Summary of Key Points
AI computing power cluster provider Jiliu Technology, which was established just three years ago, plans to list on the Hong Kong Stock Exchange. Over the past three years, it has completed 11 rounds of financing, and its valuation has soared from 29.4 million yuan to 9.16 billion yuan (an increase of 311 times), largely due to its deep partnership with Zhipu Huazhang (which is both a shareholder and a major customer). However, the company faces several challenges: its debt-to-asset ratio exceeds 136% (more debt than assets); 80% of its revenue comes from low-margin hardware integration services; its customer base is highly concentrated (relying on a few key clients); there are concerns about its independence due to overlapping customers and suppliers; and it is rushing for an IPO despite not yet having a self-sustaining business model.
I. The “Shovel Seller” in the AI Boom, Earning “Hard-Sworn Money”
Jiliu Technology plays the role of an “shovel seller” in the AI industry: it does not manufacture chips (such as NVIDIA GPUs) or develop large models (like ChatGPT). Instead, it connects thousands of GPUs into supercomputers and sells them to large model companies, research institutions, or cloud service providers, while also offering maintenance services. In the past two years, the demand for AI computing power has skyrocketed, leading to rapid revenue growth—31.8 million yuan in 2023 (10 months after establishment), 325 million yuan in 2024, and 520 million yuan in 2025. However, most of its profits come from hardware integration (assembling GPUs into clusters for sale), with a gross margin of only 16.8%—meaning it earns a small profit on each transaction. The more profitable maintenance services account for just 16.1% of revenue.
II. High Debt and Continuous Losses: A Business Model That Hasn’t Yet Turned Profitable
Jiliu Technology’s financial situation is somewhat precarious:
- Debt Exceeds Assets: As of the end of 2025, its debt-to-asset ratio was 136.61%, indicating that it owes more (1.188 billion yuan in current liabilities) than it owns (739 million yuan in current assets), nearly on the brink of insolvency.
- Continuous Cash Outflows: The company has had negative cash flows over the past three years (−9.4 million yuan in 2023, −21.2 million yuan in 2024, and −1.1 million yuan in 2025), although it turned positive in 2025, it still relies on external financing to survive.
- Losses in 2025: The main reason for the loss was changes in the fair value of preferred stocks acquired during financing, indicating that its core business has not yet become profitable, and its growth is largely dependent on capital injections.
III. Highly Concentrated Customer Base: Both Client and Supplier, a Major Risk
Jiliu Technology’s dependence on its customers is significant:
- Early Success依赖于 a Few Key Clients: In 2023, the top two clients accounted for 97.1% of revenue, and in 2024, the top five clients accounted for 98.9% (with the largest client accounting for 59%). Although this proportion decreased to 56.6% in 2025, it remains high.
- Sensitive to Client Losses: For example, a major client in 2023–2024 dropped out of the top five in 2025, causing a significant impact on revenue. Since AI computing power projects are large-scale deals, customer switches to competitors or self-built solutions can lead to a sharp drop in revenue.
- Overlapping Customers and Suppliers: Some customers act as both buyers of its products and suppliers of hardware (GPUs). While this is common in the industry, it raises questions about the company’s independence.
IV. Zhipu Huazhang: The “Backstage Driver” of the 311-Fold Valuation Surge
Zhipu Huazhang has played a crucial role in Jiliu Technology’s rapid growth:
- Continuous Capital Investment: Through its ecological fund, Xinglian Capital, Zhipu has invested in Jiliu from the angel round to the D round, and now holds 7.7% of the company’s shares.
- Business Support: Zhipu is both a major customer (48.5% of revenue in 2023, third-largest client in 2024). This partnership has fueled Jiliu’s early success and led to a 311-fold increase in valuation over three years. However, this dependency also means that Jiliu’s performance could be affected if Zhipu reduces its cooperation.
V. Rushing for an IPO After Just Three Years: Questions About Compliance and Risk Resistance
Jiliu’s decision to list so soon has sparked controversy:
- Frequent Financing: Eleven rounds of financing in three years, averaging less than four months per round, raises doubts about whether the company’s growth is driven by business performance or external capital.
- Compliance Issues: The Hong Kong Stock Exchange requires new companies to have at least three full fiscal years of operating data, but Jiliu only operated for 10 months in 2023 (from February to December), so it had to apply for an exemption to list.
- Lack of Risk Resistance: Having only three years of history and no experience with economic downturns, there are concerns about whether the company can withstand a slowdown in AI computing power demand.
Jiliu Technology’s story reflects the boom in AI computing power. To establish itself in the secondary market, it must address issues such as self-sustaining growth and reducing its reliance on customers. Otherwise, a company that has been rapidly developed by capital may not be able to withstand the long-term challenges of the market.
(The translation maintains the original Markdown structure, using clear language suitable for financial journalism, and adapts expressions to fit the target audience.)