Shanghai Securities: Enterprise-level storage is expected to maintain high prosperity, and domestic enterprise-level SSD-related manufacturers are likely to benefit

According to the Zhitong Finance APP, Shanghai Securities released a research report stating that IDC expects the market size of China's enterprise-level solid-state drive (SSD) market to reach USD 9.1 billion by 2029. With the increase in enterprise-level SSD capacity and the decrease in unit costs, the replacement of HDDs is expected to accelerate. Policies are promoting an increase in the localization rate, and domestic storage companies are narrowing the gap with international brands through technological innovations such as high-capacity products and main control chips, as well as localization of the industrial chain and services. The overall market share is expected to increase. The firm believes that enterprise-level storage is likely to maintain high prosperity, and relevant manufacturers that are laying out enterprise-level SSDs in China are expected to benefit.

The main points of Shanghai Securities are as follows:

Enterprise-level SSDs are applied in high-performance computing, data centers, and other scenarios

According to the Forward Industry Research Institute, enterprise-level solid-state drives (eSSD) refer to solid-state drives used in various enterprise-level scenarios such as high-performance computing, edge computing, high-end storage, and data centers. They have the capability for continuous operation and can handle I/O-intensive workloads, such as database files, index logs, data analysis, and other transaction processing operations that require high performance. Compared to consumer-grade SSDs (mainly used in consumer electronics scenarios such as computers, mobile phones, and external hard drives), enterprise-level SSDs target enterprise users and require guaranteed data security, speed stability, and long-term durability, with stricter requirements for high reliability in product delivery, offering stronger performance, higher reliability, and greater durability. IDC data shows that the market size of China's enterprise-level solid-state drives will reach USD 6.25 billion in 2024, a year-on-year increase of 187.9%. IDC expects the market size of China's enterprise-level solid-state drives to reach USD 9.1 billion by 2029.

HDD faces a supply-demand gap, and enterprise-level SSDs are expected to accelerate the replacement of HDDs

According to the PwC Youce WeChat public account, compared to HDDs, SSDs have significantly superior read and write speeds due to the elimination of mechanical components, completely removing rotational and seek delays, especially in random read and write rates, which have improved by an order of magnitude. Enterprise-level QLC SSDs, due to their large capacity and low energy consumption advantages, are currently the mainstream replacement for enterprise-level HDDs. HDDs still have cost-effectiveness in conventional large-scale cold data or warm data storage scenarios, and their shipment volume still accounts for a large portion. However, major global HDD manufacturers have not planned to expand production lines in recent years, failing to meet the sudden and massive storage demands stimulated by AI in a timely manner. According to TrendForce, the delivery time for NL HDDs has sharply extended from a few weeks to over 52 weeks, accelerating the storage gap for CSPs. According to the CFM flash memory market, due to the tight supply of HDDs for data centers in the North American market, some North American customers have shifted their procurement needs to enterprise-level storage. Additionally, with the increase in enterprise-level SSD capacity and the decrease in unit costs, the replacement of HDDs is expected to accelerate.

The value of eSSD in AI servers is higher than that in traditional servers, and SSD capacity increases during the AI inference phase

According to the PwC Youce WeChat public account, the large models in the AI era have higher storage requirements for data transmission and computation during training and inference processes. The stable and rapid growth of the AI server market will generate downstream market demand for large-capacity QLC SSDs, driving the growth of enterprise-level SSDs The value of enterprise-level SSDs in a single AI server is generally more than three times that of general servers. According to IDC's forecast, the global AI server market size will reach USD 34.71 billion by 2026, with a compound annual growth rate of 17.3% from 2021 to 2026. Additionally, the capacity of SSDs during the AI inference phase is expected to increase compared to the training phase. According to the WeChat public account of Zhiyan Consulting, during the training phase, SSDs are responsible not only for storing model parameters, including continuously updated weights and biases, but also for creating checkpoints to regularly save the AI model training progress, ensuring that even if training is interrupted, it can be resumed from a specific point. These functions heavily rely on the high-speed transmission and write durability of SSDs. Therefore, customers primarily choose to use 4TB and 8TB TLC (Triple-Level Cell) SSD products to meet the demanding requirements of the training process. In the AI inference phase, as more generated information is displayed in video or image form, the data storage volume also increases accordingly. Therefore, large-capacity SSDs such as TLC/QLC (Quad-Level Cell) with over 16TB have become the main products used in AI inference.

Industry Landscape: Solidigm and Samsung are global leaders, with significant room for domestic manufacturers to increase their market share

IDC data shows that from the perspective of vendor sales revenue, Solidigm and Samsung are the global leaders, holding the largest share of the Chinese enterprise-level solid-state drive market. Samsung maintains its global leadership position due to its full-stack technological capabilities and high-end products, but faces domestic substitution and cost pressures in the Chinese market. Solidigm focuses on AI-optimized storage and large-capacity products as its core, rapidly rising with the support of the SK Hynix ecosystem. At the same time, policies are promoting an increase in domestic production rates, and local storage companies are narrowing the gap with international brands through technological innovations in large-capacity products and main control chips, as well as localization of the industrial chain and services, with an overall increase in market share expected