The race for humanoid robots is quietly sweeping across China's county-level economies.
Not long ago, a local leader shared an investment attraction story during a visit to Shangyu District. During negotiations with a leading industry player, the local government proposed using specialized chemical industry scenarios as an alternative to traditional policy support. Unexpectedly, the company immediately agreed to the proposal.
Currently, many county-level regions are planning embodied artificial intelligence innovation centers. More and more industrial parks are competing for the same batch of top-tier companies, with strikingly similar preferential terms in numerous policy packages. This strategic layout for future industries is increasingly showing signs of intense competition.
While others are still competing on subsidies and land prices, Shangyu is already enabling companies to enter workshops with work orders in hand. This seemingly counter-intuitive choice precisely reveals the most brutal and authentic logic in the county-level embodied AI competition—the decisive factor in this 'street fight' is not who offers the thickest policy package, but who has the clearest strategic vision.
The County-Level Race for Embodied AI
In 2025, "embodied intelligence" was first included in the Chinese government work report. Within less than two years, over ten provinces and municipalities including Beijing, Shanghai, Guangdong, Zhejiang, Shandong, and Chongqing have intensively issued supportive policies. Statistics show that China's embodied AI market size is expected to exceed one trillion yuan this year—a pie so large that every county-level economy wants a piece.
Jiangsu's Kunshan aims to build a "new hundred-billion-level industrial cluster" for embodied AI. Shandong's Pingyin is investing in constructing the province's largest humanoid robot training ground. Guangdong's Huiyang signed agreements to introduce 13 embodied AI enterprises within a year. Fujian's Longwen held the city's first dedicated investment promotion conference for embodied AI... The list goes on.
However, warning signals are already flashing. The National Development and Reform Commission has publicly cautioned: "While this is a good thing for encouraging innovation, we must also focus on preventing risks such as the 'crowding' of highly similar products entering the market and the compression of R&D space."
This "follow-the-trend and rush-to-act" approach is making this county-level "street fight" increasingly crowded. Zhejiang is no exception. Shangyu is building a comprehensive service center for embodied AI. Jiashan has released its "15th Five-Year Plan" for intelligent robotics development. Xiangshan has established the Ningbo Embodied AI Industry Base. Longyou is independently researching quadrupedal and humanoid robots... For this hot sector, how can county-level regions achieve success through unconventional strategies?
Qian Linjiang, Secretary of the Shangyu District Committee, stated a plain truth: "County-level regions often lack the capacity to lead in R&D. They must adapt to local conditions, seek differentiated development, and focus the breakthrough point for embodied AI on industrial implementation."
Top universities, high-level laboratories, and massive venture capital—these "standard" resources for developing embodied AI are indeed not necessarily available to a single county. However, the core advantage of Zhejiang's county-level economies lies in the characteristic industrial clusters formed through the evolution of its "block economy," which has built a nationally leading manufacturing foundation and a complete industrial chain ecosystem.
Taking Shangyu as an example, the global motor leader Wolong Electric Drive has accumulated decades of expertise in servo motors and precision transmission. It holds key "hard currency" components in the robot supply chain that are in high demand, such as frameless torque motors, harmonic reducers, and six-dimensional force sensors.
Leading companies are arriving following the industrial chain. The national base for Zhiyuan Special Robots has commenced construction. UBTech has strategically acquired a controlling stake in Shangyu's Fenglong Co., Ltd.. The Yifei Industrial Robot manufacturing project has landed. To date, Shangyu has gathered 26 core enterprises in the embodied AI industry chain, with total investment exceeding 10 billion yuan.
An industry insider commented: "Traditional industries originally had low acceptance of new things, but once the embodied AI industry couples with the existing industrial foundation, the resulting chemical reaction is enormous."
Application Scenarios Become the Trump Card
Observant individuals will notice that many regions have been playing the "scenario card" this year. Ningbo issued its first batch of "Artificial Intelligence + Manufacturing" application scenario opportunity lists. Hangzhou's Binjiang District released an artificial intelligence "scenario list." Suzhou Industrial Park launched an intelligent connected vehicle scenario opening plan. Hefei's Baohe District opened test roads for autonomous driving...
Currently, many embodied AI companies are not short of funds or technology; what they lack are implementation scenarios and validation environments. When the Ningbo Investment Promotion Bureau went to Hangzhou to meet Wang Xingxing from Unitree Robotics, the first requirement raised was for scenarios.
