In China, humanoid robots are no longer just a niche industrial technology topic; they have been integrated into a key part of the country’s economic modernization strategy.
In the latest published “14th Five-Year Plan,” Beijing proposed the comprehensive implementation of the “AI+” initiative, clearly stating the need to accelerate the upgrading and application of humanoid robots and other forms of products. The broader field of humanoid robots—”embodied intelligence”—has also been listed as one of China’s “future industries,” together with artificial intelligence, expected to become a new engine for driving technological independence and economic growth.
Since this year’s Spring Festival, this technology has begun to enter the public eye. During the CCTV Spring Festival Gala, a series of humanoid-like robots appeared on stage, performing acrobatics, dancing, practicing martial arts, and participating in skits. In just a few hours, it became a hot topic on social media.
However, behind the policies and cameras, another sentiment is quietly spreading in the industry: caution.
For many practitioners, the biggest concern is not that humanoid robots are not popular enough, but that they may become too popular too quickly.
Capital First
In the past two years, capital has surged into this arena at an unprecedented speed.
According to UBS statistics, investments in humanoid robots in China have increased from 450 million RMB in 2022 to 4 billion RMB in 2024, growing nearly ninefold in two years. The China Academy of Information and Communications Technology estimates that by 2025, the global humanoid robot market will reach 17 billion RMB, with the Chinese market exceeding 8.5 billion RMB, accounting for more than half of the global share.
According to market research firm Omdia, most of the 13,000 humanoid robots delivered globally in 2025 will come from Chinese companies.
A persistent question remains amid the frenzy: what can these robots actually do for people?
“When a new product just enters the market, customers often first ask not ‘how advanced is it’ but ‘why do I need it?'” a seasoned industry practitioner told BBC Chinese. Requesting anonymity due to concerns about the company brand.
“The skepticism faced by humanoid robots entering households today is essentially the same question that delivery robots faced when they entered hotels: customers still do not recognize their value.”
The company where this practitioner works primarily develops “composite robots” widely used in delivery and hotel service scenarios. Humanoid robots build on similar capabilities, layering human-like appearance and operational abilities to attempt more complex tasks. Both types of robots belong to one of the most sought-after industrial concepts in China today: embodied intelligence.
One of the earliest entrepreneurs in the humanoid robot industry, founder of leading robotics manufacturer UBTECH Zhou Jian, also expressed doubts in a recent interview with Chinese media: If a humanoid robot can dance and kick but can’t even bring a cup of water, what is its significance?
Industry insiders worry that as capital floods in, the humanoid robot industry may soon experience homogenization and competition.
The aforementioned industry insider noted that the most common strategy among new entrants is to “exchange financing for lower prices.” In the absence of brand accumulation, algorithmic capabilities, and validation in real-world scenarios, prices often become the primary means of market entry: first, financing to produce equipment, then entering the market quickly via financing to cut prices.
This strategy may generate short-term sales but could disrupt the industry’s long-term development.
Mature enterprises place greater emphasis on operational efficiency and whether the business model can sustain itself. Robot products must be used frequently and stably in real scenarios for businesses to cover research, production, and maintenance costs through continuous orders.
If the market is dominated by short-term price competition, firms may find it harder to establish stable business models.
Policy Promoted
Behind the humanoid robot craze is a very clear national strategy.
In the recently published “14th Five-Year Plan” and government work report, artificial intelligence, the “AI+” initiative, and embodied intelligence have all been given a more prominent policy position. Beijing hopes to enhance efficiency in multiple areas such as manufacturing, logistics, education, and healthcare through AI and robotic technologies, addressing economic slowdown, aging populations, and technological competition with the United States.
This renders humanoid robots not merely a startup track in China but part of a national industrial upgrade narrative.
For Beijing, it connects chip manufacturing, AI modeling, and high-end manufacturing on one end, while addressing labor shortages, elderly care pressures, and the need for new growth points on the other.
“Currently, it is a prosperous period driven by both policy and capital,” said Georg Stieler, regional manager at consultancy Stieler Technology & Market Advisory, to BBC Chinese.
“But the future will inevitably face setbacks, and the pace of development may be slower than expected,” he said.
One significant advantage for China in this competition is its complete upstream supply chain. Regions such as Shenzhen, Shanghai, and Suzhou are forming industrial clusters covering various aspects, including sensors, motion control, and servo systems.
This complete hardware ecosystem can shorten R&D cycles, enhance mass production capacity, and reduce overall costs. Therefore, the current craze for humanoid robots in China is driven by a combination of policy, capital, and manufacturing capabilities.
Who Foots the Bill?
As the narrative of “the future is here” grows louder, a fundamental question arises: who is paying for these robots?
At least for now, the answer does not include ordinary households. Government agencies are among the earliest buyers. According to Reuters, the amount of humanoid robots and related technology procured by the Chinese government is expected to increase from 4.7 million RMB in 2023 to 21.4 million in 2024. The Economist also points out that local governments and government-affiliated institutions may continue to be significant buyers in the humanoid robot market in China.
