AMD Buys Fei-Fei Li's World Labs for $8.2B: The Race for Physical AI Heats Up
AMD will acquire Fei-Fei Li's World Labs for $8.2B in stock, making her chief scientist and betting big on "physical AI" that understands 3D space.
The deal
Advanced Micro Devices announced on September 28, 2026 that it will acquire World Labs, the spatial-intelligence startup founded by AI pioneer Fei-Fei Li, in an all-stock transaction valued at $8.2 billion. The deal is expected to close by the end of 2026, subject to regulatory approvals, and would be AMD's second-largest acquisition ever -- behind only its roughly $50 billion purchase of Xilinx in 2022.
As part of the agreement, Li -- often called the godmother of AI for her work on ImageNet and computer vision -- will join AMD as executive vice president and chief scientist, reporting directly to CEO Lisa Su.
What World Labs actually builds
World Labs does not build chatbots. It builds "world models": AI systems that generate, reconstruct, and simulate three-dimensional physical environments from text, images, and video. In November 2025 the company shipped Marble, its first commercial product, which turns a text prompt, a photo, or a video into an editable 3D environment with paid plans up to $95 a month. On September 1, 2026 it released Atlas, a more advanced model that reconstructs navigable 3D scenes from just a handful of photos and is aimed at training robots.
The two companies are not strangers. AMD invested in World Labs' $1 billion funding round earlier this year, and the teams began a deep technical partnership last year around model training and inference optimization on AMD GPUs.
Why AMD is buying instead of building
AMD is a chip company racing Nvidia. Buying World Labs gives it something money alone cannot: a window into how the next generation of AI workloads will actually behave. "Building the compute platforms for the next generation of AI requires a deep understanding of how models are evolving," CEO Lisa Su said in the announcement. The company wants World Labs' research to guide its hardware, software, and systems roadmaps as demand shifts from text-based models to physical AI and robotics.
The deal also sharpens AMD's open-ecosystem positioning: the company says it will keep supporting models with publicly available code, training data, and weights that any developer can modify -- a contrast with closed systems like OpenAI's GPT-6 Astra and Anthropic's Claude Sonnet 5.5.
What this means for your business
Physical AI is moving from research to procurement. If your company touches logistics, manufacturing, construction, real estate, or field services, the technology that lets AI understand and simulate physical space is arriving within a commercial product cycle -- not a decade.
Action steps:
1. Audit where spatial AI applies to you. Warehouse layout, facility planning, digital twins, equipment training, and virtual site walkthroughs are the first commercial use cases. 2. Watch AMD's open-ecosystem play. If AMD ships optimized open tooling for world models, small companies get enterprise-grade spatial AI without building it -- the same pattern that made cloud GPUs accessible. 3. Talent signal. Li joining a chipmaker means the best AI researchers are moving toward full-stack roles. Hiring AI talent gets harder; buying AI capability gets easier. 4. Vendor roadmap check. Ask your AI vendors when their models gain 3D and physical-world understanding. The answer tells you how future-proof they are.