UBTECH's top priority may be to maintain the original intention of scientific and technological research and development, and to ensure that it is always on the right track in the long night of waiting for large-scale commercialization.
Produced by Emperor Cai · Author: Yifan.
On December 19, Shenzhen UBTECH Co., Ltd. (hereinafter referred to as "UBTECH") disclosed the prospectus and officially launched the issuance, and the company planned to offer 1128 globally20,000 shares, with an issue range of HK$86 to HK$116 per share, raising a maximum of about 13HK$0.9 billion.
When UBTECH finally stepped into the secondary market, it was the most lively time for the humanoid robot track.
Elon Musk unveiled the humanoid robot Optimus Gen2 (Optimus Prime 2nd generation), demonstrating more intelligent capabilities. ByteDance was exposed to expand the robot team, the former Huawei "genius boy" Zhihuijun released the Zhiyuan robot, and Xiaomi invested in the construction of a humanoid robot innovation center with UBTECH and other companies.
In the prospectus, UBTECH made it clear that it had been commercialized, but the financial data showed that it was in a state of loss, and 9 rounds of financing and high debt could not avoid financial difficulties. At a time when the competition is becoming increasingly fierce, UBTECH has successfully grasped the life-saving straw of listing, which is undoubtedly a great relief.
For the market, humanoid robots are more of a cool concept than a real need. For investors,The current development of the track seems to be attractive, but the rewards are not the best.
UBTECH's landing on the Hong Kong Stock Exchange with the limelight of "the first share of humanoid robots" just provides an opportunity for people inside and outside the industry to examine the business model and future potential of humanoid robots.
01 Zhou Jian's flag missed
In 2018, Zhou Jian, the founder of UBTECH, set up a flag: "I believe that the bottom line of revenue in 2019 is 6 billion yuan, if it reaches 8 billion yuan, it will be 80 points, and 10 billion yuan will be 100 points." ”
At that time, his optimism came from that, according to publicly reported data, UBTECH's total sales in 2014 were only 2 million yuan, reached 50 million yuan in 2015, soared to 300 million yuan in 2016, and exceeded 1 billion yuan in 2017.
But the reality is far from the same, and the prospectus shows that in 2020, 2021, and 2022, UBTECH will achieve revenue of 7400 million yuan, 8200 million yuan and 10100 million yuan,Revenue for the first six months of 2023 was 2600 million yuan.
UBTECH, which has been commercialized for many years, still lacks self-hematopoietic ability.
According to the prospectus, in 2020, 2021, 2022 and the first six months of 2023, UBTECH's gross profit will be 33.1 billion yuan, 25.6 billion yuan, 29.4 billion and 05.3 billion yuan, with a net loss of 70.7 billion yuan, 91.8 billion yuan, 9$8.8 billion and $55 billion yuan, with a cumulative loss of nearly 3.2 billion yuan.
Zhou Jian is a professional in mechanical engineering, insisting on independent research and development, and the research and development funds are "spent like running water".
From 2020 to the first half of 2023,UBTECH's R&D investment was 42.9 billion yuan, 51.7 billion yuan, 4$2.8 billion and $22.4 billion yuan, accounting for the total revenue of the current period. 5% and 859%。
Since revenue can't support R&D,External borrowingIt became an important option. As of December 31, 2020, 2021 and 2022 and October 31, 2023UBTECH's bank borrowings are: 100 million yuan, 30 billion and 7500 million yuan, the overall increase.
UBTECH's asset-liability ratio has also been high for a long time. UBTECH's debt-to-asset ratio was 58 at the end of 202003%, and by the end of 2021, this value had risen to 7151%, far exceeding the recognized appropriate level of debt-to-asset ratio (40%-60%). The asset-liability ratio is too high, which means that UBTECH has almost no room for debt financing.
At the same time, in order to open up market awareness as soon as possible, the company's sales and marketing expenses remained high. From 2020 to the first half of 2023, its expenses were 31.3 billion yuan, 35.8 billion yuan, 361 and 19 billion yuanrespectively accounted for the total income of the current period. 8% and 727%。
The prospectus shows that in the first half of 2020 and 2023, UBTECH's inventory will be 41.3 billion yuan, 42.6 billion yuan, 3$3.3 billion and $4$1.6 billion.
Based on various indicators, UBTECH's financial situation seems to be deteriorating, with revenue and gross profit on a downward trend, and net losses, various expenses, and borrowings on an upward trend, which may explain why its listing process has been bumpy.
In February 2019, Zhou Jian publicly stated that he would strive to carry out an IPO in China in 2019.
