PARIS, FRANCE — June 9, 2026: In a landmark deal reshaping the artificial intelligence landscape, AMI Labs, the new AI research venture co-founded by Turing Award winner Yann LeCun, has secured a staggering $1.03 billion in funding. Announced today, the investment propels the Paris-based startup to a $3.5 billion pre-money valuation and marks one of the largest single rounds for a European AI company. The capital infusion is dedicated to a singular, ambitious goal: building advanced world models—AI systems designed to learn from and understand the physical world, moving beyond the linguistic confines of today’s large language models (LLMs). This funding signals a pivotal moment where investor confidence and scientific ambition converge on what many experts believe is the next critical frontier in artificial intelligence.
AMI Labs and the $1.03 Billion Bet on World Models
The funding round, co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions, far exceeded initial expectations. Reports from December 2025 indicated AMI Labs was seeking approximately €500 million. The final close at roughly €890 million (approximately $1.03 billion) underscores intense investor appetite for foundational AI research with long-term horizons. According to AMI Labs CEO Alexandre LeBrun, the quality of the startup’s founding team allowed it to be highly selective. “High interest gave us a chance to have our pick of investors, both in terms of expectation alignment and background,” LeBrun stated in an exclusive interview. The investor syndicate is notably diverse, including strategic corporate ventures like NVIDIA, Samsung, and Toyota Ventures, alongside prominent angels such as Tim Berners-Lee, Mark Cuban, and Eric Schmidt.
This capital provides a substantial runway to address the startup’s two primary cost centers: immense computational power and elite research talent. LeBrun outlined a deliberate, quality-first hiring strategy focused on four global hubs: Paris (headquarters), New York (where LeCun teaches at NYU), Montreal, and Singapore. The latter location is strategic for both recruiting Asian AI talent and positioning the company near future clients in the region. Despite the colossal funding, LeBrun is candid about the timeline. “AMI Labs is a very ambitious project because it starts with fundamental research,” he explained. “It’s not your typical applied AI startup that can release a product in three months.” The path from theoretical research to commercial applications for world models could span years, a timeline these investors have explicitly endorsed.
Defining the Next AI Buzzword: What Are World Models?
The core mission of AMI Labs is to develop world models, a category of AI distinct from the generative text and image models that dominate current discourse. While LLMs learn statistical patterns from vast text corpora, world models aim to build an internal understanding of how the real world works—its physics, cause-and-effect relationships, and spatial reasoning. This approach is based on Yann LeCun’s Joint Embedding Predictive Architecture (JEPA), a framework he proposed in 2022. JEPA aims to allow AI to learn predictive models of the world by observing it, much like humans and animals do, reducing reliance on purely language-based training which is prone to “hallucinations” or factual confabulations.
CEO Alexandre LeBrun, who also serves as Chairman of digital health startup Nabla (AMI Labs’ first disclosed partner), reached the same conclusion as LeCun regarding LLM limitations, particularly in high-stakes fields like healthcare. “Hallucinations could have life-threatening repercussions,” LeBrun noted, highlighting the urgent need for more reliable, reality-grounded AI. He predicts a coming wave of hype, telling TechCrunch, “‘World models’ will be the next buzzword. In six months, every company will call itself a world model to raise funding.” However, he asserts AMI Labs is fundamentally different due to its deep grounding in LeCun’s research and its commitment to open, fundamental science. The startup plans to publish papers and open-source code throughout its development, a philosophy increasingly rare in today’s secretive, commercially-driven AI race.
The Competitive Landscape for World Model Development
AMI Labs is not operating in a vacuum. The world model thesis is attracting significant capital and high-profile researchers globally, indicating a broader strategic shift in AI investment. Last month, Fei-Fei Li’s World Labs secured a massive $1 billion funding round. In Europe, SpAItial raised a notably large $13 million seed round. These ventures, while differing in specific approach, share a common belief that the next breakthrough in AI requires moving beyond pattern-matching in text to building robust, internal models of reality. The table below compares key players in this emerging space.
| Company | Key Leadership | Recent Funding | Primary Focus |
|---|---|---|---|
| AMI Labs | Yann LeCun (Chair), Alexandre LeBrun (CEO) | $1.03 Billion (June 2026) | General world models based on JEPA architecture |
| World Labs | Fei-Fei Li | $1 Billion (May 2026) | AI that interacts with and understands the physical world |
| SpAItial | Not Disclosed | $13 Million Seed (2025) | Spatial intelligence and 3D world understanding |
From Research to Reality: The Long Road to Commercial Application
The immediate plan for AMI Labs is not revenue generation but rigorous research and early partnership development. LeBrun emphasized the necessity of real-world engagement. “We are developing world models that seek to understand the world, and you can’t do that locked up in a lab. At some point, we need to put the model in a real-world situation with real data and real evaluations,” he said. This philosophy explains the strategic inclusion of potential partners like Nabla, Samsung, and Toyota Ventures in the investment round. These entities provide not just capital, but also critical domains for testing and eventual deployment.
