Speaking at VivaTech in Paris, Yann LeCun said ChatGPT, Claude and Gemini are “not a path” to human-level intelligence because current LLMs still fail to understand the physical world “as well as a rat.”
AMI Labs is pursuing that gap with JEPA-based “world models,” which aim to learn abstract representations and causal, multi-outcome reasoning rather than rely on next-token text prediction.
The Paris startup raised a $1.03 billion seed round in March at a $3.5 billion valuation, backed by Nvidia and Bezos Expeditions, giving LeCun an unusually well-funded test of his alternative AI approach.
AMI Labs plans to publish research and open-source code as development continues, though CEO Alexandre LeBrun has said world models may take years to move from research into deployed products.
Is a Paris startup's record funding the beginning of the end for Silicon Valley's AI dominance?
With a billion dollars now funding physical AI, how soon will robots truly leave the research lab?
Can AI truly understand the physical world without first mastering the language we use to describe it?
AMI Labs Raises €890M to Challenge LLMs with World Models: Yann LeCun’s Vision for the Next Generation of AI
Overview
AMI Labs, a newly established French AI laboratory led by Yann LeCun, has quickly become a major force in the AI field. The lab attracted strong investor confidence, raising about €890 million—far surpassing its initial €500 million target. This success is driven by a formidable leadership team and a bold vision to challenge the dominance of Large Language Models by focusing on world models that understand and predict the physical world. With substantial funding and ambitious goals, AMI Labs is well-positioned to pioneer a new direction in artificial intelligence.