Chip producers divert resources to AI high-performance chips, while inventory hedging and market panic exacerbate supply shortages and price volatility.

Recent trade reports indicate that global memory and storage device prices are climbing sharply as AI development demands outpace supply, fueling a worldwide shortage in DRAM and NAND chips.

The sudden price surge began after chipmakers such as Samsung and SK Hynix began prioritizing high-margin server memory and advanced chips for AI data centers, shifting capacity away from standard consumer products.

According to TrendForce, major suppliers have substantially raised DRAM contract prices, with Q4 2025 projections now forecasting a rise of 18–23% for conventional DRAM, and 25–30% for PC DRAM alone. Some US and Chinese hyperscalers are reportedly receiving only 70% of their server DRAM orders despite agreeing to contract hikes as high as 50%.​

The crunch is leading to rapid, visible fallout across the consumer hardware market. Mini PC manufacturer Minisforum announced that starting 4 November (2025) prices would rise for all models containing DDR5 memory or solid-state drives, directly citing escalating raw material costs and memory price spikes; devices sold without memory or storage were not affected. The market panic has prompted a flurry of double and triple ordering (for hedging) as firms race to secure inventory, echoing previous electronics shortages.​

The impact is especially severe for legacy DDR4 chips, whose prices have reportedly almost doubled within a week in late October 2025, from US$13 to US$25 per module. DDR5 prices, which had been declining through much of 2024 and early 2025, are now rising sharply—some reports indicate increases of 30-50% in certain regions compared with mid-2024 levels.

NAND and SSD prices have also jumped between 20% and 35%, triggered by overlapping AI expansion and resulting supply bottlenecks for both memory and storage components. Some memory chip producers have responded to the acute shortage by halting price quotes entirely.​

Industry analysts expect these supply constraints and elevated pricing to continue well into 2026, with server and advanced memory for AI workloads absorbing most of the available manufacturing capacity.

For every brick laid with AI hope, a shadow of confusion grows, as Babel’s story cautions us still.