AMD Acquires Fei-Fei Li’s World Labs in Reported $8.2 Billion Deal
AMD is set to acquire World Labs for $8.2 billion, a move designed to strengthen its AI chip ecosystem and leverage advanced deep learning models for understanding physical reality. This strategic acquisition will bring AI pioneer Fei-Fei Li to AMD as chief scientist, enhancing its competitive edge against Nvidia in the development of critical world models for robotics and generative AI.
AMD has announced its acquisition of World Labs, the AI company co-founded by renowned computer scientist Fei-Fei Li, in a deal valued at $8.2 billion. World Labs develops deep-learning models designed to understand and simulate the physical world, an area increasingly viewed as important to the development of advanced AI and robotics.
Both companies highlighted the importance of combining model research, computing systems and hardware, with AMD saying insights from World Labs’ frontier workloads will help shape its future chip-development roadmap.
As part of the acquisition, Fei-Fei Li will join AMD as executive vice president and chief scientist, strengthening an existing relationship between the companies that includes an inference optimisation and training partnership announced last year.
Li, a Stanford computer science professor, is widely known for her work in computer vision and for creating the ImageNet database, which helped drive advances in visual AI through the competitions built around it.
She founded World Labs in 2024 to develop models capable of understanding physical environments, arguing that more advanced forms of intelligence require grounding in the physical world rather than relying primarily on text.
Why World Models Matter to AMD
World Labs’ first product, Marble, generates detailed virtual environments that can be used for applications including immersive entertainment and robot training. The broader concept of a world model covers AI systems capable of understanding, generating and maintaining representations or simulations of physical environments, potentially providing synthetic data for training autonomous vehicles, industrial robots and humanoid systems.
AMD’s acquisition therefore gives the chipmaker access to technology that could complement its hardware strategy and strengthen its position in an increasingly competitive AI infrastructure market, where Nvidia has already developed its Cosmos family of world models.
The deal also comes as AI companies search for ways to overcome the limited availability of useful real-world data for training robots. World models could help generate synthetic environments and scenarios at scale, supporting the development of systems from autonomous machines to general-purpose humanoids.
The acquisition remains subject to regulatory approval and is expected to close before the end of the year, potentially marking a significant expansion of AMD’s ambitions beyond supplying the computing infrastructure on which AI models run.