Fei-Fei Li’s World Labs raises $1 billion to advance spatial intelligence, world models, 3D AI, robotics, creativity and real-world applications. Fei-Fei Li’s World Labs has raised $1 billion to accelerate the development of spatial intelligence and AI world models. The funding will support technologies designed to understand, generate and interact with 3D environments, with potential applications across robotics, storytelling, scientific discovery, design and other real-world fields.
Fei-Fei Li World Labs Raises $1 Billion for Spatial Intelligence
Artificial intelligence is entering a new phase in which understanding language and images may no longer be enough. Fei-Fei Li’s World Labs is betting that the next major leap will come from teaching AI to understand space, objects, movement and the physical relationships that shape the world around us. In February 2026, the company announced that it had raised $1 billion in new funding to accelerate its work in spatial intelligence and world models.
The funding round brings together some of the biggest names in the technology and investment ecosystem, including NVIDIA, AMD, Autodesk, Fidelity Management & Research Company, Emerson Collective and Sea. Autodesk alone invested $200 million and is expected to work with World Labs as an adviser.
For World Labs, the investment is more than another large AI funding round. It represents a major financial vote of confidence in an emerging area of artificial intelligence that aims to connect digital intelligence with the physical world.
The company was co-founded by Li, a prominent computer scientist whose work in computer vision and ImageNet helped shape modern AI. World Labs emerged from stealth in 2024 with $230 million in funding, making its latest raise a significant acceleration of its original ambition.
World Labs Funding Targets the Next Frontier of AI
The central idea behind World Labs funding is spatial intelligence — the ability of AI systems to understand and reason about three-dimensional environments rather than simply processing text, images or isolated pieces of information.
Traditional generative AI has made remarkable progress in producing text, images, audio and video. However, generating something that looks realistic is different from understanding how that environment actually works. A system capable of spatial intelligence needs to recognize where objects are located, how they relate to one another and how changes in the environment could affect what happens next.
World Labs describes its approach through “world models,” which are designed to perceive, generate and interact with three-dimensional environments.
This distinction could become increasingly important as AI moves beyond screens and into physical environments. Robots, autonomous systems, immersive technologies and advanced simulation platforms all need some understanding of space and physical relationships.
The scale of the new investment suggests that investors see this as a potentially important layer of the next AI ecosystem. Forbes reported that more than $3 billion flowed into world-model startups during the first half of 2026, highlighting growing investor interest in AI systems designed to model reality.
For World Labs, the challenge now is to convert that interest into reliable technology with practical applications.
Spatial Intelligence Could Transform How AI Understands the World
Spatial intelligence may sound like a technical extension of computer vision, but its potential reach is considerably broader. Computer vision enables machines to interpret visual information. Spatial intelligence seeks to go further by helping AI understand the structure and behavior of the environments represented by that information.
That difference matters because the physical world operates according to geometry, physics and dynamics. An AI system may be able to identify a chair in an image, for example, but a more advanced system needs to understand that the chair occupies a particular location, has a particular shape and can interact with people and other objects within a room.
This is the broader ambition behind World Labs’ world models.
Fei-Fei Li has argued that AI must eventually understand worlds rather than words if it is to become genuinely useful in real-world settings. World Labs’ approach therefore sits at the intersection of computer vision, generative AI, simulation and physical intelligence.
The implications could extend into robotics, where machines must navigate physical environments, as well as scientific research, where accurate virtual environments can support experimentation and simulation.
There is also a creative dimension. Designers, filmmakers, game developers and other creators could eventually use spatial AI to construct interactive environments instead of producing only fixed images or videos.
That possibility makes spatial intelligence an increasingly important area of competition in the AI industry.
Marble AI Shows What World Labs Is Building
World Labs is not approaching spatial intelligence as a purely theoretical research project. Its first major product, Marble, provides an early look at how the company intends to bring world models into practical use.
Launched in 2025, Marble enables users to generate persistent, high-fidelity 3D environments from inputs such as text, images and video. World Labs describes these environments as spatially cohesive and persistent, allowing users to create worlds that can be explored and developed rather than simply viewed as a conventional AI-generated image.
That distinction is significant. Many generative AI tools produce content designed to be consumed as a finished image or video. Marble is aimed at creating an environment with spatial structure.
The technology could potentially give creators new ways to develop visual concepts, virtual spaces and interactive experiences. It could also provide a foundation for simulations and other applications in which understanding a three-dimensional environment is essential.
World Labs has identified storytelling, creativity, robotics and scientific discovery among the areas that could benefit from its technology.
The commercial importance of Marble is also reflected in the company’s partnership with Autodesk. The two companies plan to explore how World Labs’ world models can work with Autodesk’s design technologies, initially focusing on media and entertainment use cases.
That partnership could eventually connect generative world models with professional design workflows.
