AMD to Acquire World Labs for $8.2 Billion, Bringing Spatial AI Research Into Its Compute Roadmap
AMD has signed a definitive agreement to acquire World Labs, the spatial-intelligence research company led by Fei-Fei Li, in an all-stock transaction valued at approximately $8.2 billion. AMD announced the agreement on September 28, 2026 and expects it to close by the end of 2026, subject to regulatory approvals and customary closing conditions.
The deal brings a frontier-model research organization directly into AMD. World Labs develops models that generate, reconstruct and simulate interactive 3D environments from text, images and video, alongside technology for robotic learning and simulation. After closing, Li is expected to become AMD executive vice president and chief scientist, reporting to CEO Lisa Su, while the World Labs team continues its AI model research.
For AMD, the technical significance is the tighter feedback loop between emerging AI workloads and the hardware, software and systems built to run them. World models place substantial demands on multimodal training, inference, memory movement, 3D representation and simulation. AMD says World Labs' research will help inform its future compute platforms as these workloads evolve.
What AMD is buying
World Labs was founded in 2024 around spatial intelligence: AI systems designed to model and reason about three-dimensional environments and the physical world. Its work now spans generative 3D environments, simulation and robotics-oriented research.
The company's September 2026 Atlas release shows the direction of that research. World Labs describes Atlas as an omni world model pretrained from scratch across text, images, video and 3D. The architecture combines those inputs into a shared spatial context and generates subsequent content while maintaining 3D consistency.
World Labs also operates a commercial product layer. Its Marble system generates explorable 3D worlds, while the World API exposes world generation programmatically for applications and downstream workflows.
That mix of model research and usable spatial-computing products gives AMD direct exposure to workloads that extend beyond conventional language-model inference.
The $8.2 billion transaction
AMD says the acquisition is an all-stock transaction valued at approximately $8.2 billion. The agreement remains subject to regulatory approval and other customary closing conditions, with completion expected by the end of 2026.
Reuters independently reported the same transaction value and closing target. The report also notes that World Labs raised $1 billion earlier in 2026 from investors including AMD, Nvidia and Autodesk.
The acquisition follows an existing technical relationship. AMD and World Labs have already worked together on model training and inference optimization using AMD hardware. Bringing the research team inside AMD turns that collaboration into a direct input to AMD's platform roadmap.
Why spatial AI matters to AMD's hardware roadmap
Spatial-intelligence systems combine several compute patterns that are strategically relevant to accelerator vendors: multimodal model execution, large generative models, high-bandwidth memory access, rendering or reconstruction pipelines, simulation and robotics workloads.
World Labs' Atlas research is particularly useful as an indicator of that workload direction. The model operates across text, image, video and 3D inputs and is designed to scale with additional training compute. Those characteristics give AMD researchers and architects a live model-development environment for studying future compute, memory, interconnect and software requirements.
AMD's stated rationale is similarly architectural. Lisa Su said building next-generation AI compute platforms requires understanding how models are evolving. The World Labs team supplies model-level research expertise alongside AMD's existing CPU, GPU, networking and software engineering organizations.
This creates a potentially shorter path between model bottlenecks and platform design decisions. A model team can identify where compute throughput, memory capacity, data movement, interconnect behavior or software tooling constrains a workload, while AMD can evaluate those requirements against future accelerators and systems.
Fei-Fei Li becomes AMD chief scientist after closing
The personnel component is unusually prominent for a semiconductor acquisition. After the transaction closes, Fei-Fei Li is expected to join AMD as executive vice president and chief scientist and report directly to Lisa Su.
Li is a Stanford computer scientist whose earlier work includes ImageNet, a dataset and benchmark that played a major role in the development of modern computer vision. She co-founded World Labs to pursue spatial intelligence and world models capable of understanding and generating physical environments.
World Labs says joining AMD provides greater engineering depth and closer access to hardware while allowing its team to continue model research.
Where this fits in AMD's AI expansion
AMD has been expanding its AI strategy beyond accelerator silicon into systems, networking, software and specialized intellectual property. World Labs adds a model-research layer to that stack.
The acquisition therefore has a different technical role from a conventional chip or infrastructure purchase. AMD is gaining a research organization whose workloads can help shape future hardware and software requirements while continuing to develop spatial-intelligence models.
The near-term product implications remain tied to the transaction closing and subsequent integration. The durable significance is the organizational link between a major compute-platform designer and a team building large multimodal world models for 3D generation, simulation and robotics.
Bottom line
AMD's proposed $8.2 billion acquisition of World Labs places spatial-intelligence research inside the company at a time when AI infrastructure is expanding from language and image generation toward interactive 3D environments, simulation and robotics.
The deal gives AMD both the World Labs research team and a direct model-development feedback loop for future compute platforms. It also puts Fei-Fei Li in the chief-scientist role after closing. The transaction is expected to complete by the end of 2026, subject to regulatory approvals and customary closing conditions.
Sources
- AMD Newsroom — AMD to Acquire World Labs to Advance the Future of AI Compute: https://newsroom.amd.com/news/amd-acquire-world-labs/
- World Labs — World Labs is Joining AMD: https://www.worldlabs.ai/blog/world-labs-is-joining-amd
- World Labs — Atlas: A World Model for Spatial Intelligence: https://www.worldlabs.ai/blog/atlas
- Reuters — AMD to buy Fei-Fei Li's World Labs in $8.2 billion bet on physical AI: https://www.reuters.com/technology/amd-acquire-fei-fei-lis-world-labs-82-billion-deal-2026-09-28/