World Labs company profile: valuation, funding & revenue

This World Labs company profile covers the company's valuation, revenue, funding rounds, investors and history. World Labs works in artificial intelligence, with a focus on spatial intelligence and 3D world models. It was founded in 2024 and is headquartered in San Francisco, California, United States. Its latest valuation on record is About US$8.2 billion (AMD all-stock acquisition agreement) (September 2026). Every figure links to a public source, and anything the company hasn't disclosed is marked as such.

Revenue

Funding history

World Labs timeline

Key dates from the sources on this page, oldest first. Metric snapshots are listed after the funding events.

• 2024 — World Labs is founded in San Francisco by Fei-Fei Li (CEO), Justin Johnson, Christoph Lassner and Ben Mildenhall.
• Sep 2024 — Equity round: US$230M, valuing the company at About US$1B. Investors: Andreessen Horowitz, NEA, Radical Ventures.
• Feb 2026 — Equity round: US$1B. Investors: AMD, Nvidia, Autodesk, Emerson Collective, Fidelity, Sea.
• September 2024 — Debut valuation: About US$1 billion after a US$230 million financing.
• January 2026 — Reported 2026 valuation talks: Talks at a valuation of about US$5 billion reported in January 2026; the valuation of the closed US$1 billion round wasn't disclosed.

Investors

Overview

World Labs builds AI models with "spatial intelligence": the ability to perceive, generate and reason about three-dimensional worlds. Rather than producing flat images or video clips, its models create explorable 3D environments from text prompts, images or video, which users can walk through, edit and export to game engines and design tools. The company's first commercial product, Marble, lets creators, game developers, filmmakers, architects and robotics researchers generate persistent 3D worlds. World Labs is one of the leading companies in the race to build world models, which many researchers see as the next frontier after language models.

Fei-Fei Li, the CEO, co-founded World Labs in 2024 with Justin Johnson, Christoph Lassner and Ben Mildenhall. Li is a Stanford professor and one of the most influential figures in AI: she created ImageNet, the dataset whose 2012 competition sparked the deep learning revolution, and co-directs Stanford's Institute for Human-Centered AI. Johnson is a computer vision researcher, Lassner worked on 3D graphics and rendering, and Mildenhall co-created neural radiance fields (NeRF), a breakthrough technique for reconstructing 3D scenes from photos. The company is based in San Francisco.

What spatial intelligence means

Language models understand and generate text, but the physical world is three-dimensional, governed by geometry, physics and the relationships between objects. Li has argued that true intelligence requires understanding space: knowing where things are, how they relate, and how they might move or be manipulated. Models with spatial intelligence could power realistic virtual worlds for games and entertainment, help architects and designers explore spaces, generate simulated environments for training robots and autonomous vehicles, and support augmented and virtual reality.

Products

• Marble: a world generation product that turns text, images, panoramas or video into navigable 3D worlds that can be edited and expanded.
• Export and integration: outputs in formats such as Gaussian splats and meshes that work with game engines, 3D tools and VR headsets.
• Editing tools: ways to change layouts, styles and objects within generated worlds.
• Research models: ongoing work on larger world models that support interaction and physical reasoning.

Funding

World Labs emerged from stealth in September 2024 with US$230 million of funding that valued it at about US$1 billion, from investors including Andreessen Horowitz, NEA and Radical Ventures. In January 2026, Bloomberg reported it was in talks to raise at a valuation of about US$5 billion. In February 2026, Reuters reported that it had raised US$1 billion from investors including AMD, Nvidia, Autodesk, Emerson Collective, Fidelity and Sea, but the company didn't disclose a valuation. Because the closed round's valuation wasn't announced, we show valuation as not disclosed and list the 2024 figure and the reported 2026 talks separately.

Business model

World Labs sells subscriptions to Marble for individual creators and teams, with enterprise and partnership arrangements for studios, design software companies and robotics firms. Autodesk's investment suggests interest in bringing world generation into design and engineering software. The company hasn't disclosed revenue, so we show it as not disclosed.

Competition

World Labs competes with video and world model developers such as Luma AI and Runway, Google DeepMind's Genie world models, Nvidia's Cosmos platform for physical AI, and startups building 3D generation tools. Robotics foundation model companies such as Physical Intelligence may use world models as simulation tools. World Labs' advantages are its founders' standing in computer vision and 3D, and its focus on persistent, explorable 3D worlds rather than video.

World Labs at a glance

World Labs is a San Francisco AI company founded in 2024 by Fei-Fei Li, Justin Johnson, Christoph Lassner and Ben Mildenhall. It builds spatial intelligence models and the Marble 3D world generator. It debuted with US$230 million at about US$1 billion in 2024 and raised US$1 billion in February 2026 from AMD, Nvidia, Autodesk and others, without disclosing its valuation. In September 2026 AMD agreed to acquire it in an all-stock deal valued at about US$8.2 billion.

Timeline

• 2024: founded; emerges in September with US$230 million at about US$1 billion.
• Late 2024 and 2025: previews research on generating 3D worlds from images.
• 2025: launches Marble publicly.
• January 2026: reported in talks at about US$5 billion.
• February 2026: raises US$1 billion from AMD, Nvidia, Autodesk and other investors.
• September 2026: AMD agrees to acquire World Labs for about US$8.2 billion in stock.

World models and robotics

Training robots in the real world is slow, expensive and sometimes dangerous. Simulated environments let robots practise millions of times, but traditional simulators require artists and engineers to build each scene by hand. World models that can generate realistic, varied 3D environments on demand could dramatically speed up robot training and testing, along with autonomous vehicle development. This potential connects World Labs to the broader physical AI boom covered in our guide to humanoid robot companies.

Gaussian splats and 3D formats

Many new 3D AI tools output Gaussian splats, a way of representing scenes as millions of small, coloured, semi-transparent blobs that can be rendered quickly and look photorealistic. This technique, which builds on NeRF research, has become popular because it is fast to render on ordinary hardware. Supporting standard formats such as meshes alongside splats lets World Labs' output fit into existing creative pipelines.

Questions for creative and technical teams

• How detailed and consistent are generated worlds across large areas?
• Which export formats and tools does your pipeline need?
• What rights do you have to use generated worlds commercially?

Latest developments (October 2026)

World Labs has agreed to be acquired by AMD. AMD announced on 28 September 2026 an all-stock deal valued at about US$8.2 billion, expected to close by the end of 2026 subject to regulatory approval, according to SDxCentral and The Next Platform. Co-founder Fei-Fei Li will become AMD's executive vice president and chief scientist. The deal follows a US$1 billion round in February 2026. Once it closes, World Labs will no longer be an independent private company. See our new AI labs guide.

What to watch

Watch Marble's adoption by game studios and designers, new world models with interaction and physics, partnerships with Autodesk and chip makers, and any disclosure of valuation in future rounds.

Takeaways

Creators and designers can generate explorable 3D spaces quickly. Founders can see how a renowned research leader attracts strategic investors. Investors should note that the 2026 round's valuation hasn't been disclosed.

Sources

Every figure on this page comes from the public sources below. Where sources disagreed, we used the company's own announcement or a primary news report. Private-company data changes; check the date beside each figure.

1. The Next Platform
2. Yahoo Finance (Reuters)
3. Bloomberg