So, Meta Platforms (META) stock is having a bit of a rough Friday. It's down sharply, and the reason is a story that feels very 2024: an internal AI project, code-named "Avocado," isn't quite ripe yet.
According to reports, this foundational AI model—the kind of big, general-purpose brain that companies are racing to build—is lagging in internal evaluations. Specifically, it's reportedly not keeping up with the newer models from rivals like Alphabet (GOOG) and the well-funded startup Anthropic. While "Avocado" apparently beat earlier versions of Google's Gemini model, it couldn't match the latest release. That's the kind of detail that makes investors nervous in a race where being second-best can feel like being last.
When Your AI Needs More Time on the Vine
Because of this performance gap, Meta has reportedly pushed the launch back. The model was originally slated for a March debut, but now it's looking like May at the earliest. In the fast-moving world of AI, a two-month delay isn't necessarily a catastrophe, but it's a signal that the competition is fierce and the technical hurdles are real. It's a reminder that for all the hype, building these systems is incredibly hard.
The $135 Billion Bet Behind the Scenes
This comes as CEO Mark Zuckerberg is making an enormous, long-term pivot. Meta has projected that its capital expenditure—the money it spends on physical assets like buildings and equipment—could hit a staggering $135 billion by 2026. The goal? To build gigawatt-scale data centers to power this next generation of AI.
It's a bet-the-company kind of move, shifting resources toward what it sees as the future. But it's also drawing scrutiny. Famed investor Michael Burry has warned that tech giants are abruptly pulling back on share buybacks to fund these soaring costs. When a company stops returning cash to shareholders to pour it into a massive, uncertain project, Wall Street pays very close attention.












