According to Business Insider, Meta’s revenue surged 24% to $59.9 billion in the final quarter of 2025, with ad impressions up 18% and ad prices rising 6%. CEO Mark Zuckerberg, however, dodged specific questions about the company’s ambitious AI reset, Meta Superintelligence Labs (MSL), led by former Scale AI CEO Alexandr Wang. He told analysts on the Wednesday call that the first models, expected in the coming months, are about showing a “rapid trajectory” rather than a single breakthrough, with the frontier being pushed steadily throughout the year. Zuckerberg admitted his answers might be “unfulfilling,” as the six-month-old AI rebuild is still in progress. Financially, the company is flush, generating $43.6 billion in free cash flow in 2025 even with heavy AI infrastructure spending, partly because AI already improved ad performance, leading to 3.5% more Facebook clicks.
The Substance of No Substance
Here’s the thing: Zuckerberg‘s performance was a masterclass in managing expectations downward. He basically said, “Don’t judge us on what we release soon; judge us on the slope of the line.” Analysts like Brian Mulberry at Zacks called out the lack of “real substance,” and you can’t really blame them. Wall Street lives on milestones and metrics, and “trust the trajectory” isn’t one. But in a way, it’s a smart, if frustrating, pivot. After the letdown of the last Llama model in April, which developers found lagging in coding and reasoning, promising another big bang would be risky. So instead, he’s setting up a narrative of steady, demonstrable progress. The question is, will investors buy a story when they can’t see the book?
The Cushion of a Cash Firehose
And this is why Meta can play this long game. Let’s be real. That financial performance is absurd. A 24% revenue jump on that scale is almost unheard of. The core ads business isn’t just healthy; it’s being directly improved by AI *right now* in the background. That’s the ultimate hedge. It gives them what Bloomberg’s Mandeep Singh called a huge “cushion.” They can afford to spend billions on AI infrastructure—those Nvidia chips aren’t cheap—while telling everyone the really cool, profit-driving AI products are still a year or more out. It’s the luxury of a company that has one massively profitable engine funding the build of a second. Not every tech giant has that.
Specialization Over Supremacy
The reported focus of their new models, “Mango” for images/video and “Avocado” for coding, hints at the real strategy. They’re not trying to build a single, all-conquering GPT-o1-preview model to beat OpenAI or Google head-on. They’re specializing. This makes total sense. Trying to beat everyone everywhere, as Singh noted, is a fool’s errand. Meta’s strengths are in visual media and its massive developer ecosystem. A model that’s best-in-class for Instagram or for helping devs build on their platforms? That has clear, monetizable paths back to the ads business. It’s a more pragmatic, and probably more sustainable, approach than a moonshot for artificial general intelligence.
The Patience Gamble
So what’s the risk? Basically, credibility and momentum. The AI race is moving at a ludicrous speed. Asking for patience until 2026 is an eternity in AI time. If competitors drop groundbreaking models every quarter and Meta’s “trajectory” looks flat by comparison, the narrative will sour fast. There’s also the internal execution risk—remember, they reportedly missed a target to release a new Llama model by end of 2025. Zuckerberg is betting Meta’s immense cash flow can buy the time needed for his new team to cook. But in tech, money can’t always buy speed or innovation. He’s playing a long game in a short-attention-span world. We’ll see if Wall Street stays for all the innings.