Autodesk’s $200 Million Investment Adds Commercial Weight
One of the most notable elements of the World Labs funding round is Autodesk’s $200 million investment. The relationship gives the funding announcement a strategic dimension because Autodesk already operates deeply within industries where spatial understanding is central.
Architecture, engineering, construction, manufacturing and entertainment all depend on digital representations of physical objects and environments. Autodesk’s software is already used to create and manipulate sophisticated 3D designs, making spatial AI a natural area for collaboration.
Under the agreement, Autodesk will serve as an adviser to World Labs, while the two companies explore potential cooperation at the research and model level. TechCrunch reported that the companies are initially looking at media and entertainment applications.
The long-term possibilities could be broader. A future workflow might allow a designer to create an environment through a world model and then move seamlessly into detailed engineering or design software. Conversely, an existing 3D design could potentially be placed inside a larger AI-generated environment for visualization and testing.
Such integration would help bridge the gap between generative AI and professional 3D workflows.
For World Labs, Autodesk’s participation is therefore important not only because of the capital involved, but also because it provides a pathway toward real-world commercial applications.
The investment signals that spatial intelligence is increasingly being considered as infrastructure for industries that already depend on digital models of the physical world.
World Models Could Open New Possibilities for Robotics
Robotics is one of the areas where spatial intelligence could eventually have its biggest impact. Unlike software agents operating inside a digital interface, robots have to function in environments governed by physical constraints.
A robot must understand where objects are, how far away they are, what can move, what cannot move and what might happen when it takes an action. It must also deal with changing conditions rather than relying on a perfectly predictable digital environment.
World models could provide part of the intelligence needed for this challenge.
World Labs has specifically identified robotics as one of the potential applications of its technology. The concept fits into a broader industry movement toward physical AI, where models are designed not only to generate information but also to understand and operate within physical environments.
The technology could potentially be used for simulation before a robot performs an action in the real world. Virtual environments could allow researchers to test navigation, interaction and other behaviors under different conditions.
However, significant technical challenges remain. A generated 3D environment that looks convincing is not necessarily a perfect simulation of reality. Physical systems must deal with uncertainty, imperfect sensors and consequences that cannot simply be corrected with another generated response.
That means spatial AI will need to become increasingly accurate, consistent and physically grounded before it can reliably support demanding robotic applications.
Fei-Fei Li’s Vision Pushes AI Beyond Language
Fei-Fei Li’s role gives World Labs an unusual position in the AI landscape. Her academic work helped establish modern computer vision as a critical component of artificial intelligence, particularly through ImageNet, the influential large-scale image dataset and research project associated with major advances in visual recognition.
World Labs represents a continuation of that broader trajectory: moving from recognizing what is visible to understanding the world in which those objects exist.
The company’s billion-dollar funding round arrives as the AI industry increasingly explores alternatives and complements to large language models. LLMs have demonstrated extraordinary abilities in language, reasoning and coding, but the physical world presents problems that cannot always be solved through language alone.
A world model attempts to address part of that gap by giving AI a representation of space and environments.
This does not necessarily mean that language models will become obsolete. Instead, the future could involve multiple forms of AI working together: language models for communication and reasoning, vision systems for perception, world models for spatial understanding and specialized systems for physical interaction.
That possibility explains why World Labs has attracted such a diverse group of investors.
The company is effectively positioning spatial intelligence as another foundational layer of AI — one that could eventually connect digital reasoning with the physical world.
What the $1 Billion World Labs Investment Means for AI’s Future
The $1 billion investment in World Labs comes at a moment when investors are looking beyond increasingly crowded generative AI markets toward technologies that could define the next stage of artificial intelligence.
World Labs has not disclosed a valuation for the new funding round. Earlier reports suggested that the company had been discussing a valuation of around $5 billion, although that figure was not confirmed by World Labs.
What is clearer is the direction of the company’s strategy.
World Labs wants AI to understand more than language and images. It wants machines to build representations of environments, reason about three-dimensional space and eventually interact with those environments. Its Marble product provides an early demonstration, while the new funding gives the company substantially more resources to expand its research and commercial ambitions.
The road ahead will not be simple. World models require enormous amounts of data, sophisticated computing infrastructure and reliable methods for representing complex physical environments. They must also demonstrate that their outputs are useful beyond impressive visual demonstrations.
Yet the momentum behind spatial intelligence is difficult to ignore. With NVIDIA, AMD, Autodesk and major financial investors backing the sector, world models are becoming an increasingly serious part of the AI conversation.
For Fei-Fei Li and World Labs, the $1 billion raise is therefore not simply a funding milestone. It is a bet that the next generation of artificial intelligence will need to understand the world itself not merely describe it.
Previous article: https://theempiremagazine.com/rezolv/
Magazine Feautures | Podcasts: Empire Global Talks
Follow The Empire Magazine on Facebook | Instagram | LinkedIn | YouTube
The Empire Magazine | Crown for Global Insights
