Demand from artificial intelligence (AI) data centers is projected to further increase memory semiconductor prices. Major cloud companies are securing long-term contracts for DRAM and NAND flash memory, which could tighten supply for consumer products such as PCs and smartphones.
On June 29, U.S. technology news outlet Wccftech reported that Jefferies, a U.S. investment bank, forecasts that DRAM and NAND flash prices will rise by 40% to 50% in the third quarter compared to the previous quarter, and by an additional 30% to 40% in the fourth quarter. The bank also expects memory prices to increase by 40% to 45% year-over-year in 2027.
The primary driver behind the price increase is investment in AI infrastructure. Jefferies noted that approximately half of the total memory production is currently tied up in long-term supply agreements (LTAs) between memory manufacturers and major tech companies. If this proportion rises to 70%, supply for memory used in general products like PCs and smartphones could diminish further.
Micron is a notable example, having signed multi-year supply contracts with key customers in the data center, consumer, and automotive sectors. These contracts reportedly include purchase obligations extending from 2026 to 2030.
Analysts also indicated that Chinese memory manufacturers may struggle to reverse the price trend in the short term. Jefferies stated that Changxin Memory Technologies (CXMT) is lagging 1.5 to 2 generations behind global leaders in DRAM technology, suggesting its impact on the market will be limited through 2026 and 2027.
Price pressures are also affecting the finished goods market. Market research firm Gartner predicts that rising DRAM and SSD prices could lead to increased average PC prices and decreased shipments. Taiwanese market research firm TrendForce also expects further increases in PC DRAM contract prices in the third and fourth quarters.
Jefferies anticipates that by 2028, new supply may lead to a decline in average selling prices (ASP). However, the ongoing demand from AI and server markets suggests that the timeline for price stabilization will depend on the pace of supply expansion.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.