Unitree listed on the Shanghai exchange this week on record order demand for its shares, the clearest signal yet that investors are pricing China's humanoid robot leader as an industrial company rather than a research project. The company's G1 and H1 platforms have spent two years racking up viral demo videos, backflips, stair climbs, martial arts routines, and that footage is doing real work now: it is the thing the prospectus points to when it explains why a company that ships mostly to research labs and a handful of industrial pilots deserves a public listing.

The mechanism worth naming is the one every humanoid maker is still stuck on: a robot that can climb a staircase once, on camera, in a controlled take, is not the same machine as one that climbs ten thousand different staircases in ten thousand different buildings without a human minder nearby. That gap, between a motion-control demo and a repeatable industrial deployment, is why Chinese humanoid makers are already being described as facing a harder second act once the IPO cash lands: the actuators, battery life and fall-recovery software that make a robot useful in a warehouse are a different engineering problem than the one that makes it useful in a highlight reel. Unitree's next real test is not another video. It is whether its order book, beyond the IPO subscription numbers, shows paying customers running its robots on shift, not just at trade shows.