Introduction: How to Navigate Industry Cycles?
No one expected that in this round of the storage industry's upswing, the company that first ignited market sentiment in the A-share market would be a storage module manufacturer.
In early July, Longsys (Shenzhen Longsys Electronics Co., Ltd.) released its semi-annual performance forecast for 2026. The company expects to achieve an operating revenue of 22 billion to 25 billion RMB in the first half of 2026, a year-on-year increase of 115.8% to 145.2%; the net profit attributable to shareholders of the listed company is expected to be 9.2 billion to 11 billion RMB, a year-on-year increase of 62204.03% to 74393.95% (a staggering growth); the net profit after deducting non-recurring gains and losses is expected to be 9 billion to 10.5 billion RMB, a year-on-year increase of 27844.32% to 32501.71%. This set of figures almost left the A-share semiconductor sector stunned.
What is even more striking is that Longsys is not a storage wafer original manufacturer with higher profit margins like Micron, SK hynix, or Samsung, but a downstream storage module manufacturer. Amid this shift, Longsys is undergoing an identity transition: moving from a leading storage module manufacturer closer to a "storage platform company in the AI era."
Core Data of Longsys's 2026 Semi-Annual Performance Forecast. Source: Compiled by eefocus Research Institute, Longsys 2026 Semi-Annual Report
The Storage Cycle Provides Longsys with a "Super Leverage"
The most easily misunderstood aspect of Longsys's performance forecast is the growth rate.
The year-on-year growth of net profit attributable to shareholders exceeds 600 times, an astonishingly massive increase. However, Longsys's net profit in the first half of 2025 was only 14.766 million RMB. With such a low base, once the industry enters an upswing cycle, the profit growth rate is amplified to an extremely exaggerated level.
Crucially, the storage industry has entered a highly prosperous phase since the second half of 2025. After years of inventory digestion, production cuts, and price declines, the supply-demand relationship for DRAM and NAND has begun to reverse. Downstream demand from AI servers, AI PCs, AI smartphones, smart vehicles, AI glasses, and industrial equipment has surged simultaneously, while upstream storage wafer capacity expansion remains relatively restrained, naturally driving prices to rise again.
Longsys also stated the reasons directly in its performance forecast: in the first half of 2026, driven by increased downstream demand and the limited overall growth of global storage wafer capacity, the global semiconductor storage industry has prospered; meanwhile, the company has successfully renewed wafer supply agreements with several major global storage wafer original manufacturers, ensuring the supply of storage wafers.
The business model of module manufacturers essentially involves purchasing storage wafers or dies, and forming products such as SSDs, UFS, eMMC, memory modules, and memory cards through master controllers, firmware, packaging and testing, module design, and brand channels. When storage prices rise, if a company has prior inventory, supply agreements, and customer orders, it can simultaneously reap the triple benefits of price increases, inventory revaluation, and shipment volume expansion.
This is also the first layer of logic behind Longsys's performance explosion in this round: the reversal of the storage cycle amplifies the profit elasticity of module manufacturers.
Four Layers of Logic Behind Longsys's Performance Explosion. Source: Compiled by eefocus Research Institute, Longsys 2026 Semi-Annual Report
Peers Are All Growing, but Why Is Longsys's Growth the Most Exaggerated?
As another leading domestic storage module manufacturer, Biwin Storage expects a year-on-year increase of 550% to 620% in net profit attributable to shareholders in the first half of 2026; DMET Memory Solutions has not yet released its official performance forecast for the first half of 2026, but the market expects a substantial year-on-year increase. In contrast, Longsys's net profit has achieved a growth of hundreds of times, far exceeding its peers, with one key reason being the low base in the same period last year.
However, explaining Longsys solely by its low base is incomplete.
From an industrial logic perspective, the elasticity of different storage companies in this cycle varies.
Wafer original manufacturers hold the most core capacity and pricing power, benefiting most directly from the price increases of DRAM and NAND; module manufacturers, if they possess low-cost inventory, long-term supply agreements, strong channels, and high-end customers, will also gain strong profit elasticity; design companies and interface chip companies benefit more from product upgrades and shipment growth, with potentially more stable performance, but they may not exhibit the extreme elasticity seen in module manufacturers during a price upswing cycle.
What makes Longsys special is that it simultaneously occupies several positions: on the one hand, as a module manufacturer, it can directly enjoy the price upswing cycle; on the other hand, it is strengthening its master controller chips, firmware algorithms, high-end packaging and testing, and AI storage solutions, attempting to increase added value; at the same time, it also has global consumer brands such as Lexar, as well as a brand matrix including FORESEE and Zilia targeting industrial and overseas markets.
