Spatial intelligence — AI that understands and builds the 3D world
HQ: San Francisco, CA, United States | Founded: 2024 | Employees: ~100+ (est.) | Stage: Late Stage VC ($1B+ new round Feb 2026) | Website: https://www.worldlabs.ai
World Labs was founded in 2024 by Fei-Fei Li, the Stanford professor whose ImageNet dataset catalyzed the modern deep-learning era and who is often called 'the godmother of AI,' alongside three co-founders who are leaders in 3D machine learning: Justin Johnson (University of Michigan professor; generative visual modeling), Ben Mildenhall (inventor of NeRF, the breakthrough technique for photorealistic 3D reconstruction from images, ex-Google), and Christoph Lassner (rendering and Gaussian-splatting pioneer, ex-Amazon, Epic Games). The thesis: language and 2D image models are only part of intelligence — real understanding requires models that perceive, generate, and reason about 3D worlds.
World Labs' models generate persistent, explorable 3D environments from a single image or text prompt — not flat images or short videos, but worlds you can navigate in a browser, export to standard 3D formats, and use for games, VFX, simulation, and robotics training. Its first commercial product, Marble, launched November 12, 2025 (TechCrunch), converting photos and prompts into explorable 3D worlds and downloadable meshes.
Capital followed conviction: a $230M debut round (September 2024) from Radical Ventures, NEA, and angels including Andrej Karpathy and Geoffrey Hinton; a reported raising at a ~$5B valuation (Bloomberg, early 2026); then a $1B round announced February 18, 2026 — bringing total capital to $1.23B — with strategic investors AMD, Autodesk, Nvidia, and Salesforce Ventures joining, each a potential distribution channel (chips, 3D design software, robotics, enterprise software).
World Labs is the highest-profile bet that the next AI frontier is 3D: large world models that perceive, generate, and let you interact with persistent, explorable 3D environments. Founded by Fei-Fei Li — whose ImageNet catalyzed deep learning — with the inventors of key 3D deep learning technology (NeRF's Ben Mildenhall, GAN-era visual synthesis leader Justin Johnson, and rendering veteran Christoph Lassner), it shipped its first commercial product (Marble) in November 2025 and raised $1B more in February 2026 with strategic investors (AMD, Autodesk, Nvidia, Salesforce Ventures) who are also distribution partners. The bull case: if AI is to act in the physical world (robots, AR, industrial design), it needs spatial understanding that 2D models lack. The bear case: 3D generation is becoming commoditized quickly, and the 'spatial intelligence' category remains undefined commercially.
Category: AI & Machine Learning Startups
Browse: Home | Blog | About | All Startups A-Z | AI & Machine Learning Startups | View full profile on StartupWiki