quarter, the fifth consecutive quarter of losses in excess of $4 billion. (Zuckerberg could have hired at least two AI researchers for that kind of money.)
The underlying message to take away from Wednesday’s communications blitz is that it helped Meta navigate what might have been a tricky transition for its AI strategy, one that could have been interpreted as an embarrassing admission that its initial approach was proving misguided.
What some analysts have described as a doubling-down from Meta on AI looks to me more like a critical pivot: After going full throttle with building an open-source AI model, Llama, Zuckerberg has now changed course and strongly hinted that this “personal superintelligence” — powered by the insights into users that only Meta has — will be closed to outsiders. The company will be “careful about what we choose to open source,” Zuckerberg wrote in his manifesto — a statement about safety, yes, but one that matters for its business model, too. Recent releases of Llama have been poorly received as its capabilities fell behind US competition while being shown up by the cost efficiencies of China’s DeepSeek. Meta has no footing in the enterprise AI space that OpenAI, Microsoft Corp., Amazon.com Inc. and others are going after — meaning it must look to the consumer market to find its much-needed return on investment. A closed model makes that aim more straightforward — why give away the fruits of all this investment to anyone else?
It’s easy for Zuckerberg to say his AI won’t be the one to replace jobs, but that’s because no companies are coming to him for that. Making lemonade from those lemons, Meta will lean heavily on that idea that it’s on the side of the everyday person. While farcical, it’s a tone that might land. The PR strategy started in earnest on Wednesday with Zuckerberg’s manifesto. “The rest of this decade seems likely to be the decisive period for determining the path this technology will take,” it read, “and whether superintelligence will be a tool for personal empowerment or a force focused on replacing large swaths of society.”
Also still to be determined: what “personal superintelligence” even means, practically speaking. I guess those details come later.