The explosion in AI computing power is driving exponential growth in demand for high-end memory chips, while a simultaneous concentration of production capacity on high-end products like HBM is exacerbating structural supply shortages. This has led the global memory chip market into a "bull market" propelled by a supply-demand imbalance. During this period of industry opportunity, many companies in the storage supply chain are seeking to grow and strengthen by leveraging capital markets. This includes companies that have achieved dual listings in A-shares and H-shares, such as Gigadevice Semiconductor Inc. (603986.SH), A-share listed companies that have submitted applications to the Hong Kong Stock Exchange, such as Montage Technology Co.,Ltd. (688008.SH), Shenzhen Longsys Electronics Co.,Ltd. (301308.SZ), and Biwin Storage Technology Co.,Ltd. (688525.SH), as well as others like LiJi Storage, Xintianxia, and Hongxinyu Electronics that have also applied for initial listings in Hong Kong. On January 27, Beijing XSKY Data Technology Co., Ltd. (referred to as "XSKY"), which specializes in providing enterprise-level AI storage solutions, submitted its listing application to the main board of the Hong Kong Stock Exchange, with Huatai International acting as its sole sponsor. According to a report by China Insights Consultancy, based on installed capacity in 2024, XSKY held a 10.4% share in China's distributed AI storage solutions market, making it the second-largest provider and the largest independent provider in this sector. In terms of financial performance, XSKY reported revenues of 167 million (Renminbi, same below) and 172 million yuan in 2023 and 2024, respectively, with losses of 181 million and 84.18 million yuan during the same periods. For the first three quarters of 2025, XSKY's revenue reached 195 million yuan, a year-on-year increase of 65.4%, and it achieved a net profit of 8.11 million yuan, turning a profit compared to a loss of 65.506 million yuan in the same period of 2024. Given its continuously improving performance and the high market enthusiasm for the storage industry, will XSKY's Hong Kong IPO attract strong investor interest? A look at the company's prospectus provides some clues.
Having undergone eight rounds of financing since its establishment in 2015, XSKY has developed over a decade from a storage technology provider into a leading player in China's distributed AI storage solutions market. Its development can be divided into three distinct stages, each accurately capturing technological trends and continuously building its core competitiveness. The first stage, from 2015 to 2017, saw the company initially enter the private cloud and virtualization market, providing stable and reliable cloud infrastructure for clients with massive data storage needs, successfully establishing a foothold in the market. The second stage, spanning from 2017 to July 2022, involved XSKY proposing the architectural concept of "one data center, one storage" and independently developing its Data Operating System (DataOS) on this foundation, enhancing its underlying support capabilities for diverse computing chips. This move enabled it to offer clients a unified, intelligent data platform, completing a strategic upgrade from storage products to a data platform. The third stage, from July 2022 to the present, marks XSKY's full transformation towards AI, focusing on building end-to-end AI data infrastructure. Its self-developed XGFS (high-performance file storage) and XEOS (object storage) have been deeply integrated into AI training and inference workflows. The company's AIMesh solution further breaks down data protocol barriers, enabling zero-copy data flow from data lakes to training grounds, significantly improving the overall efficiency of AI data processing. After a decade of continuous accumulation and strategic evolution, XSKY, with its excellent products and market performance, ranked second in China's distributed AI storage solutions market by installed capacity in 2024, firmly positioning it in the industry's top tier. This growth trajectory stems from both its keen insight into technological trends and the sustained support of capital. According to the prospectus, prior to 2023, XSKY conducted a total of eight financing rounds, with the last being an F+ round in December 2022, which valued the company at 4.58 billion yuan. After these eight rounds, XSKY's current shareholder structure presents a typical model led by founders, driven by professional investment institutions, empowered by industrial capital, and supported by state-owned capital. The founders and parties acting in concert form the single largest shareholder group, holding a combined 25.72%. Professional investment institutions include Boyu Investment, Chuo Jun (representing Northern Light Venture Capital), QM188 Limited (representing Qiming Venture Partners), Junlian Shengyuan (representing Legend Capital), and China Gold Jiazi (representing China International Capital Corporation), holding stakes of 9.62%, 7.18%, 4.92%, 3.85%, and 3.85%, respectively. Industrial capital includes Tencent Guangxi (under Tencent), Nanjing Xingcheng (representing Hundsun Technologies), and Beijing Kunlun (representing Kunlun Wanwei), holding stakes of 5.64%, 0.77%, and 1.15%, respectively. State-owned background shareholders include Shenzhen Jianwei (representing China Renaissance Capital) and Shanghai Guoxin, holding stakes of 7.5% and 1%, respectively. This indicates a notably strong shareholder base for XSKY, reflecting the confidence various capital entities have in its future development. However, it is noteworthy that XSKY's shareholding structure is relatively fragmented, and once lock-up shares are released post-listing, it could potentially exert significant pressure on the company's stock price.
