China has more humanoid robot manufacturers than any other country, over 140, according to TrendForce, and more government money behind them than any other market. Beijing has allocated at least $20 billion in subsidies since 2024, according to Reuters, and has set a directive requiring more than 10,000 humanoid robots in commercial use by the end of 2026. Barclays projects annual humanoid installations in China could reach 11 million units by 2035, with a total stockpile of 24 million, equivalent to roughly 4% of China’s effective workforce. The ambition is extraordinary. The current reality is considerably more modest.
What the Sales Data Actually Shows
Morgan Stanley estimates that only 12,000 humanoid units were sold in China last year, the vast majority to research institutions and educational organizations rather than commercial operators. Over 450,000 firms are now registered in intelligent robotics in China, more than triple the tally from four years ago, but the gap between registered companies and units actually deployed in commercial settings is enormous.
The rental market that has emerged over the past six months is the clearest signal of where commercial adoption actually stands. AgiBot, one of the market leaders, launched a humanoid robot rental platform offering units at 3,500 yuan ($517) per day, inclusive of shipping and a human operator. In three months, the platform logged more than 5,500 orders, a strong initial traction that AgiBot has used to project a rental market worth $1.5 billion by the end of 2026. But rental prices have already begun sliding as the novelty effect wears off, and the inclusion of a human operator in the rental package is a telling detail: the robots are not yet capable of operating autonomously in uncontrolled commercial environments.
The Productivity Gap
The productivity numbers are candid. UBTECH, one of China’s most technically advanced humanoid manufacturers, acknowledges that its most capable models achieve approximately 80% of human productivity — but only in specific, controlled tasks such as box stacking and package sorting. Unitree, which gained viral attention when its robots performed at China’s Spring Festival Gala and is preparing for a Shanghai IPO, has been equally direct about the timeline for broader factory deployment. “That will likely depend on technological breakthroughs and the overall development of the industry,” Unitree’s PR manager told reporters, a statement that, coming from a company preparing to go public, is a remarkably honest assessment of the sector’s limitations.
The gap between 80% productivity in controlled tasks and the flexibility required for general factory deployment is not a small engineering problem. It requires breakthroughs in AI perception, manipulation, and real-time adaptation to unstructured environments, capabilities that are advancing rapidly but have not yet reached the threshold required for mass commercial deployment. As CNN’s reporting on China’s humanoid rental market noted, the current business model relies on human operators accompanying the robots, fundamentally altering the economics of automation.
Bubble Warnings From Inside the System
The most striking aspect of China’s humanoid robot market is that the warnings of a bubble are coming from Chinese officials themselves. In November 2025, government advisors publicly warned of overcrowding in the humanoid segment, with more than 150 manufacturers competing for a market that had sold fewer than 15,000 units in total. The proliferation of startups and the concentration of government subsidies in a sector with limited near-term commercial viability has created the conditions for a classic investment cycle: enthusiasm, overcapacity, price competition, consolidation.
This dynamic is not unique to humanoid robots. EastFrontier has tracked similar patterns in China’s EV sector, solar manufacturing, and earlier waves of AI investment. China’s State Council’s national survey on AI and jobs reflects the government’s awareness that AI-driven automation, including robotics, will have significant labor market consequences, and its intent to manage the transition rather than simply accelerate it.
What Investors Should Watch
For investors, the humanoid robot sector in China presents a classic early-stage market challenge: the long-term opportunity is real and large, but the near-term path to profitability is unclear and the risk of capital destruction through overcapacity is significant. The rental market provides a useful test: if AgiBot’s platform can sustain order volumes and maintain pricing as novelty fades, it suggests genuine commercial demand. If rental prices continue to slide and the human operator requirement persists, it suggests the market is still in the demonstration phase rather than the deployment phase.
The IPO pipeline, Unitree preparing for Shanghai, AgiBot likely to follow, will provide market-based valuations that will either validate or deflate the current enthusiasm. EastFrontier’s earlier coverage of AgiBot’s factory trial at Longcheer documented a 99.99% success rate in a controlled six-day trial, an impressive result in a controlled setting that nonetheless leaves open the question of how performance holds up over months of continuous commercial operation in a real factory environment.
The gap between China’s humanoid-robot ambitions and its current commercial reality is not a failure. Rather, it is the normal state of a technology sector in the early stages of a long development curve. The question is whether the capital flowing into the sector is patient enough to wait for the breakthroughs required to close that gap, or whether the bubble warnings will prove prescient before the technology matures.