AMD has struck a deal to purchase AI research lab World Labs in an all-stock transaction valued at about $8.2 billion, the chipmaker announced Monday, Sept. 28. The deal aims to co-design hardware, software, and systems for emerging AI models and applications such as robotics and physical AI. The transaction is expected to close by the end of 2026, subject to customary closing conditions and regulatory approvals. The acquisition matters because it positions AMD to understand chip requirements better as AI advances, rather than simply buying a model company. Matt Kimball, vice president and principal analyst at Moor Insights & Strategy, said the move is about capturing compute and memory needs today so they can be baked into low-level architectural decisions and designs.
San Francisco-based World Labs develops spatial intelligence "world models" that use text, images, and video to generate and simulate interactive 3D environments, AMD said. World models differ from large language models in that they focus on environments—geometry, physics, objects, motions, and dynamics—rather than the relationship between language, tokens, and information. Kimball described the technology as "a model for the physical world," noting that any physical AI and robotics require 3D modeling, geometric understanding, and physics to work properly. As an example, training a robot to place a cup on a table requires the model to account for motions and contact dynamics so the cup lands flat rather than half on and half off.
AI pioneer Fei-Fei Li, co-founder and CEO of World Labs, will join AMD as executive vice president and chief scientist upon closing. World Labs will continue to focus on AI model research, the company said. AMD CEO Lisa Su said in a statement that building compute platforms for the next generation of AI requires a deep understanding of how models are evolving, and that Li and the World Labs team bring exceptional research leadership and model expertise. Kimball called Li "a rock star" and said she may be as much of the target as the company itself, bringing world-class talent that could help AMD find a competitive footing against Nvidia more quickly. Brendan Burke, a research director at Futurum Group, said access to talent is the biggest advantage in AI training and that "you can't hire a more accomplished researcher than Dr. Fei-Fei Li."
The deal places AMD in a stronger position in the emerging physical AI market, which AMD estimates could reach $200 billion by 2035. Burke said AMD identified a transformative category that has yet to produce significant commercial impact, and that the purchase shows the company is using its recent momentum to think long-term about the future of AI. Nvidia has already moved into physical AI with its Cosmos world model, Isaac and Omniverse platforms, and Project GR00T robotics models, giving it an early lead in the market, according to Kimball. Burke said World Labs researchers could help AMD train useful models and inform the design of AMD's rack-scale systems for those model architectures, potentially helping AMD gain market share. He added that AMD increasingly sees itself as an AI research organization, and that co-designing with World Labs could help AMD "move toward leadership in fields that aren't yet captured by leading labs."
The evidence base for this analysis comes from a single full-text source, Data Center Knowledge, which reported the announcement and included analyst commentary from Moor Insights & Strategy and Futurum Group. The source did not disclose the deal's financing structure beyond the all-stock nature and approximate valuation, nor did it provide World Labs' revenue, headcount, or prior funding details. Regulatory approval timelines and specific closing conditions were not detailed. What remains to be watched is whether AMD can integrate World Labs' research talent without disrupting its existing roadmap, and whether the co-design approach yields measurable chip design advantages against Nvidia's established physical AI stack. The transaction's expected closing by the end of 2026 leaves a window for regulatory review and competitive responses.