AMD Buys Fei-Fei Li's World Labs for $8.2 Billion in Stock, Naming Her Chief Scientist
News Summary
AMD has agreed to acquire World Labs, the San Francisco AI lab founded by Stanford computer scientist Fei-Fei Li, in an all-stock transaction valued at about $8.2 billion. The companies announced the agreement on September 28, 2026 (Eastern Time). Li will join AMD as executive vice president and chief scientist, reporting to CEO Lisa Su. The deal is expected to close before the end of 2026, subject to regulatory approval and customary closing conditions.
What Was Announced
Coverage from TechCrunch, CNBC, Bloomberg, Fortune and AMD's own newsroom agrees on the core terms. The purchase price is roughly $8.2 billion and is paid in AMD stock rather than cash. World Labs will be folded into AMD, and its founder takes a senior leadership role at the chipmaker. Reports describe it as the second-largest acquisition in AMD's history. The largest remains the roughly $50 billion purchase of Xilinx, which AMD completed in 2022.
What World Labs Builds
World Labs develops "world models," AI systems that generate, reconstruct and simulate three-dimensional environments from text, images or video. The company describes this field as spatial intelligence: teaching machines to understand and reason about physical space rather than only language. Its first commercial product, Marble, lets users build interactive 3D worlds from a prompt. Reports say the technology has uses in entertainment and games, and in simulated environments for training robots.
Who Is Fei-Fei Li
Li is a professor of computer science at Stanford University and a pioneer of modern computer vision. She is best known for creating ImageNet, the large labeled image dataset whose associated competition helped launch the deep learning boom of the early 2010s. She founded World Labs in 2024. Her move to AMD as chief scientist puts one of the field's best-known researchers in charge of the science agenda at a major hardware company.
Funding and Valuation Background
According to reports, World Labs came out of stealth in September 2024 with a $230 million round at a valuation of about $1 billion, led by Andreessen Horowitz. It is reported to have raised roughly $1 billion more about a year later, at a valuation near $5 billion. Those funding figures come from secondary reporting, not from the acquisition announcement. If accurate, the $8.2 billion price is a substantial step up from the last reported private valuation.
Why AMD Wants It
AMD and World Labs already had an inference optimization and training partnership, formed last year, and Li appeared as a guest during AMD's CES presentation earlier in 2026. World Labs said that AI development requires close collaboration across model research, systems and compute. For AMD, the practical benefit is direct access to frontier AI workloads. Understanding how cutting-edge models are trained and run helps a chip designer decide what to build next, and it is expected to shape AMD's chip roadmap.
Strategically, the deal positions AMD to compete with Nvidia not only on chips but on the wider ecosystem of models, software and systems around them. World models are widely seen as important for physical AI, meaning robots, autonomous vehicles and other machines that must act in the real world. Those systems need large amounts of simulated 3D data and heavy compute, which is where AMD hopes to sell hardware.
What Comes Next
The acquisition still needs regulatory clearance, and the companies expect to close before the end of 2026. Open questions include how Marble will be offered after closing, how World Labs' research team will be integrated, and how quickly the research will show up in AMD's product roadmap. Readers worldwide should watch for updates from AMD and World Labs as the review proceeds.
Why It Matters for Science and Technology Education
The deal shows how research in computer vision and 3D scene understanding is turning into commercial infrastructure. Concepts that were once academic topics, such as spatial reasoning, simulation and training data for robots, are now central to hardware strategy. Students and learners who follow AI can treat this as a case study in how models, data and silicon design influence one another.