Yann LeCun — Turing Award winner, pioneer of convolutional neural networks, and Meta's chief AI scientist for the past decade — has launched his own startup. And he's using it to bet $1 billion against the entire LLM paradigm.

AMI Labs

Advanced Machine Intelligence (AMI), based in Paris, announced a $1.03 billion seed round at a $3.5 billion pre-money valuation. Backers include Jeff Bezos, Mark Cuban, Eric Schmidt, Nvidia, Toyota, Samsung, and French telecom billionaire Xavier Niel. The round was co-led by Cathay Innovation, Greycroft, Hiro Capital, and HV Capital.

It's the largest seed round a European AI startup has ever raised.

The Thesis

LeCun has been saying it for years, but now he's putting a billion dollars behind it: LLMs are a dead end for human-level AI.

"The idea that you're going to extend the capabilities of LLMs to the point that they're going to have human-level intelligence is complete nonsense," he told WIRED.

His alternative: world models — AI systems built on JEPA (Joint Embedding Predictive Architecture) that learn abstract representations of physical reality instead of predicting the next token. The idea is that most human reasoning is grounded in the physical world, not language.

What AMI Will Build

AMI aims to create AI systems with persistent memory, real-world reasoning, planning capabilities, and controllable safety. Target markets include manufacturing, biomedical, and robotics — domains where understanding physics matters more than generating text.

The founding team is largely drawn from Meta's AI research org. LeCun will continue his NYU professorship while leading AMI from offices in Paris, Montreal, Singapore, and New York.

Why It Matters

This isn't just another AI startup. LeCun is one of three researchers who won the 2018 Turing Award for deep learning. When someone at that level says the current approach is fundamentally wrong and raises a billion dollars to prove it, the industry pays attention.

Whether AMI delivers or not, the bet itself is significant: the biggest counter-narrative to the LLM scaling thesis now has serious capital behind it.