Zhuī Huījūn und Wáng Xīngxīng haben einen entscheidenden Kampf um den humanoiden Roboter ausgetragen.

The biggest trend in the tech circle recently belongs solely to humanoid robots.

At the beginning of July, news spread that Unitree Technology would choose the Science and Technology Innovation Board as its listing venue. A few days later, Zhipu Robotics triggered market speculation about "whether it is backdooring a listed company" with an operation of "a unicorn acquiring a listed company".

Regardless of whether the news is true or not, a definite trend is that the battle for the "first humanoid robot stock" has begun. As the two companies with the highest valuations in the current humanoid robot field, the comparison between Unitree Technology and Zhipu Robotics has never stopped.

1. Why is the "first humanoid robot stock" so important?

As more than half of 2025 has passed, raising funds has become the KPI for humanoid robot companies.

According to statistics from Huaxin Capital, from January to July 2025, there were about 158 investment and financing events in the robot industry in the Chinese mainland, almost one event per day. The total average amount exceeded 30 billion yuan, far exceeding the total amount for the whole of 2024.

Zhipu and Unitree, as companies with a valuation of tens of billions in the humanoid robot field at present, are undoubtedly the most concerned by the market.

Whether it is Unitree, which is rumored to list on the Science and Technology Innovation Board, or Zhipu, which is using a "smart way" to pursue an IPO, in essence, they are both running towards the "first humanoid robot stock" faster.

Why is the "first humanoid robot stock" so important?

Firstly, it is to occupy the mental position of the track with the "first stock". Compared with the prospectus, which is often hundreds of pages long and full of professional terms, the "first stock of XX" is obviously more appealing.

Secondly, the essence of listing is the discounting of a company's future earnings. Through listing, a company can have access to more resources. And the hard - tech field of humanoid robots itself is a capital - intensive track. The funds raised through listing can continue to expand the company's leading position.

In addition, for most domestic venture capital institutions, listing is still the main exit path, but there is a queuing period for IPOs. According to previous statistics from "Yicai", there are already more than 20 AI companies queuing up for listing in the Hong Kong stock market, a significant increase compared with the same period in previous years.

This is also why Zhipu used flexible capital operations in the secondary market this time. The founder team established an SPV (Zhipu Hengyue and Zhiyuan Xinchuang Partnership, actually controlled by Deng Taihua, the chairman and CEO of Zhipu Robotics), and obtained control of Shangwei New Materials, a listed company, through agreement transfer and tender offer.

Zhou Fen, a partner at Taihe Tai Law Firm who has long been focused on the investment and mergers and acquisitions field, believes that Zhipu actually used a combination of "founder SPV + step - by - step acquisition + share - swap restructuring" to bypass the slow lane of IPO queuing. It not only achieved the transfer of control but also reserved space for future business injection and capital operations.

This is a more proactive and flexible exit mechanism. Although it seems alternative, Zhipu is not an isolated case. In March this year, Hello Inc. obtained the actual control of Youon, the "first shared - bicycle stock", in a similar way.

Opinions among industry insiders about Zhipu's operation are divided: some believe that in the current capital market where there is widespread valuation inversion and the reality of difficult exits, Zhipu's approach of linking the primary and secondary markets has enlightenment significance for the exit of high - valuation tracks such as robots. However, some industry insiders believe that the development of humanoid robots is still in its early stage, and this flexible operation is a double - edged sword that may boost the impetuosity of the industry.

2. "DJI and Huawei" in the field of embodied intelligence: Two development paths

Gu Jufeng, a senior partner at Rongyi Investment who has participated in Unitree Technology's financing twice, admitted at a recent conference that the robot industry will see a wave of listings in the second half of this year.

Unitree and Zhipu have fired the first shot in the battle for the "first humanoid robot IPO", but they represent two different development paths of Chinese embodied intelligence.

Besides having 90 - post founders, the two companies have common characteristics in terms of product lines and application scenarios.

As of now, both companies have relatively complete and complex product matrices. They not only produce "humanoid" robots but also sell a series of components such as robotic arms, dexterous hands, and lidars.

Zhipu, which started with general humanoid robots, currently has three product lines: Expedition, Lingxi, and Genie. In addition to humanoid robots, it also has products in various forms such as legged and wheeled robots.

Unitree started with robotic dogs, including quadruped robotic dogs (Go series, B series) and humanoid robots (H1 series, G1 series) and other complete machines.

In terms of application scenarios, both companies target both the B2B and B2C markets.

For the consumer market, their products are for entertainment and for tech enthusiasts to try out. For the B - side, both companies provide industry - specific solutions in fields such as industry. Not long ago, Zhipu Robotics jointly demonstrated the on - the - job case of its general embodied robot A2 - W with Annu Intelligence in the Mianyang factory of Fuling Precision Industry, an auto parts company.

The operation site of Zhipu Expedition A2 - W. Source: Zhipu Robotics official

However, in terms of the commercialization process, strategic approaches, and organizational structures, Unitree and Zhipu are very different.

Firstly, in the commercialization process, Unitree is relatively ahead of Zhipu. Wang Xingxing, the founder of Unitree, said at the Summer Davos Forum that Unitree Technology's annual revenue has exceeded 1 billion yuan, and it has achieved profitability for five consecutive years since 2020.

Behind the difference in commercialization speed is the difference in production scale. In 2024, Unitree delivered more than 1,500 humanoid robots and 23,700 robotic dogs. At the beginning of the year, Zhipu officially announced that it had delivered 1,000 general embodied robots, including 731 humanoid robots and 269 wheeled robots.

