AMD acquires World Labs for $8.2 billion, Fei-Fei Li appointed as Chief Scientist

AMD is making a major all-stock bet on AI’s next frontier—the understanding and simulation of the physical world.

On September 28th, AMD announced an approximately $8.2 billion all-stock acquisition of AI startup World Labs, a company co-founded and led by renowned AI research pioneer Fei-Fei Li as CEO.

Upon completion of the transaction, Fei-Fei Li will join AMD as Executive Vice President and Chief Scientist, reporting directly to CEO Lisa Su. The acquisition is expected to be completed by the end of 2026, pending regulatory approval and other customary closing conditions.

This acquisition marks a significant expansion of AMD’s AI strategy beyond its core chip manufacturing business, aiming to guide its cross-hardware, software, and systems technology roadmap through direct control of cutting-edge model research.

For investors, this deal represents a key move for AMD to counterbalance Nvidia's ecosystem advantages, as the latter already possesses a range of open-weight world models, including Cosmos.

The core rationale behind AMD's acquisition of World Labs lies in internalizing model research capabilities as a driving force for chip design.

In a statement, AMD noted that as AI expands into inference, robotics, simulation, and physical AI, the demand for computational infrastructure is becoming increasingly diverse.

By bringing in World Labs’ expertise in model development, AMD will gain deeper insights during the evolution of workloads and optimize its future technology roadmap accordingly. Lisa Su stated in the announcement:

Building the next generation of AI computing platforms requires a profound understanding of how models evolve. Fei-Fei Li and the World Labs team bring outstanding research leadership and unmatched expertise in model development.

Fei-Fei Li described joining AMD as a necessary choice to drive technological breakthroughs from the lab to large-scale application. She noted that advancing the next generation of AI technologies requires close collaboration between model research, systems, and compute. Fei-Fei Li said:

We now have a real opportunity to prove ourselves, and we want to do everything we can to accelerate the future. Achieving that demands greater scale, broader impact, and closer integration with hardware.

Founded in 2024 and headquartered in San Francisco, World Labs focuses on developing spatial intelligence models that can generate, reconstruct, and simulate interactive 3D environments from text, image, and video inputs, while also advancing related technologies for robotic learning and simulation.

Spatial intelligence differs from AI systems primarily centered on language or 2D images; its core lies in modeling how objects, spaces, and actions operate within three-dimensional environments. This capability holds broad potential in real-world application scenarios such as robotics, simulation, and design.

Earlier this year, World Labs completed a $1 billion funding round. Its first product, Marble, is geared toward entertainment experience creation, while also providing simulated environments for robot training.

Notably, world models are considered by the industry as the key infrastructure for deploying generative AI models on robotic platforms. From autonomous vehicles to industrial robots and general-purpose humanoid robots, due to the lack of effective real-world data for training general robots, synthetic data generated by world models will play a crucial role in realizing this vision.

This acquisition will directly strengthen AMD’s position in the AI chip ecosystem competition.

Currently, Nvidia has already established a relatively complete ecosystem matrix through open-weight world models like Cosmos, while the models AMD previously made available to the public are still limited to the text and video categories, and it has yet to release a world model product focused on physical world understanding.

The two companies have an existing foundation of collaboration. According to reports, World Labs and AMD established a partnership focused on inference optimization and training last year, and have maintained close ties since. Earlier this year, Fei-Fei Li was a guest speaker at AMD’s CES presentation.

Fei-Fei Li is an influential academic figure in the AI field, especially recognized for her pioneering contributions in computer vision. She is widely known for leading the creation of the ImageNet database and subsequent challenges. In 2024, she founded World Labs, advocating that true general intelligence must be grounded in physical reality and possess data comprehension and reasoning capabilities beyond text alone.

AMD stated that this acquisition will help advance its overall strategy to provide AI infrastructure for the open ecosystem.