China's humanoid robot procurement data for June 2026: over 59 projects, exceeding 1.112 billion RMB, a month-on-month surge of over 227%.
*Scroll down for this image
Image source: Reference to the Humanoid Robot Scenario Application Alliance, compiled by MIR
This time, we are not just focusing on the numbers. We want to explore the logic behind the deployment and cooperation of humanoid robots—Who is paying? Who is delivering? Which scenarios are they entering? Is the industrial chain truly up and running?
After reviewing all public projects in June and deployment cases in the first half of the year, we have identified several structural changes taking place: In June, there were only 12 government platform projects, yet they contributed about 70% of the total amount. Universities remain the most active buyers—Tsinghua University, Fudan University, Southeast University, Central South University, Nankai University, Dalian University of Technology, Harbin Engineering University... almost every 985/211 university is procuring. However, the single transaction amount for universities is relatively small, mostly ranging from 100,000 to 2 million RMB. Although there are only 12 government platform projects (accounting for 20% in number), the "big money" in the current humanoid robot market comes from local governments and industrial platforms. What they are buying is not just a product, but an entire industry—innovation centers, data bases, exhibition platforms, and industrial clusters.
In summary, there are four changes:
① The purchasing entities have changed; government platforms are replacing universities as the largest contributors in terms of amount;
② The procurement targets have changed; orders for industrial bases and data centers are beginning to replace single equipment procurement;
③ Application scenarios have changed, rapidly penetrating from laboratories to industrial manufacturing, urban services, and data infrastructure—shifting from "demonstrations" to "real work";
④ The locations of purchasing entities have changed, sinking from first-tier cities to second- and third-tier cities, and even county-level units. Data collection and training centers are becoming new infrastructure that governments and industries are competing to invest in.
01. Billion-Yuan Projects on Government Platforms and the Underlying Application Cooperation Models
The biggest highlight in June is the three major government/industrial platform projects, which reveal three typical paths for the current application of humanoid robots on government platforms.
Image source: Compiled and collected by MIR
[Yibin · 236 Million] Galbot
In 2025, Yibin listed AI (Artificial Intelligence) as the fourth major future industry, injecting 11.5 billion RMB into funds and opening a 2 billion RMB scenario list. The essence is to use robots for the intelligent integration of traditional industries. Galbot's wheeled robots (patrol/guidance/cleaning at high-speed rail stations) + heavy-duty robots (factory logistics handling) + retail cabins perfectly cover urban services + industrial logistics + commercial consumption. In July 2025, Galbot landed its Southwest Training Center in Yibin, won the bid for the first batch of procurement in August, signed a manufacturing base in March 2026, and secured a massive order of 500 units in June—completing the closed loop of "pilot → verification → capacity → scale" in four steps. This also allowed it to deeply participate in Yibin's digital transformation process, making the cooperation mutually beneficial.
[Nanyang · 130 Million] ZQ Robotics
ZQ Robotics laid out its operations in Nanyang in 2025, and this order represents a leap from pilot to base-oriented development. Once the training base is up and running, subsequent model iterations, enterprise settlement, and scenario verification will revolve around this platform—securing the data entry point first is the prerequisite for having a say in the ecosystem later. For ZQ Robotics, Nanyang is a testing ground for low-cost verification of large-scale data collection; for Nanyang, ZQ Robotics is a grip to bind the "humanoid robot data infrastructure."
[Huizhou · 213 Million] Zhejiang Zhaoju + Henan Zhongcheng
The essence of the Huizhou project is to undertake the industrial spillover from Shenzhen and seize the supporting position of the humanoid robot industry in the Greater Bay Area. Shenzhen has a large number of humanoid robot R&D enterprises (such as UBTECH and LimX Dynamics), but faces land scarcity and capacity constraints. Huizhou uses the innovation center for testing and verification, the data collection center to generate data, and the Robot World for exhibition and investment promotion.
The three functional modules—the innovation center, the data collection center, and the Robot World—form a complete industrial undertaking closed loop, marking that the layout of the humanoid robot industrial chain in the Greater Bay Area is evolving from "single-point enterprises" to "regional industrial clusters."
02. Data Collection and Training Centers Are Becoming an Infrastructure-Level Track for Humanoid Robots
The above three projects have one thing in common—they all include investments in data collection/training infrastructure. This is no coincidence.
