ch
English
In the first quarter of 2026, the global memory market ushered in a super cycle, with Samsung Electronics and SK Hynix experiencing a surge in performance due to the soaring prices of memory chips. Meanwhile, the high cost of memory began to affect the terminal manufacturing business, and the AI-driven HBM research and development competition also entered a white-hot stage.
On May 15th, two Korean companies simultaneously released their financial reports, revealing a staggering increase in storage prices. Samsung Electronics' financial report disclosed that the average selling price of mixed DRAM and NAND in Q1 2026 surged by 146% compared to the annual average in 2025. Although the report did not break down the increase for the two product categories separately, it emphasized the strong demand for high-bandwidth memory such as HBM4 and DDR5, as well as NAND for large-capacity SSDs.
SK Hynix's price increase is equally exaggerated: the average price of DRAM increased by 60% month-on-month, while shipments remained flat; the average price of NAND increased by 70% month-on-month, while shipments decreased by 10%. The decrease in volume and increase in price confirm the extreme tension between supply and demand. In terms of performance, SK Hynix's Q1 revenue was KRW 52.58 trillion (+198% year-on-year), with an operating profit margin as high as 72%, surpassing NVIDIA's level during the same period; Samsung's DS division generated KRW 53.7 trillion in operating profit, contributing 94% of the group's profits.
For Samsung, the surge in memory is akin to "making profits with one hand while losing with the other". Financial reports show that the cost of mobile memory in the DX department (terminal manufacturing) has surged by 107% year-on-year, directly driving up the material costs of products such as mobile phones and tablets.
To cope with cost pressures, Samsung's DX department has included mobile memory in its raw material procurement list for the first time. In Q1, mobile memory accounted for 9.4% of total procurement, surpassing camera modules (8.9%) for the first time and ranking only behind mobile APs (19.9%) and display screens (10.2%). More crucially, Micron has become a key supplier of mobile memory to Samsung, breaking the long-standing pattern of self-research and self-use, reflecting that internal supply can no longer meet demand or is too costly.
The intensifying competition in AI storage has driven a significant increase in research and development expenses for two companies. Samsung invested KRW 11.34 trillion in R&D in Q1 (up 25.5% year-on-year) and KRW 11.23 trillion in capital expenditure, 90.7% of which was allocated to the DS division. For the whole year, it plans to invest KRW 110 trillion in capital expenditure and R&D, with a focus on new businesses such as AI semiconductors and robotics. In terms of technological progress,
SK Hynix's R&D investment has seen an even more significant increase, reaching KRW 2.55 trillion in Q1 (+68.3% year-on-year), accounting for 4.9% of revenue. According to the product roadmap, SK Hynix's HBM4E is expected to be sampled in the second half of 2026 and mass-produced in 2027. The base chip will be customized according to customer needs, and the core chip will adopt 1c DRAM technology.
The current surge in memory prices is attributed to an AI-driven structural shortage. The memory configuration of a single AI server is 8-10 times that of a regular server, and over two-thirds of the global DRAM production capacity is absorbed by the AI and server markets. Amidst the imbalance between supply and demand, the high boom period for memory will persist for a long time, with HBM, DDR5, and enterprise-class SSDs emerging as core growth engines.
In terms of the industry chain, upstream memory manufacturers are reaping substantial profits, while downstream terminal manufacturers (such as Samsung DX) are forced to bear high costs and adjust their supply chain strategies. Meanwhile, the HBM technology competition is accelerating, with advanced manufacturing processes and packaging capabilities becoming the key to success. The focus of the global semiconductor industry is further shifting towards AI storage and computing infrastructure.
