By The TENS Magazine Editorial Staff
AMI Labs, a new artificial intelligence startup co-founded by Turing Award winner and former Meta chief AI scientist Yann LeCun, has successfully raised $1.03 billion in seed funding. The massive investment values the company, formally known as Advanced Machine Intelligence Labs, at a pre-money valuation of $3.5 billion. The funding round is designed to support the development of world models—a next-generation AI architecture that aims to understand the physical world and simulate real-world environments, moving beyond the capabilities of current text-based generative AI systems.
The seed round ranks among the largest early-stage investments in the history of the European technology sector. The financing was co-led by prominent venture capital firms including Cathay Innovation, Greycroft, Hiro Capital, and HV Capital. The round also attracted significant participation from global corporate and institutional investors, notably the United States chip manufacturer Nvidia, the Singaporean state investment firm Temasek, and Sea, the parent company of Shopee. Additional backing came from Bezos Expeditions (the family office of Amazon founder Jeff Bezos), French venture firms Daphni and Bpifrance, South Korean investor SBVA, and high-profile individual investors such as former Google CEO Eric Schmidt, internet pioneer Tim Berners-Lee, billionaire Mark Cuban, and venture capitalist Jim Breyer.
The core mission of AMI Labs is to pioneer the development of world models, an approach to artificial intelligence that diverges significantly from the Large Language Models (LLMs) that currently dominate the industry. While LLMs generate text by predicting the next word in a sequence based on vast datasets, world models are engineered to process continuous, high-dimensional sensor data—such as video, audio, and spatial information—to learn how the physical world operates. According to Yann LeCun, who serves as the executive chairman of the startup, true intelligence originates in the physical world rather than in language. The company utilizes a specialized framework known as the Joint Embedding Predictive Architecture (JEPA) to filter out background noise and focus on the underlying rules of cause and effect. This allows the system to reason, plan complex sequences of actions, and anticipate consequences under real-world constraints.
To execute this long-term scientific endeavor, AMI Labs has assembled a leadership team of seasoned AI researchers and technology executives. Alexandre LeBrun, the former CEO of the medical AI startup Nabla, has been appointed as the Chief Executive Officer of AMI Labs. The founding team also includes Chief Science Officer Saining Xie, a former research scientist at Google DeepMind; Mike Rabbat, the Vice President of World Models and a former research science director at Meta; and Pascale Fung, the Chief Research and Innovation Officer, who previously served as a senior director of AI research at Meta. Headquartered in Paris, the company is actively recruiting engineers and scientists across its four global research hubs, which include offices in New York, Montreal, and Singapore.
The transition from theoretical research to applied commercial products is expected to take several years, reflecting the fundamental nature of the technology being developed. However, AMI Labs has clearly defined the target sectors where its world models will eventually be deployed. The startup intends to focus on high-stakes industries where reliability, safety, and controllability are critical. These applications include robotics, industrial automation, manufacturing, wearables, and autonomous machines such as self-driving vehicles and drones. By teaching machines the kind of spatial intuition that comes naturally to humans, the company aims to solve complex physical tasks that current AI systems struggle to navigate.
In the near term, AMI Labs has identified the healthcare sector as its first major commercial testing ground. The company plans to partner with Nabla, the digital health startup previously led by Alexandre LeBrun, to explore how world models can assist in clinical settings. In medicine, AI systems must move beyond generating plausible text to accurately simulating how a patient’s physiology might respond to specific medical interventions. By predicting the outcomes of various treatment paths, the technology could eventually serve as a collaborative reasoning partner for physicians, helping to plan and advise on patient care while ensuring that medical professionals remain in full control of the final decisions.
The successful funding of AMI Labs underscores a broader shift in the artificial intelligence investment landscape. As the limitations of purely generative, text-based architectures become more apparent in physical and industrial applications, venture capital is increasingly flowing toward alternative frameworks. Competitors in the space have also secured billion-dollar valuations to pursue spatial intelligence and physical AI. By securing $1.03 billion before launching a commercial product, AMI Labs has positioned itself at the forefront of this emerging sector, armed with the capital and the scientific pedigree required to challenge the current paradigms of machine learning.
