Our research teams are dedicated to building world models that capture the fundamental structure of the physical world, from the laws of physics governing motion and interaction to the complex dynamics of real environments. We believe that truly intelligent systems must develop rich internal representations of how the world works, learning not just statistical patterns but the causal mechanisms and physical laws that underlie observable phenomena. By grounding AI in the principles that govern our universe, we aim to create models that can reason about unseen situations, predict future states with physical plausibility, and ultimately enable machines to perceive, plan, and act in ways that respect the deep regularities of nature. This is the path toward AI that does not merely correlate but genuinely understands.