Humanoid Robots Are “Rushing” to Market, While Users Urge Clarification of Demand and Warn Against a “Technology Bubble”

“At this stage, humanoid robots resemble the discussions about ‘autonomous driving’ two years ago, when the technology was believed to be mature but actual deployment scenarios were vastly different—an archetypal technology bubble,” remarked Jin Jing, head of the Center for Technology Innovation in Aging Societies, during the sub-forum of the 2025 Zhangjiang Embodied Intelligence Developer Conference and International Humanoid Robot Skills Competition on May 29. Jin emphasized that although humanoid robots are currently a popular concept, many products fail to address a critical issue: understanding market demand.

Many humanoid robot manufacturers lack a clear grasp of the genuine needs in elderly care. The specifics of how robots should provide “care”—including the design of hands, arms, and fingers, and the necessity of interaction—remain poorly understood and underexplored in practice.

Since the start of 2025, humanoid robots have accelerated their transition from laboratories to real-world industrial applications, particularly within elderly care and automotive sectors. In elderly care, enterprises and research bodies explore their roles in companionship, rehabilitation, and mobility assistance. In automotive manufacturing, companies such as UBTECH and Figure AI have deployed humanoid robots in final assembly workshops and logistics lines, engaging in sorting, transportation, and assembly tasks.

Discussions at the conference highlighted the industry’s active debate on the practical applications of humanoid robots across different environments. A key concern shared by many professionals is how to identify authentic needs amid technological enthusiasm and avoid falling into “technology traps.”

In January, Leju Robotics delivered its 100th full-sized humanoid robot to BAIC off-road vehicles, signaling a surge in factory deployment of humanoid robots. In March, dozens of UBTECH’s industrial humanoid robots Walker S1 successfully performed coordinated multi-task operations—such as collaborative sorting, transportation, and precision assembly—in complex automotive production settings.

Nonetheless, Xu Xiaoshun, Senior Manager of Intelligent Equipment at SAIC General Motors Powertrain Technology (Shanghai) Co., Ltd., pointed out that despite widespread efforts to find deployment opportunities, comprehensive application remains limited. This is primarily due to the complexity of automotive workshop tasks and the current maturity level of humanoid robot hardware and software. Moreover, the manufacturing costs of humanoid robots significantly exceed those of industrial robots.

“Some robotics company leaders foresee a future where industrial robots, composite collaborative robots, humanoid robots, and human labor will represent a ratio of approximately 7:1:1,” Xu noted.

Xu further explained that domestic efforts to accelerate humanoid robot applications are somewhat “rushed.” Internal research revealed that humanoid robots are best suited for repetitive and labor-intensive tasks in production stages like final assembly, effectively reducing worker fatigue. For example, humanoid robots may excel in roof inspection tasks, whereas semi-humanoid robots, which offer advantages like longer battery life and stronger navigation capabilities, present competitive alternatives.

“We consider semi-humanoid robots as accompanying applications before the official promotion of humanoid robots,” Xu disclosed. Within the automotive sector, more manufacturers are expected to deploy humanoid robots in select models this year, with numbers potentially reaching hundreds or thousands by next year.

During the roundtable discussion, Yang Yongzhe, Technical Brand and Planning Director of Great Wall Motors, and Jin Jing both cautioned the industry about the risk of falling into “technology traps” with humanoid robots.

Jin emphasized that user and industry understanding of genuine needs and caregiving workflows in elderly care remains inadequate. Robot manufacturers lack insight into what elderly care truly requires or how to implement “care.” Given that technology in elderly care should primarily assist and serve humans, current focus on the technology itself is premature.

“Only by grasping the core of ‘understanding demand’ can technology realize value. Technology is fundamentally a means, not an end,” Jin stated.

Yang echoed this view, comparing the current state of humanoid robots to the “autonomous driving” discussions from two years ago, when expectations of maturity clashed with vastly different deployment realities—an evident technology bubble. Present challenges mainly revolve around technology capabilities and enterprise cost management. Technically, the robot’s form dictates suitable applications; for example, deploying a production-line robot to perform retail sales is unrealistic.

From a cost perspective, Yang noted that automotive manufacturers extensively use robotic arms in vehicle assembly. Frontline workers have not disappeared but relocated to control rooms to operate and monitor robots. “This has not lowered costs but added new maintenance expenses,” he said. Yang stressed that the scalability of humanoid robots depends on technological maturity, cost control, and sustainable business models.