Business · Companies · 3 hrs ago
AI demand drives server prices sharply higher as memory supplies tighten
AI data centers are driving a memory-chip shortage that has sharply raised server costs.
The pressure is affecting businesses that need servers and storage, as well as makers of memory chips such as Samsung Electronics and SK Hynix.
AI systems use large amounts of high-bandwidth memory and other high-capacity chips, so chipmakers are prioritizing those products over memory for ordinary servers, PCs and phones.
Some server prices have risen to as much as four times their level a year earlier, while Gartner estimates shortages of both DRAM and flash memory this year.
New factories are being built, but construction and production ramp-up take time; Gartner expects added supply from current expansions only from the second half of 2027 onward.
High prices and long delivery times may persist at least until 2028, though Gartner expects memory prices to begin stabilizing around late 2027 to the first half of 2028.
Businesses and consumers may also face higher costs for computers and phones, and some may delay replacing their devices.
Gartner advises businesses to review memory needs, adjust replacement plans and extend the life of existing servers to manage costs.
AI data-center demand is tightening memory supplies and has driven some server prices to as much as four times their level a year earlier.
Gartner says high prices and long delivery times for servers and storage could persist at least until 2028.
Memory’s share of server manufacturing costs has risen from 39% in 2015 to 59% this year, Gartner estimates.
The shortage reflects chipmakers prioritizing high-bandwidth memory and high-capacity products for AI systems.
New production capacity is not expected to add substantially to supply until after the middle of 2027.
- Who
- AI data centers are driving demand; server buyers and memory suppliers are affected.
- What
- Memory shortages have pushed server prices sharply higher and may keep server and storage costs elevated.
- When
- Gartner’s report was released on October 16, 2026; elevated costs and long lead times could last at least until 2028.
- Where
- Not stated
- Why
- Memory makers are prioritizing production for AI systems, reducing the relative supply of conventional DRAM and NAND flash.
This story does not have two clearly opposing sides.
Memory accounted for about 39% of server manufacturing costs, according to Gartner.
Gartner estimated memory would account for 59% of server manufacturing costs.
Gartner expects ongoing capacity expansions to begin increasing actual supply.
Gartner expects DRAM prices to begin stabilizing.
- Server prices
- Some rose to as much as four times their level a year earlier
- Memory share of server manufacturing costs
- 39% in 2015; 59% in 2026
- Estimated supply shortfall
- DRAM 6%; NAND flash 9% this year
- Projected price rises
- DRAM 237%; NAND flash 293% this year
- New supply
- Expected to increase from mid-to-late 2027 onward
1 source
What people are saying
AI-triggered increases in server and storage costs in 2026 have significantly pressured corporate IT infrastructure investment budgets and derailed equipment expansion and replacement projects
Companies should budget on the assumption of high costs and long procurement lead times at least until 2028