China Is Building 100,000 Humanoid Robots. They Still Struggle With Basic Jobs.

China has emerged as the overwhelming manufacturing leader in humanoid robots, producing machines that can perform increasingly sophisticated demonstrations. But they still struggle to do many ordinary jobs reliably.

A Reuters investigation found a widening gap between China's massive investment in humanoid robotics and what the machines can actually accomplish in factories and other workplaces. Despite rapid advances in hardware, experts say today's humanoids are not advanced enough to replace workers on a broad scale.

At a robot-training center in Liuzhou in southern China visited by the outlet, more than 100 humanoids were being taught basic activities such as sorting crates, packaging noodles and making coffee. Human trainers use headsets and controllers to guide their movements and generate data that can eventually help robots learn to operate more independently.

The process remains remarkably inefficient. An inexperienced trainer might need as many as 300 attempts to produce one usable robot movement, while an experienced trainer might succeed once in 50 tries. One employee said the machines currently perform simple tasks at only about 20% of a human worker's speed or output.

"The robots' IQ is too low," Tang Wenbin, co-founder and CEO of robotics venture Yuanli Lingji, said during an industry panel cited by the outlet. "A lot of what we see is dancing disguised as working."

However, the technological shortcomings have not slowed Beijing's ambitions. Chinese manufacturers accounted for 95% of the roughly 20,000 humanoid robots shipped worldwide last year, according to BofA Global Research figures cited by Reuters. A Chinese industry ministry official said the country expects to manufacture more than 100,000 humanoids this year.

China now has more than 150 humanoid-robot companies, according to the country's National Development and Reform Commission. Government spending has also surged. National, provincial and local authorities spent at least $230 million purchasing humanoids and related products during the first half of 2026, compared with $62 million during the same period last year and just $6 million in 2024, Reuters found after reviewing nearly 1,000 government tenders.

China's Ministry of Industry and Information Technology acknowledged in June that there are still numerous obstacles. The government nevertheless wants to support deployment capacity of 10,000 humanoid robots and develop more than 100 high-value applications by the end of 2026.

Humanoid robots face a fundamentally different challenge from generative AI systems such as chatbots. Instead of simply processing language, a robot must interpret visual information and translate it into precise physical actions, from gripping fabric to tightening a screw with exactly the right amount of force.

Small changes in lighting, object position, surfaces, or equipment can cause a robot that performed successfully during training to fail in a new environment. "The problem goes from the last 10 centimeters to the last 1 centimeter, even 1 millimeter," Jia Baoxiong, a research scientist at the Beijing Institute for General Artificial Intelligence, told Reuters.

That helps explain why conventional industrial robots remain more attractive for many manufacturers. At Xiaomi's electric vehicle factory in Beijing, for example, more than 700 industrial robots already perform jobs including welding, riveting and adhesive bonding. Humanoids are being trained for tasks such as installing nuts, but traditional automation still dominates the production process.

There are signs of genuine commercial potential. Galbot has deployed robots in pharmacies in more than two dozen Chinese cities, where machines retrieve medicines and other products from shelves. The company says its robots achieve a success rate above 95% in those highly structured tasks.

For now, however, the industry's extraordinary production boom may be moving faster than its technology and its customers. Some founders have already described the sector as a bubble, and investors expect consolidation as subsidies eventually become more selective. "There will 100% be a consolidation sooner or later," Interconnected Capital founder Kevin Xu told Reuters. "It's just a matter of who and how ugly."