Yann LeCun's new AI venture, AMI Labs, has just raised $1.03 billion — and the technology it's building could be the most consequential shift in artificial intelligence since large language models exploded into the mainstream. If you've been watching the AI funding landscape, this is one story you don't want to scroll past.
 |
| Credit: Ruhani Kaur/Bloomberg via Getty Images / Getty Images |
What Is AMI Labs and Why Does It Matter Right Now?
AMI Labs is a cutting-edge AI research company co-founded by Turing Award winner Yann LeCun — one of the most respected names in deep learning — after his departure from Meta. The company is focused on building world models: a fundamentally different category of AI that learns from reality rather than from language alone.
Unlike the generative AI tools that have dominated headlines over the past few years, world models are designed to understand and reason about the physical world. Think of it as the difference between an AI that can describe a bicycle and one that understands how riding one actually works. That distinction might sound subtle, but the implications are enormous — particularly for industries like healthcare, robotics, and scientific research.
The company closed its latest funding round at a $3.5 billion pre-money valuation, cementing its status as one of the most serious contenders in next-generation AI.
From Meta to a $1 Billion Startup: LeCun's Bold New Chapter
Yann LeCun spent years at Meta pushing the boundaries of AI research. His departure to co-found AMI Labs wasn't a quiet exit — it was a statement. LeCun has been one of the most vocal critics of the limitations baked into today's dominant AI systems, particularly large language models (LLMs).
At AMI Labs, he serves as chairman, lending the company both credibility and a clear intellectual direction. The technical foundation of AMI's approach is JEPA — Joint Embedding Predictive Architecture — a framework LeCun first proposed in 2022. Rather than predicting text token by token, JEPA models learn by predicting abstract representations of the world, making them potentially more robust and less prone to the kinds of errors that plague today's AI.
This isn't a pivot for LeCun. It's the culmination of a long-held belief that the AI field has been climbing the wrong mountain.
The CEO Who Saw the Hallucination Problem Up Close
Alexandre LeBrun, AMI Labs' CEO, brings a perspective shaped not just by research but by real-world consequences. Before joining AMI Labs, he served as CEO of Nabla, a digital health startup where he witnessed firsthand how dangerous AI hallucinations can be in clinical settings.
When an AI assistant gives a doctor inaccurate information, the stakes aren't just reputational — they can be life-threatening. LeBrun reached the same conclusion as LeCun: the current generation of LLMs, for all their impressive capabilities, has a fundamental flaw that more data and more compute alone won't fix.
Now, LeBrun chairs Nabla while steering AMI Labs, and it's no coincidence that Nabla is AMI's first announced partner. Healthcare is the proving ground — the domain where getting AI right matters most and where the world model approach has the clearest immediate application.
"World Models" Is About to Become the Next Big Buzzword
If you think "world models" sounds like niche academic territory, prepare for a shift. LeBrun himself predicted with a knowing smile that within six months, nearly every AI company will start calling itself a world model company in order to attract funding.
It's a familiar pattern. We've seen it with "deep learning," then "generative AI," then "foundation models." The terminology gets colonized by marketing departments faster than the science matures. But LeBrun's point isn't cynical — it's a recognition that AMI Labs is out ahead of a wave that's about to crest.
The company isn't rushing to capitalize on hype, though. LeBrun has been candid that AMI Labs is not a typical applied AI startup that can ship a product in three months and hit $10 million in annual recurring revenue within a year. This is fundamental research with a long time horizon — measured in years, not quarters. That kind of patience is rare in the current funding environment, which makes the $1.03 billion raise all the more striking.
The Dream Team Behind AMI Labs
Part of what makes AMI Labs credible — and what likely pushed the funding well beyond its initially rumored €500 million target — is the caliber of its team. The company reportedly sought around €500 million in late 2024 but ultimately closed nearly €890 million, a testament to investor confidence in the people involved.
Beyond LeCun as chairman and LeBrun as CEO, the roster is exceptional. Laurent Solly, formerly Meta's VP for Europe, serves as Chief Operating Officer. Saining Xie, a leading AI researcher, holds the Chief Science Officer role. Pascale Fung, an internationally recognized expert in multilingual AI and natural language processing, serves as Chief Research and Innovation Officer. Michael Rabbat, another distinguished researcher, rounds out the executive team as VP of Research.
This isn't a group that assembled to chase a funding cycle. These are researchers and operators who have spent careers building toward exactly this kind of ambitious, long-horizon project.
A New Race Is Forming Around World Models
AMI Labs isn't operating in a vacuum. The world model space is quietly becoming one of the most competitive arenas in AI, even if it hasn't captured as many mainstream headlines as generative AI has.
Other well-funded players are emerging fast. A spatial intelligence startup recently closed an unusually large seed round for a European AI company. Fei-Fei Li, the pioneering computer vision researcher, secured $1 billion for her own world-model-focused venture just last month. The pattern is clear: serious money is following serious researchers into this new frontier.
What sets AMI Labs apart, according to its leadership, is the depth of its commitment to foundational research over quick commercialization. The company isn't trying to build a better chatbot. It's trying to build AI that genuinely understands cause and effect, physical space, and the logic of the real world. That's a harder problem — and potentially a much more valuable one.
Why This Matters Beyond the Funding Headline
It's easy to become numb to billion-dollar funding rounds in AI. They arrive so frequently now that they risk blurring together. But the AMI Labs raise deserves more than a passing glance, for a few reasons.
First, the scientific credibility here is unusually high. LeCun is not a founder capitalizing on a hot market — he is one of the architects of modern deep learning, with decades of peer-reviewed work to back his convictions. Second, the problem being addressed is genuine and urgent. AI hallucinations aren't a quirk — they're a structural limitation, and the healthcare space has already shown us how costly they can be. Third, the timing is significant. If world models are truly the next paradigm, the companies that get the foundational research right now will hold enormous advantages as applications emerge.
AMI Labs is betting that understanding the real world isn't just a technical upgrade — it's the entire point. And with $1.03 billion now behind that bet, the world is about to find out if they're right.
