The news of the 2025 Robotics Conference was released. Goldman Sachs's latest interpretation named these stocks

The Zhitong Finance App learned that Goldman Sachs watched the live broadcast of the 2025 World Human-like Robotics Conference (WHRG) held in Beijing last weekend (August 14-17). At this event, Goldman Sachs paid special attention to evidence of outstanding robots in terms of autonomy and generalization capabilities to confirm the importance of technological progress and its possible driving effect on the stock price performance of related stocks.

Overall, Goldman Sachs observed from this event that robots have made some positive developments in their autonomous capabilities. For example:

1) Tiangong Ultra added autonomous control capabilities to all running events. Compared with its performance using remote control in the April marathon, the competition rules encourage autonomous control. If the robot uses autonomous control, the running time can be reduced by 20%;

2) Soccer games highlight the need for complex multi-robot collaboration and team intelligence;

3) Scenario-based activities emphasize the more practical application of humanoid technology. Completing tasks requires both accuracy and speed, and autonomous robots score 50% higher than remote-controlled robots. Some of the best performing robots can complete tasks completely autonomously, recognize specific objects, make decisions, and operate them with high accuracy.

However, Goldman Sachs found that test samples during the conference were still limited, which still reflected a gap in robustness and speed compared to actual commercial applications. Additionally, performance-based competitive sports events are tests of hardware robustness and durability. In these projects, top performing robots excel in motor control and battery power management, and Goldman Sachs is positive about this.

Here are more detailed highlights:

The World Humanoid Robotics Conference is the first global multi-event competition dedicated to humanoid robots. More than 500 humanoid robots from 280 teams from 16 countries (including the United States, Germany, Japan, Brazil, etc.) participated in the event. These teams include 192 university teams and 88 private enterprise teams, but different teams can develop software or use these platforms on shared hardware platforms from 127 robot brands. Among them, platforms frequently mentioned include Unitree (Unitree), Tiangong (Tiangong), Agibot, Booster Robotics, GalaxEA, etc.

Goldman Sachs believes that performance-based competitive sports are tests of hardware robustness and durability, more specifically, including high-torque motors (in the case of sprinting), battery life and effective thermal management (in the case of a 1500 meter run), full body control, navigation, self-balancing gait algorithms (especially in obstacle running), hardware durability, and actuator strength (especially in long jump and high jump events).

Goldman Sachs is pleased to see that the top performing robots in these projects have excelled in motor control and battery power management. For example, Yushu Technology's G1 robot that participated in the 1,500-meter race implemented an “overheating protection procedure” to prevent the robot from falling or stopping operation due to the generator temperature being too high. Compared to the robot marathon held in Beijing on April 19, 2025 (at the time, all robots, including Tiangong, had to use remote control), Tiangong Ultra, which finished second in the running program this time, increased autonomous control capabilities in all running events. Xiong Youjun, CEO of the Beijing Humanoid Robot Innovation Center, who designed the robot, emphasized that this is a key differentiator. He mentioned that Tiangong Ultra can use visual perception to sense the environment and plan autonomous routes, and make real-time decisions. Yushu Technology's H1 robot, which won first place in the 400m and 1500m races, is remotely controlled. Wang Xingxing, CEO of Yushu Technology, said in an interview that Yushu Technology's robots will be able to switch to autonomous mode in the future, which should not be difficult to achieve. In fact, the conference actively encouraged a shift to a fully autonomous model as a standard for measuring the capabilities of advanced robots. Compared with remote-controlled robots, autonomously controlled robots can reduce running time by 20%, thus gaining an advantage in final performance.

