AMD is acquiring AI research company World Labs for approximately $8.2 billion. With this acquisition, the chipmaker not only gains expertise in AI models but also brings founder Fei-Fei Li on board. The acquisition is intended to give AMD greater insight into the hardware requirements of future AI applications.
World Labs specializes in so-called spatial intelligence: AI models that can generate and simulate interactive 3D environments based on text, images, and video. Among other things, the technology can be used to train and simulate robots.
From model to chip
For AMD, the knowledge behind these models is particularly interesting. The demands that AI places on processors and other infrastructure are changing rapidly. While generative AI initially focused primarily on training and running large language models, part of the development is shifting toward reasoning, robotics, and systems that must interact with a physical environment.
By bringing in-house researchers who develop such models, AMD can identify earlier on what computing capacity, memory architecture, and software future workloads will require. That knowledge can then be incorporated into the design of new GPUs, CPUs, and complete AI systems. This trend has been evident for some time: AMD previously acquired AI chip developer Taalas, which integrates models more closely with the underlying hardware architecture
The price AMD is paying for World Labs is particularly striking. Just seven months ago, the chipmaker participated in a $1 billion funding round that valued World Labs at $5 billion. Nvidia also invested in the company at that time, according to SiliconANGLE. Now, AMD is putting up $8.2 billion in stock to fully acquire World Labs.
Virtual factories for robots
Among other things, World Labs is developing Atlas, a model that can create a virtual version of, for example, a factory or warehouse based on just a few photos. Moving objects, such as conveyor belts, can also be simulated. The generated environments also contain information about depth and the position of objects.
This is particularly useful for robotics. AI models that control robots can first learn how to perform certain tasks in a simulated environment. Through repetition, they can try out different actions without the constant need for physical robots and test environments.
This application aligns with AMD’s existing activities. With the acquisition of Xilinx in 2022, the company gained access to FPGA technology used in industrial systems and robots, among other applications. AMD has also introduced the Ryzen AI Embedded X100 series, which combines CPU, GPU, and AI computing power.
Fei-Fei Li joins AMD
A key part of the deal is the arrival of Fei-Fei Li. She is best known as the founder of ImageNet, the image recognition dataset that played a crucial role in the development of computer vision and early deep learning models such as AlexNet.
Li will become executive vice president and chief scientist at AMD and will report directly to CEO Lisa Su. World Labs will continue to conduct research on AI models after the acquisition is completed.
The transaction will be conducted entirely in stock. AMD expects to complete the acquisition by the end of 2026, subject to regulatory approval. With World Labs, the company is not only acquiring new AI technology but, more importantly, expertise that will enable future hardware to be designed more closely aligned with the requirements of AI models.