China Reportedly Ships 97% of Humanoid Robots

China humanoid robot shipments have reached a headline-grabbing number: Chinese manufacturers reportedly supplied more than 97% of global shipments in the first half of 2026. But the figure does not mean that 97% of the world’s productive robot workers are operating in Chinese factories—or even that every shipped machine is performing sustained paid work.

The 97% estimate comes from market researcher Smart Analytics Global and was reported by Bloomberg on August 10. The firm counted roughly 19,100 humanoid robot shipments worldwide during the first six months of 2026, up from about 5,100 a year earlier. Shanghai-based AGIBOT was credited with 8,400 units, or 44% of the global total, while Unitree reportedly shipped 5,900.

Those numbers point to a formidable manufacturing lead. They also raise the question that matters more to customers, workers and investors: what exactly happens after a robot leaves the factory gate?

The headline numbers

Metric	Reported result	What it measures
Global H1 2026 shipments	About 19,100	Units classified and counted as shipped by Smart Analytics Global
Chinese vendor share	More than 97%	Share of the reported shipment total
AGIBOT	About 8,400 / 44%	Vendor shipments in the same dataset
Unitree	About 5,900	Vendor shipments in the same dataset

Smart Analytics Global forecasts around 60,000 humanoid shipments for full-year 2026 and about 500,000 by 2030, according to Bloomberg’s account of the data. In an earlier April market release covering humanoid and quadruped robots, the researcher described enterprise demand, maturing supply chains and falling costs as the main growth drivers.

Another research company broadly supports the regional imbalance. ABI Research says China currently accounts for 97% of humanoid robot deployments and estimates that the global installed base could rise from 24,000 in 2025 to 248,000 by 2030. That is a different dataset and a different metric, so the two 97% figures should not be treated as a single independently replicated result.

Four stages that are often confused

Humanoid-robot market claims frequently use production, shipment, installation and deployment as if they were interchangeable. They are not.

Produced: a robot rolls off a production line and passes internal checks.
Shipped or delivered: the unit moves to a distributor, laboratory, customer or project site.
Installed or deployed: the robot is placed in an operating environment, which may still be a pilot or validation project.
Productive operation: the machine repeatedly completes useful work at an acceptable cost, uptime and safety level.

A robot can be counted at the first two stages while spending months in integration, training or demonstration. Some shipments go to universities, data-collection centers and showrooms. Others support genuine industrial tasks but only inside a carefully structured workstation.

Public market summaries rarely disclose enough methodology to reconstruct every vendor count. Definitions of “humanoid” also vary. Some datasets focus on full-size, two-legged machines; others include smaller educational platforms, wheeled humanoids or broader “embodied intelligence” products. That makes cross-report comparisons difficult.

China’s production advantage is still real

The methodological caution does not erase China’s scale. TrendForce projected in April that Chinese humanoid output would grow by as much as 94% in 2026, with Unitree and AGIBOT approaching a combined 80% of shipments. It linked the acceleration to standardized supply chains, strong domestic component production and order-driven manufacturing.

Company milestones reinforce the picture. AGIBOT said its 10,000th humanoid robot rolled off the line in March and announced a 15,000th “embodied AI robot” milestone in June. The broader wording matters because AGIBOT sells several robot forms. The company also reported approximately 100 cumulative hours of livestreamed factory operation by its G2 system in a tablet quality-inspection process—useful evidence of a real task, but not the same as independently audited fleet-wide productivity.

Unitree says it is testing an embodied-intelligence model in its own manufacturing facility, where robots help assemble other robots. UBTECH reported 1,079 sales of full-sized embodied-intelligence humanoids in 2025, with more than 80% directed to industrial scenarios. Its Walker S2 is being tested with Hitachi in elevator manufacturing, but the partners explicitly describe the work as operational testing and application validation.

This is how early commercialization usually looks: production milestones, customer orders and increasingly serious pilots appear before standardized uptime, return-on-investment and safety data become public.

What are the robots actually doing?

The most credible near-term work is narrower than the “general-purpose robot worker” image suggests. Publicly described tasks include:

moving and sorting totes or components;
quality inspection on electronics and automotive lines;
loading, unloading and pallet-related handling;
collecting training data in industrial environments;
showroom navigation, reception and entertainment;
research, education and software development.

These can be economically useful, but many depend on controlled floors, predictable object placement and extensive integration. A humanoid may be valuable precisely because it can use a human-designed workstation without rebuilding the entire line. Yet a conventional robot arm, mobile robot or purpose-built automation system can still be cheaper and more reliable for repetitive tasks.

An Associated Press investigation published in June found the same tension: Chinese makers cited thousands of orders, while industry observers warned that demand could lag manufacturing capacity and that many machines remain more performative than functional in messy real-world settings.

Why China can scale faster

China combines several advantages in one ecosystem:

Component depth: motors, reducers, batteries, sensors, electronics and machining capacity are available close to final assembly.
Manufacturing feedback: robotics companies can test products inside dense industrial networks and iterate quickly.
Domestic demand: manufacturers, research institutions, local projects and state-owned companies provide early customers.
Price pressure: competition encourages lower hardware prices and standardized production.
Policy support: humanoid robots and embodied intelligence are strategic technology priorities.

The United States retains strengths in frontier AI, chips and venture funding. Europe has expertise in industrial automation, safety engineering and premium machinery. But China is currently turning more humanoid designs into physical units—and every unit can potentially generate operational data for the next software iteration.

What the 97% figure does not prove

The shipment share does not establish that Chinese robots are more autonomous, more reliable or cheaper over their full working life. It does not reveal utilization rates, human supervision, maintenance hours, task success, safety incidents or customer renewal.

It also describes an extremely young market. Nineteen thousand units would be a remarkable acceleration for humanoids, but it remains tiny beside the installed base of conventional industrial robots. A dominant percentage in a small, rapidly changing category can move sharply as soon as a major non-Chinese producer begins volume manufacturing.

For buyers, the better questions are operational:

How many hours does the robot work between interventions?
What percentage of tasks are completed without teleoperation?
How long does integration take?
What is the cost per successful task?
Can the system recover safely from errors?

What it means for Europe

Europe does not need to match China robot for robot to remain competitive, but it cannot ignore the learning effects of scale. Larger fleets improve procurement, manufacturing quality and the flow of real-world data. European suppliers risk being pushed toward a high-cost niche if Chinese platforms become the default hardware layer.

The opportunity is to compete where industrial customers are most demanding: certified safety, dependable integration, privacy-preserving data systems, maintainability and measurable return on investment. Partnerships with manufacturers can turn European factories into validation environments rather than waiting for fully mature products to arrive from abroad.

Bottom line

The reported 97% share is a powerful sign of China’s manufacturing lead, not a census of productive robot workers. The next phase of the race will be decided less by how many humanoids leave a factory and more by how many can perform useful tasks, day after day, with limited supervision and defensible economics.

Until vendors and researchers publish standardized deployment and productivity data, shipment rankings should be read as an early industrial signal—not as proof that the general-purpose humanoid era has already arrived.

Sources and further reading:

Editorial note: Shipment and deployment figures come from market-research estimates and company disclosures. Definitions differ, and Alpha Bionic has not independently audited the underlying vendor data. The feature image is an original AI-generated editorial illustration.

Bewerte den Beitrag hier!

[Total: 0 Average: 0]

Nico Nuss

Website

Author Nico Nuss has been working on mobile computing and automation software since 2001. Drawing on his experience and strong interest in future technologies, he focuses on robotics and AI.