According to Goldman Sachs, soccer games require specific types of multi-robot collaboration and team intelligence (unlike stand-alone operation and individual competition); require millisecond information synchronization to achieve effective group strategies, such as multi-robot defense and midfield blocking; and accurate visual and spatial recognition (for more complex robot-robot interactions and ball deviations). Booster Robotics' T1 is the hardware of choice for each team. The platform provides an open source code framework, enabling the participating teams to quickly build their own modules, and is equipped with an NVIDIA AGX Orin chip. Its computing power supports advanced artificial intelligence algorithms such as imitation learning and reinforcement learning, and the robot can stand up autonomously within 2 seconds after falling. These characteristics were reflected in the game. The robot showed the ability to respond quickly, perceive the environment, and correct itself, but challenges still occurred in several scenes where multiple robots gathered near the ball, causing multiple robots to collide and required human intervention to separate. Furthermore, the robot's performance near the goal is relatively stiff, which indicates that current artificial intelligence has limitations in its ability to “predict” rather than just its ability to respond, which makes defensive strategies less intelligent.

Scenario-based activities underscore the industry's primary interest in the practical application of human-like technology. Evaluation metrics generally consider both the accuracy of completing the task and the time required to complete the task. At the same time, robots that operate autonomously score 50% higher than robots controlled remotely. The top three robots in the hotel service (customer reception and cleaning) category all used AGV wheeled mounts and functional grippers instead of walking on two legs, which appears to be due to their inherent stability and positioning accuracy. This reflects a trend that Goldman Sachs observed at the recent World Artificial Intelligence Conference (WAIC), where companies are focusing on designing products that are closest to immediate commercial viability. Although wheeled robots can complete autonomous tasks such as transporting luggage, complex tasks such as accurate garbage sorting require dexterous hands and advanced software. Despite this, some teams, such as the Beijing General Artificial Intelligence Research Institute, intend to use biped robots in hotel tasks to rigorously test the “usability” of their algorithms in the real world, even if there may be a time disadvantage.

Medical scenarios involve tasks such as drug sorting and distribution, further highlighting the need for fine motor control. Since delicate items such as softgels require high-precision operation, they usually require remote operation using a VR controller. A major challenge Goldman Sachs has observed is interference with wireless communication in large venues, which affects the mobility of robots and highlights the need for strong connectivity in actual deployments.

Goldman Sachs has a positive attitude towards most robots committed to autonomous operation. For example, Lingyi Intelligent Manufacturing (002600.SZ) robots can handle material trays and measure the size of materials with the support of visual guidance, recognition algorithms, and operation capabilities. Working with both arms can complete the task in 1 minute and 13 seconds, with an accuracy rate of 95%.

However, Goldman Sachs found that test samples during the event were still limited, and in actual commercial applications, higher accuracy and robustness were needed in more tasks.

Furthermore, Goldman Sachs notes that the robotics conference may not reflect the most advanced capabilities in the industry because only a few robots are participating in the competition.

(Continued table)

Stock views

According to Goldman Sachs, the progress in autonomous capabilities observed by Goldman Sachs at the World Humanoid Robotics Conference is positive for the humanoid robot industry because it shows more use cases and room for adoption.

Sanhua Intelligent Control (02050/002050.SZ, buy/buy) Sanhua Intelligent Control is a global leader in HVAC control and thermal management components. Goldman Sachs gave Sanhua A/H a buying rating because Goldman Sachs is optimistic about its growth potential in human-like robot actuators on the basis of a stable market-leading position in its core business, and expects its revenue/net profit compound annual growth rate to reach 19%/19% from 2025 to 2030. In terms of HVAC, Goldman Sachs believes that, driven by an increase in commercial HVAC market share and a ramp up in sensor products, the company is expected to achieve higher revenue growth than the residential HVAC industry. In terms of electric vehicles, Goldman Sachs believes that the company's revenue growth will be driven by a further increase in the global electric vehicle penetration rate and a moderate increase in content value. Furthermore, Goldman Sachs sees humanoid robots as an important long-term technology trend, and Sanhua Intelligent Control is expected to occupy a key position in the supply chain and become a highly visible actuator assembler. The catalysts include technological advances in humanoid robot functions, mass production for major customers of humanoid robots, and a further increase in the penetration rate of electric vehicles in Europe. Goldman Sachs believes that compared to the average valuation of Chinese industrial technology stocks covered by Goldman Sachs, the stock's valuation is not high.

