AMD to acquire AI startup World Labs for $8.2 billion
Chipmaker AMD has agreed to acquire AI startup World Labs for $8.2 billion, a move aimed at boosting its physical AI and robotics capabilities to better compete with rival Nvidia.

AMD's acquisition of World Labs is valued at $8.2 billion and is expected to close by the end of the year, subject to regulatory approval. Founded in 2024 by computer vision pioneer Fei-Fei Li alongside researchers Justin Johnson, Christoph Lassner, and Ben Mildenhall, World Labs previously raised $230 million in funding, with AMD among its early backers. Following the acquisition, Li will step into the role of Executive Vice President and Chief Scientist at AMD, while Johnson and Mildenhall will continue to lead the World Labs team within AMD's new frontier research division.
World Labs specializes in developing AI systems designed to simulate the physical world. The startup's first public tool, Marble, utilizes Gaussian splats to generate small 3D spaces that can be exported as assets for game development and film production. The company has also introduced other tools, such as Atlas. While these early tools focused on content creation, AMD plans to leverage World Labs' expertise to bolster its presence in robotics, physical AI, and simulation.
This acquisition represents a major step in AMD's effort to challenge Nvidia's dominance in the AI hardware and software ecosystem. While Nvidia currently leads the market with its Cosmos models and synthetic data generation tools for robotics, AMD has struggled to match that comprehensive stack, despite releasing its own open-source world model called Micro-World. For AI practitioners and developers, this merger means AMD will likely integrate World Labs' technology directly into its hardware design process. This will help AMD optimize its future chips for complex physical AI workloads, potentially offering developers a viable, high-performance alternative to Nvidia's proprietary software and hardware ecosystem.
This is our own summary of reporting by Ars Technica AI



