Wang Xingxing, the founder of Yuque Technology, who has just completed its listing on the Sci-Tech Innovation Board, delivered his first public speech after the listing at the main forum of the 2026 World Robot Conference, sharing his latest insights on the future development inflection point of the embodied intelligence industry.
In her view, for the humanoid robot industry to truly experience a comprehensive boom, it must first overcome the "ChatGPT moment" specific to embodied intelligence. This critical turning point has a very clear standard: when a robot is placed in 80% unfamiliar scenarios, it should be able to independently and smoothly complete about 80% of tasks based solely on human verbal or written instructions.
Regarding the timing of this historical turning point, Wang Xingxing provided a rational forecast, believing that it could arrive within 2 to 3 years, or at the latest, within 5 to 10 years, after which the industry is expected to officially reach this inflection point.
However, the biggest core challenge preventing embodied intelligence from reaching the "ChatGPT moment" is no longer the robot's physical hardware itself, but rather the alignment challenges between AI large models and real robots. In other words, the logical instructions generated by the model in the cloud still produce some deviations when converted into real, stable, and executable actions for physical robots.
