(Yicai) March 18 -- China’s robotics industry is striving to reach a “ChatGPT moment,” as companies work to improve artificial intelligence capabilities, scale manufacturing, and accelerate real-world adoption.

That milestone would come when robots can complete 80 percent of tasks in 80 percent of unfamiliar scenarios, according to Wang Xingxing, founder of Unitree Robotics, who said the industry is working to strengthen supply chains, speed up production, improve affordability and accessibility, and establish comprehensive after-sales service systems.

Wang also sparked discussion at the Yabuli China Entrepreneurs Forum yesterday after saying robots could run faster than Olympic sprinter Usain Bolt this year, raising public questions about what such performance would mean for practical applications.

Wang said the industry still needs to overcome three key challenges: improving models’ ability to understand tasks and actions to break through generalization bottlenecks, increasing the efficiency of using diverse data sources such as video, simulations, and real-world machines to reduce reliance on large-scale physical data collection, and enabling reinforcement learning to achieve reusable and cumulative scaling effects.

Production Scaling and Commercial Adoption Remain Key Hurdles

Regarding technology pathways, Wang said he is optimistic about world models and video generation models, noting that they offer higher potential ceilings and better opportunities to leverage the vast volume of internet video and text data.

Jiang Zheyuan, founder of another robot maker called Noetix Robotics, told Yicai the company has no plans to launch new products in the near term and is targeting sales of more than 10,000 robots this year. He said the industry has entered an early stage of scaling in 2026 and that while robots are not yet widely used in households for tasks such as laundry and cooking, total robot sales will rise significantly this year.

Humanoid robots contain hundreds or even thousands of components, Jiang noted, and shortages of any single part can halt production. Scaling monthly output from hundreds to thousands of units also requires large production teams, creating management challenges. This year, Noetix aims to achieve breakthroughs in both its supply chain and manufacturing capabilities to support deliveries.

“Reaching a production scale of 10,000 units is a significant milestone for the humanoid robot industry. Noetix places more emphasis on actually delivering products to end-user families and them being actually used in various scenarios,” Jiang said.

Li Yiyan, chief executive officer of AgiBot-backed robot rental firm Bot Share, told Yicai that demand for robot rental services currently comes mainly from business clients, including brand events and shopping malls. The consumer market is beginning to emerge but remains in an early development stage, with demand mainly coming from family gatherings and weddings.

“We predict that in the coming period, the robot rental market will show a rather clear trend that the B-end will first establish a viable business model, while the C-end will gradually develop consumption habits. Both ends will jointly drive the expansion of the industry,” Li said.

Editor: Emmi Laine