AMD Agrees to Acquire World Labs in $8.2 Billion AI Deal

AMD has entered a definitive agreement to acquire World Labs, the San Francisco AI research company led by Fei-Fei Li. Reuters reports the transaction is valued at $8.2 billion and is expected to close by the end of 2026. AMD announced the agreement on September 28.

World Labs develops spatial-intelligence models designed to generate, reconstruct and simulate interactive 3D environments from text, images and video, alongside technology for robotic learning and simulation. The acquisition therefore adds model-research expertise rather than another conventional CPU, GPU or networking product line.

After the transaction closes, AMD says World Labs will continue working on AI model research and Fei-Fei Li will join AMD as executive vice president and chief scientist, reporting directly to Lisa Su. That places the acquired research organization unusually close to AMD's top-level technology strategy.

AMD frames the deal around learning from emerging model workloads earlier in the design cycle. Its stated goal is to use World Labs' model expertise to inform future hardware, software and systems. That is strategically different from claiming that World Labs already provides a shipping workload that changes the performance of current Instinct accelerators.

The acquisition follows an existing relationship rather than starting from zero. AMD Ventures said in March that it had invested in World Labs' Series B after an earlier investment and described ongoing technical and go-to-market collaboration around running spatial-intelligence workloads on AMD Instinct GPUs.

World Labs also appeared alongside AMD at CES 2026 as AMD expanded its broader AI platform story. The September agreement moves that relationship from investment and collaboration to planned ownership.

AI infrastructure is increasingly shaped by the workloads it must run. Spatial models that generate and reason about persistent 3D environments can stress compute, memory capacity and bandwidth, data movement, simulation and inference differently from text-centric language models. Owning a team that develops those workloads can give AMD earlier visibility into those requirements.

AMD explicitly says World Labs will help it understand how models are evolving and use that knowledge to shape future technology roadmaps. The practical significance is therefore vertical integration of workload research with compute-platform design. It does not by itself prove that AMD will gain a particular accelerator market share, performance advantage or revenue contribution from World Labs.

World Labs joins a growing list of AI acquisitions and team additions as AMD expands beyond selling individual processors. In August, AMD announced an agreement to acquire Taalas for specialized AI inference technology, while its broader platform strategy combines Instinct accelerators, EPYC CPUs, Pensando networking, ROCm software and rack-scale systems.

The World Labs transaction attacks a different layer: model research itself. If the deal closes as planned, AMD will have both the compute platform and an internal research organization developing emerging spatial-AI workloads. How much that influences future silicon or software will only become measurable when AMD discloses concrete roadmap changes or products.