Data from the Beijing Academy of Artificial Intelligence indicates that from April 8, 2021, to March 2025, leading enterprise Unitree’s over 100 winning projects included numerous universities as major purchasers, with nearly 30 universities having procured relevant products. By 2025, the customer structure will change, with increased orders from state-owned enterprises and industrial clients.
These changes indicate that the industry is slowly transitioning from experimental needs to industrial demands. However, it also highlights that humanoid robots have not yet genuinely entered the mass consumer market. Even in a market like China that is highly accepting of new technologies, very few individuals have actually purchased a humanoid robot to bring home.
In this sense, the current fervor surrounding humanoid robots appears more like a prosperity sustained by demonstrations and pilot projects rather than a proof of maturity for a mass market.
Real Demand or Repeated Orders?
Nonetheless, this wave is not built on fantasies; the demand for humanoid robots indeed has its realistic background.
Chinese society is aging rapidly, and many manufacturing jobs have long struggled with recruitment difficulties. Dangerous, tedious, and repetitive positions have been facing the pragmatic need for “machines replacing people.” Humanoid robots are expected to enter factories, warehouses, hospitals, and nursing facilities to take on those jobs that people either do not wish to do, find difficult to sustain, or entail risks.
Market expectations for this future prospect are heating up. Morgan Stanley raised its projection for humanoid robot sales in China in 2026 to 28,000 units in January, doubling the growth rate.
Technology giants are also increasing their investments. American electric vehicle enterprise Tesla has been placing more resources on embodied artificial intelligence and its flagship humanoid robot, “Optimus.” CEO Elon Musk has repeatedly stated that Chinese companies may become their strongest competitors. During the earnings call at the end of January, he claimed that China would be the “biggest competitor” in the humanoid robot field and suggested that external observers have underestimated China’s comprehensive strength in manufacturing and artificial intelligence.
However, even so, caution remains within the industry.
Several interviewees pointed out that almost all new technologies require a “non-fully market-driven” growth phase initially. The real test lies in whether this demand can translate into consistent, stable orders rather than merely remaining at initial procurement levels.
“The key is whether there are repeat orders from large industrial and service clients; this signals the distinction between genuine deployment and subsidized pilot programs,” Stieler said.
In other words, the commercial tipping point for humanoid robots hinges not on the first purchase from customers but on whether they will buy a second time.
Is It Still Difficult?
Even though demand exists, the commercialization of humanoid robots may be harder than anticipated.
The primary challenge is the technology itself. Compared to traditional industrial robots, humanoid robots face a more complex physical world. Industrial robots typically repeat actions along fixed paths, while humanoid robots must perceive the environment, understand tasks, make decisions, and finally execute actions. This means they represent the most intricate category of robots humans have developed to date, requiring machines to possess the collaborative capacity similar to that of a human brain, cerebellum, and body.
“This process necessitates the integration of AI, which we now refer to as the manifestation of ’embodied intelligence’ in the physical world,” Zhou Jian explained.
Another challenge is data. Training language models can rely on vast amounts of online text, but robots require real-world scenarios. For instance, training a robot to open a door may require hundreds or even thousands of repetitions, adjusting algorithms based on each failure.
However, the biggest barrier is reliability. Stieler explains that industrial applications have extremely high reliability requirements. For example, critical production processes such as automobile manufacturing usually mandate that robotic systems achieve a reliability of 99.95% or higher; even in typical industrial scenarios, reliability is often demanded at over 99.9%. In contrast, scenarios like logistics and warehousing may tolerate a slightly higher failure rate since the costs of individual failures are lower.
A recently disclosed representative datum comes from Xiaomi. According to a report released by Xiaomi in early March, its humanoid robot achieved a successful task rate of 90.2% after continuously working autonomously for 3 hours in a bolt installation position at an electric vehicle factory.
This indicates that robots are no longer just laboratory toys, but they still have a considerable gap to bridge before they can achieve the stability required for critical industrial processes.
Stieler believes that in the next one to two years, humanoid robots are more likely to first take on relatively repetitive tasks in “semi-structured environments,” such as internal logistics, quality inspection, simple assembly, and some pilot work in hospitals and care facilities. Broader applications in general scenarios may still require three to five years.
Antonia Hmaidi, a senior analyst at the Mercator Institute for China Studies (MERICS), stated that to judge whether humanoid robots have genuinely entered a phase of scale, one can observe several signals.
One of these involves whether manufacturers like UBTECH can increase revenues while lowering costs. Another crucial indicator is whether industries that deploy robots en masse actually enhance efficiency. If the manufacturing sector extensively uses robots but neither employment decreases nor output increases, it suggests that these robots have not genuinely improved efficiency.
At the current stage, there remains a gap between the high expectations and realistic demands for humanoid robots. The coming years may witness a lengthy and tumultuous verification period for the industry. Those who can navigate through this validation phase are the ones likely to define the future of humanoid robots in China.