Server Prices Soar Up to 4-Fold Amid Rising AI Investment; Gartner Says Companies Should Recalibrate Infrastructure Spending
TECHWORLD ·
✦ AI Summary
Gartner said server and storage prices could rise as AI infrastructure investment expands and memory supply is reorganized, with the trend potentially lasting through 2028.
As memory makers shift supply priorities to HBM for AI GPU and DRAM for large cloud operators, shortages of memory for general-purpose servers and NAND flash continue.
Gartner analyzed that server prices this year could rise 2 to 4 times from 2025 levels and storage prices by as much as 2 times, and recommended that companies delay server purchases and adjust budgets.
Gartner projected a sharp rise in server and storage prices as AI infrastructure investment expands. On the 18th, Gartner released a report titled "Server and Storage Costs in 2027 Amid the AI Boom" and said the upward trend in server and storage prices would continue through 2028. It cited the restructuring of memory supply centered on AI infrastructure and the concentration of memory supply toward AI data centers.
According to Gartner's analysis, as memory makers shifted their supply priorities, preferred supply was redirected to HBM for AI GPU and DRAM for large cloud operators. As a result, supply of memory for general-purpose servers has declined, leading to continued shortages of components for general server and storage use, while reduced supply for general server memory has also persisted. NAND flash is also experiencing shortages, which Gartner attributed to supply concentration driven by AI-related storage demand.
As a result of this supply reshuffling, server prices this year were analyzed to be 2 to 4 times higher than in 2025, while storage prices this year rose by up to 2 times compared with 2025. Gartner said this trend could continue through 2028 and advised companies to adjust their infrastructure budgets and investment plans.
Gartner said it expects the memory supply shortage to last for a prolonged period, putting the end of the shortage at early 2028. It said the reason is that building new facilities at Samsung Electronics, Micron, and SK Hynix typically takes 2 to 3 years.
Gartner forecast that additional supply would begin to ramp up in the middle to second half of 2027. However, it said it would take an additional 1 to 2 quarters for any price-cutting effect from increased component supply to be reflected in the supply chains of server and storage manufacturers. Accordingly, it projected that a return to historical price levels for servers and storage would be difficult.
Gartner said that even if supply improves, manufacturers are unlikely to immediately pass along lower component costs to product prices in order to preserve profitability. It projected that the price of a high-end x86 server equipped with 1.5TB of memory would rise by as much as 3.5 times compared with 2025. It also said that even in an optimistic scenario, the server's price would remain above 2025 levels throughout 2028.
As memory costs rise, Gartner said the share of server manufacturing costs accounted for by memory will also increase further. While it projected that memory's share of total server costs will rise in the future, the share of memory components in the overall cost of general servers is expected to climb from 45% in 2025 to two-thirds by 2030.
Based on this pricing environment, Gartner said in the report that companies need to factor current high price levels into infrastructure budgeting. It recommended delaying server purchases as much as possible and extending the use of existing servers by 6 to 8 years. It also pointed to 2028 as the expected timing for prices to decline and recommended postponing server purchases until then.
Source: TECHWORLD · Kim Hye-jin
Original: https://www.epnc.co.kr/news/articleView.html?idxno=408387
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.