Advanced Machine Intelligence Labs (AMI Labs), the Paris-headquartered startup founded by Yann LeCun after his departure from Meta, has closed a $1.03 billion seed round at a $3.5 billion pre-money valuation — one of the largest seed rounds in AI history.

The funding comes from a global syndicate including Nvidia, Temasek, Jeff Bezos’s Bezos Expeditions, Cathay Innovation, SBVA (formerly SoftBank Ventures Asia), and Sea. Former Meta COO Sheryl Sandberg and other high-profile figures have also joined the board.

With only 12 employees so far, AMI Labs is focused on developing world models — AI systems that learn physics, causality, and persistent memory directly from real-world sensory data rather than text alone. LeCun has long argued this approach is the missing piece for truly intelligent, autonomous agents capable of planning, predicting outcomes, and operating safely in the physical world.

What AMI Labs Is Building

Unlike today’s dominant large language models that predict the next word, AMI’s world models aim to create systems that:

• Understand cause-and-effect and spatial reasoning
• Maintain long-term memory across interactions
• Plan complex, multi-step actions in dynamic environments
• Power next-generation robotics, autonomous systems, and scientific discovery

The company already has an early strategic partnership with French health-tech startup Nabla, giving it first access to world-model technology for FDA-certifiable clinical AI.

Why World Models Are the Next AI Frontier

This raise is a major validation of the shift from language-only AI to physically grounded, agentic intelligence. As autonomous agents move from chat interfaces into real-world execution (robotics, manufacturing, healthcare, infrastructure), world models are becoming the critical foundation for reliability and safety.

LeCun’s contrarian bet — that the next leap in autonomy will come from understanding the physical world rather than scaling text prediction — now has massive capital behind it. At just one month old, AMI Labs joins the elite group of frontier AI companies (alongside World Labs, xAI, and Anthropic) that are attracting billion-dollar checks to redefine how agents interact with reality.

For the autonomous AI ecosystem, this is a pivotal moment: the infrastructure and research for production-grade, physics-aware agents is accelerating faster than ever.