Unitree CEO Outlines Three Future Directions and Announces Physical AI Competition
Unitree founder and CEO Wang Xingxing outlined three major strategic directions—mass production scaling, self-developed and partner-driven physical AI small models, and large-scale computing clusters—while announcing an annual multi-million RMB global physical AI competition. This announcement signals a major push by a leading robotics firm to scale hardware manufacturing and accelerate software iteration through automated closed-loop training and global talent recruitment. It highlights how the robotics industry is aggressively tackling the commercialization and technical bottlenecks of physical AI. The newly announced World Physical AI Competition features eight tracks ranging from mobility and combat to healthcare and manufacturing, offering tens of millions of RMB in annual prize money. Additionally, Unitree is establishing a self-evolving robotic data loop covering autonomous AI R&D, simulation training, real-world testing, and feedback iteration.
## BACKGROUND
Physical AI refers to artificial intelligence systems that combine software algorithms with hardware like robots and sensors to perceive, reason, and act within the real world. Simulation training uses virtual environments like NVIDIA Omniverse or MuJoCo to train AI models safely and efficiently before deploying them onto physical robots.