Humanoid Robots Become the 'Next Big Thing', Why Are Domestic Practitioners Shaking Their Heads?

[baoshi.rao]

The craze for humanoid robots has quickly spread from the U.S. to China. In recent days, stocks related to humanoid robots have surged in the domestic A-share market, with many hitting the daily limit-up.

According to a CITIC Securities research report, the popularity of the humanoid robot concept is mainly due to improved expectations for the progress of Tesla's humanoid robot supply chain, AI giants igniting the fire of robotic technological innovation, and continuous favorable policies and manufacturer support. Currently, humanoid robots have a wide range of application scenarios and serve as a powerful complement to industrial robots and other automation equipment. Their global market potential could reach trillions, comparable to 3C products and new energy vehicles.

However, many domestic practitioners have raised doubts.

Qiu Dicong, founder of Jacobi Robotics, which is currently seeking a new round of funding, told Sina Technology bluntly, "Future robots will definitely be oriented towards multi-scenario, general-purpose, and multi-functional directions, but they don’t necessarily have to converge into a human form, as doing so would incur high costs that may not align with ROI."

Meanwhile, Fu Sheng, Chairman and CEO of Cheetah Mobile and Chairman of Orion Star, told Sina Technology, "Robots can have many mechanical arms, but why must they have legs?"

On one side, there is fervor at the capital level; on the other, there is restraint and rationality from within the industry. Has the resurgent popularity of humanoid robots become a fleeting trend? Or do people really need humanoid robots? Two Key Factors Reshaping the Future of Robotics?

Recently, international tech companies have been announcing significant breakthroughs in humanoid robotics. American startup Figure AI secured massive funding, Optimus humanoid robots demonstrated generational improvements, and NVIDIA announced the establishment of GEAR to accelerate robotic development.

In a recent interview, NVIDIA co-founder and CEO Jensen Huang commented: "If machines can generate text and create images, shouldn't they also generate motion? The answer appears to be yes. Once machines master motion generation, they may comprehend underlying intentions and develop universal expressions. Thus, humanoid robots might arrive sooner than expected."

Currently, nearly 90% of global AI companies rely on NVIDIA's computing power. While supporting these companies, NVIDIA also leads the AI industry's direction. Similarly, numerous robotics companies depend on NVIDIA's computing solutions, especially for high-performance robotic products where NVIDIA's embedded platforms remain indispensable.

"If NVIDIA shifts focus to motion generation, this strategic change could fundamentally transform the future trajectory of the entire robotics industry." If NVIDIA represents the hardware factor changing development, then ChatGPT undoubtedly stands as the software counterpart. Qiu Dicong stated that since ChatGPT's emergence in 2023, the emergent intelligence capabilities of large language models have fundamentally altered the development trajectory of the robotics industry. This has evolved from the Robot 2.0 era of algorithm + single-task robots to the Robot 3.0 era featuring significantly enhanced interactivity, cognitive abilities, and generalization capabilities.

"Robots in the 2.0 era, such as those in catering, hospitality, or industrial scenarios, possessed some image recognition and logical operation capabilities. However, they essentially relied on programmers writing code into the robot's brain and teaching it to perform tasks. This meant that changing scenarios basically required starting from scratch," Qiu Dicong explained. With Robot 3.0, due to improved language and environmental understanding capabilities plus autonomous motion path planning, "the new generation of robots demonstrates strong generalization abilities while significantly reducing programming training costs. This creates prerequisite conditions for general embodied intelligence robots of the 3.0 era to further integrate into human life."

Do robots really need legs?

The transformation in underlying technological logic has propelled robotics into a new developmental phase. Humanoid robots, with their striking resemblance to humans in appearance, have naturally become the focal point across various sectors.

In China, the emphasis on humanoid robots has become visibly apparent in the stock market. Recently, domestic robotics concept stocks have shown sustained strength, with multiple listed companies including Huayan Precision Machinery, Oriental Precision, Bojie Shares, Standard Shares, Tian Yong Intelligent, Tianqi Shares, and Mingzhi Electric experiencing consecutive days of limit-up surges, demonstrating remarkable momentum. According to Choice data statistics, 23 A-share listed companies in the robotics sector - including Supcon Technology, Soochow Securities, YHEDA, ESTUN, and STEP - received intensive institutional research visits between January 1 and February 28. Among them, Supcon Technology recorded the highest number of institutional visits at 373.

Behind the booming capital market, a new wave of humanoid robot industrialization is gaining momentum. However, for industry practitioners, the humanoid form factor doesn't necessarily prove more attractive than specialized robotic solutions.

"Future robots will undoubtedly move toward multi-scenario, general-purpose functionality, but they don't necessarily need to converge to human form. In many scenarios, this approach incurs high costs that may not justify the ROI," explained Qiu Dicong, using supermarket restocking robots as an example. Adding legs would generate significant noise and consume excessive power, making wheeled solutions far more cost-effective. "If equipping humanoid robots with arms and legs doesn't deliver substantial value advantages, why bother with unnecessary complications?"

An industry veteran with eight years of AI+robotics engineering implementation experience told Sina Tech: "The rationale for humanoid design lies in bipedal locomotion's adaptability to complex terrains and scenarios. However, we still lack quality robotic solutions even for flat, structured environments. We should first perfect solutions for structured scenarios before tackling complex terrains."

From this perspective, the true potential of robotics lies in manipulation capabilities rather than all-terrain mobility. Current development could focus on upper-body manipulation and robotic arms, reserving lower-body solutions like bipedal locomotion for future development when actually needed. At the recent Sina Finance 'The Way of Wealth' salon, Fu Sheng, Chairman and CEO of Cheetah Mobile and Chairman of OrionStar, shared his thoughts on humanoid robots: 'Robots can have many mechanical arms, but why must they have legs? In many scenarios today, adding legs actually compromises safety, and the cost doesn’t add up either.'

Clearly, from the perspective of real-world deployment and commercialization, safety, practicality, and cost-effectiveness are the true determinants of a robot's long-term viability. Beneath their cool, human-like appearance, humanoid robots have garnered significant attention. However, truly integrating them into people's daily lives still requires substantial financial resources and faces practical challenges.