Fei-Fei Li’s World Labs Secures $1 Billion for Physical AI Vision

• $1 Billion Funding Round (February 18, 2026)

  World Labs closed a $1 billion round on February 18, 2026, backed by Autodesk, AMD, Nvidia, Emerson Collective, Fidelity, and Sea. Autodesk contributed $200 million. The company did not disclose its valuation, though reports in January indicated funding discussions around $5 billion. The startup previously raised $230 million in September 2024 at a $1 billion valuation.

• Autodesk Partnership Targets Commercial 3D Workflows

  Autodesk, a global leader in 3D CAD software used in architecture, engineering, manufacturing, and media, will collaborate with World Labs at the research and model level and serve as an adviser. The first focus is entertainment. Autodesk Chief Scientist Daron Green described workflows where users begin with a world-model layout in World Labs and refine detailed components inside Autodesk tools.

• Marble: Turning Prompts Into 3D Worlds (November 2025)

  World Labs launched Marble in November 2025 as its first commercial product. Marble generates high-fidelity, persistent, and editable 3D environments from text, images, or video. Unlike traditional image models, it produces explorable spatial worlds.

• World API Expands Platform Reach (January 21, 2026)

  In January 2026, World Labs introduced the World API, allowing developers to embed Marble-generated 3D worlds into applications, simulations, and web platforms. It supports panoramas, multi-view inputs, and video — signaling a move toward a programmable spatial AI platform.

• What “Spatial Intelligence” Means

  Spatial intelligence refers to AI systems that understand how the 3D world works — not just text or flat images. World Labs’ models can perceive, generate, and interact with environments governed by geometry and physics. Li has linked these capabilities to future use in AR, VR, and robotics.

• Autodesk Neural CAD and Physical AI

  Autodesk is developing “neural CAD,” generative models trained on geometric data that produce functional 3D designs. Combined with world models, this could connect object-level design with full environmental context. Li describes this direction as “physical AI,” where language, space, and physical reasoning converge.

• Rising Competition in World Models (2025–2026)

  Google DeepMind released Project Genie in January 2026, and Runway introduced its own world model in December 2025. Researchers such as Yann LeCun, through AMI Labs, are also exploring this field. Media and gaming are early use cases, but robotics and scientific research are emerging targets.

Why This Matters

• AI Moves Beyond Text: The $1 billion round signals that spatial AI is moving from research into commercial deployment.
• Enterprise Integration: Autodesk brings world models into professional design and production pipelines.
• Platform Strategy: The World API positions 3D world generation as infrastructure, not just content creation.
• AR, VR, Robotics Potential: Spatial intelligence supports AI systems that must reason about physics and space.
• Hardware Alignment: Backing from AMD and Nvidia highlights the computing demands of large-scale 3D AI.
• Valuation Momentum: A reported $5 billion target would place World Labs among the most valuable spatial AI startups.
• Next Phase of AI: Combining language models, neural CAD, and world models may redefine how machines simulate and interact with the physical world.

This article was drafted with the assistance of generative AI. All facts and details were reviewed and confirmed by an editor prior to publication.