Best (300580.SZ, neutral) Goldman Sachs believes that with the support of steady growth in the auto parts business (including internal combustion engines and electric vehicles), the company is expected to become a competitive supplier of humanoid robotic planetary roller screws (PRS). Given that Goldman Sachs expects the company's entry into the high-specification humanoid robot supply chain to bring potential benefits, Goldman Sachs predicts that the compound annual growth rate of global high-specification humanoid robot shipments will reach 80% from 2024-2035. Goldman Sachs expects that with precision manufacturing experience accumulated from fixture products and production capacity support from equipment procurement, the company will account for 10% of the global market for planetary roller screws used in high-specification human robots starting 2027. At the same time, the commercialization of machine tool parts progressed smoothly, and in 2024, ball screws and linear guides began to be used by some domestic machine tool manufacturers. Despite this, Goldman Sachs believes that the risk return is fair compared to Chinese industrial technology stocks and human-like robot supply chain stocks covered by Goldman Sachs, so Goldman Sachs gave it a neutral rating.

Green Harmonics (688017.SH, neutral) Goldman Sachs gave a neutral rating to Green Harmonic, a domestic leader in the Chinese harmonic reducer market. Goldman Sachs has a positive attitude towards the company because the total addressable market (TAM) for its harmonic speed reducers is expanding, thanks to a growing number of applications such as domestic and foreign industrial robots, collaborative robots, human/service robots, CNC machine tools, and other high-end equipment (solar/semiconductor/medical/aerospace). The company is penetrating overseas customers, such as starting bulk shipments to the Chinese factories of at least two of the top four international robotics brands in 2024, and is expected to make a more meaningful revenue contribution in 2025. Although Goldman Sachs is constructive about the company's long-term growth potential and believes that the humanoid robot business will gradually contribute, Goldman Sachs believes that its valuation is fair compared to the Chinese industrial technology stocks covered by Goldman Sachs.

Mingzhi Electric (603728.SH, neutral) Goldman Sachs gave Mingzhi Electric, a leader in domestic motor and drive solutions, a neutral rating. Goldman Sachs anticipates that although its motors may be less competitive than global competitors in terms of parameters such as speed and acceleration, the company will become a key player in the coreless motor supply chain for human-like robots by 2027, gaining market share in high/medium/low specification robots due to its significant cost advantages and potential to expand production capacity at the new Taicang plant. Goldman Sachs expects that Mingzhi Electric's revenue in coreless motor applications for human-like robots will continue to grow, and that in addition to individual components, it is also possible to increase its content value by providing modules. However, Goldman Sachs's main business outlook is more conservative for the following reasons: 1) Compared with its peers, the servo motor market share growth record is poor; 2) the organic growth rate in 2014-2023 was 10%, and the acquisition growth rate during the same period was 23%; 3) factory relocation/overseas capacity ramp up may cause operations to be interrupted. After fully considering the company's growth potential in the field of humanoid robots and its main business (motors and drives), Goldman Sachs believes that the risk return is fair compared to the Chinese industrial technology stocks covered by Goldman Sachs.

Ling Yunguang (688400.SH, neutral) In the rapidly growing machine vision sub-industry, Goldman Sachs expects Ling Yunguang's market share in traditional machine vision systems to grow moderately, thanks to its ability to provide solutions to major customers and selective self-developed software, cameras, and lenses. However, starting in 2025, Lianying Medical's FzMotion motion capture system began to increase its exposure in the humanoid robot data collection business, creating a second growth engine for Lianying Healthcare. The compound annual growth rate from 2025 to 2030 is expected to be 73%. Goldman Sachs expects the humanoid robot motion capture business to account for 18%/17% of Lianying Healthcare's revenue/net profit by 2030 as the high output of humanoid robots grows. Combined with other underperforming machine vision terminal markets (consumer electronics, batteries, solar energy), Goldman Sachs expects Ling Yunguang's overall revenue/net profit compound annual growth rate to reach 16%/24% from 2025 to 2030. All in all, given that Goldman Sachs's target price has a moderate amount of implied downside compared to the average of shares covered by Goldman Sachs — 4% — Goldman Sachs believes its valuation is fair.