Therefore, Longsys's explosion in this round is not due to a single factor, but the result of the superposition of cycles, inventory, supply chain, brand, product structure, and AI narratives.
Comparison of Performance Elasticity and Industrial Position of Major Storage-Related Enterprises. Source: Compiled by eefocus Research Institute, Longsys 2026 Semi-Annual Report
What Longsys Most Wants to Shake Off Is the Term "Module Manufacturer"
In the past, the outside world often viewed Longsys within the framework of a "storage module manufacturer."
The business logic of traditional module manufacturers is simple: buy dies, make modules, and sell products. When upstream prices rise, product prices rise; when upstream prices fall, inventory depreciates. The core capabilities of the enterprise are procurement, inventory management, channels, and branding. However, this business has an inherent problem: it is too cyclical, has limited bargaining power, and is easily understood by the market with a lower valuation.
Longsys is clearly unwilling to stay only at this position now. It is trying to push itself from a "storage product supplier" towards a "storage technology platform provider."
In the AI era, especially the edge AI era, storage is becoming more complex. Terminal devices such as AI PCs, AI smartphones, AI glasses, robots, and local AI NAS require not only larger capacity but also faster data retrieval, lower power consumption, better heat dissipation, higher stability, and deeper synergy with system software, AI models, and processor platforms. This means that storage is no longer just a "plug-and-play" standard component, but is beginning to enter system-level design.
Whoever can participate in system-level design has the opportunity to transform from a hardware accessory into a platform capability.
What Longsys is doing now is precisely moving up along this path. According to the annual report, Longsys emphasizes supporting the diverse and comprehensive needs of edge AI storage systematically with self-developed chips, such as SPU master controller chips; self-developed software architectures, such as HLC; and its own high-end packaging and testing capacity.
HLC can be understood as advanced caching technology. Through the synergy of master controller chips, firmware algorithms, and system architecture, Longsys enables SSD or UFS storage devices to take over part of the warm and cold data caching work originally handled by DRAM, thereby achieving the extension of NAND to DRAM.
This is crucial for AI terminals. Products like AI PCs, AI smartphones, and AI glasses cannot stack memory infinitely like servers. Terminal devices have constraints on cost, power consumption, size, and heat dissipation. The larger the model, the higher the requirements for memory capacity and data retrieval capabilities. If DRAM is simply increased, the overall cost of the device will rise rapidly; however, if the pressure on DRAM is partially alleviated through the synergy of SSDs, UFS, master controllers, and firmware algorithms, it could open up new design space for AI terminals. The annual report mentions that Longsys has completed joint optimization with AMD, enabling its SSD storage agent and HLC technology to support a reduction of about 40% in DRAM usage for edge AI products.
The story Longsys is telling is no longer "I can sell more SSDs," but "I can help AI terminals reduce DRAM usage."
In the AI Era, Storage Is Shifting from a Supporting Role to a Bottleneck
As AI moves from the cloud to the edge, storage has gradually become a critical link in the supply chain.
AI PCs require local models, local knowledge bases, file indexing, image retrieval, meeting minutes, and offline inference; AI smartphones need stronger edge inference and multimodal capabilities; AI glasses are even more extreme, requiring both thin and light designs and long battery life, as well as real-time shooting, voice interaction, image recognition, and local caching; AI NAS, local servers, and robots also cannot bypass storage.
It is understood that Longsys has launched mSSD for AI PCs, achieving chip-level packaging quality for SSDs through SiP (System in Package) packaging, and introducing VC (Vapor Chamber) phase-change liquid cooling to address the heat dissipation issues of small-sized, high-performance SSDs; in the direction of AI smart wearables, the company's ultra-small-sized eMMC and ePOP products have been applied in smart glasses and smart watches from multiple leading domestic and international manufacturers, and the new-generation ePOP5x, targeting AI smart wearables and thin-and-light terminals, can be as thin as 0.52mm; in the direction of AI servers and data centers, the company has launched DDR5 RDIMM products with a maximum capacity of 256GB, and is deploying memory expansion modules such as MRDIMM, LPCAMM2, CAMM2, and CXL 2.0.
These deployments indicate that Longsys is no longer satisfied with the consumer storage market, but is extending towards AI PCs, AI wearables, data centers, and enterprise markets.
The common characteristics of these markets are: higher technical barriers, longer customer validation cycles, higher product added value, and a greater likelihood of helping Longsys break away from the price war logic of traditional consumer storage.
