The popularization of AI triggers a change in storage demand

Although several years ago, various industries were talking about AI sinking to the edge, which would bring massive storage demands. But just before the beginning of 2025, when industry insiders mention storage and AI, they tend to lean towards the statement that 'AI cannot bring too much storage demand'.
 

The reason is that in the field of AI data centers, high bandwidth and low latency storage requirements are mainly focused on high-performance computing and large-scale data processing, and HBM (High Bandwidth Memory) precisely meets these needs.
 

As AI gradually lands on the edge, its popularity will drive more storage demand. Although HBM has excellent performance, its cost and power consumption are relatively high, making it difficult to popularize in edge devices. The storage categories that edge AI devices can leverage will be DRAM and flash memory (NAND), especially for storage solutions such as embedded storage and distributed file systems that are more suitable for edge devices.