Dialogue —— Gaining Insights into the Future of the Humanoid Robot Industry

Currently, the humanoid robot industry is undergoing a crucial transition from technological verification to commercialization. As a leader in humanoid robots and a top enterprise in intelligent service robots, Ubtech Robotics (09880.HK), an enterprise invested by Qiming Venture Partners, is one of the very few companies globally with full - stack technological capabilities for humanoid robots and has taken the lead in realizing the practical application of humanoid robots.

At the 16th RMB Fund Annual Meeting and Investor Summit of Qiming Venture Partners, Kuang Ziping, the founding managing partner of Qiming Venture Partners, had a dialogue with Zhou Jian, the founder, chairman of the board, and CEO of Ubtech Robotics, on the theme of "Insight into the Future of the Humanoid Robot Industry". The two sides shared their views on topics such as the current development status and application scenarios of the humanoid robot industry, the potential of embodied intelligence technology in improving cost - effectiveness and work efficiency, and Ubtech's core competitive advantages.

Kuang Ziping, the founding managing partner of Qiming Venture Partners (left), and Zhou Jian, the founder, chairman of the board, and CEO of Ubtech Robotics (right)

In response to the emergence of new competitors in the industry, Zhou Jian stated that Ubtech welcomes partners and competitors in the field of humanoid robots, emphasizing that the market is vast with rich segmented areas. Meanwhile, he pointed out that embodied intelligence technology requires robots to have the capabilities of autonomous recognition and understanding, environmental perception, and task execution, which poses extremely high requirements for the technological capabilities of all startups. Whether they are the "old players" or the "newcomers" in the humanoid robot industry, only those enterprises that truly achieve technological breakthroughs and realize the application of technology can gain a foothold; Ubtech firmly believes that as long as it persists in doing difficult but right things, there will definitely be opportunities.

Regarding the economic evaluation of the commercial application of humanoid robots, Kuang Ziping pointed out that it is feasible to deploy one or two humanoid robots in a factory for demonstration purposes. However, from the perspective of actual benefits, it is not very reasonable to deploy only two. A batch of robots should be placed to fully realize their real value.

The following is a carefully edited transcript of the dialogue, with editing and abridgment.

01/

The application scenarios of humanoid robots are crucial

Kuang Ziping: First of all, let me introduce. Sitting next to me is Zhou Jian, the founder, chairman of the board, and CEO of Ubtech Robotics. His WeChat name is "The Father of Robots", and he is also the "father" of the humanoid robot running outside the venue. Everyone is looking forward to hearing Zhou Jian share Ubtech's story of developing humanoid robots.

Let me start by raising a few questions for discussion. Recently, humanoid robots have become extremely popular. I personally and the Qiming Venture Partners team are very optimistic about the future development of this direction and this industry. However, there are also a lot of skeptical voices in the market, thinking that it is just a gimmick. Why spend so much money on developing a humanoid robot that can only do a somersault at most?

First, please analyze the current stage of the entire humanoid robot industry. Then, combined with what Ubtech has done, is doing, and will do, share it with us.

Zhou Jian: Thank you, Mr. Kuang. Qiming Venture Partners has accompanied us for nearly a decade. I am very grateful for Qiming's support.

I'll answer your question. There have always been many disputes in the market about humanoid robots. When it comes to the stage, there is an important topic that cannot be bypassed. Can humanoid robots be used immediately? One crucial word is "scenario".

If humanoid robots want to enter households or scenarios with extremely high requirements for generalization ability right now, to be honest, no robot company in the world can achieve this, especially when it comes to entering Chinese households. The biggest problem with Chinese households is that the scenarios are very complex. There are different types of cups, such as teacups, water cups, and even thermos cups. Such scenarios place very high requirements on the generalization ability of humanoid robots. At least with the support of large - language models today, it is still quite difficult to achieve in the field of embodied intelligence in the physical world. However, if we view the scenario in a structured way, for example, in the handling or SPS sorting scenarios in an automobile factory, for humanoid robots, by continuously training with simulation data and real - world data on - site, this scenario can be realized.

Many people have high expectations for humanoid robots. They are concerned about when this robot can be put into use. Is it too early now? And is it even necessary to make it in a humanoid form? Currently, we are collaborating with several global automobile factories, including Dongfeng Liuzhou Motor, Geely Automobile, FAW - Volkswagen Qingdao Branch, Audi FAW, BYD, BAIC New Energy, etc. Most of them generally agree that humanoid robots are the most versatile. At first, I wanted to develop a quadruped robot dog. People thought that a quadruped robot dog was more stable and would not fall or damage parts in the factory. Later, we found that if it wanted to work like a human, the quadruped robot dog would have to be as tall as a horse to screw the bolts on top.

