A newly registered embodied robotics startup Kunlun has secured billions in funding within 90 days and attained unicorn status | 36Kr Exclusive
More than a year after leaving Alibaba Cloud, Ren Geng, the former vice president of Alibaba Group and the former president of Alibaba Cloud's China region, re-entered the AI battlefield as the founder of Kunlunxing. This time, the field he targeted is the terminal for AI implementation - robots.
36Kr learned that within less than 90 days of its registration, Kunlunxing Robotics has completed three rounds of financing in succession, with a cumulative financing scale of billions of yuan. After the three rounds of financing, the company's valuation has exceeded one billion US dollars, making it a unicorn in the embodied intelligence industry. This pace has obviously refreshed the speed of becoming a unicorn in the embodied intelligence industry.
Previously released information from the company shows that Kunlunxing Robotics "targets Tesla's humanoid robot, the industry benchmark" and "adheres to the dual - wheel drive strategy of 'body + brain'". Lang Xianpeng, the former top - ranked person in Li Auto's assisted driving, is a co - founder of Kunlunxing.
A leading capital firm that has had contact with Kunlunxing judged that Ren Geng will focus more on strategic direction, organizational system building, business operations, and overall corporate operations, while Lang Xianpeng will mainly be responsible for the R & D and implementation of AI technology.
This entrepreneurial combination has made Kunlunxing highly sought after by capital. In less than three days after the company's registration, the shares of its first - round financing were snapped up by top - tier investment institutions. A person in the industry commented: "This is quite rare in the embodied intelligence track."
Currently, the shareholders of Kunlunxing include leading domestic venture capital institutions such as Hillhouse Capital, Gao Rong Capital, Zhongding Capital, Zhongke Chuangxing, Huaye Capital, Sinovation Ventures, and Xin Capital, as well as the industrial capital under the giant industrial group C&D Group. Moreover, all the investors who participated in the first round, including top - tier venture capitals like Hillhouse Capital, Gao Rong Capital, Zhongke Chuangxing, and Huaye Capital, have continuously increased their investment in Kunlunxing for three consecutive rounds.
A person in the industry close to Kunlunxing even told 36Kr: "In fact, the company's valuation had already reached unicorn level before its registration was completed."
36Kr learned from many investment institutions that the combination of Ren Geng and Lang Xianpeng has attracted much attention because in the current embodied intelligence track, their past performance accumulations are both substantial.
Ren Geng once led Alibaba Cloud to maintain the top market share in China for a long time and is a comprehensive - ability leader who has achieved tens of billions in revenue. Lang Xianpeng led a team from scratch to deliver intelligent driving solutions for over one million vehicles, laying the foundation for Li Auto's leading position in the assisted driving industry. He is a technical expert who understands cutting - edge technology and has rich mass - production experience.
"We mainly look at whether the team is strong enough."
A series of factors, such as the technological iteration of Tesla's Optimus and the rapid passing of Yushu Technology's IPO review, have ignited capital's enthusiasm for the embodied intelligence track. Although there is a large amount of hot money, "there are not many good investment targets." Therefore, an investor told 36Kr: "In the face of some good projects, investors are now more like the 'buyers' (in a weaker position)."
Regarding how to measure the quality of an investment target, a leading capital firm told 36Kr: "Whether the team is strong enough is a crucial measurement dimension. Moreover, we hope that the enterprise can balance software and hardware and find some implementable application scenarios within a certain period."
The main basis for judging the strength of a team is whether they have achieved key results in the past and how many people they have led.
A leading capital firm that has invested in Kunlunxing also frankly told 36Kr that when screening investment projects, "they mainly look at whether the team is strong enough."
In a sense, the background of Kunlunxing's founding team can be regarded as a "rare combination" among current embodied intelligence startups.
According to public information, Ren Geng, the founder of Kunlunxing, held the important position of Huawei's overseas country CEO before the age of 30. He was fully responsible for Huawei's three major business groups (carrier, enterprise, and terminal) in the overseas country, coordinating businesses from solution formulation, sales, and delivery to procurement, supply chain, and GR/PR.
After joining Alibaba in 2015, as one of the few post - 80s vice presidents of Alibaba Group at that time, Ren Geng successively held core positions such as the president of Alibaba Cloud's China region, the president of Alibaba Cloud's Video Cloud & Edge Cloud R & D, and the chief of staff of Alibaba Cloud. In terms of data, during Ren Geng's tenure at Alibaba Cloud, Alibaba Cloud maintained the top market share in China for a long time and set a record of 42.1% in Alibaba's public cloud market share in 2020.
During his tenure as the president of ENN Group, Ren Geng led this group of over 50,000 people to layout in cutting - edge fields such as industry - specific large models, digital intelligent agents, and embodied intelligence.
Beyond his public titles and resumes, some industry insiders close to Alibaba Cloud summarized Ren Geng's contribution to Alibaba Cloud to 36Kr as follows:
Around 2015, Alibaba Cloud had completed its early development stage and needed to make further progress in commercialization. "However, Alibaba Cloud's early To B approach was relatively rough. In contrast, Huawei's capabilities in To B business at that time were recognized by many large domestic enterprises."
