Yann LeCun's AMI Labs Raises Over $1 Billion in Europe's Largest Seed Round for World Models

AMI Labs, the AI startup founded by Yann LeCun after his departure from Meta's AI research division, has raised over $1 billion in what becomes Europe's largest seed round. The company is building "world models" — AI systems that learn from physical environments and sensory data rather than relying primarily on the next-token prediction approach that powers current large language models.

World Models vs. Language Models

LeCun has been a prominent critic of the dominant language model paradigm, arguing that systems trained primarily on text cannot develop genuine understanding of the physical world. World models, by contrast, learn by predicting what will happen next in physical environments — how objects move, how forces interact, how scenes change over time. The approach draws on LeCun's decades of research in self-supervised learning and his belief that the path to human-level AI runs through understanding the physical world, not through mastering language.

The Funding

The billion-dollar seed round is unprecedented in both its size and its stage — seed rounds typically range from $1-5 million for early-stage companies. The scale reflects both LeCun's stature as one of the "godfathers of deep learning" and investor conviction that world models represent a fundamentally different approach to AI that could yield breakthroughs inaccessible to language-model-focused companies. The round was reportedly oversubscribed, with investors competing for allocation.

Implications

AMI Labs' approach, if successful, could create AI systems with a physical understanding that current language models lack — systems that can reason about mechanics, predict the consequences of physical actions, and interact with the real world in ways that go beyond text generation. Applications range from robotics and autonomous vehicles to scientific simulation and industrial automation. Whether world models can deliver on this promise — and whether they represent a path to artificial general intelligence as LeCun believes — remains one of the most consequential open questions in AI research.