Yann LeCun’s AMI Labs raises $1.03 billion to build world models

AMI Labs, the new artificial intelligence venture co-founded by Turing Prize winner Yann LeCun, has secured $1.03 billion in funding at a $3.5 billion pre-money valuation. The company is tackling the development of 'world models,' a class of AI designed to learn from observing reality rather than just text. The significant capital injection signals strong investor appetite for long-term, fundamental research into alternatives for large language models, particularly for applications where factual accuracy is non-negotiable.

The company's approach is rooted in the Joint Embedding Predictive Architecture (JEPA) proposed by LeCun in 2022. CEO Alexandre LeBrun acknowledged that commercial applications are years away, positioning AMI Labs as a research-first organization. The round was co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions, with participation from NVIDIA, Samsung, Toyota Ventures, and prominent individuals including Tim Berners-Lee and Eric Schmidt. To pursue its goals, AMI Labs is establishing teams in Paris, New York, Montreal, and Singapore, and has already named digital health startup Nabla as its first partner to explore real-world deployments.

This funding event places AMI Labs alongside other heavily-backed efforts in the world model space, suggesting a new, capital-intensive front is opening in the AI industry. While LeBrun anticipates 'world models' will become the industry's next popular term, he asserts that AMI's focus on deep research and its commitment to publishing papers and open-sourcing code will differentiate it from competitors. The investment gives the company a substantial runway to cover the high costs of compute and talent, while its strategy of engaging early with partners like Nabla indicates a plan to ground its theoretical work in practical evaluation from the outset.

AMI Labs' $1.03B raise for a long-term research project without immediate commercial goals signals a strategic shift in AI investment. Top-tier VCs and corporate backers are now placing substantial bets on foundational model architectures that move beyond language, willing to underwrite years of R&D for a chance at developing AI with a genuine understanding of the physical world.