Yann LeCun Raises $1B to Prove the Entire AI Industry Got It Wrong
Yann LeCun, the Turing Award winner who spent over a decade as Meta's chief AI scientist, has raised $1.03 billion for his new venture AMI Labs at a $3.5 billion pre-money valuation. It is Europe's largest seed round ever. And his pitch to investors is essentially this: every major AI company in the world is building the wrong thing.
AMI Labs, founded just four months ago, is betting on "world models" — AI systems that learn by observing and interacting with reality rather than by processing mountains of text. LeCun's approach, called JEPA (Joint Embedding Predictive Architecture), represents a fundamental departure from the large language models that power ChatGPT, Claude, Gemini, and virtually every other AI product that has captured public attention. The investor list reads like a who's who of tech capital: Nvidia, Temasek, and Jeff Bezos's Expeditions, among others.
The Case Against LLMs
LeCun has been making this argument for years, even while drawing a paycheck from Meta. His core claim is that language models are a dead end for achieving anything resembling real intelligence. They can generate fluent text, but they do not understand the physical world. They cannot predict what happens when you push a cup off a table. They cannot reason about spatial relationships the way a toddler can after a few months of observing the world.
JEPA is designed to learn from sensory data — video, audio, physical interactions — the way biological brains do. Instead of predicting the next word in a sequence, it learns abstract representations of reality by predicting how different parts of an observation relate to each other. In theory, this could produce AI systems that genuinely understand cause and effect rather than just pattern-matching statistical correlations in text.
A $3.5 Billion Bet on Being Right
The valuation is staggering for a company that is four months old and has produced no commercial product. Alexandre LeBrun, the former Meta executive, serves as CEO, while LeCun holds the title of executive chairman and maintains his professorship at NYU. The structure suggests LeCun wants to be the intellectual architect without getting bogged down in the operational grind of running a startup.
What makes this interesting is the timing. The AI industry has collectively poured hundreds of billions of dollars into scaling LLMs. OpenAI, Google, Anthropic, and Meta itself are locked in an arms race to build bigger, more capable language models. LeCun is now the most credentialed person in the field to publicly declare that this entire approach has a ceiling — and to put a billion dollars behind an alternative.
The Skeptic's View
There are legitimate reasons to be cautious. World models have been a research concept for decades without producing anything commercially viable. JEPA is theoretically elegant, but scaling it to useful applications is an unsolved problem. And $1 billion sounds like a lot until you compare it to the $10-20 billion rounds that OpenAI and xAI have raised to pursue the approach LeCun says is wrong. If scaling is what matters, AMI Labs is bringing a knife to a gunfight.
There is also the question of whether LeCun's critique of LLMs is as devastating as he makes it sound. Yes, language models do not understand physics. But they have proven remarkably useful for coding, writing, analysis, and reasoning tasks that do not require physical world understanding. The market does not care about theoretical purity — it cares about what works today.
The Bottom Line
Yann LeCun is either the smartest person in AI or the most expensive contrarian the field has ever produced. His track record — co-inventing convolutional neural networks, winning the Turing Award — earns him the benefit of the doubt. But the history of technology is littered with brilliant researchers who were theoretically correct but commercially irrelevant. If JEPA works at scale, LeCun will have been vindicated and the LLM era will look like a very expensive detour. If it does not, AMI Labs will go down as the most prestigious science project a billion dollars ever bought.