news

Securities Times reporter Chen Shuyu Wu Qi

From August 21st to 25th, the 2024 World Robot Conference was held in Beijing. 27 humanoid robots showed their talents and became the undisputed popular kings at the exhibition.

In the capital market, the concept of humanoid robots has attracted increasing attention. Since the beginning of this year, the performance of this sector has been active, and fund managers have frequently investigated related listed companies and expressed optimism about the development prospects of humanoid robots. A fund manager said that in the next 10 years, humanoid robot investment will be a "large-scale beta opportunity". Humanoid robots are gradually moving from the laboratory concept verification stage from 0 to 1 to the commercial explosion stage from 1 to N.

Humanoid robots show their talents

Fund managers give thumbs up on the spot

2024 is considered the first year of commercial use of humanoid robots. All kinds of capital have paid special attention to it, and a new hot track has been opened.

At the 2024 World Robot Conference, 169 companies exhibited more than 600 innovative products, of which more than 60 were first-time new products. Humanoid robots were undoubtedly the most eye-catching protagonists of this exhibition, with a total of 27 models on display, the most ever. It is worth mentioning that many fund managers also witnessed the Beijing debut of Tesla's Optimus humanoid robot.

Invesco Great Wall Fund used live broadcast to share information and views on the 2024 World Robot Conference with investors. Zhang Xiaonan, manager of Invesco Great Wall Fund, said that the biggest advantage of humanoid robots lies in their versatility and huge market space. Humanoid robots can be put into production without changing existing factory equipment, and the market space is naturally much larger than that of special robots such as industrial robots. The huge shipment volume can further reduce the individual total cost of humanoid robots, allowing more people to buy them, thus forming a virtuous circle. At the same time, the huge shipment volume can also feed back the iteration of robot models. Just like the current intelligent driving of cars, it is necessary to continuously feed the driving data to the intelligent driving program for iteration in order to improve it. Humanoid robots can also continuously update and improve software capabilities during use. "Maybe it takes 10 minutes to wash dishes today, and it will only take 5 minutes tomorrow."

Zhang Lu, fund manager of Yongying Fund, told the Securities Times reporter that he had recently visited the Beijing World Robot Conference and once again felt the vigorous development of the humanoid robot industry. "I was impressed by the surgical robot exhibition area. The precise 'hand of God' can already skillfully and perfectly detach the shell of crayfish. Minimally invasive surgery performed by surgical robots has small incisions, is more accurate than human hands, and defibrillates, and the adverse reaction rate is greatly reduced. The technology of surgical robot companies, such as the integration of advanced control algorithms, sensor technology, and mechanical design, can empower humanoid robots with a pair of dexterous hands." In addition, Zhang Lu said that when investigating intelligent car companies, it is often found that the technology and supply chain of humanoid robots and car companies can be reused and transferred, and the visual recognition and scene construction of humanoid robots all use the same neural network learning technology as the car FSD (fully automatic driving). Domestic humanoid robots are expected to replicate the global competitiveness of new energy, and Chinese manufacturing companies with large value and good positioning have hardware advantages. However, in areas where the core technology barriers of humanoid robots are relatively high, such as planetary roller screws, domestic manufacturers started late and are still working hard on research and development. After overcoming difficulties, they are expected to catch up.

In October 2022, Tesla Optimus was born, and as soon as it debuted, it set off a wave of humanoid robots in the capital market. Tesla CEO Elon Musk has repeatedly expounded on his great vision, saying that "in the future, humanoid robots may exceed the number of humans." Nvidia founder Huang Renxun also said that "the next wave of AI (artificial intelligence) will be embodied intelligence," and humanoid robots are considered to be the best carriers of AI's next generation of "embodied intelligence." Humanoid robots, a new type of productivity carrier that integrates advanced technologies such as artificial intelligence, high-end manufacturing, and new materials, will undoubtedly have a subversive impact on social and industrial changes and the global competitive landscape. Based on this, humanoid robots are also regarded as another disruptive product after computers, smartphones, and new energy vehicles.

