Solidigm: Storage is Key for Agentic AI & Inference

Solidigm is focusing on the intelligence layer as agentic inference makes storage a central part of artificial intelligence infrastructure. This shift from model training means a fundamental rethinking of AI infrastructure components. The rise of sovereign AI deployments and enterprise AI clusters is increasing this pressure. Storage has moved from an afterthought to a key consideration. Greg Matson, Senior Vice President at Solidigm, states that the intelligence layer is now being housed on storage. High-capacity SSDs are becoming essential for AI inference, token economics, and sovereign compute deployments. Hyperscalers recognized this problem 12 to 18 months ago. They began replacing older storage with high-capacity solid-state drives to keep GPUs fully utilized. Matson explains that if a GPU is waiting for data, it's wasting money. A simple prompt can generate many tokens, pushing storage requirements into petabytes. Solidigm offers drives with 122 terabytes of capacity per unit to address this. This matters because efficient storage directly impacts the performance and cost-effectiveness of AI systems.