Story
September 29, 2026

AMD Bets $8.2 Billion That Fei-Fei Li Can Close Nvidia’s AI Gap

AMD sees World Labs’ research as a route to better AI hardware and systems, while Fei-Fei Li sees AMD’s scale as the way to push physical-world intelligence beyond the lab. The wager is that their combined effort can compete in an AI market where Nvidia already holds a formidable ecosystem advantage.

AMD’s push into AI took a major turn with its agreement to acquire World Labs, the physical-world AI startup founded by computer-vision pioneer Fei-Fei Li, in an all-stock deal valued at about $8.2 billion. The transaction is expected to close by year-end, pending regulatory approval.

World Labs was founded in 2024 around a larger ambition than text generation: models that can reason about the physical world. Its first commercial product, Marble, generates interactive 3D environments from prompts—technology that could serve entertainment creators but also provide simulated worlds for training robots.

The companies were not starting from scratch. AMD and World Labs had already partnered on inference optimization and training, and Li had appeared at AMD’s CES presentation earlier this year. Now Li will become AMD’s executive vice president and chief scientist, reporting to chief executive Lisa Su, while the World Labs team continues its model-research work.

For Li, the sale is framed as an expansion of the original mission rather than a retreat from it. “Now that we have tangible proof of the possibilities, we want to do everything we can to accelerate the future,” she wrote, arguing that doing so means “scaling our efforts, widening our reach, and getting closer to the hardware.”

AMD’s case is equally strategic. Su said next-generation compute platforms require “a deep understanding of how models are evolving,” and argued that World Labs’ expertise can help shape the hardware, software and systems behind them.

That puts the deal squarely in AMD’s contest with Nvidia. Nvidia already offers open-weight world models through Cosmos, while AMD’s public offerings have focused on text and video. As robotic systems—from industrial machines to autonomous vehicles—need synthetic training data, AMD is betting Li’s research can turn a hardware rivalry into a broader platform fight.