The 'godmother of AI' just dropped what could be generative AI's next big thing. Dr. Fei-Fei Li's World Labs launched Marble this week, a platform that builds downloadable 3D worlds from simple text prompts - backed by $230 million and the bold claim that spatial intelligence will define the next decade of AI development.
AI just got a new dimension, literally. World Labs launched Marble this week, its first commercial product that generates explorable 3D worlds from simple text, image, or video prompts. It's the debut offering from Dr. Fei-Fei Li's heavily funded startup, which raised $230 million last fall on the promise that spatial intelligence represents AI's next evolutionary leap.
The Stanford AI pioneer, often called the 'godmother of AI,' isn't making small bets. Li believes spatial intelligence will be the 'defining challenge of the next decade,' as she wrote in a Substack post this week. After watching generative AI sprint through chatbots, images, voice, and video, she's staking out 3D world generation as the final frontier.
'We see that world model is just as big and exciting, if not more than the previous eras,' Li told The Verge in an interview. World Labs defines these world models as generative AI that can 'perceive, generate, reason, and interact with the 3D world.'
Marble's pricing structure reveals ambitious commercial targets. The platform offers four tiers: Free (4 world generations), Standard ($20/month for 12 generations), Pro ($35/month for 25 generations plus commercial rights), and Max ($95/month for 75 generations). The paid tiers export files compatible with industry-standard tools like Unreal Engine and Unity, directly targeting game developers and filmmakers.
Early user tests show both promise and limitations. The Verge generated an 'open-air castle with waterfalls' and explored user-created ruined structures and Hobbit-like homes. But the 3D environments hit walls after a few steps - a technical constraint that co-founder Ben Mildenhall acknowledges while noting that dedicated users can still 'stage out fairly large environments.'
'Bringing 3D to life, and understanding the richness of spatial and 3D stuff, is just a whole next level beyond the baseline of most of these other single modes,' Mildenhall told . He's targeting the massive friction in traditional 3D world-building: 'It requires such a large team and so many pieces of software and so much time and effort.'
