On October 22, the 2026 World Physical AI Conference and Unitree Partner Ecosystem Conference took place in Hangzhou. During the event, Wang Xingxing, Chairman of Unitree Robotics, unveiled the company’s three primary strategies for the future. These include: bolstering robot mass-production capabilities to meet future market demands on a scale of tens or even hundreds of millions; adopting a dual approach of in-house research and collaborative efforts to facilitate the implementation of physical AI small models customized for the company’s proprietary hardware; and constructing an ultra-large-scale computing power cluster to drive technological progress. Furthermore, the company is developing a robot self-evolution system, which improves the efficiency of algorithm iteration and replaces traditional manual debugging through a closed-loop model of “AI self-research and development—simulation training—real-world testing—feedback iteration.”