2026 Memory Industry Insights
Entering 2026, the global storage industry has witnessed a long-awaited robust recovery. The latest performance of Samsung and SK Hynix has directly boosted market enthusiasm, while the tight supply-demand pattern, iteration of new technologies, and transformation of the industry's business model have made this track full of highlights. This article discusses the core changes and future trends of the storage industry in 2026.
I. Explosive Performance of Samsung & SK Hynix, Full Growth Momentum
First, let’s look at the most intuitive performance data. Both Samsung and SK Hynix achieved significant performance breakthroughs in Q4 2025. Samsung’s memory division posted an operating profit of 17 trillion Korean won, a year-on-year surge of 250%, with the average selling price (ASP) rising by over 40%, far exceeding market expectations; SK Hynix also delivered a strong performance in the same period, with revenue of 30.7 trillion Korean won, operating profit of 17.1 trillion Korean won, and an operating profit margin of nearly 56%. Among them, high-bandwidth memory (HBM) and high-density DDR5 products constituted the core growth drivers.
Behind the outstanding performance of the two companies is the support of the tight global storage market supply-demand pattern. Currently, global DRAM inventory is only 2-3 weeks, and NAND inventory is 3-4 weeks, both at historically low levels. This tight supply-demand situation is expected to last for a year, and suppliers will take a dominant position in price negotiations in the first half of 2026. On the demand side, in addition to stable server demand, the consumer electronics sector (mobile terminals, PCs) is expected to start inventory replenishment in Q2 2026, which will further consolidate the foundation for industry growth.
II. Clear Price Uptrend
The price trend is a core issue of concern to industry practitioners and investors. From the current industry forecast, the upward trend of global storage product prices in 2026 is clear.
In the DRAM field, prices are expected to rise by 30% in Q1 2026 and maintain high single-digit growth in Q2, driven by low inventory levels and sustained strong server demand. The NAND sector has outperformed market expectations, with price increases expected to be no less than 20% in Q1, significantly higher than the previous market estimate of 10%, and will also maintain high single-digit growth in Q2, with strong growth certainty under the tight supply-demand pattern. Regarding the price trend in the second half of the year that the market cares about, it is clear that the price growth rate may slow down in the second half, but will still maintain a mid-to-high single-digit positive growth, among which NAND has relatively higher growth certainty.
It is worth focusing on that NVIDIA’s launch of the Rubin platform at CES 2026 has significantly boosted the market demand for AI SSDs, while driving the market to raise expectations for the NAND industry. It should be noted that the PSLC (Pseudo Single-Level Cell) mode adopted by AI SSDs can improve product speed and durability, but it will lead to a reduction of 2/3 in actual storage capacity. This means that to maintain the same effective capacity, suppliers need to increase physical wafer input by 3-4 times, which will further exacerbate the structural supply constraints in the NAND industry. This trend has attracted widespread attention in the South Korean industry. Combined with the current low NAND inventory level of only 3-4 weeks, and the fact that Samsung and SK Hynix have continuously reduced NAND capital expenditures over the past 4-5 years and have not increased relevant investment in 2026, the pattern of limited NAND supply growth has been confirmed. If demand increases significantly, the NAND price increase is expected to exceed current expectations. As core suppliers of Rubin platform AI SSDs, Samsung and SK Hynix will be affected by this, and NAND prices are likely to rise further in 2026.
III. Production Capacity and Capital Expenditure
Faced with the industry recovery, both Samsung and SK Hynix have clear orientations in capital expenditure planning and production capacity allocation, with the core strategy of focusing on high-profit areas and implementing prudent capacity expansion.
From SK Hynix’s perspective, its capital expenditure in the memory field in 2025 was 28 trillion Korean won, of which 85%-90% was invested in the DRAM sector, and the investment related to HBM was 2 trillion Korean won, a year-on-year decrease of 50%, mainly due to concerns about the possible oversupply risk in 2027; the total capital expenditure in 2026 will exceed 35 trillion Korean won, the proportion of investment in the DRAM field will still maintain 90%, the HBM investment will increase to 2.5 trillion Korean won, and it plans to start the new Union Fab in South Korea in Q1 2027, with the preparation work for relevant equipment to start in Q1 2026. Samsung’s pace is similar: its capital expenditure in 2025 was 33 trillion Korean won, focusing on DRAM, and its NAND expenditure was even lower than that of SK Hynix; the total capital expenditure in 2026 will exceed 40 trillion Korean won, with growth mainly coming from DRAM and HBM, especially as the yield of HBM4 has improved significantly; the proportion of NAND capital expenditure is about 10%, the same as in 2025. In short, the capital expenditures of both manufacturers are focused on DRAM (accounting for about 90%), HBM investment will increase in 2026, Samsung’s total expenditure is higher, and both are relatively conservative in NAND expenditure.
