According to Fortune, AI pioneer Yann LeCun is targeting a €500 million (about $586 million) funding round for his new startup, Advanced Machine Intelligence Labs (AMI Labs), which could value it at a staggering $3.5 billion before it even launches. The company, which plans to establish its Paris headquarters early next year, aims to build “world models”—AI that understands physics and plans actions, not just predicts text. LeCun, who confirmed his departure from Meta on November 18 after 12 years, will serve as executive chairman. He has already selected Alexandre LeBrun, founder of health-tech company Nabla, as the startup’s CEO. While Meta won’t invest, the two companies plan a partnership, and Nabla will get first access to AMI’s tech for developing FDA-certifiable healthcare AI systems.
LeCun’s bet against the LLM grain
Here’s the thing: LeCun isn’t just starting another AI company. He’s making a very public, very expensive bet that the entire industry is heading down a limited path. He’s been openly skeptical, saying we won’t get to human-level AI just by scaling today’s large language models because they don’t truly understand the world. So his startup is his vehicle to pursue what he sees as the “next big revolution”: systems with persistent memory that can reason and interact with the physical world. It’s a direct challenge to the Silicon Valley playbook he says is “completely hypnotized” by current generative AI. And he’s putting his money—or rather, convincing investors to put theirs—where his mouth is.
The valuation bubble question
A $3.5 billion valuation for a pre-revenue, pre-launch company is, to put it mildly, audacious. It immediately amplifies whispers about an AI investment bubble. Can even a Turing Award winner like LeCun command that kind of premium without a product? He’s essentially selling pure vision and credibility. The funding target itself is one of the largest pre-launch raises in AI history. It tests whether investor confidence in a founder’s reputation can outweigh the complete lack of commercial traction. And it’s not happening in a vacuum—well-funded European rivals like Black Forest Labs (valued at $4B) and Quantexa are already in the arena. This isn’t just about building a new AI; it’s a high-stakes experiment in market psychology.
The Meta divorce and Paris play
LeCun’s exit from Meta is fascinating timing. It coincides with Meta’s strategic pivot toward more powerful LLMs under new chief AI officer Alexandr Wang, as detailed in a New York Times report. Basically, his research direction and Meta’s corporate strategy may have diverged. But the planned “partnership” is a classic Silicon Valley-style soft landing. It lets Meta keep a tie to one of its most famous scientists without funding his competing vision. And LeCun’s choice of Paris is deeply personal. He’s long championed European AI talent—he convinced Meta to open its FAIR lab there in 2015. Now, he’s arguing that to do truly novel research, you have to get outside the Valley’s groupthink. It’s a powerful statement.
What success actually looks like
So what does a win for AMI Labs even look like? The immediate application through Nabla in healthcare is a smart, concrete starting point. It provides a real-world testbed for “world model” tech, aiming for FDA-certifiable systems. But revolutionizing robotics, transportation, and more? That’s a decades-long research marathon, not a sprint. LeCun is effectively building a moonshot factory. The real question is whether investors pouring in €500 million will have the patience for that timeline, or if they’ll demand quicker, commercial pivots. LeCun’s announcement post emphasized continuing the research program he began at FAIR and NYU. This suggests he’s guarding his independence fiercely. He’s not just building a company; he’s trying to build a new AI paradigm. And that might be the hardest thing to put a valuation on.