According to the prospectus, XSKY primarily offers two core services to enterprises: AI storage solutions and accompanying AI storage services. The solutions contribute the majority of revenue, accounting for 80% in 2024, while storage services accounted for 20%. Specifically, XSKY's AI storage solutions aim to build a unified data foundation, comprehensively covering the AI application lifecycle from data preparation to model training and inference. These solutions consist of two key components: AI Data Lake Storage and AI Training/Inference Storage, which accounted for 40.7% and 39.3% of revenue in 2024, respectively. The AI Data Lake Storage is responsible for integrating, cleaning, and centrally managing massive amounts of raw unstructured data (such as images and videos), serving as the starting point of the AI data lifecycle. The AI Training/Inference Storage, on the other hand, caters to the model training and inference stages, providing high-throughput, low-latency data supply for high-performance computing resources like GPU clusters, aiming to improve GPU utilization and reduce inference costs. The two work synergistically to ensure efficient data flow within enterprises. The AI storage services provided by XSKY refer to post-deployment support, including end-of-warranty maintenance and renewal services, to ensure the long-term stable operation of client systems. Looking at the financial performance, the significant narrowing of XSKY's net loss in 2024 can be attributed to three main factors. The first is steady revenue growth. Benefiting from the ongoing release of downstream AI transformation demand, the number of orders for its AI Data Lake Storage solutions increased from 257 to 364 in 2024, driving revenue for this business up 13.02% year-on-year to 70.141 million yuan. Coupled with a simultaneous moderate increase in AI storage service revenue, the company's total annual revenue increased by 3.38% to 172 million yuan. The second factor is a marked improvement in gross margin. The overall gross margin reached 63.7% in 2024, an increase of over 8 percentage points from 55.4% in 2023. This was primarily due to margin improvements in the AI storage solutions resulting from product mix optimization, as well as an increase in the gross margin of AI storage services as scale effects materialized. The third factor is effective control of operating expenses. Under the company's strategy of cost reduction and efficiency improvement, the ratio of total operating expenses to revenue dropped to 119.79% in 2024, a significant optimization from 174% in 2023. This effective cost control directly accelerated profit realization, providing key support for the loss reduction. In the first three quarters of 2025, although XSKY's overall gross margin declined by 6 percentage points to 63.7% due to the company adopting competitive pricing to maintain sales volume, it achieved a turnaround to profitability thanks to a substantial increase in revenue and a continued decrease in total operating expenses. The significant revenue growth was driven by the accelerated AI transformation in downstream industries, spurring a surge in demand for advanced storage solutions. Orders for AI Data Lake Storage solutions increased from 255 to 371, and orders for AI Training/Inference Storage solutions rose from 201 to 294 during the period, leading to revenue growth of 106.8% and 54.5% for the two businesses, respectively. Combined with a 14% growth in AI storage service revenue, these three drivers propelled XSKY's revenue for the first three quarters of 2025 to increase by 65.4%. Furthermore, the ratio of total operating expenses to revenue decreased from 130.99% in the first three quarters of 2024 to 65.52% in the same period of 2025. This "halving" of the operating expense ratio acted as a catalyst for the accelerated release of profits. Therefore, the sustained improvement in XSKY's performance results from the synergistic effect of revenue growth, gross margin improvement, and expense control, demonstrating a clear trajectory of financial improvement and enhanced operational efficiency.
