Yann LeCun, Meta's former chief AI scientist and a founding father of deep learning, has raised $1.03 billion for his new startup Advanced Machine Intelligence (AMI). The company aims to build AI systems that "understand the world, have persistent memory, can reason and plan, and are controllable and safe" — a direct challenge to today's transformer-based large language models.
The massive Series A round was led by Cathie Innovation, with participation from Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions. Notable individual investors include former Google CEO Eric Schmidt and World Wide Web inventor Tim Berners-Lee. The funding amount positions AMI among the largest AI startup launches in history.
LeCun has long argued that current LLMs, despite their utility, cannot achieve human-level intelligence because they lack true world understanding. AMI plans to build "world models" that can process spatial data and understand physics — similar to approaches being pursued by World Labs (led by Fei-Fei Li) and UC Berkeley's robotics labs. The market for AI reasoning systems could reach hundreds of billions as enterprises seek more capable AI agents.
This funding signals a potential shift away from the scaling-focused approach dominating AI development toward more fundamentally different architectures. With concerns mounting about AI safety and the limitations of current models, LeCun's vision for grounded, physics-aware AI could reshape how the industry thinks about artificial general intelligence. The company's emphasis on "controllable and safe" systems also addresses growing regulatory pressures.