The global memory market is currently in the midst of an intense 'AI Supercycle', leading to a dramatic surge in DRAM and NAND prices.
This situation is primarily driven by the explosive demand for AI servers. Tech giants and cloud service providers are racing to build AI infrastructure, which requires vast amounts of high-performance memory like HBM (High Bandwidth Memory) and server DRAM. This demand is far outstripping what suppliers can produce.
The numbers are stark. First, market intelligence firm TrendForce reported that DRAM contract prices jumped by 90-95% in the first quarter of 2026, with another significant increase projected for the second quarter. Second, major memory producer Micron confirmed this reality, reporting massive quarterly increases in their average selling prices. The cumulative price hike from the end of 2025 is over 200% for DRAM.
In response, suppliers like Samsung and SK hynix are reallocating their production capacity away from consumer products toward more profitable AI-related memory. They've warned that these shortages are likely to persist through 2027, creating a supply-constrained environment.
This has led to a significant shift in how memory is sold. Instead of buying on the volatile spot market, major customers are now signing Long-Term Agreements (LTAs), sometimes lasting for years, to secure their supply. This locks in high prices for suppliers but provides stability for buyers.
However, there's a limit to how high prices can go. We're now seeing the effects of consumer elasticity. The soaring cost of memory is making PCs and smartphones more expensive. As a result, manufacturers are starting to cut back on memory specs or raise prices, which is beginning to soften consumer demand.
This combination of factors explains the current forecast: another sharp price jump in the second quarter driven by AI, followed by a moderation in the third quarter as long-term contracts stabilize the market and high prices temper consumer demand.
- AI Supercycle: A prolonged period of high demand and prices for memory chips, driven by the needs of artificial intelligence technologies.
- HBM (High Bandwidth Memory): A specialized type of high-performance memory used in GPUs and other processors for AI and high-performance computing.
- LTA (Long-Term Agreement): A contract between a supplier and a buyer to provide goods over an extended period, often locking in prices and volume.