Consumers have not bought robots the way they buy phones or appliances. The price is too high, the use cases too narrow and the technology too early to justify a household capital expenditure that could run tens of thousands of dollars. Retail prices start around $19,000 for entry models and climb past $100,000 for advanced ones.
AGIBOT, one of China’s leading humanoid robot makers, launched Sharebot in December 2025 as a direct response to that barrier. Customers can rent a humanoid robot for as low as 3,500 yuan, roughly $517 a day, including shipping and a human operator who helps program and control the machine on-site. In three months, the platform logged more than 5,500 orders. AGIBOT said the robot rental market could reach $1.5 billion by the end of 2026, CNN reported.
The model shifts the economics of physical AI in a specific way. Rather than asking customers to bet thousands of dollars on hardware that may sit idle, rental converts a capital decision into an operating one — the same shift that turned software into a subscription business — under the label robotics-as-a-service.
Rental Model Partly Solves the Adoption Problem
The use cases driving Sharebot’s early volume are not the ones robotics companies imagined. Customers are renting humanoids for corporate galas, retail promotions, trade shows, tourism, and live-stream content creation, not factory floors or household labor. Humanoids Daily reported that AGIBOT’s rental portal at MWC 2026 in Barcelona positioned robots as “cyber intelligent hosts” for events, performers for private celebrations, and traffic-driving co-hosts for social media influencers. The rental price in Europe starts at 899 euros (about $1,027) a day. The service now covers 17 countries and regions, including Spain, Germany, France, the U.K. and North America.
That positioning reflects a practical reality about where humanoid robots actually work reliably today. CNN reported that inside a state-backed training facility near Beijing, more than 120 humanoids stand in rows, each performing a single task repeatedly, guided by human trainers with handheld controllers. Robots can dance, sort packages or scoop popcorn. They cannot yet adapt fluidly to the range of tasks a household or factory floor demands. The rental model sidesteps that constraint by deploying robots in controlled, short-duration scenarios where their current capabilities match the job.
Li Yiyan, CEO of Sharebot, told CNN that the rental market is not just a revenue stream. “It ultimately helps robots transition faster from mere exhibition pieces to large-scale deployment,” he said. Each rental generates operational data from real-world environments, feeding back into the software systems that determine what the next generation of robots can do.
Each Deployment Generates Training Data That Improves the Next Robot
Physical AI has a data problem and rental deployments are emerging as one way to solve it. Virtual simulations can teach robots to perform choreographed routines, but not how to grasp and move unpredictable objects, because simulations struggle to model real-world physics accurately, MIT Technology Review reported.
Robotics companies are now spending more than $100 million per year on real-world movement data from collection firms, Micro1 CEO Ali Ansari told MIT Technology Review. Scale AI collected more than 100,000 hours of robotic demonstration data from its lab and global contributors. Every rental deployment generates the kind of physical-world interaction data that no simulation can replicate.
Nvidia CEO Jensen Huang said at CES in January 2026 that the ChatGPT moment for robotics had arrived, PYMNTS reported. Morgan Stanley estimates there could be 1 billion humanoid robots in use by 2050, representing a market of more than $5 trillion, though adoption at scale is not expected to accelerate for at least another decade.
AGIBOT’s Sharebot model generates deployment volume now, in controlled short-duration scenarios, producing the real-world data that feeds the software refinements needed to close that decade-long gap. Rental closes the ownership gap in the short term. The data it generates is what closes the capability gap in the long one.