Ji Junkai, fund manager of HFT CSI Auto Parts Theme ETF, believes that 2024 is called the first year of commercialization of humanoid robots, mainly due to the dual promotion of technological progress and market demand. The capital market's attention and expectations for the humanoid robot industry stem from its leading position in the fields of artificial intelligence and robotics, as well as its potential for wide application. At present, the industry is in a stage of rapid growth, with new technologies and new products emerging continuously, and more commercialization paths are being explored.

Regarding the great expectations that the capital market has placed on humanoid robots, Zhang Yinxian, fund manager of Ping An Fund, believes that the main reason is that humanoid robots may be the representative of the fourth industrial revolution, greatly improving labor productivity and thus solving the problem of an aging population. Once a breakthrough is made in industrial development, it may form a scale of trillions or even tens of trillions of yuan, even surpassing the current market size of the automotive industry.

Humanoid robots are evolving faster

The investment is very cost-effective after scaling up

The domestic humanoid robot industry is on the eve of a "blowout" and is one of the hottest tracks in the technology sector.

At present, humanoid robots are evolving at an accelerated pace. Through disruptive innovations in artificial intelligence algorithms, flexible mechanical design, and autonomous navigation and interaction capabilities, humanoid robots are expected to improve production efficiency and service quality in multiple fields, with huge application potential. The "Guiding Opinions on the Innovation and Development of Humanoid Robots" issued in China clearly states that by 2027, the technological innovation capabilities of humanoid robots will be significantly improved, and related products will be deeply integrated into the real economy, becoming an important new engine of economic growth.

Humanoid robots have humanoid shapes and are theoretically the most perfectly adapted production tools for human working environments. They have great application potential in industrial production, commercial services, disaster relief, elderly care, and domestic services. Zhang Lu said that humanoid robots are not just about freeing up human hands, they can "replace" and "better" humans in highly repetitive and dangerous tasks.

The development of humanoid robots is changing with each passing day. Thanks to breakthroughs in artificial intelligence technologies such as deep learning, large models, and high-performance computing, robots' learning and autonomous decision-making capabilities have been greatly improved, and they can better understand the environment and make "smart" responses. In addition, based on artificial intelligence technologies such as neural networks, graph grammar, and evolutionary algorithms, humanoid robots will be able to automatically build modules such as legs, arms, and torsos according to scene and task requirements, achieve coordinated optimization of morphology and control, and greatly improve their adaptive capabilities.

"Productivity that can improve production efficiency can be regarded as new quality productivity. There is no doubt that humanoid robots have this possibility." Zhang Yinxian said that from a hardware perspective, the number of parts of a humanoid robot is around a thousand, far lower than the tens of thousands of parts of a car, and the weight of a humanoid robot may be only 3% to 5% of a car. Therefore, after large-scale production, the hardware cost of a humanoid robot is expected to be lower than the cost of a car. "Assuming that the cost of a humanoid robot is 150,000 yuan, and the labor cost will basically show a gradual upward trend in the future, then the investment payback period of a humanoid robot may be shorter than 2 years, which is a cost-effective investment." Zhang Yinxian said.

In addition, the rapid development of artificial intelligence can greatly enhance the perception, decision-making, execution, and interaction capabilities of humanoid robots, allowing them to be integrated into unstructured scenario applications on the B-end or C-end more quickly.

"Humanoid robots are considered to be an important direction for developing new quality productivity because they can imitate human appearance and functions and perform complex tasks, thereby improving production efficiency and service quality in multiple fields." Ji Junkai pointed out that the disruptive innovations in this track include advanced artificial intelligence algorithms, flexible mechanical design, and autonomous navigation and interaction capabilities.

The application of humanoid robots in manufacturing is to improve automation and efficiency. For example, in order to improve production efficiency and safety, Tesla's Optimus robots are planned to undertake highly repetitive and dangerous tasks in automobile factories.

In addition, Amazon has tested the humanoid robot Digit in its warehouses for tasks such as unloading trucks, carrying boxes and managing shelves. It is expected to be fully available in 2025. In home applications, humanoid robots are mainly involved in household assistants, education and entertainment. For example, the Mobile ALOHA robot has shown great potential in housework and can complete tasks such as washing clothes, watering flowers and making coffee.

