World Labs, led by Fei-Fei Li, has raised $1 billion in a new funding round from a16z, Nvidia, AMD and other backers to accelerate development of world models focused on robotics and scientific discovery.

This round follows Li’s previous funding of about $230 million in September 2024 to launch World Labs, signaling a continued, aggressive investment trajectory in spatial intelligence.

The investment aims to scale large, general‑purpose models that can be applied to robotics and scientific research tasks, tying research to practical, real‑world use cases.

World Labs’ Marble multimodal world model is designed to generate 3D worlds from image or text prompts, aligning with similar efforts from Google DeepMind and AMI Labs in the world-model space.

Investors are backing a vision to build foundational and world models capable of processing visual data from physical environments to enable advanced reasoning.

The round reflects a broader push in the AI community toward embodied and multi‑domain intelligence, driven by major tech firms seeking scalable capabilities.

Participation from major tech and venture capital firms underscores strong industry interest in world models for autonomous systems and scientific applications.

World Labs is positioned as a bridge between AI research and practical robotics and discovery-oriented use cases, signaling a strategic role in deploying AI to real-world tasks.

Spatial intelligence is defined as AI that reasons about how the 3D world works, leveraging 3D perception, generation, and interaction rather than relying solely on 2D data.

No further financial terms or deployment timelines were disclosed beyond the $1 billion raise.

The company aims to build world models that process environmental visuals for advanced reasoning, with potential applications in augmented/virtual reality and robotics.

Autodesk contributed $200 million and will serve as an adviser, with other investors including AMD, Nvidia, Fidelity, Emerson Collective, and Sea.