Das Robotikunternehmen mit der höchsten Finanzierung hat in einem Jahr 150 Einheiten verkauft.

After three years of establishment, it has broken the industry financing records several times, and the cumulative financing amount is the highest in the industry. This includes raising 5 billion yuan within three months. The current enterprise value has already exceeded the mark of 20 billion yuan.

This young player in the industry named Yinhe General is, after Unitree Robotics and Zhipu Robotics, the most frequently mentioned company in the field of Embodied Intelligence.

In early March, Yinhe General announced the completion of a new round of financing worth 2.5 billion yuan, which once again increased the previous highest single financing amount in the robotics industry.

Its growth curve is steeper than that of Unitree. Compared with Zhipu Robotics, it surpasses both in terms of financing amount and enterprise value.

This rising company, founded by Wang He, a post - 90s outstanding student with a background from Tsinghua University, Stanford University, and Peking University, has won the full - scale support of "state funds + industrial capital + international institutions". It is also the first company in the field of Embodied Intelligence that the large - scale state fund has invested in. Its position in the industry speaks for itself.

As the world's first company in the field of Embodied Intelligence that has achieved comprehensive self - development from a "hundred - million - scale dataset" through an "Embodied Large Model" to a "robot body", Yinhe General follows a different development principle from other players in the industry. This is perhaps one of its most important features and the key reason why it has been recognized even without many product sales.

This super - unicorn is particularly observed by the market. The reasons behind it are also worth being examined more closely by the public and investors.

01

Remarkable

Together with large models and AI chips, Embodied Intelligence also triggers an enormous capital wave.

Since the beginning of 2026, the financing wave in the field of Embodied Intelligence has increased. Only in the first quarter, more than 50 financing events with a cumulative sum of nearly 20 billion yuan were reported, which corresponds to a 60% increase compared to the previous year.

Among them, 14 financings were over 1 billion yuan. Qianxun Intelligence raised four rounds and nearly 5 billion yuan within three months. Xingyuan Intelligence raised 1 billion yuan within 10 months. Among these companies, Yinhe General is the largest in terms of scale.

In early March, the company announced the success of a B++ round worth 2.5 billion yuan. The investors include the third large - scale state fund, Sinopec, CITIC Group, SAIC, BOC Asset Management, Yizhuang State - owned Investment, and others. This amount has also set a new record for a single financing round in the field of Embodied Intelligence. After the completion of this financing round, the enterprise value has successfully exceeded the mark of 3 billion US dollars (over 20 billion yuan).

At the end of last year, Yinhe General raised just over 300 million US dollars - about 2.1 billion yuan - in the B+ round, which at that time broke the record for a single financing round in the field. This means that it has received over 5 billion yuan in capital within just three months.

If we look at the financing history of Yinhe General, it is a rapidly rising company in the technology industry:

After its establishment in May 2023, it completed the seed round in a very short time and received the support of Matrix Partners and BlueRun Ventures;

In June and November of the following year, it successively completed the angel round with over 700 million yuan and the strategic round with 500 million yuan. Shenzhen Capital Group, IDG Capital, Meituan, and Qiming Venture Partners successively invested;

In June 2025, it received another 1.1 billion yuan in the B round. This was at that time the largest single financing amount in the field of Embodied Intelligence in China. CATL, GGV Capital, and CMBC International Capital all invested real money.

Yinhe General, which has only existed for three years, has broken the industry financing records three times and amazed all participants inside and outside the industry with a total financing amount of 7 billion yuan.

Even more remarkable is that its new financing round has won the favor of the third large - scale state fund. This is also the first time that the latter has intervened in the field of Embodied Intelligence.

The large - scale state fund is the undisputed "national team" in the chip investment industry. Behind it stand the Ministry of Finance, the State Development Bank, and state - owned enterprises such as China National Tobacco, China Electronics, and China Electronics Technology Group. As a state - owned investment institution, its investment direction also represents the direction of the national strategy.

The first prominent investment project of the first large - scale state fund was Semiconductor Manufacturing International Corporation (SMIC), whose significance in the field of wafer manufacturing in China is undisputed. This time, Yinhe General is the first "selected" object, which has a clear demonstration and guiding function.

This will also be one of the most important features that distinguish Yinhe General from other top players.

02

Race for the "Brain"

"Whoever manages to build a robot with large - model technology first is the world's leading AI company and even deserves the Nobel Prize." said Wang Xingxing, the founder of Unitree Robotics, in an interview once.

Since this year, most of the newly emerged super - unicorns in the field of Embodied Intelligence with an enterprise value of over 10 billion yuan belong to the "Embodied Brain" direction. This shows that the focus of investors has shifted from the "body" to the "brain" dimension.

The concept of the "Embodied Brain" is the large - model framework for robots, as Wang Xingxing calls it.

