AMD to Acquire World Labs in $8.2 Billion Deal to Advance AI World Models

AMD is acquiring World Labs, the AI company founded by computer vision pioneer Fei-Fei Li, in an $8.2 billion deal that combines advanced AI model research with semiconductor development.

The companies said the acquisition reflects the growing need for closer collaboration between AI model developers, computing systems, and hardware manufacturers. AMD said insights from frontier AI workloads developed by World Labs will help inform its future chip-making roadmap.

The deal will see Li join AMD as executive vice president and chief scientist. AMD and World Labs had already worked together through an inference optimisation and training partnership, strengthening ties between the two companies before the acquisition.

Fei-Fei Li’s World Labs Focuses on AI Models That Understand Reality

Li, a Stanford computer science professor, is widely recognised for her contributions to computer vision, including the creation of the ImageNet database and the research ecosystem built around it. She founded World Labs in 2024 to develop AI systems that can better understand and reason about the physical world.

In a post announcing the acquisition, Li said the partnership would help scale World Labs’ technology beyond the research environment, allowing the company to expand its work and move closer to hardware integration.

World Labs has focused on “world models,” a broad category of AI systems designed to represent and reason about real-world environments. The term covers different approaches, including models that process visual information and systems that generate realistic simulated environments.

Li has previously outlined different categories of world models, describing how these systems could support applications that require deeper understanding of physical spaces and interactions.

AMD Expands Its AI Competition With Nvidia

The acquisition is expected to strengthen AMD’s position in the AI hardware market as it competes with Nvidia’s established ecosystem. Nvidia has already introduced AI models focused on areas such as physical AI and simulation, while AMD has primarily made text and video AI models available publicly.

World models are increasingly being explored for robotics applications, including autonomous vehicles, industrial machines, and humanoid robots. These systems could help address challenges around collecting enough real-world training data by generating synthetic environments for AI development.

The acquisition is expected to close before the end of the year, pending regulatory approval.