Samsung’s pace is similar: its capital expenditure in 2025 was 33 trillion Korean won, focusing on DRAM, and its NAND expenditure was even lower than that of SK Hynix; the total capital expenditure in 2026 will exceed 40 trillion Korean won, with growth mainly coming from DRAM and HBM, especially as the yield of HBM4 has improved significantly; the proportion of NAND capital expenditure is about 10%, the same as in 2025. In terms of production capacity launch rhythm, wafer fab equipment will be intensively put into use in the first half of 2026, aiming to release production capacity as early as possible to seize the window of industry recovery.
IV. New SSD Race
In addition to price and production capacity, the technological iteration and business model transformation of the storage industry are also worthy of attention.
In the research and development of new SSDs (known as HBF in the industry), Samsung and SK Hynix have launched a “race”. Currently, Samsung is in a leading position and plans to sample new SSDs supporting PSLC mode in Q4 2026; SK Hynix’s similar products are expected to be sampled in Q1 2027 with a differentiated name, obviously in a catching-up stage. Other manufacturers such as Micron, Kioxia, and Taiwanese manufacturers are generally in a catching-up state, and it is expected that most orders will flow to Samsung. From the market feedback, customers have a very positive demand for HBF, and this technological upgrade will also drive the upstream and downstream of the industrial chain. For example, the demand for testing equipment for new SSDs will increase, and relevant equipment manufacturers are likely to benefit.
A more far-reaching change is the transformation of the industry’s business model: from the traditional “component supply” to “solution cooperation”. Simply put, Samsung and SK Hynix have begun to explore sharing risks and R&D costs with customers, somewhat similar to the wafer foundry model, customizing memory specifications according to customer needs. Many people are concerned about whether this transformation will change the company’s business model and investment cycle. At present, the industry is still in the early stage, and no specific cooperation cases have been disclosed, but SK Hynix is advancing it particularly actively, planning to announce specific cooperation details in the second half of 2026 or 2027. The core reason for this transformation is to cope with the challenge of slowing overall growth in the memory industry—except for HBM and server-related fields, growth in other market segments is not as good as before, and manufacturers need to share R&D costs and enhance competitiveness through new models.
V. What Impact Will Chinese Manufacturers Have?
As mentioned earlier, NAND supply and demand are tight, but interestingly, mainstream manufacturers are quite cautious about NAND capacity expansion and have not increased NAND capital expenditures in 2026. This raises a common question: why not expand NAND production when supply and demand are tight? There are three core reasons: first, the profit margin difference is large—the gross profit margin of DRAM can reach 70%-80%, while that of NAND is only about 25%, so DRAM is obviously more profitable; second, the current NAND capacity utilization rate has not reached 100%, and manufacturers prefer to first improve efficiency through technology migration; third, concerns about changes in the competitive landscape—the capacity expansion expectations of Chinese manufacturers (such as Yangtze Memory Technologies Co., Ltd., YMTC) may ease future supply tensions, so Korean companies are waiting to see the supply and demand changes in 2026-2027. The leading observation indicators for NAND price fluctuations mainly include three aspects: NAND inventory level (currently 3-4 weeks, at a low level), capital expenditure plans of major manufacturers, and the progress of capacity release by Chinese manufacturers.
Speaking of Chinese manufacturers, this is also one of the focuses of the industry. Many people are curious whether the expansion of Chinese storage capacity and technological development will pose a threat in the next two years. From the actual situation, the short-term impact is limited: in the DRAM field, after ChangXin Memory Technologies (CXMT) goes public in the first half of 2026, it will likely increase the supply of low-end DRAM, which has caused some concerns among South Korean institutions, but high-end demand (such as high-density DRAM needed by major U.S. cloud service providers) is still dominated by Samsung and SK Hynix, and ChangXin is upgrading from DDR4 to DDR5, with clear demand for technological transformation; in the NAND field, Chinese manufacturers account for about 10% of the market share, and their pricing strategy mainly follows the overseas market. For example, after Micron stopped quoting in September 2025, the domestic market immediately followed suit, indicating that it is still difficult to shake the dominant position of Korean companies in the short term. However, in the long run, with technological progress, the industry influence of Chinese manufacturers may gradually increase.
Overall, 2026 is a year of “recovery-driven, technology-empowered, and pattern fine-tuning” for the storage industry. Samsung and SK Hynix will still occupy a dominant position by virtue of their advantages in DRAM and HBM fields. The price increase brought by tight supply and demand, the technological race of new SSDs, and the transformation of the industry’s business model will all become key variables driving the development of the industry. For investors and practitioners, closely monitoring inventory changes, manufacturers’ capacity strategies, and the progress of technological iteration may help seize the core opportunities of this round of industry recovery.