The Architects of AMI Labs: How Yann LeCun, Alex LeBrun, and Nabla Are Reshaping AI's Future

In February 2025, during an AI Action Summit held in Paris, a significant moment unfolded when French President Emmanuel Macron announced a 109 billion euro investment initiative aimed at positioning France as a global AI research hub. Against this backdrop, Yann LeCun—Meta’s former chief AI scientist—launched AMI Labs, a venture that has rapidly emerged as one of the most intriguing players in the AI innovation landscape. The company’s leadership structure reflects a carefully orchestrated transition in the AI community, particularly through the strategic involvement of Alex LeBrun, who recently transitioned from his role at Nabla, a health-focused artificial intelligence startup. This combination of visionary leadership and practical healthcare expertise is already reshaping expectations around what next-generation AI systems should accomplish.

From Nabla to AMI: The Strategic Leadership Transition

The founding team of AMI Labs extends beyond LeCun himself. Alex LeBrun, who serves as the company’s CEO, brings a unique trajectory to the role. Previously, LeBrun co-founded and led Nabla, an AI startup specializing in medical applications with operations in both Paris and New York. His appointment to lead AMI Labs was facilitated through a formal strategic partnership between Nabla and AMI Labs—an arrangement that gives Nabla privileged access to AMI’s emerging technologies while enabling LeBrun’s transition into his new position.

LeBrun’s background at Nabla is particularly revealing about his expertise in addressing real-world AI challenges. After Facebook acquired his earlier venture, Wit.ai, LeBrun worked directly under LeCun at Meta’s AI Research division, FAIR. This employment history demonstrates a consistent thread: applying cutting-edge AI research to practical problems. At Nabla, he gained deep insights into the limitations of current AI systems when deployed in sensitive domains like healthcare. These experiences clearly informed his decision to move to AMI Labs.

LeBrun is not operating alone in this new venture. He is joined by other colleagues from his earlier collaborations, including Laurent Solly, who recently departed from his position as Meta’s European vice president to join the AMI Labs team. This concentration of talent signals that LeCun is assembling a team specifically capable of executing on the ambitious mandate of building artificial intelligence systems that interact meaningfully with the physical world—a capability that Nabla’s healthcare work had already begun to demonstrate was necessary.

World Models vs. Language Models: Why AMI Labs Takes a Different Path

The philosophical foundation of AMI Labs stems from a fundamental critique of current AI development trajectories. While the industry has largely pursued large language models (LLMs) as the pathway to advanced intelligence, LeCun and his team have articulated a contrarian position. As the company’s mission statement explains: “We believe that genuine intelligence originates from interaction with the world, not just from language.”

This distinction carries profound implications. LLMs, despite their impressive capabilities, suffer from a critical vulnerability: they frequently generate false information—a particularly dangerous limitation in fields like healthcare. LeBrun, drawing on his experience at Nabla, has been vocal about this challenge. In discussions with the media, he noted that the opportunity to deploy AMI’s world models in medical applications represented a major motivator for his joining the company.

World models, by contrast, are designed as AI systems capable of understanding and responding to real-world complexity. Rather than processing text or language alone, they integrate sensor data, temporal reasoning, and interactive learning. This approach addresses a fundamental gap in how current AI systems operate. Unlike generative models that are, in LeCun’s assessment, “ill-equipped to handle unpredictable data such as sensor inputs,” world models are engineered specifically to operate in dynamic, uncertain environments.

The competitive landscape underscores the potential of this approach. World Labs, founded by renowned AI researcher Fei-Fei Li, achieved unicorn status rapidly after unveiling its first product, Marble—a system capable of generating realistic 3D environments. World Labs is currently in funding discussions at a reported valuation of $5 billion. AMI Labs, benefiting from LeCun’s reputation and the strategic involvement of seasoned executives like LeBrun, is reportedly seeking investment at a $3.5 billion valuation, with firms such as Cathay Innovation, Greycroft, Hiro Capital, 20VC, Bpifrance, Daphni, and HV Capital engaged in discussions.

