Investors behind ZHIYUAN and Galaxy Universal: How I Keep Betting on "All-Rounders"

Chen Weiguang, a managing partner at BlueRun Ventures, still clearly remembers the first time he met Wang He, the founder of Galaxy Universal, in July 2023. They had a three - hour conversation and even forgot to eat. Wang He didn't talk much about the specific implementation path of robots. Most of the time, they discussed how to achieve the deep integration of the physical world and artificial intelligence.

“He repeatedly emphasized to me that this is a long and difficult ‘journey to the West’,” recalled Chen Weiguang, but he still felt extremely excited.

BlueRun Ventures is one of the seed - round investors in Galaxy Universal and has continuously invested in multiple rounds of the project's subsequent financing. As one of the few senior managing partners in China who are still actively involved in evaluating projects on the front line, Chen Weiguang always maintains his keenness and patience when dealing with early - stage projects.

AI and embodied intelligence are key areas for the BlueRun Ventures team. For example, BlueRun has invested in Darkside of the Moon, Genspark, Dance with Love, RockFlow, and Vivix in the model and application layers; in the field of embodied intelligence, it has bet on ZHIYUAN ROBOTICS, Galaxy Universal Robotics, Tashizhihang, Lingchu Intelligence, and Hillbot; in the direction of AI hardware, it has invested in VITURE, KeYi Technology, and Haivivi...

“Actually, some investments are still under the surface and have not been officially announced,” said Chen Weiguang. He introduced that BlueRun Ventures sees 1,000 - 2,000 projects every year, “and may only invest in 1% of them.”

Regarding the embodied intelligence industry, Chen Weiguang put forward two observations: First, the future market capacity of embodied intelligence is at least 10 times that of the automotive market. The annual automobile sales volume is about 100 million units, while embodied intelligence (intelligent robots) targets about 3.5 billion of the world's labor force. If each person is equipped with 1 unit or even 2 - 3 units, the potential demand scale can reach billions. Second, once the generalization ability of intelligent robots is improved, in the future, a single device can be reused in multiple scenarios such as factories and logistics. Compared with the past “functional robots”, the marginal cost will be significantly reduced, thus changing the future production mode.

Therefore, when being interviewed by China Entrepreneur at the end of 2025, Chen Weiguang was very optimistic about the prospects of “AI + hardware” in the next decade.

Wonderful views from the article:

1. Embodied intelligence is not equivalent to humanoid robots. Embodied intelligence is the combination of the physical world and AI. People should imagine embodied intelligence as the interaction and delivery of AI combined with hardware in the physical world.

2. Investors in the embodied intelligence track are quite enthusiastic, and the valuations and financing amounts are rising relatively quickly. Entrepreneurs should be more calm and still focus on solidifying their products and businesses.

3. Embodied intelligence is a systematic project. It doesn't just rely on algorithms or better hardware. The team needs to have strengths in various aspects including the ecosystem, scenario selection, and financing. The most feared thing is a significant weakness in one aspect. The best teams are well - rounded.

4. Among the world's 8 billion population, 3.5 billion are employed. If each worker has an embodied robot to assist in life and work, their productivity will be greatly improved. Therefore, the future market capacity of embodied intelligence is at least 10 times that of the automotive market.

The following is an abridged dialogue between China Entrepreneur and Chen Weiguang.

01

Embodied intelligence teams should be “well - rounded players”

China Entrepreneur: BlueRun started investing in robots in 2015. When did you think that embodied intelligence was a very worthy track to invest in? Was it because of GPT - 3.5, or were there any other landmark events?

Chen Weiguang: I think investment is a process of repeated exploration from the bottom - up and top - down. From a top - down perspective, we saw the inflection point of AI. Previously, there were only functional robots, just like in the mobile phone industry before 2008, there were only feature phones with single functions such as making calls or playing games. After Steve Jobs' iPhone moment, mobile phones became multi - functional.

As the capabilities of AI and machine learning increase, robots have gradually become more generalized. However, this generalization is more reflected in actions, that is, robots can perform different forms of movement such as dancing and fighting.

The emergence of large models means that robots have the ability to interact.

We saw a significant improvement in AI, just like the iPhone back then. Its software capabilities were enhanced, and things that were previously impossible to do can now be achieved.

