AMD to acquire Fei-Fei Li’s World Labs in $8.2B AI deal

AMD is acquiring World Labs, described as one of the leading developers of deep learning models intended to understand physical reality, in an $8.2 billion deal. The companies said the acquisition reflects the need for close collaboration across model research, systems and compute.

The deal is expected to close before the end of the year, subject to regulatory approval. For readers tracking AI infrastructure, the key takeaway is that AMD is moving model research closer to its chip roadmap.

World Labs founder Fei-Fei Li will join AMD as executive vice president and chief scientist. Li is a Stanford computer science professor and is considered a pioneer in AI, particularly computer vision, for her work building ImageNet and the competitions it inspired.

AMD and World Labs already had a close relationship: the companies formed an inference optimization-and-training partnership last year, and Li appeared at AMD’s CES presentation earlier this year.

World Labs was founded in 2024 to develop deep learning models with a stronger understanding of the physical world. The article notes that “world model” remains loosely defined, spanning models that understand visual inputs to systems capable of generating and sustaining high-fidelity simulations of reality.

World Labs’ first product, Marble, is pitched for creating entertainment experiences and simulated environments for robot training. That makes the acquisition relevant beyond software: world models are seen as important for robotics, autonomous vehicles, industrial robots and general-purpose humanoids.

The acquisition is likely to help AMD compete with Nvidia in building an ecosystem for AI-specific chips. TechCrunch notes that Nvidia already has open-weight world models such as Cosmos, while AMD has only offered text- and video-based models to the public.

For AMD, understanding frontier workloads from World Labs could help shape future chip design. For the broader market, the move highlights how AI hardware companies are tying compute strategy more directly to model development.