China’s Humanoid Robots Face the Factory Test

HONG KONG — Chinese humanoid robotics companies are shifting attention from public demonstrations to commercial viability, with top developers emphasizing hard manufacturing metrics and strict industrial benchmarks for the next phase of deployment.

The World Humanoid Robot Games in Beijing last August featured prototype machines attempting sprinting, weightlifting, and kickboxing. Manufacturers across China, however, are directing investment toward factory floors and service sectors as they seek to justify heavy capital investments.

Counterpoint Research figures released in late August showed that Shanghai-based robotics startup AGIBOT led global humanoid robot shipments in the first six months of the year, delivering 9,700 units. AGIBOT is currently considering an initial public offering in Hong Kong and is targeting industrial lines where operational precision is paramount.

At the Fortune Leaders Forum in Macau on Sept. 8, AGIBOT co-founder Yao Maoqing said commercial adoption in the industrial sector depends on rigorous operational standards rather than novel hardware designs. “It’s genuinely difficult for today’s humanoid firms to develop a robot that can run stably inside a factory,” Yao said. “Industry players are brutally honest: They care about four metrics—success rate, cycle time, stability, and cost—rather than what technology is used. From that standpoint, we absolutely have to train our robots to 100% efficacy first.”

AGIBOT tested that consistency through a six-day continuous livestream in late June, during which its robots completed more than 64,000 manufacturing tasks with a recorded success rate of 99.99 percent. According to Yao, reaching that performance level required intensive testing, including month-long nocturnal preparation sessions operating eight hours per night.

The company’s development strategy applies artificial intelligence scaling laws to physical machines, a field known as embodied AI. Those laws correlate increased compute power, larger parameter counts, and expansive training datasets with leaps in performance. Yao projected that embodied AI will experience its “GPT-3.5 moment” within three to five years. The benchmark refers to OpenAI’s late-2022 release of GPT-3.5, the foundation model for ChatGPT that established general-purpose utility across diverse everyday tasks with reliability rates between 80% and 90%.

“Most industry teams believe that embodied AI will follow the exponential scaling law we discovered in digital intelligence and large language models,” Yao said. “As data volume rises and your model’s parameter count grows, there will definitely be a step-up in intelligence.”

Established robotics firms are also pursuing commercialization through large-scale standard industrial environments. UBTech Robotics, a Shenzhen-based firm that completed a listing on the Hong Kong Stock Exchange in December 2023 under stock code 9880.HK, has directed significant resources toward automated industrial applications.

UBTech has piloted its industrial humanoid series, Walker S, in automotive assembly plants operated by NIO, Dongfeng Liuzhou Motor, and FAW-Volkswagen. The robots perform quality inspections, component assembly, and material handling. “We’ve invested very heavily in developing robots for the industrial sector,” said Jianxin Pang, vice president and vice dean of research at UBTech. “But first, we’re starting with relatively common scenarios with a big enough market—that helps us ensure return on our investment.”

Commercial service firms are pursuing semi-structured environments such as hospitality and catering. Shanghai-headquartered Keenon Robotics has deployed specialized service hardware for hotel and restaurant management, while its chief operating officer, Wan Bin, cited a Shanghai hotel using a fleet of five robots for front-desk guest greetings, room service deliveries, room maintenance, and restaurant floor logistics.

“This scene is one that will become increasingly common in the future,” Wan said.

China’s push to commercialize embodied AI operates under broader national policy frameworks. In November 2023, China’s Ministry of Industry and Information Technology (MIIT) published policy guidelines designating humanoid robots as a disruptive technology comparable to smartphones and electric vehicles. The MIIT roadmap set targets for China to establish an initial humanoid robot innovation system and achieve mass production by 2025, followed by plans to establish a secure, globally competitive industrial supply chain by 2027.

As hardware costs decline and foundation models integrate multimodal sensing, Chinese manufacturers continue to vie with international rivals—including U.S. developers like Tesla with its Optimus platform, Boston Dynamics, and Figure AI—to secure commercial contracts across manufacturing, logistics, and service operations.