Unitree (宇树科技) founder: humanoid robots will outperform humans by mid‑year, but a 'ChatGPT moment' for embodied AI still needs 2–3 years

It has been reported that Wang Xingxing (王兴兴), founder of Unitree (宇树科技), predicted humanoid robots will begin to outperform humans on certain tasks by mid‑year. Bold claim. Big implications. Unitree is best known in China for affordable legged robots and has been ramping R&D on humanoid platforms — a sector where hardware, software and real‑world safety all collide.

The claim and how to read it

Reportedly, Wang framed the forecast in task‑specific terms rather than a blanket takeover: improvements in repeatable, structured work where robots can be precisely controlled and tested. What does “outperform” mean in practice? For Western readers, think factory repeatability, logistics handling or inspection routines — not sweeping human replacement across creative or highly social roles. It has been reported that he expects those narrow gains to arrive quickly, but cautioned that the broader transformative moment for embodied intelligence still lies ahead.

Why the 'ChatGPT moment' for robots is farther off

Wang reportedly estimated the equivalent of a “ChatGPT moment” for embodied AI needs another two to three years. Why the lag? Language models benefited from a software‑centric ecosystem and massive public datasets; embodied systems require robust hardware, reliable sensors, real‑time control, and safe field testing. Geopolitics matters here too: export controls on advanced chips and sensors, and broader trade frictions, shape supply chains and access to high‑end compute — factors that could slow or redirect development trajectories.

What's at stake

If Unitree’s timeline holds, commercial use of humanoid robots will accelerate in China’s manufacturing and logistics sectors, raising questions about regulation, workforce transition and safety standards. It has been reported that Unitree and peers are pushing both technological frontiers and the policy debate. Skepticism remains healthy; ambitious forecasts are easy to make, harder to validate. The coming months and years will show whether mid‑year milestones are hype or the start of a new era for embodied AI.