In terms of future cost, floor space, and the flexibility of the entire production line, in the intelligent manufacturing scenario, the quadruped robot dog is still inferior to the humanoid robot. So in this situation, after discussing with all our customers, we found that the humanoid robot is the most versatile and at the same time most in line with ergonomics. For example, it can raise its hand to pick up goods from a high - level shelf and can also bend down to pick up a box from the ground. At present, the humanoid robot seems to be the best solution.

But there is also a very important question: Can customers accept its cost? Last time, the chairman and board members of Mercedes - Benz also came to Ubtech for exchanges. Currently, the labor cost in Europe and the United States is much higher than that in China. So they think that a humanoid robot priced at around 400,000 - 500,000 RMB is completely acceptable, with a pay - back period of about two years. Moreover, humanoid robots have many advantages. For example, they can work 24 hours a day as long as the battery allows.

So in this situation, I think this year is an important stage for the mass production of humanoid robots. Ubtech plans to produce 1,000 humanoid robots in 2025, with an expected delivery of several hundred units. In 2025, if everything goes smoothly in various scenarios, optimistically, the delivery of humanoid robots will reach several thousand units in 2026 and is expected to reach the level of ten thousand units in 2027.

I still want to emphasize that the first large - scale application of humanoid robots will be in structured scenarios, such as intelligent manufacturing scenarios, followed by commercial service scenarios, and finally household scenarios.

02/

Humanoid robots need to help humans solve practical problems

Kuang Ziping: Let's go back to the industrial scenario. There are many misunderstandings in the outside world about the application of humanoid robots in this scenario. Humanoid robots were not invented last year. From a purely economic perspective, it seems difficult to recover the cost. It's okay to place one or two humanoid robots in a factory for demonstration purposes, but if only two are placed, it's not very reasonable from the perspective of actual benefits. In fact, a batch of robots should be placed to fully realize their real value. What is the specific number of this "batch"? Also, I've seen your introduction. In the factory, your humanoid robots are collaborating with automated guided vehicles (AGVs) to form a new - era unmanned logistics system, and humanoid robots are an important part of it. How to calculate the economic benefits?

Zhou Jian: First of all, I want to emphasize that the application of humanoid robots in factories cannot be remote - controlled. A truly advanced humanoid robot must be based on embodied intelligence, that is, the VLA model. This model combines vision, language, and action, enabling the robot to perceive, recognize, understand things, then conduct decision - making and reasoning, and finally form a closed - loop of actions. If this route is established, there is no need for continuous manual assistance or remote - control operation of the robot. Pre - programmed robots cannot meet the requirements of complex scenarios such as automobile production lines. The advantage of humanoid robots is that if this limitation can be overcome, they will show great potential in terms of cost - effectiveness and work efficiency. The key question to consider currently is whether the working rhythm of humanoid robots can match the production efficiency of the factory.

For example, when moving a box from point A to point B, the current efficiency of a humanoid robot is 30 - 40% of that of a human worker. In the future, if a humanoid robot can reach about 70% of a worker's efficiency, the factory will consider large - scale application. Currently, no humanoid robot company can complete such tasks on a large scale because it also involves tasks such as de - palletizing and palletizing, which require the boxes to be neatly stacked. Otherwise, during the towing process of the trailer, the boxes may collapse. If a humanoid robot can reach a certain efficiency and can also achieve autonomous recognition, being able to quickly complete the task regardless of where the material box is placed, basically, customers will be willing to pay.

Kuang Ziping: This is very interesting. Sometimes when we watch pre - recorded videos, we feel that humanoid robots are very flexible and can do these tasks almost as well as humans. But when observing them in real - world scenarios, we find that they are relatively slow. However, humanoid robots have unique advantages: they don't need to rest and can work continuously around the clock, which gives them significant cost and efficiency advantages in long - term operation.

Zhou Jian: Currently, some of the humanoid robots that highlight their movement abilities actually have removed sensors, reduced degrees of freedom, and are only about 1.2 meters tall. They are unable to perform tasks on the production line and don't even have the possibility of bringing you a glass of water.

Moreover, these humanoid robots do not have embodied intelligence and cannot perform intelligent mobile operations. I don't think this is the right direction for the development of humanoid robots in the field of embodied intelligence. Humanoid robots need to have high - level intelligence, perception, decision - making, and reasoning abilities and be able to help humans perform tasks and solve practical problems in the physical world.

A full - size humanoid robot should have at least 40 degrees of freedom. A humanoid robot with an autonomous battery - changing function may have even more degrees of freedom. The load capacity of both arms should be at least 15 kilograms. Automobile companies have also tested humanoid robots from other companies, but none of them could meet the factory's requirements. Finally, they chose Ubtech's humanoid robots for practical training.

