Rising even more than HBM! Enterprise-level SSD contract prices soared by more than 80% in a single quarter

Kuai Technology reported on July 25 that over the past year or so, AI data centers’ crazy purchases of HBM and general-purpose DRAM have led to tight supply and soaring prices. Now this price increase is spreading to enterprise-class solid-state drives (eSSD). According to data from TrendForce and multiple institutions, the contract price of enterprise-class SSDs in the first quarter of 2026 has increased by 80% to 100% from the previous quarter, and the prices of some popular models have nearly doubled compared with the end of 2025. The contract price of large-capacity QLC eSSD also increased by 80%. The demand for AI inference is passing the price increase baton of memory chips from memory to flash memory, and the demand for storage in AI data centers is also undergoing structural changes. As generative AI moves from model training to large-scale inference deployment, GPU clusters need to store massive KV caches (Key-Value Cache) to maintain efficient workflow. With the advantages of low cost and ultra-large capacity, large-capacity QLC eSSD is evolving from a simple storage medium to a device that can offload and support high-load AI tasks. This role change has directly contributed to the price surge. The changing role of large-capacity QLC eSSD stems from the fact that AI hardware manufacturers such as NVIDIA are redefining the relationship between storage and computing power. The AI ​​inference context storage platform launched by NVIDIA expands GPU memory through a shared POD-level context layer optimized for KV cache. The eSSD capacity corresponding to a single GPU has increased significantly. In terms of data, Morgan Stanley's model shows that AI's proportion of NAND demand will rise from 18% in 2025 to 41% in 2027, and AI NAND demand will increase by 60% year-on-year. And eSSD is becoming another key link in AI computing power investment after HBM and DRAM.