Who’s behind AMI Labs, Yann LeCun’s ‘world model’ startup

Yann LeCun’s new venture, AMI Labs, has attracted significant attention since the renowned AI scientist departed Meta to help establish the startup. This week, AMI Labs formally confirmed what it is building, while several vital details that had long circulated quietly are now out in the open.

Through its newly launched website, the company outlined its ambition to develop so-called “world models,” aiming to build intelligent systems capable of understanding the real world. The company’s name had already suggested the emphasis on world models — AMI stands for Advanced Machine Intelligence — but the announcement places the startup squarely among the most closely watched AI research ventures.

Efforts to create foundational models that connect artificial intelligence with physical reality have become one of the most compelling directions in the field. The area has drawn leading researchers and well-funded investors, even as many of the startups involved remain light on finished products. One prominent rival is World Labs, founded by AI pioneer Fei-Fei Li. World Labs reached unicorn status shortly after emerging from stealth and, following the launch of its first product, Marble — which generates physically accurate 3D environments — is reportedly in discussions to raise additional funding at a $5 billion valuation.

Given that backdrop, there is little doubt that investors would be similarly eager to back LeCun. That expectation has fueled reports suggesting AMI Labs may be seeking funding at a valuation of around $3.5 billion. According to Bloomberg, venture firms in talks with the company include Cathay Innovation, Greycroft, and Hiro Capital, all of which have LeCun as an advisor. Other investors reportedly considering participation include 20VC, Bpifrance, Daphni, and HV Capital.

One detail investors may want to note is LeCun’s role within the company. He has been clear that he is serving as executive chairman, not chief executive. The CEO position is held by Alex LeBrun, who previously co-founded and led Nabla, a healthcare-focused AI startup with operations in Paris and New York.

LeBrun’s move from Nabla to AMI Labs stems from a partnership announced last December. Nabla, which develops AI assistants for clinical care and counts LeCun as an advisor, agreed to support LeBrun’s transition from CEO to chief AI scientist and chairman in exchange for privileged access to AMI Labs’ world model technology. That arrangement cleared the way for LeBrun to take on leadership of the new venture.

At AMI Labs, LeBrun will be working alongside familiar collaborators. After Facebook acquired his earlier startup, Wit.ai, LeBrun worked under LeCun at Meta’s AI research group, FAIR. Reports also suggest that Laurent Solly, who stepped down as Meta’s vice president for Europe last December, will join the effort.

The overlap between AMI Labs and Meta may extend further. LeCun told the MIT Technology Review that his former employer could become AMI’s first customer. At the same time, he has openly criticised several strategic decisions made at Meta under its former CEO, Mark Zuckerberg. More broadly, the Review has characterised AMI Labs as a contrarian response to the current dominance of large language models.

LeCun has frequently highlighted shortcomings of LLMs, including their tendency to hallucinate — a risk that becomes especially serious in areas such as medicine. LeBrun echoed those concerns in comments to Forbes, noting that the opportunity to apply world models to healthcare was a significant reason he accepted the CEO role. Beyond healthcare, the company plans to focus on other high-stakes applications.

In its mission statement, AMI Labs said it aims to advance AI research and build systems for use cases where reliability, controllability, and safety are critical, including industrial process control, automation, wearable technology, robotics, and healthcare. “We share one belief: real intelligence does not start in language. It starts in the world,” the company wrote.

Unlike generative approaches, which the team argues are poorly suited for unpredictable inputs such as sensor data, AMI Labs says its systems will be designed to understand the physical world, retain persistent memory, reason and plan, and remain controllable and safe. The company intends to license its technology to industry partners while also contributing to the broader AI ecosystem through open research publications and open-source work.

LeCun plans to retain his academic role at New York University, where he teaches one class annually and supervises PhD candidates and postdoctoral researchers. He will remain based in New York but told the MIT Technology Review that AMI Labs will be a global company headquartered in Paris.

That decision was welcomed by Emmanuel Macron, who said he was proud that the Turing Award-winning researcher chose Paris and pledged support for the company’s success from France. In addition to Paris, AMI Labs plans to maintain offices in Montreal, New York, and Singapore. By selecting Paris as its base, the startup further strengthens the city’s position as an AI hub, joining companies such as H and Mistral AI, as well as international research labs including FAIR. As LeCun has noted, it is perhaps fitting that AMI is pronounced “a-mee,” echoing the French word ami, meaning “friend.”