AMD Just Bought Fei-Fei Li’s World Labs for $8.2 Billion
AMD confirmed yesterday that it has agreed to acquire World Labs, the spatial intelligence AI startup co-founded by Fei-Fei Li — in an all-stock deal valued at $8.2 billion. The transaction is AMD’s second-largest acquisition in history, behind only its $50 billion purchase of Xilinx in 2022. It is expected to close by the end of 2026, subject to regulatory approval.
Li will join AMD as Executive Vice President and Chief Scientist, reporting directly to CEO Lisa Su. The World Labs team will continue operating as a new frontier research group inside AMD, working on the same problems they were working on before — just with considerably more resources and a direct line to one of the world’s largest chip designers.
Who Is Fei-Fei Li and Why She Matters
Fei-Fei Li is one of the most influential AI researchers alive. She is best known for creating ImageNet — a dataset of more than 14 million labelled images that became the foundation for the deep learning revolution in computer vision. The 2012 ImageNet competition, where a neural network trained on her dataset dramatically outperformed everything else, is widely credited as the moment modern AI became a serious field. She has been called the “godmother of AI” for that contribution.
After leading the Stanford AI Lab and serving as Chief Scientist of AI at Google Cloud, Li co-founded World Labs in 2024 alongside three other Stanford researchers — Ben Mildenhall and Justin Johnson among them. The company raised $230 million at founding and $1 billion in a follow-on round earlier this year, with AMD participating as an investor. That relationship clearly deepened over time.
What World Labs Actually Builds
World Labs works on spatial intelligence, a field focused on teaching AI systems to understand and reason about three-dimensional environments, rather than processing text and flat images. The goal is AI that can model how objects move, how physical spaces are structured, and how actions play out over time.
On September 1, less than a month before the acquisition was announced — World Labs launched Atlas, a multimodal world model capable of generating, reconstructing, and simulating interactive 3D environments from just a few photos. Li described the potential directly: AI agents that learn in digitally simulated physical worlds before being deployed into the real one, making them significantly safer and more capable.
The applications are broad: robotics, autonomous vehicles, simulation tools for engineering and architecture, surgical training, and game development. Anywhere that AI needs to understand how the real world works in three dimensions — not just recognise a photo of it — spatial intelligence is the foundational capability required.
Why AMD Spent $8.2 Billion on a Company With No Revenue
World Labs has no reported revenue. It is a pure research organisation with a product (Atlas) that launched weeks ago. $8.2 billion for a two-year-old company generating nothing is, on its face, an extraordinary number. The logic behind it is strategic rather than financial.
AMD’s core business is chips — GPUs, CPUs, and AI accelerators. The market AMD competes in is increasingly defined not just by hardware performance but by understanding what AI workloads will look like in two, three, and five years. The company that best anticipates the next generation of compute requirements has a structural advantage in designing and selling the hardware that runs them.
Physical AI is the next major wave of AI compute demand. Robots, autonomous vehicles, and simulation-heavy applications require AI that understands 3D space — and that AI requires different, more complex compute patterns than large language models. By acquiring the researchers who are building the frontier of spatial intelligence, AMD gets a direct window into exactly those future workload requirements.
How It Fits AMD’s Broader Strategy
This is AMD’s third major AI-related acquisition in 2026. Earlier this year, it invested $5 billion in Anthropic as part of a computing power deal that committed Anthropic to running Claude on AMD’s MI400 chips. It also completed its $2.4 billion acquisition of Alphawave Semi for high-speed connectivity technology used in data centres.
The pattern is consistent: AMD is building a strategy that extends beyond chip hardware into the AI software and research ecosystem. Rather than just selling GPUs and hoping developers build things on them, it is acquiring the people and companies shaping what gets built — ensuring those workloads are designed with AMD hardware in mind from the ground up.
AMD passed a $1 trillion market cap earlier this month, reflecting investor confidence in exactly this kind of strategic positioning. Whether World Labs delivers on the spatial intelligence promise is a longer-term question. What’s clear today is that AMD is betting heavily on physical AI being the next major compute frontier — and is willing to pay $8.2 billion for a research team it believes can help define it. For more on AMD’s AI strategy, see our earlier coverage of AMD’s $5 billion investment in Anthropic and our breakdown of the Qualcomm-Tenstorrent deal — the two other major chip-plus-talent acquisitions defining the AI hardware race.
Read more tech related articles here
