The current wisdom in chipmaking is that building good chips requires understanding how the AI models running on them are evolving. Nvidia already has an answer to that in its own open-weight world model, Cosmos. AMD had only text- and video-generation models. Now AMD is closing that gap by buying the whole lab.
AMD (NASDAQ: AMD) said it has entered a definitive agreement to acquire World Labs, the AI research lab led by Dr. Fei-Fei Li, in an all-stock deal worth approximately $8.2 billion. The transaction is expected to close by the end of the year, subject to regulatory approval and customary closing conditions. Li will join AMD as executive vice president and chief scientist, reporting directly to CEO Dr. Lisa Su.
San Francisco-based World Labs builds “spatial intelligence” models that generate, reconstruct, and simulate interactive 3D environments from text, image, and video inputs. Li, a Stanford computer science professor, pioneered computer vision by building the ImageNet database; co-founders Justin Johnson, Christoph Lassner, and Ben Mildenhall each bring backgrounds in 3D graphics, spatial computing, and NeRF (Neural Radiance Fields) research. Founded in 2024, World Labs emerged from stealth that September with $230 million at a $1 billion valuation, then launched its first commercial product, Marble, in November. Marble turns text, images, or 360-degree panoramas into editable, downloadable 3D environments — pitched for entertainment content as much as for simulated training environments for robots.
AMD and World Labs weren’t strangers before this. The two formed a model-training-and-inference-optimization partnership on AMD GPUs last year, and Li appeared as a guest at AMD’s CES presentation earlier this year. In February, World Labs raised $1 billion in fresh funding at a reported $5 billion valuation — with both AMD and Nvidia among the investors, alongside Autodesk, Fidelity, and Emerson Collective. AMD is now buying outright a company its biggest rival also had a stake in.
“Building the compute platforms for the next generation of AI requires a deep understanding of how models are evolving,” Su said. “Fei-Fei and the World Labs team bring exceptional research leadership and model expertise. Together, we can use that insight to develop the hardware, software and systems that will power the next generation of AI and strengthen the open AI ecosystem.” Li, writing on her own blog, said World Labs now has “tangible proof of the possibilities” and wants to scale its efforts by “getting closer to the hardware,” adding that “without having a focused hardware effort, AI is hobbled in efficiency.”
The deal lands right after AMD unveiled a full-stack AI infrastructure push — Helios racks and MI400-series GPUs — that won backing from OpenAI (a 6-gigawatt supply commitment), Meta, and Anthropic. AMD now has the hardware lineup it needs. What it lacked was the research pipeline to know what comes next — particularly for training AI on physical-world tasks like robotics and autonomous driving, where real-world data is scarce and synthetic data generated by world models is expected to fill the gap.
MORE FROM THE POST
- Hugging Face Explores Sale at $13 Billion or More
- Nvidia’s $7B Poolside Courtship: License, Hire, Invest — Not Acquire
- Nvidia in Talks With Korean AI Chip Designer Rebellions
- SpaceX Acquires Cursor for $60B in All-Stock Deal
- Naver and Dunamu Merge to Build AI-Powered Web3 Platform with $6.8 Billion Investment