Breakdown of Longsys's AI Storage Deployment. Source: Compiled by eefocus Research Institute, Longsys 2026 Semi-Annual Report
Longsys and Its Peers Reaping Massive Profits, While Downstream Hardware Manufacturers Struggle
The rise in storage prices is a bonus for Longsys, but for downstream hardware manufacturers, it could be a survival-related cost pressure test.
Old Li (pseudonym), a practitioner in the AI NAS industry, told the author that the recent surge in storage prices has directly disrupted the pace of product planning. In the early stages of product development, the team would reverse-engineer product positioning based on BOM (Bill of Materials) costs, projected sales volume, channel prices, and gross profit margins. For a router, NAS, or consumer hardware with storage attributes, if the shipment volume reaches a certain scale, retaining a gross profit of tens of RMB per unit might be enough to support the team's operations. However, when key materials such as DDR, Flash, SSDs, and PCBs (Printed Circuit Boards) rise in price simultaneously, the originally calculated gross profit is quickly eroded.
What makes it worse is that this round of price increases is not mild or linear, but characterized by leapfrog changes. Old Li described the frontline procurement experience as "memory prices change every day" and "procurement staff don't even know when prices have gone up." For large companies, they can lock in orders in advance, stock up early, and buffer costs through long-term agreements and supplier relationships; but for small teams, a delay of a few days in procurement could render the previously agreed-upon prices invalid.
This means that the rise in storage prices changes not only BOM costs but also the pace of hardware entrepreneurship.
In the past, consumer electronics teams might have had the time to slowly modify IDs, adjust structures, optimize molds, and calculate selling prices; but in a storage price hike cycle, hesitation itself is a cost. The BOM calculated today may be invalid next month; the price quoted to a customer today may see another round of upstream price increases before the contract is even signed.
This is also why consumer products are becoming harder to make. Just last month, Apple and its supplier Micron engaged in a public dispute online. If even a global top-tier terminal client like Apple can barely bear the pressure of rising storage costs, let alone other terminal clients.
Therefore, while the rise in storage prices allows companies like Longsys to reap cyclical bonuses, it is also re-screening downstream hardware teams. Only truly capable clients can navigate through this price hike cycle.
How to Navigate the Cycle?
Ultimately, the storage industry is a cyclical and fluctuating one. When prices rise, profits are rapidly amplified; when prices fall, risks are quickly exposed.
The annual report mentions that Longsys's net cash flow generated from operating activities in 2025 was still -1.201 billion RMB, and it also accrued 314 million RMB in inventory depreciation losses in 2025. This shows that for storage companies, the income statement and cash flow are not always synchronized, and inventory management remains a core risk.
Longsys currently benefits from recovering demand, rising prices, guaranteed wafer supply, and product structure upgrades. However, if upstream original manufacturers expand capacity in the future, downstream demand slows down, AI hardware shipments fall short of expectations, or storage prices enter a down cycle, the profit elasticity of module manufacturers could also be released in reverse.
Longsys is now standing in a very subtle position.
Looking to the left, it remains a company deeply affected by the storage cycle. When the industry prospers, its profits can be rapidly amplified; when the industry declines, it will also face inventory, price, and cash flow pressures.
Looking to the right, it is indeed trying to transform itself into a storage platform company in the AI era.
Looking back at the previous down cycle, which started in mid-2022, bottomed out in 2023, and finally began to recover in 2024 driven by the AI wave, we can clearly see a series of textbook response measures taken by major global storage manufacturers (Samsung, SK hynix, Micron) for survival and recovery.
The market recovery since 2024 eloquently proves the decisive role of high-value-added products on corporate profitability. While the market prices for general-purpose DRAM and NAND were sluggish, HBM memory for AI servers was in short supply, with prices remaining consistently high. The fierce competition between SK hynix and Samsung around HBM3, HBM3E, and even the next-generation HBM4, as well as the accelerated penetration of DDR5 in the server and high-end PC markets, all demonstrate the importance of high-end product structures. Manufacturers must prioritize resource allocation to these product lines that can bring higher profit margins and stronger customer stickiness.
At the bottom of the previous storage industry cycle, Longsys's acquisition of Micron's consumer brand Lexar was equivalent to buying at the bottom, so in this upswing cycle of the storage industry, Longsys has reaped substantial returns. During a down cycle, manufacturers should proactively shift limited capacity from low-profit traditional products (such as DDR4) to high-demand new products (such as DDR5 and HBM). This not only improves profitability but also accelerates the market adoption of new technologies, forming a positive cycle.
All signs indicate that Longsys is utilizing the profits and market attention brought by this round of price hikes to accelerate its transformation. It wants to become a platform company that transcends industry fluctuation cycles.