The biggest bottleneck for current humanoid robots is not hardware, but data. Real physical interaction data, multimodal operation data, and long-tail fault data—these critical data that determine whether a robot "can do the work" cannot be provided by simulation environments. The core pain point solved by data collection centers is precisely this: placing robots in real factories, logistics parks, and commercial scenarios to collect "working" data 24/7 without interruption, and then feeding it back into model training.
Three driving forces are turning "data collection centers" into an independent infrastructure track:
Local government perspective: Whoever builds the data infrastructure first will be able to attract robot enterprises to settle in and take the lead in formulating industry standards. Data centers are "asset-light and leverage-heavy"—investing in venues and computing power in exchange for enterprise aggregation and technological precipitation.
Robot enterprise perspective: The cost of building an in-house collection team is extremely high, and leveraging government platforms is the most cost-effective choice at present. This is equivalent to the government bearing the fixed costs, while enterprises only need to contribute technology and operational capabilities.
Industrial ecosystem perspective: Once data collection centers reach a certain scale, they will naturally become hubs for model training, scenario verification, and talent aggregation. Enterprises that occupy the data entry point first will gain a structural advantage in subsequent competition.
03. Insights into the Top Five Industries with the Largest Scale of Humanoid Robot Deployment in the First Half of the Year
On the enterprise side, winning bids in June accounted for only about 6% of the data, but the actual deployment cooperation goes far beyond these public tenders. A large number of cooperations are conducted through business negotiations, sole-source procurement, and internal enterprise projects. We have reviewed the confirmed factory-entry cases in the first half of the year industry by industry:
Industry 1: Automotive Manufacturing (the industry with the largest deployment scale) High degree of production line standardization, calculable ROI per unit, and mature safety redundancy design;
Industry 2: 3C Electronics Manufacturing (the scenario with the highest precision requirements) The flexible advantages of humanoid robots are fully utilized here;
Industry 3: NEV (New Energy Vehicle) Batteries (CATL's self-research, self-production, and self-use) For other battery enterprises, since CATL has made it work, it is only a matter of time before they follow suit;
Industry 4: Aerospace Manufacturing (high value-added scenarios);
Industry 5: Special Industries (large-scale potential demand such as power grid inspection).
The following is a review of specific confirmed cooperation cases in the first half of 2026 by industry, showcasing the complete landscape of current industrial collaboration:
Image source: Compiled and collected by MIR
From the perspective of cooperation case density, automotive manufacturing is currently the track with the most deployed enterprises and the most sufficient competition. Multiple enterprises such as UBTECH, AGIBOT, Xiaomi, Galbot, and Beijing Humanoid Robot Innovation Center have already entered; 3C electronics and NEV batteries are in a critical stage of transitioning from pilot verification to large-scale deployment; special industries such as power grid inspection and water conservancy inspection currently have a small number of orders, but the individual volume is extremely large. Once they enter the release cycle, the procurement scale will significantly exceed existing industrial scenarios.
04. Procurement Entities Flourishing in Multiple Locations
The distribution of projects in June clearly outlines the current geographical landscape of the humanoid robot industry—it is no longer dominated solely by Beijing, Shanghai, Guangzhou, and Shenzhen, but is flourishing across multiple locations, each with its own characteristics.
Image source: Compiled and collected by MIR
The Pearl River Delta is the biggest winner in June—Huizhou with 213 million RMB, Foshan with 167 million RMB, plus multi-university procurement in Guangzhou, totaling nearly 400 million RMB, far exceeding other regions. This is strongly correlated with the depth of the manufacturing base in the Pearl River Delta itself and the strong support of local government industrial funds. A noteworthy trend is forming: first-tier cities serve as R&D headquarters, while second- and third-tier cities serve as bases and demonstration applications.
The core signal conveyed by the geographical landscape is that the humanoid robot industry has entered a nationwide competition stage. Local governments are no longer waiting and watching, but taking direct action—whoever has more complete industrial supporting facilities and more flexible landing policies will be able to attract more enterprise resources.
Conclusion
In the second half of 2026, the humanoid robot industry will face verification pressure on two paths simultaneously: on the government side, whether the "10,000-unit level" real-scenario training goals can be released in a concentrated manner in Q3-Q4; on the enterprise side, whether manufacturing leaders can move from "R&D verification" to "mass production procurement."
Looking ahead, enterprises that can both secure the data entry point in government projects and accumulate real ROI in industrial scenarios are most likely to navigate through cycles and become the leaders in the next stage.