Memory semiconductor prices are showing a steep upward trend in the new year. NAND flash commodity product prices have risen nearly 65% in just one month, while next-generation DDR5 DRAM module prices have surged close to 90%. The rally is attributed to increased data processing demand driven by artificial intelligence expansion and production cuts by manufacturers.
According to market research firm DRAMeXchange on January 30, the average fixed transaction price for 128-gigabit (Gb) multi-level cell (MLC) NAND flash, a commodity product used in memory cards and USB drives, reached $9.46 in January, up 64.83% from the previous month. Beyond the 128Gb product, the 32Gb MLC product rose 52.19% to $5.57, and the 64Gb product climbed 53.01% to $7.02, with major MLC NAND product lines recording increases exceeding 50%.
Single-level cell (SLC) NAND flash, which had shown relatively smaller price fluctuations, is also trending upward. All SLC products ranging from low-capacity 1Gb to 32Gb showed increases of around 10% month-on-month. TrendForce analyzed that "suppliers have focused on 3D NAND and high-capacity product production, reducing wafer input for mature process (legacy) products such as SLC and MLC, leading to supply constraints in the market."
In the DRAM market, DDR5 prices showed the largest gains. The January average price for PC DDR5 8-gigabyte (GB) modules (SO-DIMM/U-DIMM) surged 89.87% to $75 from $39.5 the previous month. DDR5 16GB modules also jumped 86.11% to $134. The price increase, which had been in the 30-40% range in Q4 last year, expanded significantly in January.
DDR4 DRAM prices, the existing mainstream product, continued their upward trajectory. The January average fixed transaction price for DDR4 8Gb 1Gx8 chips, a PC commodity product, rose 23.66% from the previous month to $11.5. The DDR4 average fixed transaction price broke through the $9 barrier for the first time since June 2016 last month at $9, and set a new record this month at $11. DDR4 16GB module prices also rose 83.91% to $160, narrowing the gap with DDR5.
Prices across all memory products have surged as AI-related data storage and processing demand increases. In AI environments that require storing and processing massive data, such as the "Inference Context Memory Storage (ICMS)" mentioned by Nvidia CEO Jensen Huang, high-capacity NAND storage is essential. Major suppliers including Samsung Electronics (005930.KS) and SK hynix (000660.KS) are focusing on producing high-value-added products like High Bandwidth Memory (HBM) while reducing commodity product supply, creating supply-demand imbalances.
TrendForce forecast that "supply-demand imbalances will continue for the time being as demand from AI servers and industrial equipment persists," adding that "NAND and DRAM prices are expected to remain strong until early 2026."