In fact, the substantial revenue growth of XSKY in the first three quarters of 2025 suggests it is already benefiting from the "bull market" in memory chips. Reviewing this round of price increases, the memory market began to bottom out and recover at the end of 2024, setting the stage for the current cycle. By April 2025, the industry saw its first round of tentative price hikes. Merely five months later, the price surge entered a second round of large-scale increases, and by the fourth quarter, tight supply-demand conditions again pushed prices upward rapidly. For example, the unit price of a 256GB DDR5 server memory module for data centers has exceeded 40,000 yuan, while the price of a consumer-grade 16GB DDR5 laptop memory module surged from around 380 yuan to nearly 1,400 yuan within just a few months. Research institutions generally predict that memory chip contract prices will continue to rise rapidly in the first quarter of 2026, with DRAM prices potentially increasing by 55%-60% quarter-on-quarter, and NAND Flash prices likely rising by 33%-38%. From a longer-term perspective, the overall supply-demand imbalance in the market is expected to persist at least until the end of 2026, with UBS believing the DRAM supply shortage could extend into the first quarter of 2027. For high-end HBM, due to extremely high technical barriers and slow capacity expansion, the strong market conditions are expected to continue until 2028. The CEO of Micron Technology also anticipates that the tight supply of HBM will persist until 2028. This memory chip bull market is poised to benefit XSKY. The underlying logic is that XSKY does not manufacture hardware chips but provides software-defined distributed AI storage solutions. Its core value lies in its intelligent Data Operating System (DataOS), which helps clients manage and utilize existing, heterogeneous storage hardware resources more efficiently and flexibly, thereby improving storage utilization and performance. When the capital expenditure (CAPEX) for clients to procure new hardware storage rises sharply due to chip price increases, purchasing XSKY's solutions to optimize the efficiency of existing storage infrastructure, extend its lifecycle, and reduce the total cost of ownership (TCO) becomes an highly attractive alternative or complementary strategy. This essentially creates a strong demand for "making existing hardware work smarter" versus "buying more expensive hardware," from which XSKY stands to benefit. However, the potential risks and challenges facing XSKY are also quite evident. The first is intense market competition. Although XSKY held the second position in China's distributed AI storage solutions market with a 10.4% share based on 2024 installed capacity, the top player is Huawei, with a 14.7% share, more than 4 percentage points higher than XSKY. The third and fourth positions are held by Inspur Information and New H3C Technologies, with market shares of 9.3% and 9.2%, respectively, both relatively close to XSKY's share. XSKY competes directly with comprehensive tech giants like Huawei, Inspur Information, and New H3C Technologies, which typically offer full-stack solutions from hardware to software and possess significant advantages in underlying hardware synergy optimization and brand channels. Furthermore, public cloud service providers like Alibaba Cloud are also key competitors, as their cloud storage services pose a substitution threat to on-premise deployment solutions. As an independent software vendor, XSKY needs to continuously demonstrate the unique value of its solutions compared to the ecosystems of these giants. Secondly, amidst this fierce competition with giants, XSKY's R&D expenditure has been consistently declining, which is not conducive to the company's long-term stable development. According to the prospectus, R&D expenses for 2023, 2024, and the first three quarters of 2025 were 110 million yuan, 87.959 million yuan, and 57.461 million yuan, accounting for 65.9%, 51%, and 29.5% of total revenue, respectively. Not only has the absolute value of R&D spending been decreasing, but its proportion of revenue also shows a clear downward trend. While shifting towards a "profit and efficiency-oriented" lean operation model during an IPO push is a common strategy to optimize financial statements, it also implies a slowdown and potential discontinuity in technological investment, which is detrimental to building technological competitiveness. If R&D spending is not increased subsequently, the company risks gradually losing its product competitiveness. Additionally, the AI storage solutions XSKY provides to clients are delivered primarily in two forms: integrated appliances and pure software. The integrated appliances require the procurement of upstream storage hardware. If the proportion of client purchases of integrated appliances increases significantly and XSKY cannot effectively pass the upstream hardware costs downstream to clients, it could squeeze the company's gross margin, thereby creating substantial cost control pressures.
Comments