AMI Labs: Who’s Behind Yann LeCun’s World Model Startup
What is AMI Labs building? Who’s leading Yann LeCun’s new AI venture? And why are investors lining up for a piece of it? The answers are now coming into focus. After months of speculation, AMI Labs—founded by Turing Award winner and Meta veteran Yann LeCun—has officially revealed its mission: to develop “world models” that enable machines to understand and interact with the physical world. With LeCun as executive chairman and former Nabla CEO Alex LeBrun at the helm, the startup is positioning itself at the forefront of next-generation artificial intelligence.
What Are “World Models”—and Why Do They Matter?
At the heart of AMI Labs’ ambition lies a concept rapidly gaining traction in AI research: world models. Unlike today’s large language models that excel at pattern recognition but lack true understanding, world models aim to simulate how the real world works—predicting outcomes, reasoning about cause and effect, and grounding AI behavior in physical reality.
Think of it like this: if current AI systems are brilliant students who’ve memorized every textbook but never stepped outside the classroom, world models would be the ones who’ve actually lived through experiences, learned from trial and error, and can navigate complex environments intuitively.
This shift could be transformative. From robotics and autonomous vehicles to personalized healthcare and scientific discovery, systems equipped with world models could act more safely, efficiently, and autonomously. That’s why top researchers—and deep-pocketed investors—are betting big on this approach.
Yann LeCun Steps Back From the CEO Role
While Yann LeCun’s name carries immense weight in the AI community—he’s often called the “godfather of deep learning”—he’s chosen not to serve as AMI Labs’ CEO. Instead, he’s taken on the role of executive chairman, focusing on long-term vision and technical direction while leaving day-to-day operations to someone else.
That someone is Alex LeBrun, a seasoned entrepreneur with a track record in applied AI. LeBrun previously co-founded and led Nabla, a Paris- and New York-based startup building AI assistants for clinicians. Under his leadership, Nabla developed tools that help doctors manage patient records, schedule appointments, and even draft clinical notes—all while maintaining strict privacy and regulatory compliance.
LeBrun’s move to AMI Labs wasn’t abrupt. In December 2025, Nabla announced a strategic partnership with the newly formed startup, granting it “privileged access” to AMI’s emerging world model technology. In return, Nabla’s board supported LeBrun’s transition from CEO to chief AI scientist and chairman, paving the way for his full-time role at AMI.
The Strategic Link Between Nabla and AMI Labs
The connection between Nabla and AMI Labs isn’t just personal—it’s deeply strategic. Healthcare is one of the most promising early applications for world models. Imagine an AI that doesn’t just transcribe a doctor’s notes but understands the progression of a disease, anticipates complications, or simulates treatment outcomes based on real-world physiology.
By anchoring AMI’s foundational research in a high-stakes, real-world domain like medicine, LeCun and LeBrun ensure their models are tested against tangible challenges—not just theoretical benchmarks. This alignment with practical use cases strengthens AMI’s credibility and accelerates its path from lab to market.
Moreover, Nabla’s existing infrastructure, regulatory expertise, and clinician relationships offer AMI a rare testing ground. Few AI startups have immediate access to such a controlled yet complex environment where safety, accuracy, and interpretability are non-negotiable.
Investor Frenzy Builds Around World Model Startups
The buzz around AMI Labs isn’t just academic. Venture capital firms are already circling. According to Bloomberg, AMI is in talks to raise funding at a staggering $3.5 billion valuation—a figure that underscores both LeCun’s reputation and the perceived potential of world models.
Reported interested parties include Cathay Innovation, Greycroft, Hiro Capital (where LeCun serves as an advisor), 20VC, Bpifrance, Daphni, and HV Capital. These aren’t just check-writing institutions; many bring sector-specific expertise in deep tech, European innovation ecosystems, and AI commercialization.
The frenzy mirrors what’s happening across the world model landscape. Rival startup World Labs, founded by Stanford’s Fei-Fei Li, recently became a unicorn after launching Marble—a system that generates physically plausible 3D environments. It’s now reportedly seeking fresh capital at a $5 billion valuation.
What’s driving this surge? Investors see world models as the missing link between narrow AI and more general, adaptable intelligence. While generative AI has dazzled with text and images, it struggles with consistency, physics, and long-horizon planning. World models promise to fix that—and whoever cracks the code first could redefine entire industries.
Why AMI Labs Stands Out in a Crowded Field
In a field attracting Nobel-caliber scientists and billions in capital, what makes AMI Labs different? Three factors stand out.
First, LeCun’s foundational work on self-supervised learning and energy-based models gives AMI a unique technical edge. For years, he’s argued that predictive world models—not just scaling data—are the key to human-like AI. Now, he’s finally building them.
Second, the leadership pairing of LeCun and LeBrun blends visionary research with product discipline. Too many AI labs produce breakthroughs that never leave the lab. With LeBrun’s experience shipping regulated, user-facing AI products, AMI is structured to bridge that gap from day one.
Third, its deliberate go-to-market strategy—starting with healthcare via Nabla—grounds its ambitions in reality. Rather than chasing flashy demos, AMI is embedding its models in workflows where failure isn’t an option, forcing rigor and reliability.
Challenges and Expectations
Of course, the path won’t be easy. World models remain largely experimental. Training them requires massive compute, novel architectures, and vast amounts of real-world interaction data—none of which are trivial to acquire or process. Regulatory scrutiny, especially in healthcare and robotics, will also be intense.
And then there’s competition. Beyond World Labs, companies like Google DeepMind, OpenAI, and Tesla are all exploring similar concepts under different names—“simulation engines,” “physics-aware models,” or “embodied AI.” AMI may have intellectual firepower, but it lacks the infrastructure of Big Tech.
Still, independence could be its advantage. Free from quarterly earnings pressure or platform agendas, AMI can pursue long-term bets. And with LeCun openly critical of current LLM limitations, the startup is positioned as a purist alternative in an increasingly commercialized AI landscape.
A New Chapter in the Quest for Real Intelligence
AMI Labs represents more than just another AI startup. It’s a declaration of intent—that the next leap in artificial intelligence won’t come from bigger datasets or flashier chatbots, but from systems that truly understand how the world works.
With Yann LeCun providing the scientific compass and Alex LeBrun steering execution, AMI Labs is assembling the rare mix of theory, talent, and real-world validation needed to turn world models from research curiosity into transformative technology.
As the startup emerges from stealth, the AI world is watching closely. Because if AMI succeeds, it won’t just build smarter machines—it could redefine what “intelligence” means in the age of artificial minds.
