Morgan Stanley pushed back on recent losses in U.S. memory-chip shares in a client note on Thursday, arguing the decline is best read as the market "pricing in of durability concerns" rather than evidence of a fundamental drop in consumption.
Analyst Joseph Moore reaffirmed Overweight ratings on Micron Technology and SanDisk, telling clients that the core strength in memory markets is "more durable than the market thinks." Moore acknowledged that a set of classic cycle-peak indicators - including flattened spot pricing, stepped-up capital expenditure and productivity improvements - have materialized and prompted profit-taking. Yet he emphasized that "this is anything but normal."
Morgan Stanley described memory as "THE bottleneck" for artificial intelligence deployments and next-generation central processing unit builds. The bank said shortages have intensified, with corporate buyers now "prepaying for large volume deals."
The note also addressed valuation concerns, cooling second-derivative signals and reports of memory-optimization efforts such as Google’s TurboQuant. Morgan Stanley characterized those optimization moves as "just normal course productivity improvement" that do not signal a lessening of memory demand.
Moore highlighted the growing share of semiconductor spending directed to AI workloads, estimating it could be "well north of 50%" - a dynamic the firm says will make it difficult for rising supply to keep pace with demand. With the available DRAM buffer effectively exhausted, he wrote, "everywhere we look we see indications that it is a true bottleneck."
The bank said it does not view current peak gross margins near 81% as a permanent state, but it sees little reason to expect imminent margin erosion. Morgan Stanley pointed to the sector’s potential for meaningful free-cash-flow generation and concluded that "duration is all that matters," with the relevant indicators "all appear[ing] positive."
Separately, the note addressed investor questions about whether Micron remains an attractive buy today. The research highlighted that dedicated analytics platforms evaluate MU across numerous financial metrics to identify risk-reward profiles, and noted prior examples of stocks identified through such approaches that subsequently produced large returns. Those references were presented in the context of tools that screen for fundamentals, momentum and valuation rather than as a forecast specific to any single outcome.
Context and market implications
Morgan Stanley’s stance frames the recent pullback as a market correction driven by cyclical signals and profit-taking, while maintaining that structural demand - particularly from AI-related spending - underpins the sector. The firm’s view implies that memory-supply expansion may struggle to catch up with fast-growing AI demand, keeping pricing power and cash generation prospects intact for memory suppliers over the relevant duration.
What to watch next
- Movement in spot memory pricing and whether flatness persists or reverses.
- Capital expenditure announcements from major memory manufacturers and any changes in reported productivity gains.
- Reports of customer prepayment activity for large-volume deals, which Morgan Stanley cites as evidence of persistent tightness.
Bottom line
Morgan Stanley argues the recent selloff in memory stocks is not a signal that demand has structurally weakened. The bank retains Overweight ratings on the names cited, while cautioning that very high gross margins may not be permanent even as near-term deterioration appears unlikely. The firm’s core thesis rests on AI-related semiconductor spending and evidence of constrained DRAM supply, which it says together create a durable backdrop for memory makers.