Chinese robotics firms stopped selling products and started buying companies
A Chinese collaborative robot maker spent three years passing fifty rounds of Toyota assessment to become its sole global supplier. That is not a price play, and it is the pattern European industry should be reading.
When Germany's chancellor visited China in February 2026, the first stop was not a ministry or a carmaker. It was Unitree Robotics in Hangzhou.
That choice says more about where Chinese robotics has got to than any capacity statistic, and it points at the shift that matters: these firms are no longer trying to sell cheap machines into Europe. They are buying European companies, building European factories, and entering European supply chains on quality terms.
Where self-sufficiency actually stands
Domestic market share for complete machines reached 55.3% in 2025, up from 31.4% in 2020. That headline conceals a very uneven picture underneath, and the unevenness is the commercially useful part.
Localisation of core components sits between 45% and 70%, depending on the component:
- Harmonic reducers — over 60% domestic share, led by Leader Drive
- Controllers — 52% in general-purpose products
- Servo systems — 40%, with Inovance at around 20% domestic share
- RV reducers — roughly 33%, and still the weakest link in the chain
China now holds over 190,000 valid robotics patents, close to two-thirds of the global total, and became a net exporter of industrial robots in 2025. Exports rose 48.7% year on year; the first half alone saw 94,200 units shipped, worth $746 million and up 59.7%. The destinations tell you the use cases: Vietnam for electronics, Mexico for automotive, Thailand for auto parts.
Where China leads outright: AGV and AMR robots at 55% global share, collaborative robots, and humanoids — the overwhelming majority of humanoid robots delivered worldwide in 2025 were manufactured in China. Where it does not: high-precision articulated industrial robots, where Japanese and German suppliers still hold around 70% of the high-end market, and surgical robotics.
A note on robot density, because the figure is often misquoted. The IFR's World Robotics 2025 report puts China at 166 robots per 10,000 employees, ranking it 22nd globally — behind Germany at 449 and Japan at 446. Earlier reporting placed China far higher on a different denominator. The more meaningful number is installed stock: roughly two million units, the largest in the world and around four and a half times Japan's. China's automation is enormous in absolute terms and still thin relative to its workforce, which is exactly why the runway is long.
The policy shifted from subsidy to procurement
The instruments changed, and the change is instructive.
"Embodied intelligence" is now designated one of ten priority future-industry tracks, alongside integrated circuits and biomanufacturing, and robotics runs as a connective thread through the AI PLUS modernisation strategy. The ROBOT+ programme pushes deployment across ten sectors including manufacturing, agriculture, healthcare and logistics.
But the mechanism moved from R&D and equipment purchase subsidies to scenario-driven procurement and application demonstration. State and public institutions buy roughly 20% of newly released humanoid robots each year, manufacturing the initial demand rather than subsidising the supply. Over 40 robot training and data collection centres were built by the end of 2025. State-linked funds running to ¥1 trillion back AI and robotics startups. A dedicated export tax code for intelligent bionic robots took effect on 1 January 2026, enabling targeted rebates.
Creating a guaranteed first customer is a materially different intervention from paying for the factory, and it produces different companies — ones that have shipped and iterated rather than ones that have built capacity.
The unresolved problems are equally clear: high-end core components still imported, overcapacity in the low end, and a fragmented market with over 140 domestic complete-machine manufacturers chasing the same customers.
Three companies, three routes into Europe
The outbound strategy is best read through specific firms rather than policy documents.
Estun — acquire the technology. Estun bought Trio Motion, a British motion control specialist, in 2017, filling a controller gap. In 2020 it acquired Cloos, the century-old German welding robot brand, gaining automotive welding technology outright. It now runs four brands — Estun, Cloos, Trio and M.A.i — across different segments. By 2025 overseas revenue was 30% of the total, at a 36.75% gross margin, nearly ten points above its domestic business. That is the whole argument in one number: this is not a low-price strategy. Its Polish factory began production in July 2025, serving European automotive customers, and it has entered Volkswagen and BMW supply chains. Estun listed in Hong Kong in March 2026, the first A+H company in China's industrial robot sector, targeting 40% overseas revenue by the end of the year.
JAKA — earn the qualification. JAKA leads the domestic collaborative robot market and ranks second globally behind Denmark's Universal Robots. Its significant achievement is not a sale but an approval: three years and nearly fifty rounds of Toyota quality and technical assessment to become Toyota's sole global supplier of collaborative robots. Hundreds of its machines now run drive shaft grinding, loading and inspection at Toyota's Tianjin plant, raising efficiency 20% and cutting defects below 0.3%. In April 2026 it opened a base in Toyohashi, Aichi — under an hour from Toyota's headquarters. Products now sell in over 100 countries, with overseas revenue up more than 60% year on year in Q1.
UBTECH — enter the hardest room. In January 2026 UBTECH signed a service agreement with Airbus, delivering Walker S2 industrial humanoids to core plants in Hamburg and Toulouse for precision aircraft assembly. The machine works to ±2 mm with a 15 kg payload and holds AS9100 aerospace quality certification. On the pilot line it raised single-process efficiency 47% and cut defects 18% against manual work. A March 2026 agreement with Siemens brings in digital manufacturing software toward a 2026 production target of 10,000 humanoids. Overseas orders now exceed ¥1.1 billion.
Chinese industrial humanoids are inside European aerospace manufacturing. That is a fact worth sitting with.
How the state actually steers
Robotics is not on China's sensitive investment list. Most overseas projects go through a filing process, and 2026 online approvals cut processing times further. Acquisitions of high-end technology or transfers of core components still trigger national security review, as in any high-tech sector.
The interesting influence operates outside the regulations.
The China Robot Industry Alliance, under MIIT, functions as more than a trade body — it shares global market intelligence, coordinates overseas expansion, and works to prevent firms undercutting each other abroad. High-level diplomatic visits routinely include robotics executives, and that endorsement opens doors no sales team could. Large state investment funds visibly favour companies with global expansion plans and acquisition capability, using capital allocation to steer behaviour toward national strategy without instructing anyone.
Guidance through alliance coordination, diplomatic access and capital preference is harder to see than a regulation, and considerably harder to counter.
What this means if you are on the European side
The competitive threat is no longer price. Estun earning higher margins abroad than at home, and JAKA passing fifty rounds of Toyota audit, describe companies competing on capability. Strategies built on "they are cheap but we are better" have a shortening shelf life.
The acquisition targets are specific and identifiable. RV reducers remain the weakest link at around 33% localisation; high-precision articulated robots and surgical robotics remain European and Japanese strongholds. Those are precisely the gaps that Trio and Cloos were bought to fill. If you own one of those capabilities, expect an approach — and decide in advance what you would say.
Supplier qualification is the real moat, and it is depleting. Toyota's fifty assessment rounds took three years to pass. That barrier is genuine, and it has now been passed. European OEMs running dual-sourcing mandates should understand that qualifying a Chinese supplier is no longer theoretical.
The trajectory has moved from selling products, to acquiring technology, to localised production and global R&D networks. The government's contribution is policy direction, capital, diplomatic endorsement and industry coordination — not instruction. That combination is harder to compete with than a subsidy.