The partnership with Nabla is particularly illustrative. As a digital health company, Nabla faces the acute challenge of deploying AI in clinical settings where accuracy is non-negotiable. Access to AMI Labs’ early world models could pave the way for AI assistants that reason about patient symptoms, medical literature, and biological processes in an integrated, reliable manner. LeBrun confirmed Nabla is the “first disclosed partner expecting to access these early models, but definitely not the last.” This model of co-development with industry partners mitigates the risk of building technology in isolation and ensures research is guided by tangible, high-value problems.
Building an All-Star Team for a Decadal Challenge
Beyond LeCun and LeBrun, AMI Labs has assembled a formidable leadership team that blends academic excellence with operational scale. Laurent Solly, Meta’s former Vice President for Europe, joins as Chief Operating Officer, bringing experience managing large, complex organizations. The research leadership is equally impressive, featuring Saining Xie as Chief Science Officer, a renowned computer vision researcher; Pascale Fung as Chief Research & Innovation Officer, a leading expert in NLP and AI ethics; and Michael Rabbat as VP of World Models, an authority on efficient machine learning. This concentration of talent is a key asset that likely contributed to the funding round’s oversubscription and sets a high bar for execution in a fiercely competitive talent market.
Conclusion
The $1.03 billion investment in Yann LeCun’s AMI Labs represents more than just a large funding round; it is a strategic bet on a specific future for artificial intelligence. It validates the scientific direction of world models and demonstrates that leading investors are willing to fund long-term, fundamental research with uncertain but potentially transformative outcomes. While the commercial applications in healthcare, robotics, and beyond may be years away, the race to build AI that genuinely understands the world has now been catapulted into the mainstream with unprecedented financial backing. The success of AMI Labs will hinge not just on its brilliant team and ample resources, but on its ability to navigate the long, challenging path from theoretical architecture to reliable, real-world systems. For the broader AI ecosystem, this funding serves as a clear signal: the next chapter of AI is being written, and it will be grounded in the physics of reality, not just the patterns of language.
Frequently Asked Questions
Q1: What exactly are “world models” in AI?
World models are a class of artificial intelligence systems designed to learn an internal representation of how the real world operates. Unlike large language models (LLMs) that learn from text, world models aim to understand physical concepts, cause-and-effect relationships, and spatial reasoning by observing data from the environment, similar to how humans learn.
Q2: Why did AMI Labs raise so much money ($1.03B) for research?
Building foundational world models requires immense computational resources (compute) and the ability to attract and retain top-tier AI research talent in a competitive global market. The funding provides a multi-year runway to pursue this long-term, capital-intensive research without the immediate pressure to generate revenue.
Q3: How is AMI Labs’ approach different from companies like OpenAI or Anthropic?
While companies like OpenAI focus heavily on scaling language models, AMI Labs is prioritizing a different architecture (JEPA) aimed at learning from multimodal data (video, sensor data) to build a grounded understanding of reality. Its philosophy also emphasizes open research and publishing, contrasting with the more closed approaches of some other leading AI labs.
Q4: What is the first practical application for AMI Labs’ technology?
The first disclosed partner is Nabla, a digital health startup. The initial goal is to develop world models that can assist in healthcare, where understanding complex, real-world medical contexts reliably is critical and where current LLMs’ tendency to hallucinate poses serious risks.
Q5: Who are the main investors behind this massive funding round?
The round was co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions. It also included participation from corporate ventures like NVIDIA, Samsung, and Toyota Ventures, as well as individual investors like Tim Berners-Lee, Mark Cuban, and Eric Schmidt.
Q6: When can we expect to see products or services based on this technology?
CEO Alexandre LeBrun has been clear that this is fundamental research with a long horizon. It could take several years before world models move from theory to viable commercial applications. The focus for now is on research, partnership development, and iterative testing with real-world data.
