AMD acquires Fei-Fei Li's World Labs for $8.2 billion
AMD is acquiring spatial intelligence startup World Labs for $8.2 billion, positioning the chipmaker to challenge Nvidia by integrating physical-world AI models with its hardware.
AMD has agreed to acquire World Labs in an all-stock transaction valued at approximately $8.2 billion. The deal, which is expected to close by the end of 2026 subject to regulatory approval, will bring World Labs founder and AI pioneer Fei-Fei Li into AMD's executive team. Li will serve as Executive Vice President and Chief Scientist, leading frontier AI research and reporting directly to AMD CEO Lisa Su. Founded in early 2024 in San Francisco by Li, Ben Mildenhall, and Justin Johnson, the 70-employee startup specializes in spatial intelligence.
World Labs focuses on building world models that generate, reconstruct, and simulate three-dimensional environments using text, image, and video inputs. In September, the startup unveiled Atlas, its first world model capable of generating complete spatial scenes from just a few images. This technology addresses a critical limitation in current multimodal language models, which struggle with spatial tasks like estimating distances, orientations, and sizes, or predicting basic physics.
The acquisition represents a massive valuation jump for World Labs, which previously raised a $230 million Series A in 2024 from investors including Andreessen Horowitz, Nvidia, and AMD. In early 2025, the startup raised an additional $1 billion from backers like Autodesk, bringing its valuation to around $5 billion. For AMD, which has a market cap of $1 trillion compared to Nvidia's $5 trillion, the acquisition is a strategic move to counter Nvidia's ecosystem play, which recently included a $13 billion acquisition of Hugging Face.
For AI practitioners and developers, this acquisition signals a shift toward hardware-software co-design for physical-world AI. As Li noted, scaling frontier research requires tight hardware integration, warning that "AI is hobbled in efficiency" without it. By bringing World Labs' spatial intelligence expertise in-house, AMD aims to optimize its future silicon roadmaps specifically for 3D simulation and physics-based AI workloads. This could eventually provide developers with more efficient, specialized hardware alternatives to Nvidia for training and deploying robotics, science, and entertainment models.
This is our own summary of reporting by The Decoder
