Omdia: global humanoid robot shipments rose more than 5x YoY in 2025 to ~13K units; China accounted for the vast majority, led by AgiBot shipping ~5,168 units
Context & Ripple Effects
The reported shipment surge turns humanoids from a largely prospective category into a measurable hardware market, with China supplying most of the volume and AgiBot emerging as the largest named shipper in the earlier Omdia shipment estimate.
Later coverage adds important qualification: education and R&D accounted for 42% of China’s 2025 humanoid shipments, indicating that much of the early installed base is still being used to develop capabilities rather than deployed in mature commercial roles.
First-order effects
- China-based humanoid makers gain a concrete scale signal: roughly 13,000 global shipments in 2025, with AgiBot responsible for about 5,168 units and China accounting for the overwhelming share.
- Buyers and developers in China receive a larger deployed fleet for experimentation, data collection and human-robot interaction, even as the reported use mix suggests limited evidence of broad end-market deployment.
Second-order effects
- Rival humanoid vendors face greater pressure to demonstrate shipment scale and credible customer use cases, rather than prototype performance alone.
- A larger Chinese installed base can reinforce demand for components, integration and robot-training services; the later breakdown of education and R&D-led demand suggests these adjacent markets may develop before repeatable workplace automation does.
Third-order effects
- If shipment growth persists, humanoid robotics could increasingly be shaped by manufacturers that can pair hardware production scale with access to deployment environments for iterative learning.
- The concentration of early volume in one country raises the prospect that regional ecosystems—not only individual robot designs—will determine who sets the cost, supplier and application benchmarks for the category.
The trend: Humanoid robotics is entering an early industrialization phase in which shipment scale and real-world learning loops may matter as much as headline demonstrations.