Prospect of New Stocks | Soaring Revenue, Just Turned Profit, Starry Sky Convergence's Confidence and Concerns in Sprinting for Hong Kong IPO

After large models enter the stage of large-scale training and inference, the market's understanding of AI infrastructure is changing. In the past, capital markets were more easily focused on GPUs, servers, and computing power centers. However, as model parameters, training data, and inference requests continue to expand, the efficient storage, management, and delivery of data to GPUs with sufficiently low latency are becoming significant bottlenecks in AI infrastructure. As China's largest independent distributed AI storage solution provider, Beijing Xingshentianhe Technology Co., Ltd. (hereinafter referred to as "Xingshentianhe") has officially submitted its prospectus, seeking to go public on the main board through the Hong Kong Stock Exchange 18C (Specialized Technology Company) channel. According to the prospectus, the company is projected to be China's largest independent distributed AI storage solution provider by installed capacity by 2025, capturing approximately 10.6% market share; however, when considering the entire Chinese data storage market, the company's revenue ranking is about 10th to 15th, with a market share of only around 0.1%. In the distributed AI storage market, including both independent and non-independent vendors, its revenue share is about 2.7%. As an independent storage vendor, Xingshentianhe faces dual scrutiny in the capital markets: on one hand, within an ecosystem of powerful giants and closed-source bundling, can the survival space and premium logic of independent storage be established? On the other hand, the company is just on the verge of turning a profit in 2025, but its operational cash flow continues to drain, and the volatility in upstream hardware prices raises questions about the "quality" of its financials. Betting on "data gravity"? Can independent storage withstand the giants attacks? It is understood that there exists a recognized classification line in the enterprise-level IT storage market: one category is "non-independent vendors," which provide bundled storage tightly integrated with their proprietary hardware, with software usually only compatible with self-produced servers; the other category, represented by Xingshentianhe, is "independent vendors," focusing on software-defined storage (SDS), capable of elastic deployment across third-party hardware from different manufacturers and heterogeneous computing platforms. From the monetization loop perspective, Xingshentianhes core business logic lies in deploying its proprietary distributed storage software stack, taking on end-to-end responsibilities for system design, hardware selection, procurement deployment, and after-sales maintenance. Its delivery models are categorized into "software-based solutions" (only delivering the software layer, with clients providing their hardware) and "integrated machines" (where self-developed software is pre-integrated into designated hardware as a single deliverable). Typically, clients often cannot derive tangible benefits from simply purchasing hardware or software; hardware and software collaborate closely throughout the entire solution. Moreover, the integrated machine devices are priced and contracted as singular deliverables. Furthermore, subsequent AI storage services, including warranty, maintenance, renewals, and capacity expansion as value-added services, create high stickiness in the follow-up monetization loop. With the explosive growth of AI model parameters, it is estimated that data volume in mainland China will soar from around 50 ZB in 2025 to over 150 ZB by 2030, with a significantly high proportion of unstructured data. Xingshentianhes distributed architecture supports linear scaling of storage pools to PB and even EB levels, directly driving existing clients to make "capacity additions" due to the explosive growth of data volume. The prospectus shows that the company's overall net dollar retention rate (NDR) is impressive, increasing from 46.5% in 2023 to 91.9% in 2024, and reaching 132.5% in 2025, with a historical high of 160.0% in the first half of 2026. This retention performance confirms the commercial law of "data gravity" in enterprise-level storage: once clients core business data and AI workflows become deeply embedded in the platform, the cost of migration becomes prohibitively high, leaving clients no choice but to continuously augment their purchases as data volume grows. In terms of average unit price, according to the prospectus, the average selling price of the companys AI data lake solution dramatically increased from RMB 192,700 in 2024 to RMB 255,300 in 2025, and rose to RMB 497,600 in the first half of 2026; the average selling price of the training and inference storage solution similarly surged to RMB 494,900 in the first half of 2026. Behind the increase in unit prices is the accelerated investment in AI-related data infrastructure by AI clients, leading to larger procurement volumes per project and the continual expansion of project scales. Meanwhile, the share of one-stop "integrated machine" deliveries has significantly risen. It is understood that in the first half of 2026, the proportion of integrated machines in the company's total revenue soared from 42.7% in 2025 to 55.3%. Since