Insured robots becoming irreversible wave
Flexible policies cover design defects, errors in installation, operation; serve as hedge against any financial losses
The scale of coverage is already becoming visible at the platform level. Sharebot said that by April 24, more than 1,000 robots on its platform were insured, with aggregate cover exceeding 200 million yuan. According to domestic media reports, the platform purchased annual policies at premiums typically running to several thousand yuan per robot.
"Embodied robotics insurance is a prerequisite for a viable robot rental business model. It is more than a policy, but also an essential part of the industry's infrastructure, connecting insurers, robot manufacturers, service providers, partners and end customers," said Li Yiyan, CEO of the robot leasing platform. "For us, every robot on the platform must be covered by insurance and brought into our risk management system."
Pricing unfamiliar risk
Writing a policy requires an estimate of a loss that has not yet happened. For insurers entering robotics, there is relatively little past precedent to base assessments upon.
Guo Jinlong, a researcher at the Chinese Academy of Social Sciences' Institute of Finance and Banking, pointed to the absence of a public risk database, which makes emerging risks difficult to quantify, while limited product standardization and high costs make it harder to match coverage to customers' needs.
"Moving from insuring individual machines on an ad hoc basis to covering batches of robots through rental arrangements can help insurers draw on data from actual use to overcome data barriers and fragmented demand," said Guo.
He added that by pooling data from rental platforms and operating environments, insurers and their partners can jointly develop models that better capture risk, supporting dynamic pricing and real-time risk management, which would help refine pricing methods that have traditionally relied on manufacturers' data and past experience.
Determining liability is another challenge, as responsibility can be difficult to untangle when developers, manufacturers, rental companies, users and algorithm providers are all involved, particularly in the absence of dedicated regulations and judicial precedents.
In this regard, a consortium could provide a framework for insurers to share risks and develop services jointly.
"It brings several insurers to a business that would otherwise place the burden of underwriting on a single company," Chen at PICC said.
The Beijing intelligent robot insurance consortium now extends to arranging repairs and replacements when equipment fails. By bringing financial protection and after-sales support together, Chen said, the shared approach helps address both the concentration of risk in extended-warranty business and the practical difficulties of delivering such coverage.
Building a lasting market
Government measures are beginning to address existing barriers from both sides of the insurance market, helping insurers build the data needed to assess and price risk while making coverage more affordable for the businesses buying it.
In March, four central government departments, including the Ministry of Science and Technology and the National Financial Regulatory Administration, issued measures encouraging dedicated insurance for embodied intelligence. They called for model policy clauses, better data accumulation and differentiated pricing, as well as professional insurance pools in selected technology fields with substantial, insufficiently dispersed risks.
"It systematically answers whom to cover, what to cover and how," Guo said, adding that they also targeted the problem of fragmented policies by establishing coordination across departments.
Guo also highlighted provisions for insurers to share major risks and combine their underwriting capacity, as well as simpler, more affordable products intended to bring smaller technology companies into the insurance market.
In addition, local measures are reducing the cost of taking out coverage. Ningbo, Zhejiang province offers qualifying businesses subsidies of up to 80 percent of premiums for humanoid robot application insurance, capped at 2 million yuan. The policy took effect in July 2025 and runs through the end of 2027.
Elsewhere, Zhejiang's Hangzhou has an ordinance on embodied intelligent robotics, effective May 1, which encourages insurance products suited to the industry and market-based risk sharing across research, production, sales and application. Shanghai's technology insurance guidance also identifies humanoid robotics as a field for tailored protection.
"Driven by continued policy support and expanding industry demand, technology insurance is growing in scale, offering a wider range of products, developing a deeper supportive ecosystem and taking increasingly distinct regional forms," Guo said.
The effort is part of the nation's broader expansion in the protection of technological innovation, with policies reaching into research spending, the commercialization of new technologies and intellectual property.
Tian Lihui, director of Nankai University's institute of finance and development, expects the measures to change how that protection is delivered. Tian said he sees a shift from "passive claims settlement" to "active support", with insurers helping businesses manage and reduce risks before losses occur.
According to NFRA data, China's technology insurance sector provided about 8 trillion yuan in coverage in 2025, while premium income rose 44 percent from a year earlier, providing a solid ground for the policies now being written for robots and their components.
Tian also anticipates a move from "working alone" to "ecosystem collaboration", as insurance pools and shared data platforms help providers tackle pricing difficulties.
Looking forward, for a business putting a robot to work, those changes may be felt in practical ways: whether suitable cover is available, whether the premium is affordable and how efficiently a damaged machine can be assessed when a claim is made.
lijiaying@chinadaily.com.cn



