On the eve of the explosion of intelligent level

Fund managers explore investment opportunities

Zhang Lu said: "In the next 10 years, humanoid robot investment will be a 'large-scale beta opportunity'. Humanoid robots are gradually moving from the laboratory concept verification stage from 0 to 1 to the commercial explosion stage from 1 to N."

"At the current stage, the robot market is more focused on AI+robots." Zhang Lu believes that the essence of humanoid robots is a combination of artificial intelligence, perception and execution units. The motion planning and processing system depends on the progress of the underlying AI chips, algorithms and models. The core bottleneck restricting the implementation of robot applications is also the software algorithm. With the release of Sora, Gemini1.5Pro, etc., the multimodal field of AI has achieved leapfrog development. At the same time, NVIDIA announced the establishment of the "GEAR" department to conduct research on general embodied intelligent bodies, which can build a robot-specific algorithm computing platform, drive more manufacturers to try to use large models for training "embodied intelligence" products, and accelerate the commercialization of robots in the future. In short, after the humanoid robot parts solution is gradually determined and implemented, we will see how the AI ​​technology revolution will bring about a new trend of intelligent robots with "brains". Compared with the L1~L5 level of intelligent driving, the current level of intelligence of humanoid robots is roughly equivalent to the L1~L2 level, or it is on the eve of an outbreak.

Zhang Yinxian said that China is a global manufacturing power and is gradually becoming a manufacturing power. In recent years, the export volume of the automobile industry has jumped to the first place in the world, relying on a large number of high-quality domestic manufacturing companies. These companies have the ability to mass produce, have high consistency and respond quickly, and these capabilities and advantages can be reused in new productivity tracks such as humanoid robots and low-altitude economy, which is undoubtedly extremely important for the breeding of growth investment opportunities. In terms of OEMs, China has successively produced many domestic robot manufacturers, and the growth and rise of these manufacturers will definitely drive the growth of companies in the industrial chain.

Ji Junkai believes that on the investment side, investment opportunities in the humanoid robot industry are mainly concentrated in core component suppliers, system integrators, and companies that provide software and algorithms. As the industry matures, companies in these fields are expected to achieve performance growth.

With the continuous advancement of humanoid robot technology, the industry's demand for high-performance core components is growing, including but not limited to motors, joints, sensors, controllers, reducers, etc. The quality and technical level of these components will directly affect the overall performance and stability of the robot. Of course, system integrators also play an important role in the humanoid robot industry chain, responsible for integrating various components and technologies into a complete humanoid robot system, which requires strong technical capabilities and in-depth understanding of the industry. The investment opportunity for system integrators lies in providing customized solutions and services to meet the needs of different industries and application scenarios. The level of intelligence of humanoid robots depends largely on software and algorithms, including advanced technologies in the field of artificial intelligence such as machine learning, computer vision, and natural language processing. Software and algorithm companies enable humanoid robots to better understand and respond to the environment and improve their autonomy and interaction capabilities by providing advanced algorithms and software solutions.

Zhang Xiaonan also believes that in the future, as humanoid robots increase in volume, manufacturers of component design, manufacturing, and production will be good investment targets. He introduced that in order to meet the computing needs of large models, each humanoid robot will be equipped with a set of edge computing equipment, among which the AI ​​chip has the highest value. In terms of mechanical parts, according to the disassembly information of Tesla Optimus, without considering the dexterous hands, its whole body includes 14 rotary joints and 14 linear joints, including 28 frameless motors, 14 harmonic reducers, 14 planetary roller screws, 42 encoders, 14 force sensors, and 14 torque sensors. The two dexterous hands are mainly composed of hollow cup motors, planetary reducers, and worm gears. In addition, the robot also carries a 2.3kWh battery pack.

Focus on risk-return ratio

Tracking industry trends from multiple perspectives

Humanoid robots are in the spotlight, and it is undoubtedly feasible to actively seek investment opportunities. However, the current industry is more of a thematic investment stage, and there are still many uncertainties. What investment risks should investors pay attention to? How do fund managers balance the relationship between risk and return?

