On July 27, at the "AI Women's Excellence Forum" held in conjunction with the 2025 World Artificial Intelligence Conference (WAIC), Wang Xingxing, founder of Unitree Robotics, shared his deep insights into the humanoid robot industry and the development of AI technology. He pointed out that with the rapid development of general AI, humanoid robots are facing unprecedented growth opportunities, and AI's capabilities in code generation have made breakthrough progress.
Wang Xingxing introduced that Unitree Robotics initially focused on high-performance robotic dogs, and in recent years has actively entered the humanoid robot field to meet the urgent market demand for general-purpose robots. He emphasized that the integration of robots with general AI is a global hotspot, and it is expected to play a significant role in areas such as industry and home services, although it is still in a relatively early stage.
Although still in the early stages, Wang Xingxing observed that the humanoid robot industry is experiencing explosive growth. He estimated that in the first half of this year, the Chinese intelligent robot industry, including both complete machine manufacturers and component suppliers, achieved an average growth rate of 50% to 100%. Since 2024, almost every day has seen the launch of a new robot, and the pace of industry implementation and shipping has significantly accelerated. He particularly pointed out that China's manufacturing advantages are obvious, and most of the popular robots on the market are Chinese brands.
Wang Xingxing also shared his "experience" in using AI for code. He mentioned that when he tried to use AI to write code last year, the success rate was low and there were many errors; however, by the first half of this year, the success rate had soared to over 90%.
"Now AI performance is very good. For example, if I want to generate a lottery software, it can help me generate (code) comprehensively. The current success rate is very high, basically reaching ninety percent or more. Now I am being lazy, and I don't want to write many code texts anymore. It's better to tell the large model what function I need, and it can automatically write it for me," said Wang Xingxing. This has led him to directly entrust many of his codes to be generated by large models.