Samsung Electronics has reportedly doubled the supply price of its general-purpose DRAM for the first quarter, a move that signals a dramatic shift in the memory market.
The primary driver behind this price surge is the AI boom. Let's break down this causal chain. First, the explosive growth in AI infrastructure, underscored by NVIDIA's record-breaking earnings, is fueling unprecedented demand for High Bandwidth Memory (HBM), a specialized memory for AI processors. Second, to meet this lucrative demand, memory manufacturers like Samsung are reallocating their production capacity (CAPA) from general-purpose DRAM—used in PCs and smartphones—to HBM. Third, this strategic shift creates a supply bottleneck for general-purpose DRAM. With less available supply and steady demand from device makers, prices are forced upward.
This situation didn't develop overnight. Tracing back the events, we see clear warning signs. As early as January, major tech companies like Microsoft and Google were reportedly in South Korea to secure DRAM supplies preemptively. Before that, in late 2025, both SK Hynix and Micron posted stellar financial results, confirming that the market was already tilting in favor of suppliers. Market research firms like TrendForce had been forecasting a tightening market and price increases for months, citing low inventory levels and the shift to HBM production.
The numbers paint an even clearer picture. According to reports, the spot price for PC DRAM soared from around $1.65 in April 2025 to $13 by February 2026—a nearly 700% increase. Gartner also projected that the share of memory costs in a PC's total bill of materials (BOM) would jump from 16% to 23% in 2026. This extreme supply-demand imbalance has fundamentally altered business practices, shortening negotiation cycles from the traditional quarter to a month. This gives suppliers immense leverage to adjust prices frequently, reflecting the volatile market conditions.
- General-purpose DRAM: Memory chips widely used in consumer electronics like PCs, laptops, and smartphones. They are produced in high volume and are more price-sensitive than specialized memory.
- HBM (High Bandwidth Memory): A high-performance memory standard used for applications requiring massive data throughput, such as AI accelerators and high-end graphics cards. It is more complex and expensive to produce.
- CAPA (Capacity): In manufacturing, this refers to the maximum production output that a facility can achieve. Shifting CAPA means reallocating production resources from one product type to another.