The Register's storage editor examines how artificial intelligence has become both the biggest growth driver and the biggest risk for the data storage industry. Since ChatGPT's launch just over three years ago, storage vendors have had to overhaul their technology to feed data quickly enough to power-hungry GPUs, while also grappling with new dangers from AI agents that can access, alter or leak data incorrectly or maliciously.

The piece identifies five points where storage and AI intersect: supplying data to AI systems, protecting that data and the actions AI agents take on it, extending cyber-resilience to govern AI's data access, storage vendors using AI within their own operations, and defending storage itself against AI-driven attacks. It traces how technologies such as NVMe, GPUDirect and disaggregated storage architecture (pioneered by VAST Data) emerged to keep GPUs supplied with data, with newer players like DDN, Pure Storage and WEKA rising alongside established vendors Dell, HPE and NetApp, while monolithic array makers such as Hitachi Vantara, IBM and Infinidat have struggled to keep pace.