The difference in the speed of commercialization is due to different strategic choices.

Unitree emphasizes hardware - driven development. In the early years, Wang Xingxing self - developed core components such as motors and lidars and integrated the hardware supply chain to reduce costs. This is also the key to its being called the "price butcher" and achieving rapid mass production. He once said in an interview: "Cost has always been our KPI for everything. The core is to make money."

Zhipu is more inclined to a combination of software and hardware, engaging in full - stack technology R & D and the entire industrial chain, which means its business model is heavier.

The development of embodied intelligence is inseparable from ecosystem building. In terms of ecosystem strategy, Unitree is more like product - driven, while Zhipu is more like platform - driven. The former is like "DJI in the field of embodied intelligence", and the latter is like "Huawei in the field of embodied intelligence".

The so - called product - driven means that Unitree outputs "finished robot products" to the ecosystem. For example, in April, Great Wall Motor reached a cooperation with Unitree, and the two parties focused on the exploration of "off - road vehicles + robotic dogs".

Zhipu uses "platform - driven" development. Since this year, Zhipu has accelerated its commercialization through joint ventures and investments. For example, it jointly established the Softcom Tianqing Robot with Softcom, and also established joint - venture companies with enterprises such as Dafeng Industry, Fuling Precision Industry, Lens Technology, and Junpu Intelligence.

When being interviewed by the media recently, an ecosystem partner described Zhipu's role as "focusing on providing the best basic ontology and model algorithms in the direction of the ontology and algorithms". In other words, Zhipu is more like a water - seller in the field of embodied intelligence, taking on the role of an infrastructure platform.

Behind the strategy is the change in organizational form. Many of Zhipu's senior executives have a "Huawei - style gene". The chairman and CEO is Deng Taihua, the former vice - president of Huawei and the former president of the computing product line. The chief operating officer, Qiu Heng, is the former CMO of Huawei's China enterprise business. The partner and vice - president of marketing, Jiang Qingsong, is the former head of the P & S solution product management department of Huawei.

In contrast, many of Unitree's senior management team are old acquaintances of Wang Xingxing. The co - founder, Chen Li, is Wang Xingxing's classmate and once worked at Hikvision. Yang Zhiyu, the mechanical director, graduated from the Department of Mechanical Engineering and Automation of Zhejiang University and is also one of the "founding members" of Unitree.

Although both are startup companies, the forms of the two enterprises are diverging. Zhipu is a startup company with the experience of a large company, while Unitree is more like a typical startup enterprise.

Using the high - profile approach of a large company, it is inevitable that in the early stage of the development of embodied intelligence, Zhipu has to fight on multiple fronts. Commercialization is just one of the "tasks", and more resources need to be invested in systematic projects including underlying technology, hardware, software, algorithms, data, supply and manufacturing.

Running in the form of a startup company, it is inevitable that Wang Xingxing needs to put making money on the table.

3. The entrepreneurial journey of the "genius kid" and the "little inventor": Different paths, same destination

In 2016, both Wang Xingxing and Peng Zhihui faced a choice in life - whether to continue entrepreneurship.

Wang Xingxing, who successfully got an offer from DJI after graduation, chose to resign and founded Unitree at that time. Peng Zhihui, who was still in graduate school, also tried his first humanoid robot startup and received more than 5 million yuan in angel investment, but he withdrew from the project and went to work in large companies such as OPPO and Huawei. Later, people were more familiar with his identity as the "Genius Kid from Huawei" and the B - station UP master "Zhi Huijun".

The different choices stem from different personality traits, although they now share the identity of "embodied intelligence entrepreneurs". Wang Xingxing is defined as a "little inventor". He has a serious academic bias but has strong hands - on ability since childhood. In a conversation with Yu Minhong, he described his upbringing as "free - range education": "It has helped me a lot and greatly improved my execution ability."

Wang Xingxing once mentioned in an interview that his idol is Lei Jun because Lei Jun always goes with the trend.

From being a self - media person to working in a large company, Peng Zhihui, as "Zhi Huijun", has always been in the spotlight and has never lacked attention. He has followed a typical elite growth path. However, he once revealed that he is an introvert and started his B - station account purely out of interest, never expecting it to grow so big. Different from Wang Xingxing, Peng Zhihui admires Elon Musk and Liu Cixin. He has read "The Three - Body Problem" more than 10 times.

People always compare the two, but in fact, in the entrepreneurship of embodied intelligence, the "genius kid" and the "little inventor" are actually on different paths but heading towards the same destination.

In addition to the financing pressure behind the battle for the first stock, as an early - stage industry, the development of embodied intelligence still has many uncertainties. On the one hand, although the technologies of many domestic - made components have gradually matured, the pressure of domestic integration of the supply chain and cost reduction still exists.

On the other hand, in the "brain" and "cerebellum" aspects, the technical paths are more complex. In addition to motion control and operation ability, especially in the recognition and learning ability of the "brain" end, it not only requires underlying general - purpose technologies as a foundation but also needs to obtain data in real - world scenarios.

Whether these common challenges can be solved is not only related to the survival of entrepreneurs but also determines whether embodied intelligence can truly enter all industries.

References:

1. Huxiu: Visiting this factory in Sichuan, I witnessed the birth of Zhipu Robotics

2. LatePost: LatePost's conversation with Wang Xingxing: As long as you are focused enough, I don't believe there are any problems that can't be solved

3. 36Kr: Dissecting "Zhipu Robotics": The capital game and the "Huawei - style" operators

This article is from the WeChat official account "Silicon - based Research Lab", author: kiki, published by 36Kr with permission.