At present, humanoid robots are accelerating their evolution and continuously iterating and upgrading, breaking through the limitations of laboratory environments and infiltrating a large number of real-life scenarios, demonstrating enormous potential for large-scale applications. At the recently held 2025 World Robot Conference, Shanghai Securities News interviewed several corporate executives, industry experts, and investors and learned that the commercialization of embodied intelligence has significantly accelerated due to multiple factors such as technological progress, policy drivers, and capital surges. Currently, the industry is in the stage of exploring application scenarios, with hardware costs showing a downward trend and model capabilities needing further improvement. Industry insiders predict that 2026 may be the first year of mass production delivery. The application scenarios are becoming more diversified, such as the vigorous posture of bow archery, the flexible rhythm of passionate dance, the accurate processing of precision parts, the skillful skill of stringing Tomatoes on sticks, the flexible cooperation in football matches... From traditional agricultural production to precision industrial manufacturing, from professional guided welcome to warm life company, the 2025 World Robot Conference comprehensively demonstrated the active exploration in various application scenarios in the industry. Many industry insiders believe that technological innovation must be rooted in application scenarios, allowing robots to step out of the laboratory and solve problems in multiple scenarios, giving embodied intelligence a "warmth". Approaching the booth of Yuejiang Technology, the robot DOBOT Atom holds a wrench in its left hand to tighten the mechanical arm bearings to within a tolerance of 0.1 millimeters, and synchronously sorts different specifications of parts into the six grid material box with its right hand; In the high dynamic task of "picking and playing popcorn", DOBOT Atom can not only flexibly turn between narrow workstations, smoothly and accurately pour popcorn into boxes and hand them to the audience, but also quickly respond to interactive needs such as waving, heart to heart, and clapping from the audience. It is reported that this is a robot that can achieve millimeter level accuracy and also has dual brain collaborative decision-making function. At this conference, robots have become "versatile workers". Galaxy General has launched its first wheeled humanoid robot GALBOT G1, which enables product picking in unmanned pharmacy environments; At the Xinghai Tu booth, with a command of "please make the bed", the R1 Lite robot immediately entered working mode; The world's first autonomous battery swapping humanoid robot Walker S2 brought by Ubiquitous, with its 52 degree of freedom industrial grade "skillful hand", can complete autonomous battery swapping in 3 minutes. More and more companies are no longer showcasing the single machine performance of 'showing off technology', but actively building simulated industrial, reception, medical and other scenarios, emphasizing the usability and substitutability of products in real tasks. ”Kang Yu, General Manager of the Board Office of Shoucheng Holdings, stated that this reflects that the market's expectations for embodied intelligence are no longer limited to technology, but focus on application efficiency and alternative value. The positive signal of commercialization has emerged. "This year, the parts of humanoid robots are too rolled up, and core components such as screws have been significantly reduced in price. ”Several industry chain participants reported this phenomenon to reporters. The decrease in hardware costs signifies the maturity of the supply chain and indirectly confirms the acceleration of commercialization in the embodied intelligence industry. Industry insiders say that currently, the maturity speed of embodied intelligent hardware is faster than that of models, and hardware needs to wait for algorithms to mature before it can truly achieve commercial implementation. Yushu Technology CEO Wang Xingxing also believes that low-cost hardware is an inevitable trend for the industry to achieve commercialization in the future. Policies are also continuously being implemented. In 2025, "embodied intelligence" will be included for the first time in the government work report, clearly listed as a key development direction for future industries; The State Council executive meeting held at the end of July proposed to deepen the implementation of the "Artificial Intelligence+" action and vigorously promote the large-scale commercial application of artificial intelligence. At this conference, Beijing Yizhuang released the "Several Measures for Promoting the Innovative Development of Tailored Intelligent Robots in Beijing Economic and Technological Development Zone", proposing to provide full chain support in stages from the first trial and use of undifferentiated products, first unit (set) rewards, to benchmark scene construction, and accelerate the promotion and application of robot products. In the opinion of Liu Ying, a researcher of Chongyang Institute of Finance, Renmin University of China, promoting embodied intelligence commercialization is a specific measure to implement the "AI+" action. It is expected that with the accelerating progress of commercialization, more support policies are still on the way. The 'child rearing rate' of this conference is very high, which means that the market's attention and tolerance are increasing. ”Kang Yu shared an interesting observation of the conference with the reporter. Kang Yu said that many parents bring their children to the exhibition, which shows that even if the current robots are not intelligent enough, as long as they are interesting and cute, the public's acceptance will increase, laying the foundation for the expansion of future retail and B2C scenarios. The feeling of attending the conference this time is that the commercialization progress of humanoid robots has significantly accelerated, and 2026 may be the first year of mass production. ”An investor attending the conference told reporters that 2026 is expected to be the first year of mass production in the industry, and in the future, the focus will be on the actual progress of top enterprises in mass production and investment opportunities in upstream components and downstream applications of the industrial chain. Several leading companies revealed the progress of mass production and delivery during the conference. According to Tan Min, Chief Brand Officer of Ubixion Technology, the company plans to deliver 500 industrial humanoid robots to the intelligent manufacturing industry this year; Tiangongxing, which focuses on scientific research and education, has received over a hundred orders and is expected to deliver over 300 units this year; The Xingdong Q5 robot from Xingdong Era has received dozens of orders and is expected to deliver 100 units this year; The AlphaBot series of Zhifang has received nearly 500 orders this year and is currently being put into use in factories such as Dongfeng Liuqi and Jingneng Microelectronics. Jiang Zheyuan, founder of Songyan Power, introduced that the company's core work is currently focused on mass production and delivery, with the education industry being the field with the largest order volume, covering multiple segmented scenarios such as K12, vocational education, higher education, and research institutions, and the demand for different scenarios varies significantly. Wang He, founder and CTO of Galaxy General Robot, revealed that the company's robots have been deployed in 10 unmanned pharmacies in Beijing and will expand to 100 by the end of the year. Despite the vigorous development of scene verification and mass production delivery for numerous leading robot ontology enterprises, there are also industry opinions that there is currently a homogenization phenomenon in robot products, and their application scenarios are relatively limited. Regarding this, Wang Qian, the founder of the independent variable robot, stated that it is not a hardware issue, but rather because the intelligence level of the robot's large model has not been reached. Yao Song, Director of the International Advanced Technology Application Promotion Center (Shenzhen), said that the development path of embodied intelligence is like autonomous driving. From the early high expectations for L4 level full automation, to the actual implementation of L2 level scenarios, the industry is gradually moving towards practicality. He believes that most robots are still in the L1 stage, and the evolution towards higher levels requires a focus on real, urgent, and commercially feasible scenarios. (New Society)