According to TechCrunch, Meta’s chief AI scientist Yann LeCun is planning to leave the company in the coming months to build his own startup. The Financial Times reported that LeCun, who won the prestigious A.M. Turing Award and serves as a professor at New York University, is already in talks to raise capital for his new venture. His startup would focus on continuing his work on “world models,” which are AI systems that develop internal understanding of environments to simulate cause-and-effect scenarios. The departure comes as Meta has invested $14.3 billion in data-labeling vendor Scale AI and hired over 50 engineers from competitors to build a new AI unit called Meta Superintelligence Labs. This reorganization has reportedly created chaos within Meta’s AI division, with new talent frustrated by bureaucracy while the previous generative AI team saw its scope limited.
Meta’s AI identity crisis
Here’s the thing about Meta’s current AI situation: they’re trying to play catch-up while their foundational research leader might be walking out the door. LeCun has been running FAIR (Fundamental AI Research Lab), which focuses on long-term research that might not pay off for 5-10 years. But after their Llama 4 models failed to keep up with rivals, Mark Zuckerberg decided to overhaul everything with this new MSL division.
So now you’ve got this classic big company problem. The long-term research vision that made Meta interesting to AI purists is getting overshadowed by the “we need to compete with OpenAI yesterday” panic. And honestly, can you blame LeCun for wanting out? When the company brings in 50 new people from competitors and puts Scale AI’s CEO in charge of a new division, it sends a pretty clear message about where priorities lie.
The world models gold rush
LeCun’s planned startup isn’t just some random AI play – he’s going after what many consider the next frontier. World models are basically AI systems that can understand how the world works well enough to predict outcomes. Think about teaching an AI the concept of “if I drop this glass, it will break” without having to actually break thousands of glasses first.
He won’t be alone in this space. Google DeepMind and startups like World Labs are already working on similar technology. But LeCun brings serious credibility – we’re talking about one of the godfathers of modern AI here. His recent tweet sums up his philosophy pretty well: “It seems to me that before ‘urgently figuring out how to control AI systems much smarter than us’ we need to have the beginning of a hint of a design for a system smarter than a house cat.”
What this actually means for AI
Look, when someone like LeCun leaves a company like Meta to start his own thing, it’s a pretty big deal. This isn’t just another executive departure – it’s a signal that the most forward-thinking AI researchers might be losing faith in big tech’s ability to do truly groundbreaking work.
And you know what? He might be onto something. The current AI landscape is dominated by these massive language models that are incredibly impressive but also… kind of limited in fundamental ways. They don’t really understand cause and effect. They can’t reason about the physical world. Basically, we’ve built amazing parrots rather than creating intelligence.
If LeCun can actually build a company that moves beyond the current LLM hype cycle, he could fundamentally change the direction of AI development. But the bigger question is whether Meta’s pivot toward immediate competitive pressure will cost them their long-term vision. Sometimes the most expensive departures aren’t the ones that hit your stock price today, but the ones that leave you without the people who were thinking about tomorrow.
