Once again, we return to a topic that is probably as popular in the PC world as artificial intelligence (and is close buddies with it): memory prices. TrendForce has released their outlook of how DRAM prices—that is, system memory and graphics memory—are expected to develop in Q1 2026. Unfortunately, it appears that the explosive price surge is not staying, but the three months we are now entering could be the worst yet.
In a report dated February 2, TrendForce states that demand from the data center market—particularly AI—continues to worsen the market imbalance, with DRAM supply falling short. According to the law of supply and demand, this is pushing prices upward. TrendForce analysts have therefore revised their original outlook for price developments in the first quarter of 2026—January through March—for the worse.
Originally, TrendForce predicted a 55–60% increase in contract prices for Q1 2026. However, as of early February, the firm updated its forecast and now expects average contract prices to rise by 90–95%. That is nearly a doubling of prices in just three months. Whichcomes on top of the 45–50% increase that already occurred in the fourth quarter of last year (bringing the combined increase to 176–192%, or nearly triple).
This price surge level is affecting virtually the entire market. For PC memory—namely DDR4 and DDR5—prices rose by 38–43% in Q4 2025, and TrendForce now expects a further +105–110% increase in Q1 2026. PC memory will therefore be hit the hardest. Manufacturers are likely assigning it the lowest production priority, while at the same time this segment experienced relatively smaller price increases last year, so it may simply be catching up with other DRAM types. Combined across Q4 and Q1, this represents a 2.83× to 3.0× increase.
In the server segment, DDR4 and DDR5 already rose by 53–58% in Q4, which may explain why in Q1 2026 they are projected to increase “just” by 88–93% (for a total of 2.88× to 3.05×).
Mobile memory for smartphones and notebooks essentially mirrors the situation—older LPDDR4X rose by 48–53% in Q4, while newer LPDDR5X increased by 43–48%. Now, in Q1 2026, both types are expected to rise by +88–93%, again nearly doubling quarter-over-quarter. In total, this represents a 2.69× to 2.95× increase. This will inevitably be reflected in the prices of finished devices on the market.
The sheer scale of the projected increase—around 100% just between January and March—almost certainly means the estimate currently carries a higher-than-usual margin of error, likely greater than the stated range. The roughly +100% figure should therefore be taken as rough ballpark rather than a precise prediction.
SSD prices set to continue their own catastrophe
Incidentally, according to TrendForce, the outlook for NAND flash—used as the base material for SSD storage—has also worsened. Previously, a 33–38% increase (during Q1 2026 alone) had been expected, but TrendForce has now revised this to +55–60%. Combined with the 33–38% increase in Q4 2025, this results in a total price growth of 2.06× to 2.21×.
NAND prices are suffering because demand similarly far exceeds supply, and DRAM prices are rising so sharply that memory manufacturers have an incentive to shift production lines previously dedicated to NAND toward DRAM. Supply to the market being reduced then further stimulates price increases.
The fact that fourth-quarter research shows increases of only 50–60%, while retail module prices have risen severalfold (some reports mentioned up to sixfold), is likely due to prices rising at different rates for different customers, with the table showing an average. The larger and more important the client, the better pricing they were likely able to negotiate during the shortage, meaning their increases may have been limited to tens of percent, while smaller companies selling standalone modules in the retail market may have faced supplier price hikes of hundreds of percent.
Module manufacturers such as Kingston or Adata would then likely apply similar logic when adjusting their own product pricing—granting smaller increases to large OEMs purchasing in volume, while retail-bound modules would sit at the bottom of the priority list. Fewer units would be released into the channel, causing the usual market mechanisms to drive end customer prices in retail and etail the most—so a sixfold difference between Christmas 2025 and the previous year could indeed occur in certain cases.
Spot market prices, which means immediate purchase prices of memory traded at the memory market, tend to see sharper increases, whereas the TrendForce table reflects contract prices effective for longer-term agreements, where price adjustments typically proceed more gradually. Moreover, memory prices had already begun rising in Q3 (especially DDR4), meaning the overall deterioration is the cumulative effect of increases across all the recent quarters.
The worst single-quarter increase we can remember
What the table projects for the first quarter of this year represents, incidentally, by far the most brutal figures we recall seeing in TrendForce reports. Although pricing cycles have occurred in previous decades—particularly for DRAM (and later for NAND, albeit typically somewhat flatter), and indeed the price differences between cycle lows and highs for DDR3 and later DDR4 reached as high as 2.5× to 3.0×, it is important to remember that in these cycles, the increases and subsequent declines always unfolded gradually over multiple quarters. Even during the most rapid upswings, quarterly increases wold reach perhaps 15–30%, with the large total swings always resulting only from the cumulative effect.
A 100% jump in a single quarter appears essentially unprecedented (it’s perhaps only comparable to the shock HDD price spike caused by Thailand floods in 2011). Given the scale and impact, the current DRAM cycle can safely be described as extreme and outside anything previously known in the market.
Source: techPowerUp
English translation and edit by Jozef Dudáš
⠀