From a bottom - up perspective, in the process of contacting entrepreneurs, we saw teams that could better combine the software “brain” and “cerebellum” with hardware. If the team combination is not strong enough, or not what we often call “well - rounded players”, that is, they understand the “brain”, the “cerebellum”, and the body, even if we detect the inflection point, we can't invest. Coincidentally, at that time, teams with well - rounded capabilities emerged, so we continued to invest. We invested in ZHIYUAN ROBOTICS and Galaxy Universal around the same time.

China Entrepreneur: How did you discover these two projects?

Chen Weiguang: I'm quite familiar with Yin Fangming, the CEO of Galaxy Universal. I also supported his previous projects. So when he decided to enter the field of embodied intelligence, we knew about it immediately. At that time, we had seen the inflection point of AI, but we weren't completely sure if he could make it happen.

Later, I talked with Wang He, another co - founder and CTO, for three or four hours and even forgot to have dinner. He rarely talked about how to make robots specifically but spent a lot of time discussing how to deeply integrate the physical world with artificial intelligence. I thought that Galaxy Universal could combine software and hardware to achieve embodied intelligence, so I decided to invest.

Regarding the ZHIYUAN ROBOTICS project, our team noticed Peng Zhihui (co - founder, president, and CTO of ZHIYUAN ROBOTICS) early on. After hearing that he might start a business, we actively contacted him. ZHIYUAN ROBOTICS originally only focused on hardware. Later, when we talked, they had started to think about further deepening in the software field, which was quite in line with our understanding.

ZHIYUAN ROBOTICS. Photography: Kong Yuexin

Although Galaxy Universal and ZHIYUAN ROBOTICS seem to be doing the same thing, they focus on different scenarios.

China Entrepreneur: You're still continuously paying attention to embodied intelligence. Has the investment logic changed since you first invested in this field?

Chen Weiguang: I think it's still about scenarios. There are billions of workers around the world, and there are many scenarios for different occupations. Currently, the manufacturing, logistics, and daily - life scenarios are quite clear, and there are great opportunities in these three scenarios. After establishing a foothold in one vertical field and getting the data flywheel spinning, it's relatively easy to expand to the next vertical industry.

For example, Tashizhihang has broken through in flexible and precise manufacturing and solved the “Goldbach's Conjecture” in the field of industrial automation - wire harness manufacturing. This is not just a success in a single scenario. It provides a verifiable and replicable industrial - level practical sample for the industrialization of the embodied intelligence Scaling Law.

As for whether these embodied intelligence companies will compete with each other, I think the market is too large. Although they are all robots, they replace different jobs or improve the efficiency of different occupations.

China Entrepreneur: But there should be quite common competition for talents among these companies, right?

Chen Weiguang: For ordinary engineers, I think it's okay. After all, China still has an advantage in the engineer dividend in the next five years. The competition for top - notch and highly - skilled talents is definitely more intense.

China Entrepreneur: You were among the first institutional investors in embodied intelligence. Generally, investing in the early stage is very risky. Have you had any failed cases in embodied intelligence?

Chen Weiguang: Not yet. All the projects in embodied intelligence are developing well, and subsequent financing has also been fast. We spent a lot of time researching who the top - tier people in the market are, had in - depth conversations, and had a wide range of contacts. So we can roughly know the level of each project.

China Entrepreneur: How do you design a framework based on some variables or parameters to screen out projects and entrepreneurs with a relatively high success rate? Are there any “minefields” for you?

Chen Weiguang: If the team has a strong scientific research background but lacks expertise in other areas, we may be more cautious. Because embodied intelligence is a systematic project. It doesn't just rely on algorithms or cutting - edge technologies. Of course, the team needs to have strengths in various aspects, but the most feared thing is a significant weakness in one aspect.

China Entrepreneur: How do you evaluate projects when looking at them?

Chen Weiguang: For example, if there are six indicators, we don't require each indicator to score 90 points, but at least each should score 70 - 80 points.

Of course, this scoring can't be based on products or revenues because when we invest, the product may just be a movable sample that may not even be able to perform tasks. We may score based on these six indicators - such as what they've done in the past, the core points and thinking points in their communication with us. A strong team will think more deeply and comprehensively.

China Entrepreneur: What are the key capabilities you value in the field of embodied intelligence?

