Who is Alexandre LeBrun and what is AMI Labs?

AMI Labs (Advanced Machine Intelligence Labs) is a Paris-based AI startup founded in December 2025 that quickly became one of the most talked-about projects in the global AI industry. Behind its creation is Yann LeCun — Turing Award winner, long-time head of AI research at Meta, and one of the "fathers" of modern machine learning.

The CEO is Alexandre LeBrun, who previously co-founded and led Nabla — a healthcare AI startup with offices in Paris and New York. LeBrun is therefore someone with a very concrete view of where and how AI should actually work. Not in abstract philosophical discussions about the future, but in hospitals, clinics, and industrial operations.

The AMI Labs team is exceptional. Alongside LeCun and LeBrun, it includes Saining Xie as Chief Scientist, Pascale Fung as Director of Research and Innovation, and Michael Rabbat as Vice President for World Models. The company has offices in four key locations: Paris (headquarters), New York (where LeCun teaches at NYU), Montreal, and Singapore.

A billion dollars for Europe: the largest seed round in history

In March 2026, AMI Labs closed a seed funding round of $1.03 billion — at a pre-money valuation of $3.5 billion. This is the largest seed round in the history of European venture capital. The company originally planned to raise just €500 million, but investor interest exceeded expectations.

Among the investors are names that speak for themselves: Bezos Expeditions, Jeff Bezos's fund, Cathay Innovation, Greycroft, HV Capital, Hiro Capital, and a number of angel investors including web founder Tim Berners-Lee, investor Jim Breyer, Mark Cuban, French billionaire Xavier Niel, and former Google CEO Eric Schmidt.

Such interest shows that part of the global technology establishment believes the path to more powerful AI does not necessarily lead through even larger language models — but through a completely different approach.

AGI, superintelligence — why does LeBrun reject these words?

Alexandre LeBrun's statement is straightforward: terms like "AGI" (artificial general intelligence) or "superintelligence" are, to him, imprecise marketing jargon, not scientific concepts. While OpenAI proclaims it is on the path to superintelligence, Anthropic talks about "generally intelligent systems," and thousands of startups slap the acronym AGI onto every second pitch deck — AMI Labs deliberately avoids using these terms.

LeBrun even predicted a future trend with cynical humor: "In six months, every company will claim it's building a world model to get funding." He is at least consistent — he acknowledges that even the term "world model" can become an empty buzzword if there is no concrete scientific approach behind it.

This philosophy is a direct reflection of LeCun's long-held conviction: large language models (LLMs) like GPT-4, Claude, or Gemini are impressive, but they have fundamental limitations. They do not understand the physical world, they cannot plan with causal understanding, and their "intelligence" is largely statistical interpolation of text.

What are world models and how do they differ from ChatGPT?

This is the key technical difference that explains why AMI Labs was founded. Conventional large language models (LLMs) have learned to predict which word best follows the preceding text. That is a powerful trick — but it's just a trick with language.

A world model is an AI system that instead learns how the physical world works: how things look, how they move, what happens after a specific action. It processes sensory data — images, video, sound, LiDAR — and builds abstract representations of reality.

AMI Labs builds on an approach called JEPA (Joint Embedding Predictive Architecture), which LeCun proposed in 2022. Instead of an AI model generating pixel by pixel or word by word, JEPA works in an abstract "embedding" space, where it learns to predict the structure of reality, not just its surface.

The practical impact? Such an AI could:

• Control robotic systems that can adapt to unexpected situations
• Plan in real time in industrial operations
• Help doctors with what goes beyond the capabilities of today's diagnostic tools
• Function in wearable devices where sensory data needs to be processed efficiently

First partner: healthcare

AMI Labs does not yet have a product for sale — and LeBrun says so openly, without apology. The company is in the research and development phase. The first partner will be Nabla, the healthcare startup that LeBrun previously led himself. This is no coincidence — healthcare is a field where reliability, safety, and the ability to work with real physical data, not just text, matter.

Planned deployment areas include industrial automation, robotics, wearable technologies, and healthcare. All of these are domains where today's LLMs hit their limits — where merely "generating text" is not enough, and where AI must truly understand context and physical reality.

Challenging the dominance of OpenAI and Google: a European perspective

AMI Labs is also significant from a geopolitical standpoint. It is one of the first serious attempts to build next-generation AI infrastructure in Europe, not in Silicon Valley. Headquarters in Paris, European leadership, an international team — all of this fits into the broader EU effort for technological sovereignty.

For the Czech and Slovak markets, AMI Labs does not yet have a direct product — the company is still building its foundations. But the direction it represents is crucial for Europe: if "world models" truly surpass LLMs in practical applications, it will matter where this technology is developed and who controls it.

The approach that AMI Labs advocates is a direct challenge to the dominance of American giants. OpenAI and Anthropic have bet everything on scaling LLMs — bigger models, more data, more compute. LeCun's thesis says this path has a ceiling. If he is right, AMI Labs could be one of the companies that reshapes the balance of power in the AI industry.

Science vs. marketing: a lesson for the entire industry

LeBrun's stance on terminology is refreshing at a time when words like "AGI," "superintelligence," or "transformative AI" fly through the air every day. It's not just about semantics — it's a question of responsibility. When a company claims it is building AGI, it creates expectations that may be unrealistic. When it says it is building a world model with concrete applications in industry and healthcare, that is a more grounded message.

"Real intelligence doesn't start with language. It starts in the world," says the AMI Labs motto. It's a bet on a different AI philosophy — and a billion dollars from top-tier investors suggests this philosophy is far from marginal.