Fei-Fei Li's World Labs, backed by $1.2B, unveils Atlas world model
What's the deal? World Labs, the AI startup co-founded by computer vision pioneer Fei-Fei Li, on Tuesday unveiled Atlas, a new world model that generates, reconstructs, and simulates 3D environments from text, image, and video prompts. The company has raised $1.2 billion from backers including Nvidia, AMDDealroom has a profile for this one. Try Dealroom →, and AutodeskDealroom has a profile for this one. Try Dealroom →.
What's the endgame? Atlas is what World Labs calls an "omni model," pretrained from scratch on text, images, video, and 3D. Built as a multimodal autoregressive diffusion transformer, it lets users create a scene and view it from any camera position and angle they specify.
How does it work? Atlas can generate up to one minute of video at 1440p from a single image, with what the company describes as "pixel-perfect camera control." It also reconstructs real scenes from as few as one image, outputting 3D assets that maintain geometric consistency.
What's the endgame, part two? World Labs said Atlas has applications in visual effects and game design, but its real target is robotics. The model supports "Real-to-Sim" workflows, where developers capture a physical space with an ordinary smartphone and rebuild it as a simulation robots can navigate and interact with.
Why it matters: The release advances Li's theory of "spatial intelligence" — the idea that AI cannot replicate the physical world without reasoning natively about 3D objects, environments, and how they interact. Li, former director of StanfordDealroom has a profile for this one. Try Dealroom →'s AI Lab, launched World Labs in February 2024, arguing that artificial general intelligence is impossible without such capabilities.
What could go wrong? World Labs faces growing competition. Atlas competes directly with GoogleDealroom has a profile for this one. Try Dealroom →'s Gemini Omni and overlaps with video generation models such as ByteDance's Seedance 2.5 and Google's Veo. The startup also confronts pressure from China, where numerous startups founded this year are racing to build world models.
In a blind human evaluation of camera-path adherence, World Labs said Atlas was preferred over competitors including Gemini Omni Flash and FLUX. Atlas will power future versions of Marble and other World Labs products.
The signal: World Models have become a distinct front in the AI race, with the biggest players targeting the same terrain. World Labs is betting that spatial reasoning — not just text or video generation — is the missing layer for both robotics and creative tools, and its $1.2 billion war chest gives it room to make that case.
Read more: worldlabs.ai, siliconangle.com, The Information
Image credit: Generated with Gemini