integrated machines include hardware value, their total contract amounts far exceed pure software projects, thus increasing unit price performance in sales terms. Overall, Xingshentianhes monetization loop is a rolling business model characterized by "replacing hardware with software, locking in existing stock with data gravity, and increasing unit prices through a one-stop integrated machine." Financial quality and hidden concerns behind hardcore technology Assessing a rapidly growing specialized technology company eligible for the "18C chapter," the focus of financial analysis typically lies in balancing "revenue elasticity, gross profit structure, R&D efficiency" against "cash burn rate, capital structure, liquidity security." Reviewing Xingshentianhes prospectus, the companys finances present a compelling picture: on one hand, the company is expected to hit a "high point" in 2025, achieving profitability with a net profit of RMB 7.137 million for the year, and continuing to maintain slight profitability in the first half of 2026 (RMB 307,000), primarily driven by revenue surges and significant operational leverage due to the explosion in AI demand. According to the prospectus, as core underlying technologies mature, the company's R&D expense ratio sharply decreased from 65.9% in 2023 to 28.9% in 2025, further diluted to 22.3% in the first half of 2026. This is mainly attributed to the companys implementation of unified and modular R&D processes across various business lines, allowing for extensive reuse of core components, which greatly enhances each engineer's output efficiency and reduces the incremental costs of developing new solutions. However, behind this impressive achievement in turning losses into profits lies deep concerns about sharp declines in gross profit margins, ongoing cash flow losses, and potential off-balance-sheet obligations. Firstly, there is the deterioration of gross profit structure. During the performance records, Xingshentianhes comprehensive gross profit margin stabilized around 63% in 2024 and 2025, but plunged to 51.8% in the first half of 2026. The fundamental reason for this margin "slip" lies in the product mix shifting towards lower-margin integrated machines. It is understood that Xingshentianhes pure software solutions boast a gross profit margin exceeding 99%, while integrated machines, due to comprising many purchased third-party hardware components (such as SSDs, HDDs, memory, CPUs, and network cards), have a very low gross profit margin, which even dropped to 18.9% in the first half of 2026. Simultaneously, the global AI boom has triggered strong demand for high-performance storage media, leading to structural imbalances in supply and demand and continuous price increases. Xingshentianhe has been unable to fully pass cost increases onto downstream customers and has instead been forced to increase hardware procurement reserves to mitigate supply risks, squeezing its gross profit margin space. Furthermore, in the fiscal years 2023-2025 and the first half of 2026, Xingshentianhes operating cash flows were -RMB 153 million, -RMB 39.26 million, -RMB 25.46 million, and -RMB 93.48 million, respectively, with the operational activities never generating positive cash flows. As of June 30, 2026, the companys cash and cash equivalents had dwindled to only RMB 26.335 million, putting it in a severely strained financial position. To maintain operations and inventory reserves, the company had to incur an additional RMB 28.019 million in bank loans in the first half of 2026. In terms of competition, although Xingshentianhe captured 10.6% market share to become "China's largest independent distributed AI storage provider" and "the second-largest distributed AI storage provider" based on projected installed capacity in 2025, its revenue share in the overall data storage market is only 0.1%, and merely 2.7% in the distributed AI storage market. This indicates a gap in market influence and bargaining power for hardware procurement compared to non-independent giants that have server production lines and vast customer networks. Additionally, the prospectus reveals that the concentration risk among clients is high; the revenue from the top five clients accounted for an astounding 55.5% in the first half of 2026, with the largest client making up 22.1%, most of which are distributors. Any fluctuations in the stability of distribution channels or the procurement cycles of major end projects in the future could directly trigger serious shocks to the companys revenue. On this trip to Hong Kong, Xingshentianhe's plan is to use the net proceeds to enhance the research and development of the next-generation AI-native storage platform, tailor solutions for vertical industries, expand ecosystem compatibility (joint certification with leading GPU/CPU suppliers), and build local sales and demonstration centers in major economic hubs. Overall, Xingshentianhe is portrayed as a hardcore specialized technology entity racing in the gold track of AI, with robust technological barriers and high customer stickiness, yet simultaneously grappling with cash flow pressures, restructuring gross profit dynamics, and competitive assaults from giants. Its post-listing performance will be an important litmus test for the true valuation anchor of China National Software & Service-defined storage in the AI era.