Overview
Morgan Stanley’s investor note warns that surging artificial‑intelligence (AI) demand is turning memory into a structural bottleneck, creating a multi‑year supply shortage in the global memory market. The report highlights that memory prices—including high‑bandwidth memory (HBM), DRAM and enterprise solid‑state drives—have risen more than sixfold over the past twelve months, a sharp reversal of the long‑term trend of declining semiconductor costs.
Demand Drivers and Supply Constraints
The acceleration in demand is driven by hyperscalers ramping up spending on AI infrastructure, which is increasing consumption of high‑bandwidth memory, DRAM and enterprise SSDs. Manufacturers are prioritising higher‑margin products for data‑center and AI systems, thereby reducing the volume available for traditional end‑markets such as smartphones, personal computers, automotive electronics and industrial equipment. The note estimates that, if current trends persist, smartphones and PCs could experience memory supply shortfalls by 2027.
Market Dynamics
Large cloud providers are increasingly securing memory capacity through long‑term supply agreements and pre‑payments, which tightens availability for other customers and diminishes the influence of traditional commodity pricing in the memory market. The report also records modest price movements for related equities, noting DRAM down 0.17% and AI‑focused equities up 1.26% on the day of reporting.
Economic Implications
Rising memory costs are already reflected in producer‑price inflation (PPI) and corporate cost structures, although the direct impact on consumer‑price inflation is expected to remain limited. Elevated memory prices are likely to increase hardware costs, pressure margins for companies producing consumer hardware, and potentially slow the deployment of new technologies across several industries.
Investment Outlook
The Morgan Stanley note favours memory manufacturers and related infrastructure suppliers, citing stronger pricing power and clearer earnings visibility. Conversely, companies exposed to consumer hardware and other non‑AI end‑markets are identified as facing the greatest risk from higher memory costs and constrained supply.