The 2026 World Physical AI Conference and Unitree Partner Ecosystem Conference was held today in Hangzhou, Zhejiang. Wang Xingxing, Chairman, General Manager and Chief Technology Officer of Unitree Technology, took the stage and broke down the company's future strategy into three clear lines.
The first is production capacity. Wang Xingxing emphasized raising the robot's large-scale mass production capability, aiming at a future market of tens of millions or even hundreds of millions. In his vision, robots must go from "being able to make a few units" to "being able to make millions of units," with scale itself serving as the moat. The second takes a compound route of "in-house development + cooperation," driving the real implementation of physical AI small models adapted to the company's own hardware. The third bets on infrastructure—building an ultra-large-scale computing cluster to lay the foundation for technological iteration.
Direction alone is not enough; Unitree is also building a robot self-evolution system, using the closed loop of "AI-driven R&D—simulation training—real-world testing—feedback iteration" to replace the old path of repeated manual debugging, pushing up the efficiency of algorithm iteration.
Spending big on the World Physical AI Competition
Even more eye-catching is the competition. Wang Xingxing officially announced the "World Physical AI Competition" on site, pledging to put up tens of millions of RMB in prize money every year, opening the doors to teams and individuals worldwide.
There are eight tracks in total: free movement, track and field, combat, dexterous manipulation, production, sports events, multimodal interaction, and health care. There are hardcore contests like combat and track and field, as well as practical scenarios such as health care and production, balancing showmanship with utility. Refining the competition to specific capability dimensions has a clear purpose—to gather the world's smart minds around physical AI and continually push the boundaries of the technology.