Fei-Fei Li’s artificial intelligence startup, World Labs, has raised $1 billion in a fresh funding round as it intensifies its work on advancing what it calls “spatial intelligence.” The funding marks a significant milestone for the young company as it scales its efforts to build AI systems capable of understanding and interacting with the three-dimensional world.
The round drew backing from a mix of technology leaders and investment firms. Investors include chipmakers Advanced Micro Devices and Nvidia, software company Autodesk, Emerson Collective, Fidelity Management & Research Company, and Sea, among others. Autodesk committed $200 million as part of the funding and will also serve in an advisory capacity to the startup, according to the design software maker.
Spatial intelligence represents a next-generation approach to AI. Unlike traditional systems that primarily process two-dimensional inputs such as flat images or text, spatial intelligence focuses on reasoning about how the 3D world functions. World Labs is developing foundational models that can perceive, generate, and interact within three-dimensional environments, opening new possibilities for immersive and interactive technologies.
Widely known as the “godmother of AI,” Li previously raised $230 million in September 2024 to launch World Labs. Her broader vision is that spatial intelligence models could eventually power applications in augmented reality, virtual reality, and robotics, as well as scientific discovery.
World Labs is among a growing group of AI developers building so-called “world models.” These systems aim to process visual data from the physical environment and use it to develop advanced reasoning capabilities. Google DeepMind is working on similar technologies, including its Genie family of models, which can generate and simulate 3D environments.
Meanwhile, Yann LeCun, formerly Meta’s chief AI scientist, has left the Facebook parent company to launch his own startup, AMI Labs, focused on developing AI systems based on world models.
Late last year, World Labs introduced its first product, Marble, described as a multimodal world model capable of creating 3D worlds from image or text prompts. The company has said Marble can generate immersive digital environments by interpreting visual or textual instructions, reinforcing its commitment to building AI that understands space and context rather than just pixels and words.
Li’s earlier contributions to the AI field laid important groundwork for many of today’s breakthroughs. She was part of the team behind ImageNet, an academic project that compiled millions of labeled images and significantly advanced computer vision by improving how machines recognize objects.
With the new $1 billion infusion, World Labs is expected to accelerate research and product development, particularly in robotics and other domains that rely on machines being able to interpret and act within real-world, three-dimensional settings.
