The Zhitong Finance App learned that Dongwu Securities released a research report saying that the storage industry is essentially a strong cycle industry driven by “fluctuating demand and lagging supply.” In the main cycle since 1985, the price decline was usually triggered by a combination of high production expansion, process volume and weakening demand, while bottom recovery depended on production cuts, capital expenditure contraction, and cost and capacity withdrawal. Successive in-depth clean-ups have promoted an increase in industry concentration, and supply discipline from leading manufacturers has gradually been strengthened.
The main views of Dongwu Securities are as follows:
The main demand line continues to migrate
Early storage demand was mainly driven by PCs. After 2012, smartphones and cloud computing relayed, and 2020-2023 experienced pre-pandemic demand and inventory reversal. Currently, AI is further expanding demand to HBM, server DDR5, and enterprise-grade SSDs, and storage requirements are shifting from consumer electronics cycle repair to structural increases brought about by AI infrastructure.
Supply and demand resonance in the current cycle
Original factory production cuts and capital expenditure contractions from 2022 to 2023 have reduced the supply elasticity of the industry; HBM's high wafer consumption and advanced packaging restrictions further overwhelm the general DRAM supply. At the same time, AI training and reasoning are jointly driving demand for HBM, server memory, and enterprise-grade SSDs. Supply contraction and rising AI demand are the core reasons why the current boom is stronger than the previous consumer electronics cycle.
Economic Verification and Follow-up Observation
Micron's FY26Q3 DRAM and NAND prices rose by about 60% and 85%, respectively, and all four major business units recorded record revenue. The future should focus on tracking DRAM/NAND contract prices and spot prices, original factory inventory and capital expenses, wafer delivery and process migration, HBM capacity conversion, and cloud vendor capital expenditure and AI accelerator shipping pace.
Risk warning: AI data center capital expenditure falls short of expectations; supply release is faster than expected due to expansion of original factory production or process migration; rapid rise in storage prices suppresses terminal demand; macroeconomic, geopolitical and export control risks; risk of short-term stock price fluctuations.