Unitree's Wang Xingxing Outlines Three Future Strategic Directions

At the 2026 World Physical AI Conference and Unitree Partner Ecosystem Conference held in Hangzhou, Unitree Technology Chairman, General Manager and CTO Wang Xingxing laid out three strategic directions for the company's future.

First, Unitree aims to enhance its large-scale robot production capacity to meet market demand that could reach tens of millions or even hundreds of millions of units. Second, the company will pursue both in-house development and external partnerships to deploy physical AI small models adapted to its own hardware. Third, it plans to build ultra-large-scale compute clusters, aiming to secure a foundation for continuous technology iteration.

Wang indicated that the three directions form a coordinated framework spanning manufacturing, models and computing. The move suggests Unitree is extending from a robot hardware maker toward full-stack physical AI capabilities, addressing production scale, embodied intelligence and computing infrastructure as humanoid robots move toward commercialization.

Meanwhile, Unitree is constructing a robot self-evolution system. Through a closed-loop model of "AI independent R&D—simulation training—real-world testing—feedback iteration," the system is designed to replace traditional manual debugging paths and improve algorithm iteration efficiency, thereby supporting ongoing product evolution.