--- title: "400 Humanoid Models, 80% of Global Shipments: Inside China's Robot Industry Scale-Up" date: 2026-09-08 category: Humanoids & Robotics site: NeuroAI canonical: https://neuroai.site/a/na-humanoid-400-models language: en --- # 400 Humanoid Models, 80% of Global Shipments: Inside China's Robot Industry Scale-Up > Chinese makers now ship more than 400 humanoid robot models — over half of the world's total — and account for roughly 80% of global humanoid and quadruped sales. Scale, not spectacle, is the new story. For most of the last decade, the public image of humanoid robotics was a single company's YouTube channel. In 2026, the image is a spreadsheet. Chinese industry data cited by state media in August 2026 puts the number of domestic humanoid robot products at **more than 400 models** — over half of all humanoid models worldwide — with Chinese humanoid and quadruped robots accounting for about **80% of global unit sales**. That is a different kind of claim than "our robot can do a backflip." It is a claim about industrial capacity, and it is the reason the centre of gravity in robotics has moved. ## Key takeaways - **Model count:** 400+ domestic humanoid products, more than half of the global total (as reported by Chinese state media, August 2026). - **Market share:** roughly 80% of global humanoid and quadruped robot sales by volume. - **Company count:** more than 6,600 AI companies in China as of June 2026, about 15% of the world's total, spanning base infrastructure, model frameworks and industry applications. - **Commercial proof:** at the 2026 World Artificial Intelligence Conference in Shanghai, 177 international procurement delegations attended, with intended purchase value reported at about **20.36 billion yuan**, up roughly 25% year on year. ## Why the supply chain decides this race A humanoid robot is roughly 40 categories of precision components: harmonic reducers, planetary roller screws, frameless torque motors, dexterous hands with tactile sensing, six-axis force sensors, and the batteries to make all of it mobile. China's advantage is not that it invented these parts. It is that it already manufactures them at consumer-electronics cost curves. The same industrial belt that produces smartphone actuators and electric-vehicle drivetrains can produce a robot knee actuator at a fraction of the cost of a bespoke Western equivalent. This is why the price of a capable Chinese humanoid has fallen into the range of a mid-sized car, while comparable Western platforms remain priced like laboratory equipment. Price is not a marketing detail in robotics — it determines whether a university can buy twenty units or one, and whether a factory can justify a pilot line at all. ## From stages to shop floors The criticism of humanoid robots — that they are performing rather than working — is fair, and the industry's own deployment data confirms that most units today go to **research, education and public exhibition** rather than production lines. But the trajectory matters more than the snapshot. Chinese state broadcaster reporting in 2026 documents industrial cases that were proofs-of-concept two years earlier: - A "navigator-level" smart factory in Xuzhou, Jiangsu, received simultaneous international orders for nine crane models. Its production system replanned thirty days of operations autonomously, and production lines reconfigured in **about ten minutes** — a changeover that previously took five to six hours. - Midea's humanoid robot "Meiluo 2" works quality inspection on a water-heater line: production cycle time cut by **20%**, production loss down **50%**, daily capacity up **33%**. - Acoustic large models and inspection robots at a Ningxia wind farm are projected to cut inspection labour by about **3,000 hours a year** and reduce labour cost by roughly **60%**. None of that requires a robot that can dance. It requires one that can show up for a shift. ## The honest limits Two constraints are real and openly discussed inside China's own policy research: **Bottleneck technology.** High-end chips, core algorithms and basic software remain described as "chokepoint" problems in mainstream Chinese analysis. The robot body is not the hard part forever; the compute inside it is. **Deployment depth.** Official commentary notes AI in Chinese manufacturing is still moving from local pilots to scaled deployment. Smart factory counts are impressive — 15 "navigator-level", 500+ "excellent-level", 8,200+ "advanced-level" plants per Ministry of Industry and Information Technology data — but penetration is uneven across sectors. The fair reading: China has won the manufacturing layer of humanoid robotics, is mid-transition on the deployment layer, and has not yet won the intelligence layer. That is exactly the sequence a hardware-first industry follows. *Figures are as reported by Chinese state media, MIIT disclosures and conference organisers in 2026.* --- Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-humanoid-400-models Free to quote with attribution and a link to the original.