Chen Weiguang: I think embodied intelligence is different from other sub - sectors. For example, when investing in AI applications, it's more about whether the company can capture user needs and develop quickly. But embodied intelligence is a systematic project. If there are obvious weaknesses in capabilities, problems may easily occur.

For example, the founder needs to understand hardware, software, and AI algorithms, and also have the ability to raise funds and build an ecosystem. At the current stage, the most crucial thing for entrepreneurs is to find suitable scenarios, design products based on scenarios, and establish differentiated advantages to fully release the latest capabilities of AI. After the industry becomes relatively mature, the ability to build an ecosystem will become even more important.

02

The future trend is hardware + AI

China Entrepreneur: Against the backdrop of the continuous heating - up of the industry, are you still looking at embodied intelligence?

Chen Weiguang: We're still continuously looking at embodied intelligence. However, I think the real big trend is the combination of hardware and AI. For example, a car is also a robot. China is very strong in consumer hardware. DJI has proven this, but it may represent the previous generation. In the future, there will be many projects that combine consumer hardware and AI to meet daily - life needs, and we're also looking at this area.

For example, we've also made some investments in AI glasses. People may think that embodied intelligence is equivalent to humanoid robots. I think this is too limiting. People should imagine it as the combination of AI and hardware.

Galaxy Universal Robotics. Photography: Kong Yuexin

Life doesn't just rely on AI to give us an answer. Eventually, we need to implement this answer into results. These results may require hardware to complete, or an agent to help with analysis. The analysis results can generate files such as PPTs and Excel spreadsheets, or structured reports suitable for board meetings. This is where AI can truly produce results and effects.

The same goes for the combination with hardware. There are still many hardware devices that require human operation. For example, current mobile phones are not truly AI - enabled. Suppose I'm going on a business trip to Africa. I need to search for information, buy tickets, and arrange my itinerary after arrival on my phone by myself. In the future, can I just say “arrange a business trip” to my phone, and it will automatically buy tickets and even arrange a robot to pick me up when I land?

In the future, whether through the combination of software and hardware or better agents, many tasks that we used to handle ourselves will no longer be necessary.

Many people may have concerns. For example, if a boss has 40 employees and uses AI, will these 40 employees lose their jobs? I don't think so. With cheaper robots and more useful agents, these 40 people can become bosses themselves.

If they can become bosses through AI and robots, they can earn money and then consume. Others will use their agents and robots to serve them. So it's not as pessimistic as people think.

Of course, those who don't learn about AI or understand robots will definitely have a harder time.

China Entrepreneur: Do you feel there are any chaos or bubbles in the industry?

Chen Weiguang: Currently, a lot of capital is flowing into the robot industry, so there may be cut - throat competition.

China Entrepreneur: Have you ever experienced competition for projects?

Chen Weiguang: Definitely, especially for top - notch projects.

China Entrepreneur: Given the current investment and financing environment, what suggestions do you have for the embodied intelligence projects invested by BlueRun?

Chen Weiguang: To be honest, we mostly suggest that founders don't raise funds too quickly before achieving good commercial results because the current track is quite hot, and we don't know what the situation will be next year. Once they raise a large amount of money but fail to meet the expectations, it will be quite troublesome later. For teams that have already raised a relatively large amount of money, we suggest they focus on commercialization. After all, they have enough money and should spend more time on business. For teams that haven't raised much money and still need time to develop their products, we encourage them to raise as much as possible.

China Entrepreneur: Previously, people thought that the capital enthusiasm in this wave might subside by the end of 2025 or the beginning of 2026. Do you think it has cooled down recently?

Chen Weiguang: It's not that it has “cooled down”. From an investment perspective, the next team you invest in should have significant differences in either the selected scenario or technology compared to the previous ones because most of the good projects have already received investment.

As long as there is a better team that finds a scenario not covered by others before and has significant technological changes, I think it can still raise funds. Currently, embodied intelligence technology is not yet mature. It can operate in factories, but it's difficult to use it to take care of the elderly at home. If you can develop a robot technology that is inexpensive and can take care of the elderly at home, this disruptive technology will definitely be able to attract investment because others can't do it now.

03

China's advantages in the global competition

China Entrepreneur: After a year of rapid development in the embodied intelligence industry, do you think there are more or fewer industry consensuses? What consensuses have been formed, and what may still be non - cons