World Labs, the spatial-intelligence company co-founded by Stanford professor Fei-Fei Li, raised over $2 billion at a valuation reported above $10 billion, according to Reuters, to build models that reason about three-dimensional space rather than language alone.

The bet is that the next meaningful category in artificial intelligence is not a larger language model but a model that understands physical environments, a category adjacent to the language race and arguably a precondition for robotics.

Why the founder matters

Fei-Fei Li's work on ImageNet effectively created the data regime that defined the last decade of computer vision. Investors are funding the assumption that she has identified the next one.

The technical risk

Spatial intelligence is a harder problem than language in a specific way: the training data does not exist at internet scale. The companies that solve it will be the ones that generate the data, not the ones that train on it, and that is a slower, more capital-intensive path than the language labs have walked.