Building Practical AI: Healthcare, Robotics, and Beyond

AMI Labs’ technology roadmap explicitly identifies specific sectors where world models will deliver value. According to the company’s official positioning, the technology will be deployed across industrial automation, robotics, healthcare, and wearable technology—domains where safety, reliability, and continuous learning are paramount. This sectoral focus reflects lessons learned from Nabla’s work in healthcare as well as broader industry needs.

The healthcare application deserves particular attention. The limitations of LLMs in medical contexts—where false information can have life-or-death consequences—have become increasingly apparent. World models offer a potential solution by grounding AI decision-making in verifiable interactions with real-world medical data, patient histories, and clinical outcomes. This differs fundamentally from language-only systems, which lack mechanisms for validating their outputs against empirical medical reality.

Beyond healthcare, the robotics and industrial automation sectors present equally compelling use cases. In these domains, AI systems must perceive their environment through sensors, plan actions over time, and adjust responses based on real-time feedback. These are precisely the capabilities that world models are designed to provide—features that language models simply were not built to support.

How Nabla’s Healthcare Expertise Shapes AMI Labs’ Vision

The involvement of Alex LeBrun and the partnership between Nabla and AMI Labs is not merely symbolic. By establishing a strategic partnership with Nabla, AMI Labs gains access to institutional knowledge about deploying AI in regulated, high-stakes environments. Healthcare represents one of the most complex regulatory landscapes for AI deployment; Nabla’s experience navigating this terrain directly informs AMI Labs’ go-to-market strategy.

The technology licensing model that AMI Labs plans to pursue—offering its world models to industry partners for practical deployment—was undoubtedly shaped by LeBrun’s understanding of how healthcare systems and other enterprises actually adopt AI solutions. Nabla’s experience scaling AI applications in medical settings provided a blueprint for addressing the gap between research breakthroughs and real-world implementation.

Additionally, Meta’s anticipated role as AMI Labs’ first client adds another dimension to this strategic arrangement. While LeCun has publicly offered criticism of certain Meta business decisions, the potential for Meta to adopt AMI’s world models for specific applications represents a significant early-stage partnership. This hybrid model—licensing technology to industry partners while maintaining a research-focused, open-source commitment to the broader AI community—reflects the balanced approach that both LeCun and LeBrun advocate.

The Paris Hub: Why France Became AMI Labs’ Home

Although LeCun maintains his base in New York and his academic position at NYU, where he continues teaching and mentoring graduate students, he has made a deliberate choice to anchor AMI Labs in Paris. This decision reflects both the availability of AI talent in France and the government’s strategic commitment to establishing the nation as an AI innovation center. President Macron’s 109 billion euro initiative, announced alongside the company’s launch, provided institutional backing for this choice.

AMI Labs’ presence in Paris strengthens the city’s emerging role as a hub for AI research and development. The company joins other significant players including Mistral AI and Meta’s FAIR research division in establishing Paris as a global center of AI gravity. The decision also carries symbolic weight—the company’s name, AMI, is pronounced identically to the French word for “friend,” a nod to its commitment to the French ecosystem.

The company’s global expansion strategy extends beyond Paris. In addition to its headquarters in the French capital, AMI Labs will establish offices in Montreal, New York, and Singapore. This geographic distribution reflects both talent concentration patterns in the AI industry and strategic access to key markets and research communities. New York connects AMI to the broader U.S. AI ecosystem and maintains proximity to LeCun’s continued academic work. Montreal represents a recognized hub for deep learning research. Singapore provides a gateway to Asian markets and Asian talent pools.

This multi-hub approach suggests that AMI Labs intends to operate at global scale from inception, avoiding the geographic limitations that have constrained some earlier AI ventures. Combined with the strategic partnership involving Nabla and the leadership team’s experience at Meta and other leading firms, AMI Labs appears positioned to move rapidly from research to real-world deployment in multiple sectors simultaneously.