This time, the flag fell short again. From July 2019 to November 2020, UBTECH submitted several IPO applications without success, and then changed the counseling agency to no avail, impacting A-shares without follow-up. It was not until the beginning of this year that UBTECH switched to Hong Kong stocks and restarted its listing journey.
Although it has been cultivating the robot track for a long time, its competitive advantage is slightly insufficient. According to Frost & Sullivan, UBTECH will occupy China's intelligent service robot solution industry in 2022ranked third in the market share.
The market share of the first and second place is 64% and 46%, and 2 for fourth and fifth place7% and 222%。From the perspective of market share alone, UBTECH has a large gap from the top two, and it is relatively easy to be overtaken by the companies behind.
Today's achievements were not easy to come by, and Zhou Jian almost exhausted his family property for this reason, so as to attract the attention of external investors. In October 2013, after getting 20 million yuan in angel round financing, UBTECH had the strength to breathe.
Tianyancha data shows that since its establishment, UBTECH has completed a total of 9 rounds of financing, with investors including Tencent, CDH Investment, iFLYTEK and other well-known enterprises and institutions. At present, Tencent Holdings holds 648% of the shares, the largest institutional shareholder.
At the same time, the company's shareholders include many local state-owned assets, such as Hangzhou Youzhi, a subsidiary of Hangzhou Yuhang**, and investment companies in many provinces and regions, such as Pingyin Yuandao.
As the product is launched in iterations, its valuation continues to climb. In 2015, the A round was 300 million US dollars, and in 2018, it completed an 8 billion investment led by TencentAfter a $200 million Series C financing, UBTECH's valuation exceeded $5 billion, making it the world's highest-valued AI startup.
On the other hand, the valuation is continued to be optimistic, because UBTECH has been unable to make a profit for a long time, and the investors behind it may intend to cash out, which is also an important reason for its urgent listing and opening up the financing channels of the capital market as soon as possible.
After the Series D financing, the company has had two equity transfersThe investor, Shenzhen Unicorn Company, was priced at 62. per share49 yuan, 6588 yuan**, the equity will be transferred to Shanghai Youjue, Puyang Financial Holding, ** are less than 7788 yuan.
02 BlazingTracks
As Zhou Jian sought to go public, the track became more and more crowded. As a start-up company, UBTECH has to face a group of powerful technology giants.
Tesla CEO Elon Musk once said: "In the future, robots will be as ubiquitous as mobile phones." ”
Nvidia CEO Jensen Huang said that the next wave of AI is embodied intelligence (that is, the combination of AI brain and robot body). Lei Jun said when referring to humanoid robots"We need to use the Internet thinking to make hardware. ”
Humanoid robots are the most difficult to develop among all robots, and involve very complex technologies, which are called them"The jewel in the crown of robots".。At present, the level of each player is uneven, and there is a situation of "one super and many strong" in technology.
Over the years, the industry has been focusing on the technical difficulty is the degree of freedom, the more joints, the more flexible the robot's action, and the degree of freedom of the joints is proportional to the complexity of the machine, which is extremely testing the manufacturer's research and development capabilities.
Tesla's latest release of the Optimus Gen2 is even more impressive than the first generation, with increased neck and hand freedom, and a weight reduction of 10kg. Able to support on one leg, squat up and down, do basic yoga stretching work, can walk normally like humans, and grab, pick up, and put down eggs is no problem.
In comparison, the CyberOne has only 21 degrees of articulation freedom. The gap in the degrees of freedom of the joint is illustratedTesla is better than Xiaomi.
UBTECH's representative humanoid robot, Walker X, has 41 degrees of freedom in joints, can walk smoothly on stairs, gravel, lawns and other terrains, can control household appliances such as refrigerators, coffee machines, vacuum cleaners, and provide services such as massage, bottle capping, tea and water.
In terms of degrees of freedom, Boston Dynamics joints are less than Optimus Gen2, CyberOne, and Walker X, so why is the flexibility much higher?Fundamentally, it's still a technical issue at the bottom.
At present, there are three main technical routes for joint drive control: hydraulic drive, motor drive and pneumatic drive, among which motor drive is the mainstream, and the difficulty of implementation is low, and Optimus Gen2 adopts motor drive.
Hydraulic drives are more technically difficult and cumbersome to maintain. This technology is achieved by high-precision cylinder block and piston, and the hydraulic oil is transported by hose from the hydraulic pump to the hydraulic cylinder, which may restrict the drive due to oil leakage at the interface and cracks in the pipe.
Through 3D printing, Boston Dynamics embeds all power unit parts such as oil pipes into the mechanical structure of the legs, just like human blood vessels, and no longer needs external oil pipes. It is reported thatAltas has a total of 28 hydraulic cylinders throughout the body, which are characterized by low hydraulic impact and fast response, making it more flexible than a human.
