AIToday
AI Business & IndustryFortune AIPublished: Sep 29, 2026, 10:01 JST

AMD buys Fei-Fei Li's World Labs for $8.2 billion

AMD buys Fei-Fei Li's World Labs for $8.2 billion

3 Key Points

  1. What happened

    AMD is acquiring World Labs, the two-year-old physical AI startup cofounded by Fei-Fei Li, in an $8.2 billion all-stock deal. World Labs' team will join AMD, and Li will become executive VP and chief scientist reporting to CEO Lisa Su.

  2. Why it matters

    Buying World Labs gives AMD technology in physical AI, the field aimed at powering robots and self-driving cars, and the acquisition is expected to close by the end of the year pending regulatory approvals.

  3. What to watch

    The deal still needs regulatory approvals to close, so the timing hinges on that review. Lisa Su said the combination would pair World Labs' world models with AMD's compute to strengthen the open AI ecosystem.

WHO IT HITSAMD's chip designers and product planners gain world-model technology for AI aimed at robots and self-driving cars, while World Labs staff move under a new chief scientist in Fei-Fei Li. Competitors building similar physical-AI capabilities may face a better-resourced rival in AMD.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

AMD's move is framed by its CEO as a way to broaden where the company's GPU chips are used. Lisa Su said the deal would combine World Labs' expertise in world models with AMD's compute to power the future of AI and strengthen the open AI ecosystem. World Labs was founded only two years ago and raised roughly $1 billion in funding, making it one of several richly funded startups betting that world models can extend AI beyond text.

That bet rests on a contrast Fei-Fei Li drew in her own post: language alone is not enough. World models are trained on data such as video showing the physical, three-dimensional world, and the goal is software that understands how objects move—for instance how a ball bounces, Li has said. She argued that without a focused hardware effort, AI remains limited in efficiency and scale and trapped in the digital world. That framing helps explain why AMD, which makes the GPU chips central to training and running AI models, is buying into the approach.

The acquisition also brings Li, known as the "godmother of AI" for creating the ImageNet visual database two decades ago, into an executive role alongside Su. World Labs and AMD had already worked together under a "deep technical partnership" started last year on model training and inference optimizations for AMD's GPU chips, and both sides describe the combination as a natural extension of that work. The deal is expected to close by the end of the year subject to regulatory approvals.

FAQ
How much is AMD paying for World Labs?
AMD is paying $8.2 billion in an all-stock deal, and World Labs' team will join AMD as part of it.
What will Fei-Fei Li's role be at AMD?
Li will become executive VP and chief scientist at AMD, reporting to CEO Lisa Su.
What are world models?
World models are trained on data such as video that depicts the physical, three-dimensional world, aiming to understand how objects move—like how a ball bounces—rather than relying on text alone.

Get the latest AI Business & Industry news every morning

For example, today's edition would include:

  • Anthropic ships Claude Sonnet 5.5, 30%+ faster at same priceITmedia AI+ · 26m ago
  • Nvidia unveils Open Agent Safety PlatformITmedia AI+ · 26m ago
  • Nvidia bets $150 billion on AI buildout and buybacksSemafor Tech · 26m ago

AI-summarized, only the topics you pick: one digest a day via Email, LINE, or Slack.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleFeature Store Now Essential for Enterprise ML in 2026