Yann LeCun raised $1B for world models - Europe's largest seed ever

Apr 23, 2026 · #yann-lecun #ami-labs #meta #ai #world-models #jepa #paris #europe

Yann LeCun raised $1B for world models - Europe’s largest seed ever

Yann LeCun, Turing Award laureate and former Chief AI Scientist at Meta, has raised $1.03 billion for a new startup, AMI Labs. The largest seed round in European history. Before any product.

Yann LeCun world models AI

Yann LeCun. Turing Award laureate - the equivalent of a Nobel for computer scientists. Twelve years at Meta as Chief AI Scientist.

In November 2025 he left. The official reason: “growing differences over the direction of Meta’s AI strategy.”

The unofficial reason - and it’s worth noting this was deliberate: LeCun has publicly said for years that ChatGPT and Claude are heading down a dead end.

AMI Labs - first round $1.03B

March 2026. LeCun announces a new startup - AMI Labs (Advanced Machine Intelligence). Headquartered in Paris, with offices in New York, Montreal, and Singapore.

First funding round - $1.03 billion. Pre-money valuation - $3.5 billion. All before any product.

Investor list:

Bezos Expeditions (Jeff Bezos personally)
Cathay Innovation
Greycroft
Hiro Capital
HV Capital

This is the largest seed round in European history. Period.

What AMI is building - world models

For eight years LeCun has been publicly criticizing today’s AI models - GPT, Claude, Gemini. His argument:

“Language models learn from text. Text is just a shadow of reality. True intelligence has to understand the physical world - the way a child learns, not from reading, but from touching, trying, observing.”

AMI is building world models - AI models trained on physical reality, not just text. The architecture is called JEPA (Joint Embedding Predictive Architecture).

Use cases per LeCun:

Robotics - robots that understand their environment and predict consequences of their actions
Healthcare - AI that understands biology and medicine “from first principles”
Manufacturing - autonomous systems for factories

Sound familiar? Because these are exactly the areas Tesla (Optimus), NVIDIA (Omniverse), and every other major player is moving into.

Why this matters

Three reasons.

First - LeCun isn’t a hype-monger. He’s a serious scientist with a track record predating the AI boom. If he’s putting his reputation on the line saying LLMs are a dead end, it’s worth listening.

Second - investors put up a billion dollars. Bezos personally. These aren’t people who throw money at hype. They’re people with information access at the level of Meta’s CEO - and they chose to bet against the current wave.

Third - it’s happening in Europe. In Paris. The first serious AI competition for Silicon Valley and China is being born on our continent.

My take

LeCun has always played the long game. His convolutional neural network research from the 1980s now powers all of computer vision. He waited thirty years for the world to understand what it was about.

Maybe with world models it’ll be similar. Maybe AMI Labs will build something that changes how we think about AI in 5-10 years. Maybe not.

But one thing is certain - in 2026 we have the most diverse AI ecosystem in history. OpenAI plays cloud. Anthropic plays safety. Google plays ecosystem. Meta plays open-source. Tesla plays physical world. NVIDIA plays hardware. China plays independence. And LeCun plays world models.

Each one has a sensible hypothesis on its own. The question is which of them will be right in five years.