Considering that the PC era was mainly dominated by Intel chips and the Microsoft Windows operating system; the smartphone era was characterized by Qualcomm for chip supply, Foxconn for manufacturing, and Apple and Google for the iOS and Android systems, in the era of Embodied Intelligence, the "body" (the shell of the robot) will ultimately be led by China's strong supply - chain tension. Whoever masters the universal Embodied Brain has the ultimate power in this multi - billion - dollar ecosystem.

Currently, the robot large model is the biggest obstacle to the mass application of humanoid robots. The biggest problem as to why robots are not intelligent enough lies in the model, not in the data.

At the CCTV Spring Festival Gala in 2026, many robot companies were represented. The Galbot of Yinhe General independently performed tasks such as turning walnuts, collecting glass pieces, folding clothes, and skewering sausages without prior programming. These tasks were not "preset". Ultimately, the core lies in the fact that it has a "brain" that is becoming increasingly intelligent.

Currently, most companies in the industry develop the "brain" (task planning), the "cerebellum" (motion control), and the "limbs" (perception and execution) separately, and there is a lack of an end - to - end universal large model.

This leads to difficulties in seamlessly connecting the modules, resulting in delays and weak generalization ability.

Due to technology and the high costs of training and inference, most robot companies do not have the ability to develop their own large models. Many players in the field of Embodied Intelligence have to rely on the universal large models of giants such as Alibaba, ByteDance, and Huawei. This "survival with a borrowed brain" is difficult to maintain.

Yinhe General has chosen a more difficult path. It has integrated the brain, the cerebellum, and the nerve control and developed the end - to - end large model AstraBrain.

AstraBrain is one of the few Embodied Large Models that can establish the connection between "brain - cerebellum - nerve control". Its greatest advantage lies in its high generalization ability. Traditional robots only learn what they have seen and learned. AstraBrain enables the robot to conduct millions of "trial - and - error and play" in the virtual world and have the ability to "see, think, and act" like a human being.

In March of this year, Yinhe General presented a humanoid robot that can play tennis. This is due to the ability of the algorithm LATENT to control the intelligence of the entire body in real - time. As soon as this video was released, it received praise from Elon Musk on X.

It should be noted that this robot is not from Yinhe General itself, but the G1 from Unitree. Yinhe General has been a major customer of Unitree and generated a turnover of over 18 million yuan for Unitree in 2025.

The purchasing behavior of products from competitors is a strategy of Yinhe General. By collecting a large amount of data, Yinhe General can learn complex and non - simplifiable movement abilities.

In contrast to Tesla's Optimus, which relies on manual data collection, Yinhe General chooses a path based on synthetic simulation data and uses data collection from real machines as a supplement. Internally, this is referred to as the "Stellar - Works Data Pyramid".

Wang He, the founder of Yinhe General, once said that the training success rate is only 58% with 100,000 datasets; only when the data reaches 1 billion can the success rate rise to 87%. This figure is obviously unacceptable.

It can be seen that it is impossible to collect data only through manual remote control.

The method of synthetic data is relatively practical in terms of cost and scalability and has no time and space limitations. Through training with popular robots on the market and strengthening with extensive data, Yinhe General's own Galbot will become even more intelligent.

03

Behind the Capital Raising

Yinhe General was founded by two outstanding students in the field of robotics.

One is Wang He, who was born in 1992. He completed his bachelor's degree in the School of Electronics at Tsinghua University in 2014 and obtained a doctorate in electrical engineering from Stanford University in 2021. He was a student of the American Tri - Academy academician Professor Leonidas J. Guibas. After returning to China, he entered Peking University and was appointed as a lifetime assistant professor at the Peking University Frontier Computing Research Center - at the age of 34, he is already a supervisor of doctoral students.

Wang He also founded the Embodied Perception and Interaction Laboratory (EPIC Lab) at Peking University and has a deep academic foundation in the field of Embodied Intelligence.

Source: Peking University

The other founder, Yao Tengzhou, completed his studies at the Robotics Institute of Beihang University and was a student of the robotics expert and President of the Zhongguancun Zhiyou Institute, Professor Wang Tianmiao. He worked in the robotics research center of ABB Group in Shanghai and has experience in the mass production and sales of intelligent hardware products.

In the current world of Embodied Intelligence, an academic background is a passkey. Xu Xiaoping, the founder of ZhenFund, coined the term "Southern Xiang and Northern Miao" in 2016. "Southern Xiang" refers to Professor Li Zexiang from the Hong Kong University of Science and Technology, who has promoted well - known technology projects such as DJI and Googol Technology; "Northern Miao" is Wang Tianmiao. So far, he has listed three companies on the STAR Market - Tianzhihang, Effort and Ninebot. The founders of Elite Robotics and Yinhe General are his students.