In 2022-2023, Musk exhibited the Optimus Prime prototype at AI Day, and subsequently released videos of Optimus Prime's iterative training. The market has given a relatively optimistic expectation for this, and it is expected that there will be a fixed-point event catalyzing humanoid robot-related parts from 2023 to the first half of 2024, and humanoid robots will even be mass-produced, so the performance of the humanoid robot sector is also relatively active during this period.

After entering 2024, the actual progress of robot research and development was slower than expected, resulting in a large decline in the robot sector. Zhang Yinxian analyzed that the industry is indeed still in its early stages, but society is constantly moving forward, and the pursuit of a better life by humans is an irreversible trend. "In the future, labor costs will continue to rise, and humanoid robots that can alleviate this problem are 'late but late', and we have patience and confidence in the industry." Zhang Yinxian said.

Zhang Lu said that the process of incubation from 0 to 1 and fission from 1 to N of emerging industries is often long. Compared with mature industries, the return on investment is obviously less clear, but due to the many variables involved, opportunities also coexist with challenges. This is a "transaction" with relatively high risk and return. Investors investing in emerging fields should first carefully assess their own risk preferences and tolerance, because to dare to earn "future money", not only must they have a keen observation of the future, but also have patience and confidence beyond the norm.

Ji Junkai mentioned three areas of concern: First, the accuracy and reliability of core components. The motion accuracy and stability of humanoid robots depend to a large extent on their core components, such as servo motors, reducers, sensors, etc. The performance of these components directly affects the flexibility and accuracy of humanoid robots. At present, there is still a certain gap between domestic enterprises and the international advanced level in the development of these core components, especially in terms of accuracy, durability and reliability. Second, artificial intelligence and perception processing. Humanoid robots need to have a high degree of intelligent perception and processing capabilities, including environmental perception, object recognition, natural language understanding, etc. These capabilities are crucial for the application of humanoid robots in home services, medical assistance and other fields. At present, artificial intelligence technology still faces challenges in big data processing, algorithm generalization capabilities, etc. Third, cost control and mass production. The manufacturing cost of humanoid robots is high, which limits their market penetration. How to reduce costs and achieve mass production while ensuring performance is the key to the industrialization of humanoid robots.

Zhang Xiaonan believes that when investing in the humanoid robot track, one should pay more attention to whether the long-term logic is correct. On the one hand, the logic of humanoid robots replacing human labor is difficult to disprove; on the other hand, the capabilities of humanoid robots themselves are indeed evolving and iterating rapidly, and entering the market at a low point is a good option. However, risks should also be considered, such as the difficulty of humanoid robots landing and the fact that domestic suppliers are not entering the supply chain as expected. It is important to avoid blindly investing due to short-term overheating, and be wary of purely conceptual targets.

For investors, it is undoubtedly necessary to closely follow the industry dynamics of humanoid robots in the future. In this regard, Zhang Lu gave three suggestions: First, pay close attention to every breakthrough and progress of the industry vane Tesla Optimus, and focus on targets with strong supply certainty and marginal improvements in supply chain status, such as from no supply to supply, from Tier2 to Tier1, from components to integration, etc.; second, pay attention to how the AI ​​technology revolution brings about the new trend of intelligent robots with "brains", such as the software algorithms, intelligent platforms, and system integration involved in humanoid robots. AI technology may become the technical moat of "embodied intelligence" humanoid robot manufacturers; third, pay attention to how to map A-shares after the overseas supply chain solutions are gradually clarified, so as to discover high-quality domestic manufacturers with high value, in the core supply chain and with core technologies.

Ji Junkai further stated that tracking the dynamics and investment opportunities of the humanoid robot industry can also be done by paying attention to industry reports, attending professional conferences, and monitoring the financial reports and news releases of related companies. In addition, communication with industry experts and analysts is also an important way to obtain information. Investors can pay attention to the information officially released by Tesla. Musk often reveals the latest progress of the humanoid robot Optimus through social media and official press conferences. In terms of supply chain dynamics, the production of Tesla humanoid robots depends on its supply chain, including suppliers of key components such as batteries, motors, sensors, and reducers. Paying attention to the news releases, financial reports, and industry reports of these suppliers can find clues to investment opportunities. Finally, investors also need to pay attention to industry policies and capital trends, because the government's support policies and capital investment in the robotics industry are also important factors in promoting the development of the industry.