AI · Sep 30, 2026
The four rules in the White House AI commitment, and the missing fifth: penaltiesAMD buys Fei-Fei Li's World Labs for $8.2 billion to challenge Nvidia in world models and physical AI
· Updated · Signal Desk newsroom
Analysis of reporting by Ars Technica. How we work

AMD will acquire World Labs, the world-model startup founded by Fei-Fei Li, in a deal valued at $8.2 billion and expected to close by year-end pending regulatory approval, Ars Technica reports. Li becomes AMD's executive vice president and chief scientist, in a move aimed at Nvidia's lead in simulation and robotics.
Key takeaways
- The deal is valued at $8.2 billion and expected to close by year-end, pending approval.
- World Labs was founded in 2024 by Fei-Fei Li, Justin Johnson, Christoph Lassner and Ben Mildenhall.
- It raised $230 million, including capital from AMD.
- Its Marble tool generates exportable 3D Gaussian-splat spaces for film and games.
- Li becomes AMD EVP and chief scientist; Johnson and Mildenhall continue leading the team.
The deal
Ars Technica reports that AMD will acquire World Labs in a deal valued at $8.2 billion, expected to close by the end of the year pending regulatory approval. Fei-Fei Li founded World Labs in 2024 with Justin Johnson, Christoph Lassner and Ben Mildenhall, backed by $230 million in funding that included investment from AMD.
Li will become AMD's executive vice president and chief scientist, while Johnson and Mildenhall will continue to lead the World Labs team inside AMD. The structure suggests AMD is buying both research talent and a product line, and keeping the founding team in charge of the work it started. AMD's earlier investment in World Labs means the companies already had a relationship before the acquisition.
What World Labs builds
World Labs develops world models, AI systems that predict and simulate physical environments. Its Marble tool generates 3D spaces as exportable Gaussian splats for use in film and games, and later tools such as Atlas extended the lineup. A core part of the investment thesis is synthetic data for training robots, letting machines learn in simulated worlds before acting in real ones.
That is an area where Nvidia leads with its Cosmos platform and broader physical AI work, such as Skild AI teaching robots from a single video. Nvidia executives have argued that robots are waiting for a ChatGPT moment, and both chipmakers see simulation as a driver of compute demand.
Competitive stakes
The deal is part of a wave of large AI acquisitions by chipmakers. Nvidia agreed to buy Hugging Face for $12.9 billion earlier in September. For AMD, World Labs offers a software and research foothold in physical AI that could make its hardware more attractive to robotics and simulation customers.
Open questions include whether World Labs' public tools, such as Marble and Atlas, remain broadly available or become tied to AMD hardware, and whether regulators impose conditions given growing concentration in AI. AMD's roadmap language connecting World Labs to future silicon will indicate how central the acquisition is to its strategy. More in our robotics coverage and startups coverage.
Who feels it
- Robotics and simulation teams
- Watch whether World Labs tools stay platform-neutral.
- Nvidia and Cosmos rivals
- A well-funded competitor with Li as AMD chief scientist.
- Regulators
- An $8.2 billion deal in physical AI may draw scrutiny.
What to watch
- Regulatory approval timing and any conditions.
- Post-close status of Marble and Atlas.
- AMD roadmap links between World Labs and its chips.
FAQ
How much is AMD paying?
The deal is valued at $8.2 billion.
Who founded World Labs?
Fei-Fei Li, Justin Johnson, Christoph Lassner and Ben Mildenhall, in 2024.
What role will Li have?
AMD executive vice president and chief scientist.
When will it close?
By year-end, pending regulatory approval.
Companies: NVIDIA
More on NVIDIA
AI · Oct 7, 2026
Microsoft prices its first RTX Spark PCs; the agent sandbox coming to all Windows 11 users may matter moreAI · Sep 24, 2026
Nvidia, DeepMind and EMBL-EBI release predicted structures for protein complexes from 2,800-plus virusesAI · Oct 15, 2026
NVIDIA’s Nemotron 3.5 lineup spans a 120B open agent model and Bolt edge models that fit in 24GB and 4GBAI · Oct 13, 2026
NVIDIA ties UK deployments in health, banking and cities to a $250 million agentic security compute pledge