Unitree's investors have joined forces to heavily invest in an embodied team.

A historic moment for China's embodied intelligence industry — on August 19, Unitree Robotics officially listed on the Sci-Tech Innovation Board, becoming the "first humanoid robot stock" on the A-share market.

This marks a remarkable come-from-behind story for China's robotics sector: over the past decade, Wang Xingxing started from scratch, growing Unitree from a single quadruped robot product to the world's largest humanoid robot manufacturer by shipment volume.

The investment community has noticed a notable detail: the three largest external shareholders of Unitree — Meituan, HSG, and Matrix Partners China — have all jointly placed a bet on a very low-profile embodied intelligence startup, Motion Tech. This scenario is extremely rare. According to the prospectus, Meituan is the largest external shareholder of Unitree; HSG not only invested early, but also became the independent private equity fund with the highest shareholding ratio in Unitree through multiple heavy rounds of investment. Matrix Partners China also led Unitree's early financing round and still holds more than 5% of its shares to date.

Like Wang Xingxing, the two co-founders of Motion Tech, Gao Jianrong and Yang Shuo, both used to work at DJI. Yang Shuo is the only founder in China's embodied intelligence circle who comes from the core team of Tesla Optimus. A closer look shows that Motion Tech and Unitree share quite a few common points: for example, both adhere to self-developed motors, attach great importance to full-stack software and hardware capabilities, and remain product-oriented and user-oriented. They also have some differences: Unitree's product portfolio is "quadruped robot dog + humanoid robot", while Motion Tech's product portfolio is "two-wheeled legged robot + small humanoid robot".

When the three institutional investors that previously backed Unitree all appear behind Motion Tech again, it inevitably sparks widespread speculation.

A Rare Team with DJI and Tesla Backgrounds

DJI alumni-founded startups have been a hot theme in the primary market investment over the past two years. A number of DJI alumni startups including Unitree, Bambu Lab, EcoFlow, and Cowa Robot have successively become leaders in their respective fields, drawing widespread attention. It is reported that Bambu Lab has set its 2026 revenue target at 24 billion yuan, and plans to launch its IPO process in 2027. This further convinces the market that DJI alumni companies hold the key to success in the hardware sector, and DJI has become the "Huangpu Military Academy" of the hardware industry.

It is understood that Gao Jianrong, CEO of Motion Tech, worked at DJI for 9 years. In the early stage, he was one of the most capable assistants to DJI founder Frank Wang, and later moved to specific business areas, successively taking charge of the RoboMaster robotics competition, DJI's supply chain, DJI's global marketing, and an independent business unit, DJI Education.

According to investors who have long observed DJI, it is extremely rare for executives inside DJI to span so many business areas and always report directly to Frank Wang.

Yang Shuo, CTO of Motion Tech, studied under Professor Li Zexiang. He joined DJI during his master's degree, working as an algorithm engineer for the key flight control module, and made core contributions to the Phantom product line. In 2016, Yang Shuo and Gao Jianrong jointly took charge of the RoboMaster competition, which marked the starting point of their partnership. In 2018, Yang Shuo left DJI to pursue his PhD at CMU, focusing on the control of legged robots.

After graduating with his doctorate, Yang Shuo joined Tesla's Optimus team, where the world's top robotics technical talents gather. He quickly became one of the highest-ranked Chinese engineers in the team.

"The Optimus team reports to Elon Musk every two weeks, and I attended every session," Yang Shuo said. "The deeper I got into the work, the more I realized that robots cannot do without China's industrial chain system. Instead of working in Silicon Valley, it is better to return to China, where the iteration speed is several times faster."

The recent boom in embodied intelligence was almost single-handedly driven by Elon Musk. The Motion Tech team, with first-hand insights into the embodied intelligence sector and DJI's hardware productization capabilities, received investments from top-tier institutions shortly after its establishment, which is not surprising at all.

First Product Unveiled, Breaking the Global Crowdfunding Record

One month before Unitree's listing, Motion Tech officially launched the company's first product Beni on the crowdfunding platform Kickstarter. Beni is a two-wheeled legged robot, and also the world's first camera robot.

At present, more than 5,000 people have paid to purchase this product. According to industry insiders, Beni has become the robotics category crowdfunding record holder in Kickstarter's history in terms of total funding amount.

During the crowdfunding period, Marques Brownlee (MKBHD), the most influential tech blogger on YouTube, reviewed and scored a number of recently launched tech products. Among more than a dozen products, Beni was the only one that got a full score of 10/10.

With more than 20 million followers, Marques Brownlee is widely recognized as the world's top tech blogger, with a dominant leading influence in the technology field. It is reported that MKBHD is almost not influenced by brands. In this episode of review, the lowest score was 2/10, so Beni getting a full score is of extremely high value.

The huge success of Beni is exactly the victory of Motion Tech's strategy of "cutting-edge technology + top-tier product strength". The wheel-legged robot form has actually existed for a long time, but the industry has never solved the productization problem. Beni is the world's first consumer-facing wheel-legged robot, which comes from the team's deep insight into user needs and profound understanding of robot platforms.

