Yann LeCun has raised $1.03 billion for a new AI startup built around the idea that large language models will not be enough to reach artificial general intelligence.

Advanced Machine Intelligence Labs, co founded by the former Meta chief AI scientist, secured the funding at a $3.5 billion pre money valuation. The round is one of the largest early stage financings in AI and the biggest seed round recorded in Europe.

AMI Labs is building world models, AI systems designed to learn from the physical world rather than only from text. The approach reflects LeCun’s long running critique of large language models, which generate answers by predicting the next token but do not develop the kind of physical understanding needed for planning, reasoning, and robotics.

The company is targeting areas where spatial awareness and real world understanding matter, including robotics, healthcare, transportation, and industrial systems. Those markets could be more attractive to companies that have been skeptical of chatbot driven AI but need systems that can operate in physical environments.

The investor list shows the scale of that bet. Backers include Nvidia, Temasek, Samsung, Toyota, Bezos Expeditions, Cathay Innovation, Greycroft, Hiro Capital, and HV Capital.

LeCun is not a fringe critic of the current AI boom. He shared the 2018 Turing Award for his work on deep learning and helped pioneer convolutional neural networks, which became central to modern computer vision. He also spent more than a decade leading Meta’s Fundamental AI Research lab before leaving to launch AMI.

The timing is notable. Meta has invested heavily in Llama and other large language models, while OpenAI, Google, Anthropic, and xAI continue to scale text based systems. AMI is effectively raising capital to test whether the next big leap in AI comes from understanding the physical world rather than making chatbots larger.

Toyota’s participation is especially important. Automakers and industrial companies have clear incentives to back AI systems that can reason about motion, physics, and space. For them, world models are not just a research thesis. They could become the foundation for robots, autonomous systems, and factory automation.

The funding gives AMI a rare amount of capital before it has proven its technology in the market. It also turns LeCun’s critique of the dominant AI model strategy into one of the most expensive scientific wagers in the industry.