The New Frontier of Spatial Intelligence and the end of Text-Only

With a billion-dollar investment, pioneer Fei Fei Li reveals the foundation of Large World Models and explains why true innovation requires machines to understand the laws of physics.

During a revealing interview for Bloomberg Live with journalist Emily, scientist and World Labs co-founder Fei Fei Li presented the next definitive leap in global technology. While public and market attention remains fixated on language models that merely generate words on a screen, her new company raised a billion dollars to build a much more ambitious concept. Her team's main goal now is to develop and master spatial intelligence.

The Weight of Biological Evolution

The expert explained that limiting innovation to text creation ignores our own biological foundation. Animal intelligence began five hundred million years ago, driven by the ability to see and move in physical space. Since most human work and life require perceiving, reasoning, and interacting with three-dimensional environments, unlocking this same visual perception in machines has become the great scientific priority of this decade.

The True Taxonomy of World Models

To organize the market and steer clear of confusing technical jargon, Fei Fei Li established an official taxonomy for the so-called Large World Models, dividing the technology into three fundamental layers.

The first layer is the Renderer. This is a system designed to enchant human eyes, generating the impressive videos and pixels that dominate social media today. Although visually flawless, it still has no real commitment to geometry or object physics.

The second layer is the Planner. This is a logical brain built directly for the robotics sector. It receives current environmental data and calculates the exact next physical action a machine must perform in real life.

The major game-changer of the entire operation is the third layer, dubbed the Simulator. This central pillar strictly respects structure, laws of physics, and world dynamics in three and four dimensions. It is exactly this deep structural foundation that will enable unparalleled advances in industrial design and the creation of perfect virtual environments, serving both humans and robots.