bytevyte
bytevyte
Language
ai-beats —

China's Grip on Humanoid Robot Shipments Tightens as Global Volumes Rise 432%

humanoid robot shipments

Chinese manufacturers supplied more than 95% of the humanoid robots shipped worldwide in the first half of 2026, extending a lead that has turned the category from a Western research race into a Chinese production contest. Global humanoid robot shipments reached almost 25,000 units in the first six months of the year, a 432.1% increase over the same period in 2025, according to IDC. China-based operations accounted for more than 19,000 of those units, or 77.9% of the global total.

The two share figures describe different things, and both matter. The 77.9% measures where robots left the factory; the 95% measures who owns the brand, regardless of assembly location. Western developers including Apptronik, Agility Robotics, 1X and Boston Dynamics remain in a pre-scale phase, refining hardware and software before committing to volume production.

Revenue is climbing almost as fast, though not quite. IDC puts global humanoid robot revenue above $740 million for the half, up 322.7% year over year. Unit volumes grew faster than sales value, a gap that points to falling average selling prices as Chinese vendors compete primarily on price. That pressure is the part Western entrants will feel first.

The Humanoid Robot Shipments Surge in Numbers

The scale of the jump is clearest against the prior year. IDC measured Chinese vendors at roughly 90% of a global market that shipped about 18,000 units across all of 2025. The first half of 2026 alone cleared that full-year total by thousands of units, and China's share widened from about 90% to more than 95% along the way.

MetricH1 2026Comparison
Global shipments~25,000 units+432.1% year over year
China shipments>19,000 units77.9% of global
Global revenue>$740 million+322.7% year over year
IDC forecast for 2030>750,000 unitsRaised from prior estimate

IDC's forecast for humanoid robot shipments in 2030 now exceeds 750,000 units, roughly thirty times the current half-year run rate. Closing that gap requires the market to sustain the compounding it has shown for two consecutive periods, which in turn depends on deployments moving from pilots into daily operations at factories, warehouses and logistics sites.

The cost curve is moving with the volume. Average selling prices fell between the two comparable half-year periods even as revenue climbed, and that decline is what makes a business case possible for buyers who could not justify a humanoid at earlier price points. Lower prices widen the addressable market, which generates more volume, which pushes prices lower again.

Two Vendors, Half the Market

Concentration inside China is nearly as sharp as the concentration between China and everyone else. IDC counts AGIBOT at more than 8,600 units in the first half, about a third of the global total, putting it ahead of Unitree Robotics as the volume leader. Unitree shipped an estimated 5,900 units over the same stretch. Together the two vendors account for well over half of worldwide shipments, based on the published counts.

Independent counts diverge. Counterpoint Research puts AGIBOT at roughly 9,700 units and more than 43% of a market it sizes above 22,000 units. Smart Analytics Global counts about 8,400 units and a 44% share from roughly 19,100 units industrywide. IDC's own figures land between them. The spread reflects unsettled definitions of what counts as a humanoid robot and which deliveries get logged, and it shows how young the category is: even its baseline is still in dispute.

The direction is identical across the datasets. Growth rates reported by these firms range from roughly 270% to 432%, and each describes the same shift from prototypes to purchase orders, crediting Chinese suppliers with the bulk of the volume.

Why Volume Wins

The lead compounds through manufacturing economics. A vendor shipping eight or nine thousand units a year buys actuators, harmonic drives, sensors and batteries at prices a developer shipping a few hundred cannot match. That cost advantage feeds back into pricing, and pricing wins the contracts that fund the next production run. Each cycle widens the gap instead of closing it.

China's industrial base hands domestic vendors infrastructure Western startups must assemble themselves: component suppliers within a short drive of assembly lines, a large home market willing to trial the hardware, and state-backed financing that tolerates thinner margins during a land grab. Bank of America has estimated that China supplied 95% of the roughly 20,000 humanoid robots delivered globally in 2025, so the current share builds on ground already secured.

The second-order effect lands on suppliers. Component makers that win design slots with high-volume Chinese vendors scale alongside them, while suppliers tied to low-volume Western programmes carry higher unit costs and less predictable order books. That pulls engineering capacity toward the vendors already shipping in the tens of thousands, deepening the same advantage one layer further down the stack.

Western developers are still working through the pre-scale phase. Apptronik, Agility Robotics, 1X and Boston Dynamics are perfecting their systems before shipping in volume, which means they will enter the commercial fight after Chinese rivals have anchored price expectations with buyers. For their investors, the arithmetic complicates the fundraising conversation: backing a developer that ships hundreds of units means underwriting a path to a cost structure rivals already operate at scale.

Few competitive levers work quickly against that gap. Software differentiation, service coverage and safety certification can justify a premium for some buyers, but none of them offsets a component cost disadvantage that widens with every production run. The realistic path runs through either matching Chinese scale or competing on capabilities the shipment data does not measure.

Shipments are not the same as deployed labour, and the distinction matters for anyone reading the headline number as evidence of a factory takeover. A unit counted as shipped may sit in a distributor's inventory, run in a controlled pilot, or exist purely as a demonstration model. The revenue figure of more than $740 million across roughly 25,000 units implies an average selling price near $30,000, a level consistent with early commercial and research buying, well short of mass industrial replacement.

Demand mix matters as much as volume. Research institutions and pilot programmes absorb a large share of early units, and that demand is finite by nature. Industrial deployments are where repeat orders and pricing power eventually sit, so vendors that convert pilot placements into recurring factory and logistics contracts will pull away from those selling one-off units to labs.

For procurement teams, the immediate effect is choice. Buyers who once faced a handful of Western vendors quoting six-figure systems now have Chinese suppliers offering comparable form factors at a fraction of the price. The trade-off runs past hardware into data governance, service networks, software ecosystems and supply-chain security, and some industrial buyers will weigh those factors as heavily as unit cost.

Supply agreements signed during this volume phase tend to lock in for years. If the 2030 projection holds, the vendors setting component standards today will define the interfaces, tooling and service models that later entrants inherit, which makes the current half-year data more consequential than a single period of sales would suggest.

Why this matters

The humanoid sector is now a manufacturing contest, and the current scoreboard favours the side with factories. For Western developers, the challenge is financing a cost base that can compete with rivals shipping thousands of units a quarter. For buyers and investors, the concentration of humanoid robot shipments in China is a strategic variable that shapes contract terms, component sourcing and the resilience of the hardware pipeline they will depend on.

✔Human Verified


Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.