Last August, Boston Dynamics released Atlas' Parkour**, which showcased Altas' near-human bounce and body stability.
While other players were still worried about the robot's plane walking, Altas was already able to run, jump, and backflip. Boston Dynamics has created a solid technical barrier on hydraulic drive, and even Musk has made it his development goal.
Although UBTECH has a first-mover advantage in the field of humanoid robotsHowever, it does not have advantages in the computing power of large models and training dataAnd these factors can quickly make up for the shortcomings of emerging players.
UBTECH has not yet established a moat, and will face the pressure of consumers to "shop around" in the future to deal with more dangerous competition.
03 Sell 10 units in three years, where is the imagination in the future?
According to Zhou Jian's own ideals, he is actually more willing to do the C-end humanoid robot market. However, for the survival of the company, it has to exert its strength on the B-sideThe Walker series is part of the To B business line.
The Walker series is now the only humanoid robot in the world to be mass-produced and delivered, and it is also the first product in the world to reduce the cost of humanoid robots to less than $100,000.
The cost of 600,000 or 700,000 yuan for CyberOne seems to be about the same as that of the Walker series, but Lei Jun has admitted that it cannot be mass-produced for the time being. And the cost of Atlas is as high as $2 million, and it is destined to only "survive in the laboratory".
UBTECH Walker series is indeed the closest to commercialized humanoid robots, but such ** is still too expensive for C-end consumers, and the demand for B-end is not large, industrial robots don't care whether they are "humanoid", low price and high efficiency are hard indicators.
Sales are a direct indication of the market response, with only 10 units of the Walker series sold in three years. According to the prospectus, UBTECH will sell a Walker-2 robot in 2021, one Walker, one Walker-1, two Walker-2 and four WalkerX in 2022, and one Walker-2 in the first half of 2023The cumulative income is about 16 million yuan, and the contribution to the total revenue is not large.
The real revenue pillars are:Educational and consumer robots, but they are biased in both function and appearanceRobot "toy".Application-level robots such as logistics robots and health care robots are basically not compatible with humanoids, and intelligent household robots are mainly vacuum cleaners and cat litter box products.
Even Boston Dynamics, the industry benchmark, was resold three times in 7 years due to its difficulty in making a profit, and was successively "abandoned" by Google and Japan's Softbank, and is now taken over by South Korea's Hyundai.
SoftBank Robotics, which once produced 27,000 humanoid robots, announced two years ago that it would discontinue the humanoid robot Pepper.
Although UBTECH's walker walked out of the laboratory,But the commercialization scenario is very limitedMainly for educational purposes and hospitality services.
If we look at humanoid robots from the perspective of the "innovation return curve" proposed by the British economist Mazucato, we will find that they are still in the early stage of innovation returns, and they are still far from the "singularity" of commercial applications.
According to Mazucato, the return on innovation over the life cycle of a technology has an S-shaped curve. In the early days of technological innovation, the risk is extremely high and the reward is extremely low. As the technology matures, the closer it gets to the "singularity" of commercial applications, the return on innovation begins to grow exponentially.
The emergence of AI large models has added more possibilities to the commercialization prospects of humanoid robots. The so-called humanoid robot, the most important thing is not to have human movement ability, nor to blindly pursue the similarity with human appearance and posture, intelligent brain is the foothold.
It is generally believed in the industry thatLarge model technology can be combined with humanoid robots from many aspects such as language, vision, and decision-making, and humanoid robots are regarded as the ultimate application of AI+.
Technology giants have gone deep into the chess game, which is expected to promote the technological breakthrough of humanoid robots and the upgrading of the industrial chain, so that more and more humanoid robots can go out of the laboratory.
Among them, the industry has high hopes for Optimus, Musk said at the 2023 World Artificial Intelligence ConferenceThe final ** of Optimus is likely to be less than $20,000 and is expected to be mass-produced within 3-5 years. Once the product matures, the number of mass production may reach 10 billion to 20 billion units, far exceeding the number of human beings.
Whether it is Tesla's electric car or DJI's drone, it relies on long-term research and development, and timely seizes the opportunity for commercialization to stand at the top of the industry.
Zhou Jian once mentioned that the vision is for humanoid robots to enter the public home, which is very expensive. UBTECH's top priority,Perhaps it is to maintain the original intention of scientific and technological research and development, and make sure that you are always on the right track in the long night of waiting for the outbreak of large-scale commercialization.
After all, the core of the competition between technology companies is the competition of technology.
Note: The data is derived from public disclosure and does not constitute any investment advice, investment is risky, and you need to be cautious when entering the market.