China's Humanoid Robot Hype vs. Reality Gap: Only 12,000 Units Sold Last Year as Factories Already Highly Automated
0xBroomberg
China has poured at least $20 billion in subsidies into humanoid robots and registered over 450,000 related companies — yet only about 12,000 units sold in all of 2025. The gap between supply-side frenzy and demand-side indifference is now the sector's central pricing risk.
What does 12,000 units actually mean?
Morgan Stanley analysts estimate China's total humanoid robot sales in 2025 at roughly 12,000 units, with the vast majority going to research labs and educational testing — not commercial or industrial buyers.
This means → at least $20 billion in government subsidies since 2024 has produced almost no real end-market demand.
In plain terms = the money and the companies showed up, but the buyers have not.
150 makers chasing one market — how obvious is the bubble signal?
Chinese officials issued a rare warning last November: roughly 150 humanoid robot manufacturers are competing for the same market.
By end-2024, over 450,000 smart-robotics-related companies had registered — more than double the figure four years earlier.
Bloomberg reports that by June this year, nearly 50 robotics startups had filed for Hong Kong IPOs. This reflects exit anxiety in the capital market spilling into the listing pipeline.
Where is the technology stuck?
The core bottleneck has a name: "Moravec's paradox" — machines excel at what humans find hard (complex calculations) yet struggle with folding laundry or climbing stairs.
Even when these skills are mastered in the lab, there is no guarantee they transfer to unstructured real-world settings.
This means → for the foreseeable future, humanoid robots will be largely confined to controlled environments — factories, warehouses, retail stores — far from everyday households.
Factories are already automated — do they even need humanoid robots?
The International Federation of Robotics puts China's industrial robot installed base at over 2 million units, concentrated in electronics and automotive.
These are mostly fixed robotic arms and gantry systems designed for welding, material handling, and other repetitive tasks — faster and more reliable than general-purpose humanoids.
In plain terms = factories want a "specialist" that is fast and consistent, not a "generalist" that can do everything but excels at nothing. These machines also last about 15 years on average — more than double a humanoid's expected lifespan — so the upgrade incentive is weak.
What supports the bullish forecasts?
Barclays analysts project China's annual humanoid robot installations could reach 11 million units by 2035, with a cumulative installed base of 24 million — roughly 4% of the effective labor force.
The rest of the world's combined forecast for the same period: just 2 million units — China dominates the optimistic scenario.
This means → this projection bets simultaneously on a technology breakthrough and mass commercialization, neither of which has been validated so far.
From "preschooler" to "teenager" — the truest footnote on where the industry stands?
Beijing's five-year economic plan lists robotics as a core growth engine, driven by accelerating population aging and a shrinking workforce.
Morgan Stanley analyst Tim Shen forecasts sales exceeding $15 billion by 2030, with state-owned enterprises such as State Grid and China Post expected to place large orders — government procurement will play a key role.
Yet one government official conceded that this year's humanoid robot soccer competition is expected to advance "from preschooler level to teenager level." In plain terms = even officials acknowledge the technology is still at a very early stage. The scissor gap between oversupplied makers and stalled commercialization is this sector's most critical pricing risk right now.
Content is for reference only, not financial advice.