Story
September 29, 2026

AMD’s $8.2B Fei-Fei Li Deal Takes Its Nvidia Fight Beyond Chips

AMD sees World Labs as a route from supplying AI compute to shaping the models that will demand it, while Fei-Fei Li argues that physical-world intelligence needs far tighter integration with hardware. The acquisition puts AMD more directly against Nvidia in the emerging race for robotics, simulation and spatial AI.

AMD has agreed to acquire World Labs, the two-year-old physical-AI startup co-founded by Fei-Fei Li, in an all-stock transaction valued at about $8.2 billion. The deal makes a forceful strategic point: the contest with Nvidia is no longer only about whose chips train the biggest models.

World Labs was founded around the view that AI must move beyond language. Its models are trained to interpret the three-dimensional physical world — how objects, spaces and motion behave — rather than simply generate or analyze text. Its product, Marble, creates persistent 3D environments from images, video, text and layouts.

Li said World Labs had already begun a deep technical partnership with AMD last year, focused on training and inference optimization on AMD GPUs. As the work progressed, she said, the companies concluded it was a “natural fit” to unite their software, hardware, foundation-model and application efforts.

The transaction will make Li AMD’s executive vice-president and chief scientist, reporting to chief executive Lisa Su, while the World Labs team continues its model-research work. Li framed the move as a way to reach the compute and engineering scale needed to push physical and spatial AI forward; she has argued that without a focused hardware effort, AI remains constrained in both efficiency and scale.

For AMD, the prize is insight as much as research talent. Su said building the next generation of compute platforms requires “a deep understanding of how models are evolving,” and that World Labs can help shape the hardware, software and systems behind them.

That ambition overlaps squarely with Nvidia’s expansion into world models, robotics and simulation. Supporters cast the acquisition as an offensive against Nvidia’s advantage in physical AI; one AMD welcome message put the appeal more simply: Li “taught machines to see” and is now teaching them “to understand the world.”

The companies expect the deal to close by year-end, subject to regulatory approvals, though one account says the expected closing is by the end of 2026.

Story coverage