Yann LeCun’s AMI Labs Secures Record $1.03B Seed Round for World Models

The artificial intelligence landscape shifted significantly this week as Yann LeCun, the Turing Award winner and former Chief AI Scientist at Meta, announced a staggering $1.03 billion seed round for his new venture, Advanced Machine Intelligence (AMI) Labs. This capital injection is not merely notable for its size—making it the largest seed round in European history—but for the fundamental challenge it poses to the dominant Large Language Model (LLM) paradigm. While the industry has spent the last three years scaling Transformer-based architectures, LeCun’s AMI Labs is betting that the path to true Artificial General Intelligence (AGI) requires a complete architectural pivot toward what he calls 'world models.'

LeCun has long been a vocal critic of the limitations inherent in generative AI, arguing that LLMs lack a fundamental understanding of the physical world, persistent memory, and the ability to reason through complex sequences of actions. AMI Labs is built to commercialize the Joint Embedding Predictive Architecture (JEPA), a concept LeCun championed during his final years at Meta. Unlike LLMs that predict the next token in a text string, world models attempt to learn internal representations of how the world works, allowing AI to predict the consequences of actions and plan toward goals in a manner more akin to biological intelligence.

“

AMI Labs is built to commercialize the Joint Embedding Predictive Architecture (JEPA), a concept LeCun championed during his final years at Meta.

The strategic composition of the investor syndicate underscores the high stakes of this pivot. Nvidia’s participation suggests that while world models may move away from the specific software logic of current chatbots, they will remain intensely compute-heavy, requiring the next generation of Blackwell and Rubin GPU architectures. Meanwhile, the involvement of Temasek highlights the global geopolitical interest in securing a foothold in the next wave of AI infrastructure that moves beyond simple text and image generation toward autonomous agents capable of operating in the physical world.