Is the 20-billion-yuan valuation of Galaxy Universal driven by massive actual orders or capital hype?

Investors are major clients, and the shareholder list highly overlaps with the customer list.

In September, Galaxy Universal was pushed to the cusp of public opinion by two incidents at the same time. One was that it won the 236 million yuan embodied intelligent equipment procurement project of CATL on September 10; the other was that the company issued a solemn statement on September 12, reporting the revenue doubts circulating on the Internet to the public security organ — previously, the founder of its peer Mech-Mind had publicly questioned the authenticity of its reported revenue.

This is the most distinctive company in China's humanoid robot track: it raised 7 billion yuan in three years with a valuation of over 200 billion yuan, ranking first in the industry in both financing scale and valuation, with the National Integrated Circuit Industry Investment Fund, CATL, China Mobile and Sinopec all participating in the investment; but at the same time, it is also the company that has been most fiercely questioned. The real value of Galaxy Universal needs to be calculated separately.

How the "Financing Machine" Was Built

Let's first sort out its financing history, which itself is already a phenomenal case.

Founded in May 2023, it received the seed round investment from Matrix Partners China and BlueRun Ventures just one month later; in June 2024, it raised 700 million yuan in the angel round, with Meituan, BAIC Group, BlueRun Ventures, Qiming Venture Partners and IDG Capital entering the market; in November of the same year, it raised 500 million yuan in Series A, with SAIC Motor, Hong Kong-based investment firms and Shenzhen Venture Capital participating; in June 2025, it raised over 1.1 billion yuan in Series B, led by CATL — this is the first time the leading power battery enterprise has made a heavy investment in embodied intelligence; in December 2025, it raised over 300 million US dollars in Series C, led by China Mobile, followed by CICC Capital, the Chinese Academy of Sciences Fund and CCTV Converged Media Fund; in March 2026, it raised another 2.5 billion yuan, led by the National Artificial Intelligence Industry Investment Fund, which is the first investment of the third phase of the large fund in the embodied intelligence sector.

In three years, it raised 7 billion yuan in financing, with three generations of capital including national-level funds, industrial giants and first-tier US dollar funds standing on its shareholder list.

The resume of founder Wang He lays the foundation for such financing capability: Doctor of Stanford University, studied under Leonidas Guibas, a famous expert in computer vision, researcher at Peking University, and founder of the first lab named after embodied intelligence in Chinese mainland.

The peak of its capital story appeared in the Spring Festival Gala of the Year of the Horse — the robot of Galaxy Universal appeared on the stage as "the first independently working robot", performing walnut spinning, picking up glass fragments and stringing sausages, hundreds of orders poured in within 24 hours, and the same model launched on JD.com after the gala was sold out in several minutes.

The Other Side of Orders: Investors Are Major Clients

The core of the controversy is also hidden here. The largest client of Galaxy Universal is CATL — Galbot S1 operates 7*24 hours in CATL's battery factories, with nearly 100 units deployed. It is the only humanoid robot that has realized normalized operation in CATL, and won the 236 million yuan procurement project again in September. But CATL is also the lead investor of its Series B round.

There are similar structures: Meituan is a shareholder, and Galaxy's robots have entered Meituan's smart pharmacies; BAIC Group and SAIC Motor are shareholders, and Galaxy's robots are running on their production lines.

Investors are major clients, and the shareholder list highly overlaps with the customer list.

How to interpret this model? The observer cited by Phoenix Finance put it fairly: the orders are real, and the revenue has also been confirmed by audit, but the terminal demand and sustainability of the revenue dominated by shareholders and related parties depend on the specific scenarios and repurchase situations.

Another more hidden practice in the industry has been pointed out: local governments take the lead in building data collection platforms, and enterprises sell robot equipment while the platforms purchase data — this is reasonable in the early stage of the industry, but it is easy to be alienated into "inflating the turnover".

The report statement issued by Galaxy Universal on September 12 targets the online doubts, and the doubts themselves just show that the market has begun to look at this track with a new set of standards.

