Analysis: contract prices for NAND chips are up 600%+ since September 2025's end, while DRAM chip prices are up nearly 400%, and analysts expect more increases
The worsening shortage in global memory chips due to the artificial intelligence buildout is driving a widening gulf in corporate results and stock performances.
Context & Ripple Effects
Related coverage had already established that major memory makers were operating near capacity, with much of 2026 output effectively committed to AI-driven demand and limited new supply expected before 2027. The latest contract-price moves show that tightness is no longer merely a capacity forecast but is being transmitted directly through both DRAM and NAND pricing.
Earlier reports pointed to sharply stronger earnings expectations for Micron and Samsung while warning that higher memory costs could reach smartphones, PCs and other consumer electronics. This price escalation connects those divergent outcomes: memory suppliers gain pricing power as downstream device makers face higher input costs.
First-order effects
- Memory-chip producers immediately benefit from substantially higher realized pricing, reinforcing the earnings and stock-performance gap between suppliers with available memory output and companies that must buy it.
- Data-center operators and device manufacturers face higher DRAM and NAND procurement costs, while already-constrained supply leaves limited room to offset those costs by switching suppliers or expanding volumes.
Second-order effects
- PC, smartphone and consumer-electronics vendors are more likely to pass through some memory-cost inflation, consistent with prior expectations for higher device prices, or to protect margins through lower-spec configurations and tighter component purchasing.
- The concentration of high-end memory supply in data centers intensifies competition between AI infrastructure buyers and other memory customers, making allocation—not only price—a determinant of which products can ship at planned volumes.
Third-order effects
- If capacity remains constrained until new production arrives, memory shifts from a cyclical input-cost issue to a strategic bottleneck in the AI buildout, favoring vertically integrated or supply-secured buyers over firms purchasing opportunistically.
- Sustained price dispersion between AI-oriented memory and broader commodity memory could reshape capital-allocation priorities at chipmakers, although the durability of that shift depends on whether AI demand remains strong as additional capacity eventually comes online.
The trend: The AI infrastructure buildout is turning memory availability and pricing into a central divider between semiconductor suppliers, data-center buyers, and consumer-device manufacturers.