From a professional perspective, embodied AI "craves scenarios" because without real-world scenarios, it cannot obtain the multimodal dynamic data required for training, cannot validate generalization capabilities, and finds it difficult to form a commercial closed loop.
Whether it's the nationally top-ranked Shangyu Fine Chemical Industrial Park or the operational site of the world's largest port, Ningbo-Zhoushan Port... these are the testing grounds that AI companies dream of.
When attracting a Shanghai-based robotics company, Shangyu adopted a "reverse切入" strategy—leveraging the local chemical park's needs for intrinsic safety and green, low-carbon transformation to lock in special scenario applications, which happened to be precisely what attracted the target company.
"You provide the manufacturing, I provide the scenarios. It was a perfect match," said a person in charge involved in the investment promotion.
Chemical industry scenarios were identified as the first breakthrough point. In the Shangyu Fine Chemical Industrial Park, there are many links involving "difficult recruitment and high risk," such as high-risk inspections, hazardous material handling, and packaging discharge. These are the most fitting "rigid demand scenarios" for embodied AI.
In the liquid chlorine warehouse of Zhejiang Runtu New Materials Co., Ltd., a quadruped robot equipped with multiple sensors shuttles through the storage tank area according to preset paths, constantly monitoring leakage risks.
"Even if this trial is successful, the company might not buy many robotic dogs, but we were still very willing to take the initiative to come," said an engineer from Zhiyuan Robotics.
Earlier this year, Shangyu made a somewhat "incredible" decision—establishing a Scenario Cultivation and Application Office and unveiling a "Scenario Opportunity List" worth 3 billion yuan. Daily, a large number of enterprise demand lists flow into the newly established Scenario Office and are then precisely directed to robotics companies that have settled in Shangyu.
The person in charge, Jin Quanwei, joked that the office functions like a "matchmaking agency."
The "Juechen C5" commercial cleaning robot produced by Zhejiang Zhiding Robot Co., Ltd. was initially just a laboratory-grade product. After being "polished" for several months in Kanglongda's workshop, the product matured and became popular in the market. This year, over 2,000 units were sold, with orders worth over 40 million yuan from overseas markets including Southeast Asia.
The greatest significance of scenario-based investment attraction is shifting the optimization of the business environment from the traditional "policy红利" to the more attractive "opportunity红利."
Forging the Ultimate Advantage
Having manufacturing and scenarios alone is not enough. Other places can replicate these with enough determination. It's best to also forge one's own unique strengths.
In early July, the Yangtze River Delta Center for the Embodied AI Testing Laboratory opened to the public in Shangyu. This is one of the first laboratories in the country to conduct testing under new standards. Relying on the technical strength of the China Academy of Information and Communications Technology, it specifically provides authoritative testing services for embodied AI enterprises.
What seems like an ordinary laboratory is actually Shangyu targeting a more specific and practical entry point—testing and certification services.
The embodied AI industry chain is long. Why specifically choose the "testing" link? The answer lies in the industry's collective pain points.
Currently, the embodied AI industry is at a critical juncture of transitioning "from prototype to product." A large number of robots are getting stuck in stages like prototype testing and product certification on their way to the market.
"With so many robots on the market, how do we judge if they are qualified? We are unsure and dare not buy," said Xia Weiping, General Manager of Zhejiang Weihua New Materials, expressing the concerns of many on the demand side.
This testing laboratory happens to solve this problem. In real-world applications, robots cannot always operate under ideal conditions—emergency stops, turns, posture deviations, center of gravity instability, external interference. These seemingly basic performances need to withstand rigorous testing. The same applies even more to complex terrain navigation, special scenario adaptation, and dynamic interaction response.
Wang Yao, Senior Engineer at the AI Research Institute of the China Academy of Information and Communications Technology, believes that although the industry has not yet formed a mandatory batch sampling inspection system, with the rapid development of the industry, obtaining a "qualified report" from an authoritative institution before products enter the market will be an inevitable process.
"Testing and certification are like issuing a 'birth permit' for robots. Whoever establishes the standards first gains the power of discourse."
Leading companies have already taken action upon hearing the news. Zhiyuan Robotics has connected with the testing laboratory and plans to be the first to conduct comprehensive testing for its chemical inspection robots.
In this county-level embodied AI "street fight," the real decisive factor is no longer who follows the closest, but who can find their own "irreplaceable" advantage earliest in the crowded track.