The Industry's Measurement Standards Have Changed

The situation of Galaxy Universal should be viewed against the background of the industry inflection point. On August 28, the National Development and Reform Commission clearly stated that the robot industry should "prevent blind following and swarming", and emphasized forming an application closed loop around real demand; at the same time, the IPO review of humanoid robots is tightening, and the focus of regulatory inquiries is real revenue, related party transactions and performance capabilities.

With the 73-day SciTech Innovation Board listing approval speed of Unitree Robotics as a precedent, the market will only scrutinize subsequent entrants more strictly — more than 30 robot companies in the Hong Kong stock market are queuing to submit listing applications, and none of the listed ones are profitable, the measurement standard of review is changing from "storytelling" to "cash flow".

In terms of technical routes, Galaxy Universal has always been at the center of controversy. Wang He is a firm advocate of the "synthetic data school" — using simulation environments to generate tens of billions of data to train models, while opponents such as Gao Jiyang from Star Seaview argue that the first stage must rely on real data combined with imitation learning.

There is no conclusion in this route dispute, but it determines the report logic of Galaxy: if the synthetic data route is proven to be feasible, Galaxy's advantage in data cost will become its moat; if not, the early investment will need to be revalued. The Spring Festival Gala stage verified the lower limit of its generalization capability, and the three-shift operation in factories is the verification of its upper limit.

Fairly Speaking, Its Foundation Is Not Poor

Putting the controversy aside, Galaxy Universal does have some hard strengths. In terms of shipment, the shipment of Galbot G1 exceeded 1,200 units in 2025, and the cumulative shipment in the first half of this year exceeded 1,100 units, ranking third in the industry; in terms of clients, the list of CATL, Bosch, Toyota, BAIC Group and SAIC Motor is real, and the "Galaxy Capsule" smart retail has been deployed in more than 100 stores in over 20 cities, and its cooperation with Meituan's unmanned pharmacies is expanding from 7 stores in Beijing to 100 stores in Beijing, Shanghai and Shenzhen, which can grab 5,000 SKUs of medicines. Wang He once said a very sober sentence: "We do not pursue robots that can do 100 things, but make them do 10 things up to industrial standards."

The only real question left is: among these orders, how many come from shareholders, and how many come from unrelated clients. This proportion is the most critical field in Galaxy Universal's prospectus — it determines how much of its 200 billion yuan valuation comes from capital support and how much comes from real orders.

So back to the original question: is Galaxy Universal's 200 billion yuan valuation supported by orders or by capital? The fair answer is both, but the specific proportion is not clarified — and the significance of IPO is exactly to force it to clarify. Related party transactions are not original sin, the structure that CATL is both a shareholder and a client even shows that industrial players are voting with real money; what really needs to be verified is whether orders from unrelated clients can be sustained when the shareholder list no longer expands.

For Galaxy Universal, there are two upcoming examinations: the customer structure inquiry at the Hong Kong stock listing hearing, and the repurchase rate from non-shareholder clients on the production lines. For the entire embodied intelligence industry, the result of this examination will become the reference for the pricing of subsequent entrants — the first company that is forced to make the overlap degree between shareholders and clients public is often the starting point for the industry to shift from storytelling to credit building.

(Risk Warning: This article is only an analysis of the industry and the company, and does not constitute any investment advice.)

References:

1. Huxiu: "The IPO Threshold of Embodied Intelligence Is Tightening: The Industry Questions Real Revenue, Related Party Transactions Arouse Controversy"

2. Changjiang Daily: "Galaxy Universal Raised 7 Billion Yuan in Three Years with a Valuation of 200 Billion Yuan"

3. Sina Finance: "Galaxy Universal Announced the Completion of a New Round of Financing of 2.5 Billion Yuan"

4. Phoenix Finance: "Galaxy Universal Reports the False Information on the Internet to the Police"

5. Southern Metropolis Daily: "Galaxy Universal Robotics Raised 1.1 Billion Yuan in New Financing, Led by CATL"

6. OFweek Robotics: "Humanoid Robot Industry Breaks Out into a War of Words"

7. 36Kr: "The Mass Production Watershed of Humanoid Robots"

8. Official Disclosure of Galaxy Universal and Public Reports