Ren Geng's joining "promoted Alibaba Cloud's business model to shift from the previous extensive mode to a systematic one." "For example, the mutual responsibilities of R & D, pre - sales, sales, and delivery in To B business, as well as the entire set of mechanisms including product management, sales operations, price management, project operations, and supply chain management, were gradually standardized."
As Ren Geng emphasized, for an embodied intelligence company to truly achieve mass production and complete the business cycle, in addition to AI technical capabilities, it also needs a complete business system from business opportunities to payment collection, a mature supply chain ecosystem, and a service system.
Compared with entrepreneurial teams with a pure technical executive background, Ren Geng's global business vision, experience in building To B business organizational systems, and end - to - end business organization process management can bring to Kunlunxing the scarce end - to - end overall operation ability in the industry.
Lang Xianpeng, the co - founder of Kunlunxing, is another core force of the enterprise.
An investor commented to 36Kr: "At present, in addition to intelligent technology, automobile companies may also need to have other capabilities such as product capabilities. However, the current competition in the embodied intelligence track largely focuses on the R & D and implementation capabilities of AI technology." And intelligent driving is the smallest implementation unit of physical AI. The technology chain of "perception - decision - execution" in the automotive field is highly homologous to that of robots.
As the "first employee" of Li Auto's assisted driving, Lang Xianpeng built Li Auto's assisted driving R & D system and assembled the core technical team. During his time at Li Auto, with an annual R & D budget of about 10 million yuan, he launched the self - research of assisted driving and finally delivered high - level assisted driving solutions for 1.5 million vehicles. Lang Xianpeng himself once publicly summarized Li Auto's assisted driving R & D process as "a latecomer's counter - attack to the first echelon."
This is an achievement that other executives who have crossed over from assisted driving to the embodied intelligence field do not have. Now that he has switched to the field of embodied intelligence, some capital firms judge that Lang Xianpeng's R & D and delivery capabilities in large models and assisted driving will also empower Kunlunxing.
In addition, Kunlunxing also has a core team from leading enterprises in the industry such as Huawei, Li Auto, and Alibaba. They have long been able to take charge of their own work in technology, business, and other aspects.
Targeting Tesla, Kunlunxing Bets on the Ultimate Form of Embodied Intelligence
In terms of the business path, Kunlunxing is telling a story about general humanoid robots. The company targets Tesla's humanoid robot, the industry benchmark for the mass production and implementation of physical AI.
Tesla's Optimus is one of the most highly valued humanoid robot projects in the world today. Optimus is positioned as a general humanoid robot. Tesla stated in a recent earnings conference call that it will fully release the robot Optimus Gen3 to the public in the middle of this year and start small - batch mass production at the Fremont factory from July to August 2026.
For a long time, humanoid robots have been regarded by the market as the third - generation computing terminal after PCs and smartphones. The well - known investment bank Morgan Stanley predicted at the end of last year that by 2050, its global market scale could reach 5 trillion US dollars. It is not difficult to see the company's ambitious business prospects from Kunlunxing's bet on humanoid robots.
In order to make the business narrative of humanoid robots work, Kunlunxing refers to the practical experience of Tesla's humanoid robot and autonomous driving implementation, and tries to build the company's competitive barrier through the integration of software and hardware and full - stack self - research.
The ability to integrate software and hardware is the differentiating position that Kunlunxing has found.
In the current humanoid robot track, some companies focus on hardware and control algorithms, or their product positioning is for guiding, and their AI algorithm development has just started.
Since its establishment, Kunlunxing has simultaneously laid out the "brain" and the body of the robot.
Regarding the robot's "brain", Kunlunxing believes that the lack of physical causal ability, poor scene generalization, and the "black - box" nature of the model decision - making process are common pain points of current robot brains. Therefore, Kunlunxing has proposed a dual - system robot intelligent architecture centered on the Kunlun World Model (KWM) to improve the model's iteration efficiency and generalization ability.
At the hardware level, Kunlunxing follows the idea of algorithm - defined hardware, hoping to overcome the problem of the separation between software and hardware. In addition, Kunlunxing has self - developed key components and motion control algorithms, and through modular general design, it has reduced the threshold for the model to adapt to different robots.
It is worth noting that although general humanoid robots are generally regarded by the industry as the ultimate form of embodied intelligence, there are still some problems to be solved for their mass production and implementation.
Just the hardware supply chain of general humanoid robots has not yet emerged from the fog.
Taking the easily worn - out dexterous hands and mechanical arms as an example, public information shows that the service life of robot dexterous hands is often less than two months. This means that the cost of component wear is high, and it is difficult to apply them to mass - produced products in the short term.
If the hardware is already like this, let alone the robot's "brain", which is even more difficult to implement.
Kunlunxing is not the first to bet on general humanoid robots, and it will definitely not be the last. In terms of team background and financing progress, Kunlunxing is half a step ahead of most embodied intelligence enterprises, and all these may help Kunlunxing cross the mass - production gap of humanoid robots earlier.
Author's WeChat: luckg17305264638