Fei-Fei Li’s AI Startup World Labs in Funding Talks at $5 Billion Valuation
- ByStartupStory | January 23, 2026
Fei-Fei Li’s World Labs Eyes $5B Valuation in $500M Funding Talks
AI pioneer Fei-Fei Li is in advanced discussions with investors to raise hundreds of millions for World Labs, her startup pioneering “large world models” that enable AI to navigate and make decisions in 3D environments, potentially valuing the company at $5 billion – a massive leap from its $1 billion mark set during a $230 million raise in 2024.
The anticipated round would inject about $500 million onto World Labs’ balance sheet, building on backing from Andreessen Horowitz, NEA, Radical Ventures, Nvidia’s venture arm, Sanabil Investments, Temasek Holdings, and luminaries like Google DeepMind’s Jeff Dean, actor Ashton Kutcher, and machine learning godfather Geoffrey Hinton. While terms remain fluid and unconfirmed, Li’s hand-picked investor strategy – as she shared in a recent interview – underscores her vision for spatial intelligence beyond text-based LLMs powering ChatGPT.
From ImageNet to 3D World Models
Known as the “godmother of AI” for spearheading the 2006 ImageNet project – a 15-million-image visual database that taught computers human-like vision – Li now co-directs Stanford’s Human-Centered AI Institute while driving World Labs’ flagship Marble product. Launched late last year, Marble generates interactive 3D worlds from text or image prompts, positioning the startup at the frontier of investor frenzy around multimodal AI surpassing language models.
Rivalry in the Spatial AI Race
World Labs joins Yann LeCun’s AMI Labs, which recently drew $3.5 billion valuation interest from Cathay Innovation, signaling VC bets on world models as the next AI paradigm for robotics, gaming, and autonomous systems. With deal details pending finalization, Li’s fundraising talks prove visionaries commanding premium multiples – when ImageNet unlocked computer vision, $500 million fuels the 3D reasoning stack transforming AI from chatbots to world-understanding agents.




