AMD is acquiring World Labs, a leading developer of deep-learning models designed to understand physical reality, in an $8.2bn deal, the companies announced on Tuesday.

World Labs said the acquisition would strengthen the collaboration between AI model research, computing systems and infrastructure, which it described as essential to advancing AI development.

AMD said the acquisition would also help shape its chip development roadmap by giving it deeper insight into emerging AI workloads, including the advanced models being developed by World Labs.

As part of the acquisition, World Labs founder Fei-Fei Li will join AMD as executive vice president and chief scientist.

The companies formed an inference optimisation and training partnership last year, further strengthening their relationship.

Li also appeared as a guest during AMD’s CES presentation earlier this year.

Li, a Stanford University computer science professor, is regarded as a pioneer in artificial intelligence, particularly computer vision.

She is best known for creating the ImageNet database, which helped drive the development of AI image-recognition systems and inspired a series of influential AI competitions.

Li founded World Labs in 2024 to develop deep-learning models capable of building a deeper understanding of the physical world. She has argued that achieving true artificial general intelligence would require AI systems to be grounded in physics and capable of understanding and reasoning about information beyond text.

In a post announcing the deal, Li said the partnership with AMD was driven by World Labs’ ambition to scale its technological breakthroughs beyond the research laboratory.

“Now that we have tangible proof of the possibilities, we want to do everything we can to accelerate the future,” Li wrote in the post. “To do this requires scaling our efforts, widening our reach, and getting closer to the hardware.”

World Labs’ first product, Marble, is designed to create immersive entertainment experiences and simulated environments that can also be used to train robots.

The acquisition could strengthen AMD’s efforts to compete with long-standing rival Nvidia in building an ecosystem around chips designed specifically for artificial intelligence workloads.