In terms of technical route, Motion Tech firmly chooses the direction of using reinforcement learning to train all functions of the robot. This new direction has been applied to humanoid robots, but has never been used for user-facing functions such as follow-up shooting before. A bigger challenge is how to maintain the consistency of these neural networks on mass-produced products — it is easy to make one robot jump 25 centimeters, but mass production requires the consistency of tens of thousands of robots. If the robot's jumping performance is inconsistent in different environments and under different loads, the structural service life will be greatly reduced. To this end, the team has carried out a large number of research and experiments at the algorithm and hardware levels, gradually narrowing the Sim-to-Real Gap, ensuring that the neural network can perform consistently on tens of thousands of robots.

Why did Motion Tech choose to enter the seemingly insignificant scenario of "low-angle follow-up shooting"? Motion Tech's logic is purely product-centric: shooting devices are naturally divided by perspective, drones cover the sky, smartphones cover portable scenarios, action cameras cover sports scenarios, but no one has ever filled the gap of "ground-level first-person perspective". The running of cats and dogs on the ground, the low perspective of children, the low-speed impact feeling of skateboards and tennis balls... These moments used to be missed. Beni does not replace any existing devices, but fills the blank in the market.

As the veil is lifted, Beni brings a brand new experience: it is the world's first portable all-terrain follow-up shooting robot, weighing less than 2 kilograms, capable of jumping up to 25cm, reaching a top speed of 28km/h, and it can also autonomously avoid obstacles, jump over steps, and get up by itself after falling. With Beni, users no longer have to worry about missing wonderful moments, as it can automatically help capture those fun moments in daily life.

At present, Beni already has its first batch of users, including pet owners, families with children, runners, roller skaters and many other groups. Combined with feedback and opinions from test users, Beni is still under continuous iteration. This attitude of emphasizing user experience and polishing the product together with users also distinguishes Motion Tech from its peers.

It is understood that Beni is scheduled to be officially released and sold globally in October after the end of the crowdfunding campaign. The Motion Tech team stated that its internal goal is to sell 1 million units within 3 years. If this goal is achieved, Beni will become the most popular robot product in history.

Moving Towards Humanoid Products, Adhering to Full-stack Capabilities and First Principles

Motion Tech's product development pace is similar to that of Unitree — neither company launched a humanoid robot first, but started with non-humanoid forms to develop products.

Yang Shuo further explained that the team insists on thinking about this issue with first principles.

At the hardware level, the core capability of robots is guaranteed by motors. Without mature motor design capabilities and production experience, startups often need to cooperate with suppliers through ODM or OEM methods. Although this approach is fast, it will raise costs on the one hand, and prevent tight collaboration between software and hardware on the other. Unitree's leading global robot body capabilities are precisely derived from the motor design and production accumulation brought by the mass shipment of its early quadruped robot dogs.

After accumulating motor capabilities through Beni, Motion Tech has now designed motors whose angle accuracy and torque density are no worse than those of Unitree. Especially under the same cost scheme, the team has improved the motor scheduling accuracy by an order of magnitude, which proves the team's solid hardware design capabilities.

In terms of the "big brain" and "small brain" system, Motion Tech adheres to the whole body intelligence route. "Many people gave up the humanoid biped direction because they were worried that they could not handle whole body intelligence," Yang Shuo said. "In the C-end home environment, trafficability is the first test for robots. It is hard for me to imagine how non-humanoid robots can adapt to the house scenarios in Europe and the United States."

To this end, Motion Tech has made a large number of scientific research breakthroughs in the past year. Based on the world model, Motion Tech has launched two brain technology architectures, DiT4DiT and MotionWAM, which gradually solve the industry-wide difficulty of upper and lower limb collaborative operation for humanoid robots. In terms of the cerebellum, Motion Tech's Perceptive BFM is the first in the industry to solve the problem of anthropomorphic gait walking on complex terrain.

In terms of products, Motion Tech has a mature set of methodology. "We insist on thinking about products based on scenarios and demands. Users do not care how the technology is implemented, they only care what the product can do, and whether its value matches its price," said Gao Jianrong. Through Beni, Motion Tech's products have entered families with children, pet-owning families and sports families. Among them, core users will become early co-creators of Motion Tech's first home embodied intelligence product. Designing embodied intelligence products close to users and scenarios is probably one of the most anticipated points of Motion Tech's home embodied product.

In addition, Motion Tech attaches more importance to consumer-grade product definition, user experience and data closed loop, and extends this set of capabilities from Beni to humanoid robots. Under the premise of compliance, Motion Tech can obtain data from users, and continuously iterate the product itself. With the arrival of mass production, Motion Tech expects the daily data volume to exceed 100,000 entries by the end of the year.

Witnessing the current hot scene of the track, the Motion Tech team always maintains the sobriety of entrepreneurs. They believe that the market overestimates the importance of operation models to robots. The real important issues are the safety and reliability of robots, as well as "a large number of reliability engineering work from Demo to mass production".

This is also reflected in Motion Tech's pace: it does not chase high valuations, and is willing to achieve steady growth in 3 to 5 years, "proving itself through one product after another". In Yang Shuo's view, human beings' expectation for "mechanical partners that are just like themselves" has lasted for thousands of years. This is a track worth devoting 10 years to, and there is no need to rush.

This article is from the WeChat official account "Pedaily" (ID: pedaily2012), written by Wu Qiong, authorized to be released by 36Kr.