Yann LeCun's AMI Labs Secures $1 Billion for World Model Development

AMI Labs, the new AI venture co-founded by Turing Award recipient Yann LeCun following his departure from Meta, has secured $1.03 billion in funding at a pre-money valuation of $3.5 billion. The company is developing world models—AI systems that learn from real-world interactions rather than solely from language data.

This specialized field currently has fewer participants than generative AI, though that may soon change. "I anticipate 'world models' becoming the next major industry buzzword," AMI Labs CEO Alexandre LeBrun shared with TechCrunch. "Within six months, nearly every AI company will likely reposition themselves as world model developers to attract investment."

LeBrun delivered this observation with a smile, confident that AMI Labs maintains a fundamental distinction: its core mission focuses on genuine world understanding. This approach has promising applications in healthcare, demonstrated by its inaugural partnership with digital health startup Nabla, where LeBrun now serves as chairman.

During his tenure as Nabla's CEO, LeBrun reached similar conclusions to LeCun regarding large language models' limitations—particularly concerning hallucinations that could pose serious risks in medical contexts. However, he acknowledges the considerable timeline required to develop a practical alternative using JEPA (Joint Embedding Predictive Architecture), the framework LeCun introduced in 2022.

"AMI Labs represents an exceptionally ambitious undertaking since it begins with foundational research," LeBrun explained. "This isn't a conventional applied AI startup that can launch a product in three months, generate revenue in six, and achieve $10 million in annual recurring revenue within a year." Indeed, transitioning world models from theoretical concepts to commercial implementations might require multiple years.

Despite extended development horizons, world model specialists are attracting substantial investments. SpAItial secured a $13 million seed round—notably large for European startups—while Fei-Fei Li's World Labs obtained a remarkable $1 billion just last month. AMI Labs now joins this elite group with even greater funding than initially speculated.

Though the French AI lab was reportedly seeking approximately €500 million last December, it ultimately raised about €890 million, likely bolstered by its exceptional team. Beyond LeCun's involvement as chairman and LeBrun's entrepreneurial background, the organization features Meta's European VP Laurent Solly as COO, alongside distinguished researchers Saining Xie as Chief Science Officer, Pascale Fung as Chief Research & Innovation Officer, and Michael Rabbat as VP of World Models.

LeBrun noted that strong investor interest enabled the startup to selectively choose backers aligned with both expectations and expertise. The funding round was co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital and Bezos Expeditions, with participation from multiple funds, industry-connected investors, and prominent individuals including Tim and Rosemary Berners-Lee, Jim Breyer, Mark Cuban, Mark Leslie, Xavier Niel and Eric Schmidt.

Beyond strategic value, this capital provides substantial runway for AMI Labs' primary expenditures: computational resources and talent acquisition. LeBrun emphasized quality-focused team building across four key locations: Paris (headquarters), New York (where LeCun teaches at NYU), Montreal (Rabbat's base), and Singapore (proximity to Asian talent and future clients).

While AMI Labs doesn't anticipate immediate revenue generation, it plans early engagement with potential customers. "We're creating world models designed to comprehend reality, which can't be achieved in isolation," LeBrun stated. "Eventually, we must test our models with real-world data and evaluations in authentic scenarios."

When ready for deployment, AMI Labs will collaborate with partners—Nabla being the first confirmed partner to access early models, though certainly not the last. "This strategic approach likely explains the strong interest from industrial players and potential partners in our investment round," LeBrun observed.

Alongside lead investors and angel supporters, AMI Labs counts NVIDIA, Samsung, Sea, Temasek and Toyota Ventures as backers, plus French entities Association Familiale Mulliez, Groupe Industriel Marcel Dassault and Publicis Groupe. Additional participants include Aglaé Lab, Alpha Intelligence Capital, Artémis, Bpifrance Digital Venture, New Legacy Ventures, SBVA and ZEBOX Ventures.

Commercial applications may require patience, but consistent with LeCun's principles, AMI Labs will publish research findings throughout its development process.

"We'll also open-source significant portions of our code," said LeBrun, who previously worked at Meta's FAIR research lab. Though open research practices are becoming "increasingly rare," the founders maintain their commitment. "We believe progress accelerates through openness, and building a collaborative research ecosystem around our work serves our long-term interests."