DapuStor, a Chinese maker of enterprise solid-state drives and data-center storage products, plans to seek a Hong Kong listing after reporting a sharp first-half recovery. Its board has authorized a proposed Main Board float in Hong Kong through H shares, while leaving the offering’s timing and size for later approvals and filings.
Figures cited by TechNode put DapuStor’s first-half turnover at 4.72 billion yuan, a 531.17% year-on-year increase. The company recorded 1.33 billion yuan in attributable profit after reporting a 353.8 million yuan loss in the comparable 2025 period. The figures make the planned listing an AI-infrastructure story, although DapuStor is not an AI-model company and does not sell storage only to AI customers.
The relevance is storage. AI training and inference systems create, move, and retain large amounts of information, from source data and model weights to checkpoints, logs, and generated outputs. GPUs receive the attention, but a data center’s ability to store and retrieve information efficiently can influence the usefulness of the whole system.
Enterprise SSD Demand Tracks the Growth of Data-Center Workloads
DapuStor develops enterprise SSDs and related products for data-center workloads. An enterprise drive is not simply a consumer storage device placed in a server. Large customers need predictable performance, endurance, security features, and compatibility with systems that run continuously. As cloud providers, AI developers, and businesses expand computing environments, they also need more reliable storage capacity.
The company’s H1 results show a dramatic rebound, but they do not by themselves identify the exact share of revenue tied to AI. The distinction is familiar across the hardware chain: an infrastructure supplier can benefit from AI demand without depending on a single model developer or data-center customer. The appropriate conclusion is that DapuStor operates in a layer of infrastructure that becomes more valuable as data-center workloads grow. AI is one driver of that growth, alongside cloud services, enterprise computing, databases, and other digital workloads.
The planned listing could give DapuStor another route to international capital markets. MarketScreener also reported the revenue growth and return to profit. But the company has not disclosed an offering size, valuation, timetable, or final approval. Those details should not be assumed from the board’s decision.
Storage Is the Quiet Layer of China’s AI Infrastructure Buildout
AI infrastructure is often framed as a race for chips. That is incomplete. A useful cluster needs processors, networking, power, cooling, servers, and storage. Each layer can become a constraint. Training a large model may require repeated reads of data, frequent checkpoints, and storage for intermediate results. Serving a model can require fast access to knowledge bases, user files, and logs.
This is why storage suppliers matter to the wider AI economy. Their products may not determine a model’s reasoning ability, but they help determine whether a data center can operate efficiently at scale. EastFrontier’s report on Mucang’s AI-cluster interconnect financing described the networking layer that connects accelerators. DapuStor represents another part of the same picture: persistent data capacity.
China’s push for more domestic technology capacity adds another dimension. Local AI infrastructure is not only about finding enough accelerators. It also involves developing component suppliers that can support server and data-center systems. A company with enterprise storage products can benefit if customers prefer a more domestic supply chain, although DapuStor has not presented its listing plan as a direct response to a government mandate.
A Strong Half-Year Does Not Set the IPO Terms
The revenue and profit change gives DapuStor a stronger financial narrative than it had a year earlier. A 531.17% revenue increase and a swing from loss to profit are notable figures. But an IPO depends on more than a single reporting period. Investors will examine whether the growth can continue, how exposed the company is to memory-price cycles, what its customer concentration looks like, and how much of its demand is tied to volatile data-center investment.
The board decision is an early corporate step rather than a completed market transaction. TechNode noted that the company must still obtain further approvals and complete the relevant filings. That means the company has stated its intention but has not yet provided the details needed to evaluate a transaction.
DapuStor’s story is therefore best read as a marker of infrastructure demand rather than a pure AI boom signal. Its role sits beside the domestic accelerator and data-center expansion described in EastFrontier’s report on state-funded AI chip procurement, where a usable domestic stack depends on many components beyond processors. The company has reported a strong half, plans to seek Hong Kong capital, and makes products that support data-center workloads.
The same expansion in storage, networking, and compute also means purchasers will increasingly assess whole-system resilience, from the availability of drives to the ability to recover data when a large training or inference job is interrupted. Its performance shows how the AI buildout can reach beyond model labs and chip designers into the suppliers that store the information an AI economy depends on.