World Labs builds world models, AI systems designed to understand how objects and spaces behave, allowing them to simulate what might happen when something moves or an action is taken.
Chip maker Advanced Micro Devices (AMD) is spending $8.2 billion (€7bn) on a company that builds virtual worlds, entering a field where its rival Nvidia is developing AI models to train robots.
ADVERTISEMENT
ADVERTISEMENT
AMD said this week it had agreed to buy World Labs, co-founded by Fei-Fei Li, one of the most influential researchers in artificial intelligence (AI).
Li, a Stanford University computer scientist sometimes called the “godmother of AI,” will become AMD’s executive vice president and chief scientist when the acquisition closes, according to a statement from Li.
Li is best known for establishing ImageNet, a vast collection of labelled images that helped drive advances in computer vision.
She co-founded World Labs in 2024 to develop what she calls “spatial intelligence” — AI that can understand and work with three-dimensional spaces.
What are world models?
World Labs builds world models, AI systems that can understand how objects and spaces behave, allowing them to generate interactive 3D environments from text, images and video.
One potential use is creating virtual environments where robots can practise tasks before attempting them in the physical world. World Labs is researching that application, although accurately simulating real-world behaviour remains a challenge.
Other companies are working on world models, though their approaches differ. Google DeepMind’s Genie 3 creates virtual environments users can explore.
Meta’s V-JEPA 2 research focuses on understanding video, predicting outcomes and planning actions. Nvidia has developed its Cosmos models for applications including robotics.
Why is AMD buying World Labs?
Training and running world models requires computing power. AMD and World Labs began working together last year to train and run the company’s models on AMD chips.
AMD says the acquisition will help it understand what emerging AI systems need and develop future hardware and software around those demands.
“Building the compute platforms for the next generation of AI requires a deep understanding of how models are evolving,” AMD chief executive Lisa Su said in the company’s announcement.
“Now that we have tangible proof of the possibilities, we want to do everything we can to accelerate the future,” Li wrote in a blog post.
“To do this requires scaling our efforts, widening our reach, and getting closer to the hardware.”
