
Understanding this overcapacity risk is crucial for investors, policymakers, and tech firms as it signals potential market corrections and the limits of state‑driven industrial policy. The episode sheds light on how China's approach to emerging technologies could shape global robotics supply chains and influence future innovation trajectories.
Humanoid robotics in China is already moving toward overproduction. Government subsidies and private capital are driving the rapid expansion of new factories. At the same time, demand remains limited and many applications remain in the pilot stage. This is widening the gap between capacity and actual market demand.
[

](https://substackcdn.com/image/fetch/$s_!vT9Z!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd1e45a1b-a710-48dd-988e-67a19fbbf739_1536x1024.png)
Robots play hardly any visible role in everyday life in Europe and the USA. Outside of promotional videos, technology trade fairs, or science fiction series, they rarely appear. In restaurants, people serve customers; at airports, employees clean the terminals; and even in highly automated factories, many work processes remain recognizably human. Technological visions of the future exist there primarily on screens.
The picture is different in China. In larger cities, passers-by already encounter service robots in restaurants, cleaning machines in airports, and mobile surveillance units that patrol alongside police officers. Technology has become visible, even if it does not yet shape everyday life. However, humanoid robots, i.e., machines with human-like appearance and mobility, remain largely confined to demonstrations and pilot projects.
This is precisely why it seems remarkable that Beijing has now declared this area a key industrial policy project and is openly warning against overproduction and speculation.
At the end of November, the National Development and Reform Commission spoke out publicly for the first time, warning of structural risks in the humanoid robotics sector. Spokesperson Li Chao pointed out that more than 150 companies were now working on similar models and in many cases developing almost identical products, resulting in increasing overlap in investments, supply chains, and research efforts. In the sober language of the planning authority, this suggested that the market was growing faster than its real economic basis.
The underlying mechanism is familiar from previous cycles. As soon as Beijing defines an industry as strategic, local governments, state banks, and investors initiate a coordinated expansion. Provinces attract companies with low-interest loans, subsidized building land, and special tax regulations, while state funds and private investors mobilize billions. Within a short period of time, entire production chains emerge, often with capacities far exceeding current demand.
This pattern has already shaped the rise of the solar industry, battery manufacturing, and electromobility. In all cases, the phase of rapid expansion was followed by price declines, bankruptcies, and politically driven market consolidation, leaving only a few national champions at the end. Humanoid robotics has so far followed the same development path.
New manufacturing facilities for bipedal robots are currently being built in technology centers such as Shenzhen, Hangzhou, and Shanghai. Manufacturers such as Unitree, UBTech, and AgiBot are expanding their production capacities and planning to produce four-digit quantities. At the same time, extensive venture capital is flowing into the industry, fueled by the expectation that a global mass market will establish itself in the medium term.
However, this market remains fragmented. The majority of the systems delivered are used in pilot projects, research environments, or high-profile demonstrations. Humanoid robots play hardly any role in industrial series production, care facilities, or private households. Prices often range between US$100,000 and US$200,000 per unit, plus integration, maintenance, and training costs, which further complicate their economic use.
There are also considerable technological limitations. While the systems work reliably in controlled environments, they quickly reach their limits in open, unstructured spaces. Problems with grasping, walking steadily on uneven ground, or dealing autonomously with unexpected obstacles are still part of everyday life. Several high-profile demonstrations have made it clear how much many models still rely on human support.
At the same time, competitive pressure is increasing. Many manufacturers are using similar sensor technology, drive systems, and software modules, which means that hardware and functions are becoming increasingly similar. Margins are already falling in the emerging rental and leasing market, while intermediaries and integrators are skimming off a growing share of the added value. This creates price pressure for manufacturers even before stable sales structures have been established.
Political leaders are now taking corrective action. They are deliberately making it more difficult for individual companies to gain preferential access to IPOs, linking new support programs more closely to specific application scenarios, and seeking to enshrine stricter market access and exit rules in the upcoming five-year plan. The aim is to prevent a repeat of the extreme overproduction seen in previous industrial cycles.
The interventions to date have done little to curb the drive for expansion. Capital continues to flow in, new factories are being built, and numerous business models continue to be based on the assumption that demand will automatically follow after a time lag. For investors, this reinforces the characteristics of a speculative phase in which valuations are driven more by promises for the future than by actual earnings.
In the medium term, consolidation seems inevitable. Only those providers that combine technological substance, long-term customer contracts, and political backing are likely to survive, while the majority of today's startups will disappear through acquisitions or insolvency. For China, this would not be a break with its previous development model, but rather its logical continuation, with a phase of exaggeration followed by a phase of selection.
The humanoid robotics sector is thus developing into another testing ground for the question of whether state-controlled industrialization will this time lead more quickly to viable business models than to new overcapacities.
Comments
Want to join the conversation?
Loading comments...