An independent view on technology · Monday, 12 October 2026 Telegram channel · @aihubaz · Baku · Azerbaijan
AIHUB.az
Artificial intelligence. Public procurement. Local context.
Our projectAZTender AI ↗
AI world

AMD to acquire Fei-Fei Li's World Labs for $8.2 billion in physical AI push

AMD said on 28 September 2026 it will acquire physical-AI startup World Labs, co-founded by Fei-Fei Li, in an all-stock deal worth $8.2 billion; Li will join AMD as executive vice president and chief scientist.

AMD said on 28 September 2026 that it will acquire World Labs, a physical-AI startup co-founded by Fei-Fei Li, in an all-stock deal worth $8.2 billion, the two companies said, according to TechCrunch and Fortune. World Labs co-founder Fei-Fei Li will join AMD as executive vice president and chief scientist, reporting to chief executive Lisa Su, Fortune reported.

What the deal covers

  • World Labs, which Li co-founded in 2024, builds deep learning "world models" intended to give AI a grounded understanding of physical reality, according to TechCrunch.
  • AMD chief executive Lisa Su said in a post on X that the companies would "combine World Labs' deep expertise in AI and world models with AMD's compute leadership to power the future of AI and strengthen the open AI ecosystem," Fortune reported.
  • Li wrote in a post that without a focused hardware effort, AI "remains trapped in the digital world," according to Fortune.
  • AMD and World Labs formed an inference-optimisation and training partnership last year, and ties have remained close, TechCrunch reported.
  • The acquisition is expected to close before the end of 2026, subject to regulatory approval, according to TechCrunch.
  • World Labs' first product, Marble, is pitched as a tool for creating entertainment experiences and simulated environments for robot training, TechCrunch said.

Why it matters

The deal gives AMD its own world-model technology to compete with Nvidia, which already offers open-weight world models such as Cosmos, as chipmakers position themselves for a coming wave of AI-controlled robots and autonomous vehicles that need synthetic training data grounded in physical reality, according to TechCrunch. For companies and investors in the Caspian region tracking the AI hardware race, the deal signals that access to compute is increasingly being bundled with the software models designed to run on it, a dynamic that will shape which vendors can offer viable alternatives to Nvidia's dominant ecosystem.

Other languages: AZ · RU

Don't miss a story

New pieces go out on our LinkedIn page and via RSS. Tips and corrections are welcome by email.