Fei-Fei Li's World Labs raises $1B to develop spatial intelligence AI systems

World Labs, the AI startup founded by computer vision pioneer Fei-Fei Li, has secured $1 billion in new funding to advance spatial intelligence—AI systems designed to understand and navigate three-dimensional physical environments.

Funding and Strategy

The funding round targets the development of world models, a category of AI architecture that learns representations of physical space and how objects interact within it. According to the company, this approach differs from traditional language-based or vision-only AI by building systems that can reason about spatial relationships, physics, and real-world dynamics.

Fei-Fei Li, who previously led AI research at Stanford University and served as VP of AI/ML at Google Cloud, founded World Labs to focus specifically on enabling AI systems to perceive and interact with the physical world. The company views spatial intelligence as foundational infrastructure for robotics, autonomous systems, and embodied AI applications.

Market Context

The $1 billion raise positions World Labs among the best-funded AI startups in the post-2023 funding environment, competing with other well-capitalized ventures building foundation models and AI infrastructure. The investment reflects increased attention from venture capital and strategic investors toward specialized AI capabilities beyond language models, particularly those addressing robotics and physical world understanding.

World models—systems trained to predict and understand three-dimensional space—represent a distinct technical approach from the transformer-based architectures that dominate current large language models. The architecture aims to solve specific challenges in robotics, autonomous vehicles, and industrial automation where understanding spatial relationships is critical.

What This Means

World Labs' $1 billion funding demonstrates sustained investor confidence in AI startups addressing narrow but technically challenging problems. Spatial intelligence remains largely underdeveloped compared to language and vision capabilities in current AI systems. If World Labs can deliver effective world models, the technology could become essential infrastructure for any application requiring AI to physically interact with environments—a market that spans robotics, autonomous systems, and industrial automation.

The funding also signals that large capital raises continue flowing to founders with deep technical credentials and proven track records, even as broader venture funding in AI has become more selective.