Summary of Key Points
Shangtang Technology has established an ecosystem that covers the entire AI industry chain by splitting off internal subsidiaries and supporting former employees to start their own businesses, earning it the nickname "Huangpu Military Academy" of China's AI field. Companies within this ecosystem, such as Xiwang Chips, Daxiao Robotics, and MiniMax Large Model, have secured substantial funding or gone public. However, beneath the surface of this success lie challenges including difficulties in generating profits, weak ecological synergy, and talent loss that pose threats to their competitiveness.
I. The Shangtang Ecosystem: A Network Spanning the Entire AI Industry Chain
The Shangtang ecosystem consists of two main categories: internally spun-off subsidiaries and companies founded by former employees, covering almost every segment of the AI industry:
- Internal Subsidiaries: Notably, Xiwang Chips has raised over HK$4 billion in seven rounds of financing in less than a year and half, with a valuation exceeding HK$10 billion; Daxiao Robotics (focusing on embodied intelligence) has secured hundreds of millions in funding within six months. Other subsidiaries include Jueying (working on intelligent vehicles, already mass-produced by GAC Trumpchi), Yuanluo Bo (selling intelligent chess and card-playing robots, leading sales on two platforms for three consecutive years), and Shangtang Shanhui (partnering with Midea and Meiyijia in smart retail).
- Founded by Former Employees: Momenta (autonomous driving) became the first "physical AI" company listed on the Hong Kong Stock Exchange; MiniMax (large model) saw its stock price rise by over 109% on its debut day, with a market value exceeding HK$41 billion; Vivix AI (multimodal solutions for overseas markets) was valued at $1.32 billion just one year after its establishment. Other companies include Huixi Intelligence (embodied computing power) and Zhiyan Huisheng (open-source models), operating in areas such as 3D generation and enterprise services.
These companies, like dandelion seeds, have spread across various corners of the AI industry, becoming a crucial part of China's AI landscape.
II. Why Has Shangtang Been Able to Produce So Many "Unicorns"?
The rise of the Shangtang ecosystem is not accidental and is supported by three key factors:
1. Strong Technical Foundation: The founding team comes from the MMLab at the Chinese University of Hong Kong, one of the world's top AI research institutions. The engineers trained there have extensive academic backgrounds and practical experience in large-scale AI projects, enabling them to successfully bring technologies to market.
2. The "1+X" Strategy Provides Institutional Flexibility: As of the end of 2024, Shangtang restructured its business into a core business (AI cloud and large models) and subsidiaries (such as Jueying and Yuanluo Bo). These subsidiaries can independently raise funds and make decisions, freeing them from the constraints of a larger company structure and allowing them to quickly capture market opportunities. The parent company can then focus on its core areas.
3. Shared Technical Infrastructure: Shangtang established the "SenseCore" computing power center in 2018, which serves as a foundation for AI development. In 2023, it launched the "Riri Xin" large model. Subsidiaries can leverage this infrastructure without having to start from scratch, significantly reducing startup costs by optimizing their solutions for specific use cases (e.g., using Riri Xin with medical data in Shangtang Medical).
III. Behind the Bright Financing Figures: Unproven Profit Models
Despite impressive financing and valuations, these companies face significant challenges in generating profits:
- Parent Company Shangtang: Revenue in 2025 was HK$5 billion (a 32.9% increase), but net losses amounted to HK$1.78 billion. Although the losses have decreased, they continue to erode investor patience, and its market value is now only a fraction of its peak (around HK$600 million).
- Subsidiaries: Xiwang Chips had revenue of HK$240,000 in 2024 with a loss of HK$190 million; Daxiao Robotics had no revenue in the first quarter of 2025 and incurred a loss of HK$22.85 million. Daxiao Robotics' retail inspection solutions have been implemented in just over a hundred Rosengarten stores, far from its goal of reaching a thousand stores within a year. Shangtang Medical has entered top-tier hospitals, but the high regulatory barriers and low willingness to pay for AI services make it harder to generate profits on a large scale.
Investors are now more interested in revenue and profitability than just valuations, and these companies still need to prove their commercial viability.
IV. Lack of Ecological Synergy: The Spreading Stars Have Not Yet Created a Firestorm
Despite the size of the Shangtang ecosystem, there is little collaboration among its companies:
Xiwang focuses on chips, Daxiao on robotics, Jueying on autonomous driving, and Shangtang Medical on AI in healthcare—there is little overlap in their operations, and they do not purchase goods or collaborate with each other. The parent company primarily provides technology, talent, and funding, acting more as a support system rather than a central coordinating entity. The AI industry has not yet reached a stage where complex synergies are possible, so the individual companies' efforts do not combine to create significant value.
V. Talent Loss: Cultivating Competitors Within the Company
A problematic aspect of the "Huangpu Military Academy" model is that its graduates often compete with the company:
- MiniMax's large model competes directly with Shangtang's Riri Xin.
- Momenta's autonomous driving technology competes with Shangtang's Jueying for customers.
- Vivix AI's AIGC products overlap with Shangtang's own solutions (e.g., "Miuhua" and "Ruying").
This suggests that Shangtang's incentive mechanisms and platform may not be attractive enough to retain top talent, as the companies it has trained end up competing with it. This is an issue that Shangtang needs to address.
Conclusion
The rise of the Shangtang ecosystem reflects the development of China's AI industry, but the underlying challenges in profitability, synergy, and talent management are common across the entire AI sector. For the ecosystem to truly take shape, Shangtang must not only share its technology and talent but also find ways to ensure that these resources work in its favor rather than becoming threats. After all, having the reputation of a "Huangpu Military Academy" is not enough; it must transform this ecosystem into a true competitive advantage.