Western Digital's Chief Product Officer Says AI Storage Should Balance HDD and Flash

Ahmed Shihab, Chief Product Officer of Western Digital, stated that the key to AI storage lies in continuously expanding capacity at an affordable cost, rather than solely pursuing the fastest medium. He pointed out that many architectures perform well in the early stages, but may expose issues when data scales from several petabytes (PB) to hundreds of PB or even exabytes (EB) due to uncontrolled costs. Flash is suitable for high-performance, low-latency scenarios, such as model weights and GPU overflow, while HDDs are more appropriate for large-scale, long-term storage of cost-sensitive data, such as training datasets and compliance records. He summarized, "Flash handles the present, while HDD manages the entire lifecycle." Under AI scale, storage costs will become an architectural issue, as prolonged use of high-performance media will squeeze other budgets. Future AI storage should be designed in layers based on performance, cost, power consumption, density, reliability, and data lifecycle, emphasizing the need to match appropriate media for different workloads to ensure business sustainability.