AMD has agreed to acquire artificial intelligence startup World Labs in an US$8.2 billion (A$11.7 billion) all-stock deal, giving the chipmaker access to a prominent team of AI researchers led by Stanford computer scientist Fei-Fei Li.

The transaction is expected to close by the end of the year, subject to regulatory approval.

The acquisition marks another major push by AMD to strengthen its position in artificial intelligence as it competes with Nvidia in the lucrative market for AI accelerators and data centre infrastructure.

World Labs specialises in so-called world models, AI systems designed to understand and generate three-dimensional environments. The technology could eventually be used in areas including robotics, scientific research and industrial equipment, where AI systems need to interact with or reason about the physical world.

For AMD, acquiring the startup provides more than another set of AI models. The company will also gain researchers who can help it understand how increasingly sophisticated AI systems are being developed, potentially influencing the design of future processors and computing platforms.

“The more you understand end to end, the better system you are going to build,” AMD CEO Lisa Su told Bloomberg Television.

“That’s why we are acquiring World Labs,” Su said, pointing to the opportunity to combine the startup’s research talent with AMD’s hardware, software and systems expertise.

Li, who has served as World Labs’ CEO, will join AMD as executive vice president and chief scientist once the deal is completed. She will report directly to Su.

Li said World Labs’ ambitions have increased as the company has grown, bringing with them a greater need for computing resources.

Joining AMD could create a closer relationship between the development of AI software and the hardware required to run it, she said, helping accelerate improvements on both sides.

World Labs was founded in 2024 and employs around 70 people. Its workforce also includes robotics specialists who joined following its acquisition of SceniX in July. SceniX develops software for training robots using realistic virtual environments.

The acquisition should also help AMD broaden the AI products it can offer customers. The chipmaker has been expanding beyond individual processors as it seeks a larger share of spending by data centre operators, which have become its biggest source of revenue.

AMD’s strategy increasingly involves supplying more of the technology needed to build and operate AI infrastructure rather than simply selling the accelerators used to train and run models.

Su said AMD expects to contribute to both open and proprietary AI models as those efforts expand.

“Our ambition is to define the future of AI compute,” she said.

The approach puts AMD more directly in competition with Nvidia, which has built its dominance partly by combining AI chips with networking equipment, software and other services that allow customers to deploy AI infrastructure more quickly.

Nvidia has also been expanding its AI capabilities through acquisitions, recently agreeing to buy AI startup Hugging Face in a deal reportedly worth around US$13 billion.

AMD, meanwhile, has undergone a major transformation during Su’s 12 years as CEO. The company has gained substantial ground against Intel in traditional processors and is now attempting to replicate that success in AI computing.

Nvidia remains comfortably ahead in AI accelerators, but AMD has introduced a growing range of products that it says can compete with, and in some workloads outperform, Nvidia’s hardware.

Investor enthusiasm around AMD’s AI prospects has sent its shares up nearly threefold this year, lifting the company’s market value to around US$1 trillion.

Nvidia remains significantly larger, with a market capitalisation above US$5 trillion. Its shares have gained around 22 per cent this year, however, leaving the company behind many other major chipmakers in terms of share-price growth during 2026.

By bringing World Labs’ AI researchers and models in-house, AMD is betting that closer integration between cutting-edge AI research and semiconductor development can help it narrow Nvidia’s lead — and give the company a bigger role in determining what the next generation of AI computing looks like.