In 2025, humanoid robots seem to be a hot trend in China, with almost every humanoid robot company showcasing their capabilities through videos of boxing, dancing, folding clothes, and cleaning tables. However, a recent research report on the Chinese humanoid robot industry released by Goldman Sachs revealed the true situation in the industry: while companies are expanding their production capacity, they are facing a shortage of orders, raising concerns about potential “over-capacity” issues in the future.
Goldman Sachs recently published a field research report on the Chinese humanoid robot supply chain, which has quietly circulated in the capital market, leading to consecutive three trading days of decline in the Robot Concept Index and Humanoid Robot Concept Index. On November 7th, 10th, and 11th, the Robot Concept Index fell by 1.15%, 0.32%, 0.24% respectively, and the Humanoid Robot Concept Index dropped by 1.36%, 1.23%, 0.44% respectively.
According to the report, out of the nine companies in the industry chain that were surveyed, none have confirmed receiving substantial orders, nor could they provide a clear production schedule.
On November 12th, Goldman Sachs’ Director of Chinese Industrial Technology Research, Du Xi, explained in detail in an interview with “First Financial” that the lack of substantial orders is mainly due to the fact that the scale of humanoid robot orders has not yet formed. “For industrial-grade humanoid robots, we believe that a production scale of over ten thousand units is needed to establish a certain volume of the supply chain system. If it’s just a few hundred or one or two thousand units of scattered orders, it cannot support the development of the entire supply chain.”
Reports from “Economic Observer” have corroborated the claims that Chinese companies lack substantial orders. Some orders come from shareholder companies or group companies, or are signed in the form of framework agreements, lacking specific delivery plans and pacing. These signs collectively point to a reality: the industrialization process of humanoid robots is still supply-driven, and real demand has not been widely established.
Goldman Sachs’ research on the entire humanoid robot industry chain occurred from November 3rd to 6th, focusing on visits to nine key companies including Sanhua Intelligent Control, Top Group, Rontai Health, Double Ring Transmission, Minsheng Group, Juncheng Electronics, and Zhaowei Electromechanical.
In terms of application scenarios, robots can be divided into industrial robots, service robots, and humanoid robots. Industrial robots focus on tasks like car welding and chip packaging that require high intensity, high repetition, and high precision; service robots are more involved in vertical fields like delivering food in hotels and household cleaning; while humanoid robots are more like “jacks-of-all-trades”, adapting to unstructured environments in homes, factories, and providing care for the elderly, assisting in disaster relief, among others.
According to a report by Jiemian News on Tuesday (November 11th), information obtained from companies such as Sanhua Intelligent Control, Top Group, Juncheng Electronics, Best, Zhejiang Rontai, and Shuanglin Corporation confirmed the contents of the Goldman Sachs report. Looking at the actual landing of orders, some companies claim to have products deployed, while others are sending samples to customers and engaging with domestic and overseas clients, but many indicate a lack of substantial orders or are waiting for demand from major North American clients.
A robotics expert pointed out that “listed companies have the need to boost stock valuation through the robotics concept”, and that the so-called “mass production” is mostly for small-scale demonstration projects, rather than market-driven real demand.
Zhang Xiaorong, Director of the Institute of Deep Technology Research, told Jiemian News that the current breakthroughs are mainly focused on motion control, but the universal artificial intelligence (environment perception, decision-making) that determines its value remains a significant bottleneck. The industry is still in the early stages of technical exploration, largely in the laboratory phase, and far from reaching a critical point of large-scale development.
In a report by Securities Times in July of this year, it was mentioned that the humanoid robot industry is still in the “0 to 1” stage of research and development, with a long way to go before real industry implementation. Therefore, there are currently no A-share companies specializing in the research and manufacturing of humanoid robots.
At the global technology conference hosted by NVIDIA in July of this year, Pieter Abbeel, a professor in the Department of Electrical Engineering and Computer Science at the University of California, Berkeley, stated that there are no truly humanoid robots in the world yet.
Although these humanoid robot companies have not yet secured large-scale orders, most companies are expanding their production capacity domestically and overseas. The Goldman Sachs report indicates Thailand as a prime location for expansion, with Mexico also being considered. The current annual production capacity plans of various companies range from 100,000 units to 1 million units (robot equivalent units). This is significantly more optimistic compared to Goldman Sachs’ forecast of 1.38 million units of global shipments by 2035.
According to market rumors, Top Group plans to establish humanoid robot-related production lines in Thailand, Mexico, and the United States, with the Thai factory designed for an annual production capacity of up to 1 million units, with an expected total investment of 7 to 8 billion yuan.
Sanhua Intelligent Control has purchased about 200,000 square meters of land in Thailand, specifically for assembly of humanoid robot actuators. The company has already initiated the use of the Thai base for humanoid robots. In addition, Minsheng Group has completed the construction of a production line for 10,000 sets of head and face assemblies, aiming to achieve mass production in the first quarter of 2026.
Shuanglin Corporation plans to expand the production capacity of planetary roller screw and rod ends to 300,000 sets by the end of 2025 to meet customer demands at home and abroad.
Goldman Sachs’ report directly addresses the most sensitive issue in the capital market – the visibility of performance realization. In the absence of clear demand, whether such aggressive production layout will evolve into an “over-capacity” situation akin to intensified competition in the industry.
The report from Jiemian News mentioned above stated that Zhang Xiaorong, Director of the Institute of Deep Technology Research, pointed out in an interview that the core “robot intelligence” bottleneck has not been overcome, and prematurely planned production capacities will inevitably lead to serious overcapacity. Many entrepreneurial projects will face intense industry reshuffling and eliminations. The high costs and limited functionalities raise doubts about the return on investment, and a commercialized loop has yet to form. “There is a clear overheating of market sentiment, mistaking technical milestones for commercial maturity.”
Data from Tonghuashun shows that there are a total of 947 listed companies in the A-share robot concept, with 252 companies in the humanoid robot concept. At the 2025 Huaying Capital Annual Conference, Wang Qian, founder and CEO of Netease Technology and Yinshu Robot, stated that, taking the development history of the mobile phone and automobile industries as an example, there are now only six or seven major manufacturers in the mobile phone market, but there are still dozens in the automobile market. Robotics should fall somewhere in between. The global market is expected to eventually shrink to about 10 to 20 major manufacturers.