LeCun’s AMI Labs targets €3bn to build world-model AI systems

LeCun’s exit from Meta to build AMI Labs is one of the most consequential talent moves in the AGI race this year. Unlike most frontier players doubling down on autoregressive LLMs, he is explicitly betting on ‘world models’—systems that learn dynamics of the physical world, not just token sequences. That is a direct challenge to the current paradigm and, with a targeted €3 billion valuation before launch, signals that capital markets are willing to finance an alternative path to superintelligence led by a single, highly trusted researcher.

Strategically, AMI Labs plus Nabla is an intriguing combination. Nabla’s healthcare documentation business gives AMI a regulated, high‑stakes domain in which to prove deterministic, auditable world‑model agents. If those architectures work in medicine, they will be compelling in robotics, transportation, and industrial control—areas where today’s LLMs struggle. The move also fragments Meta’s AI influence: instead of one integrated platform, we now have Meta’s Scale‑AI‑centric superintelligence push on one side and LeCun’s open‑ish, world‑model lab on the other. That increases competitive diversity, which is healthy for the field. If AMI can turn its conceptual critique of LLMs into state‑of‑the‑art systems, it might reset expectations about what kinds of architectures are needed to reach AGI.