In the industrial scenario, Tesla and Figure in the United States have also released some training videos of humanoid robots. Personally, I think Ubtech has more advantages in this aspect. Some humanoid robots' bionic design of lower - limb joints draws on the biological movement characteristics (such as ostriches), with a different technical route from ours. We combine planetary reduction and linear joint technology to achieve our goals and have very close cooperation with upstream and downstream supply chains.

Kuang Ziping: You just brought up another topic I want to discuss. In the large field of embodied - intelligent humanoid robots, what are the differences between China and the West? I think China is even leading. By comparing China and foreign countries, what are our advantages? Where have we reached now? What is the approximate level of the most advanced international technology? And to what extent do they rely on China's supply chain for further development?

Zhou Jian: This is a very good question. We've been thinking about it internally. Currently, Tesla's Optimus robot shows a very high level of whole - body movement coordination ability, not limited to the movements of the lower limbs or upper body. From the perspective of algorithm separation, if we have to select the leader in the first - tier of humanoid robots globally, I personally think it should be the Optimus robot.

Kuang Ziping: Tesla's Optimus robot?

Zhou Jian: Yes, but I haven't seen its latest progress in the integration of large - scale models and vertical - domain models. In previous demonstrations, we've seen that it already has basic functions such as object sorting. Companies like Figure are also promoting similar work. Different from the common high - reduction - ratio planetary gear structure in China, the linear actuator design of the Optimus robot has certain advantages in energy - efficiency ratio and dynamic response, which helps to improve the endurance and movement efficiency of humanoid robots. The Optimus robot is also closer to the human body's force - exerting mode in tasks such as heavy - object lifting. This design concept is leading.

When it comes to the supply chain, more than 80% of the component suppliers of companies like Tesla and Figure are in China. I think everyone in the industry can't bypass China's supply chain. Currently, only China has the ability to produce according to their requirements. Tesla is willing to invest a large amount of money, and many factories in the Yangtze River Delta in China are willing to open molds and produce for it. I think China's supply chain has a core position in this field.

03/

Do difficult but right things

Kuang Ziping: I want to raise a somewhat controversial topic. Ubtech used to be a well - known national brand and appeared on the Spring Festival Gala four times. Whenever people talk about humanoid robots, they always mention Ubtech. In the past two years, there have been many emerging enterprises focusing on the field of humanoid robots. This is a sensitive topic. Usually, people may ask privately, but I'm asking you directly today. Why do people now talk more about those new companies? Ubtech is one of the very few humanoid robot companies that are actually moving boxes in factories. It seems that people think Ubtech is a "yesterday's" company, and now they are talking about "today's" companies. I always tell these friends that new companies may have new attempts, but Ubtech already has the strength and is also making these attempts. However, the areas where real implementation can be achieved are still those we've been working on for a long time. How do you view the situation of new comers replacing the old?

Zhou Jian: First, we are not the "old players". The humanoid robot industry has just started and is still in a very early stage. The market is very large, and we warmly welcome all partners and even competitors in the field of humanoid robots. Take Apple as an example. Currently, Apple's market value is about 3.7 trillion US dollars. In a three - person family, if each person uses an Apple phone worth 3,000 - 5,000 US dollars. The future unit price of a humanoid robot is about 20,000 - 30,000 US dollars. Even if a family only has one humanoid robot, not including the large number of humanoid robots used in factories and commercial scenarios, theoretically, the humanoid robot market will be 5 - 10 times the size of the mobile phone market in 5 years. In other words, in the future, there will be a humanoid robot company in the world that will occupy a position similar to Apple's in the mobile phone industry. The market is large enough, with many segmented areas, and no single company can monopolize the market.

Second, I still repeatedly emphasize that we are not the "old players"; we are still the "newcomers". Just like BYD took nearly 30 years to become a leader in the new - energy vehicle industry. The difficulty of the humanoid robot industry far exceeds that of the automotive industry. Because the automobile is a product of dual - integration, and the data of intelligent driving can be shared among different brands. However, if we apply Ubtech's data to another humanoid robot company, the degrees of freedom and movement range of the humanoid robot will be affected, and the applicability of the data will be greatly reduced. This requires a decoupling process, so each company has its own advantages.

Third, humanoid robots cannot compromise on hardware. Recently, in collaboration with the Beijing Humanoid Robot Innovation Center, we released the full - size scientific research and educational humanoid robot Tian Gong Xing Zhe, which is 1.7 meters tall. Not only does it not compromise on hardware and has strong performance, but its price is only 299,000 RMB. The price of full - size (over 1.6 meters tall) humanoid robots of the same level in the industry is no