Morgan Stanley pushed back against recent selling in memory-related equities, reaffirming an overweight rating on Micron Technology and SanDisk and describing the pullback as a corrective move rather than evidence of a fundamental market reversal.
Analysts at the firm said supply remains constrained in ways that matter for large-scale artificial intelligence deployments. According to the bank, shortages are intensifying and customers are responding by prepaying for substantial volume agreements on the expectation that limited supply will persist.
Under typical industry patterns, a combination of flat spot prices, increased capital expenditures and productivity gains would often signal a cycle peak. Morgan Stanley, however, distinguishes the present backdrop from historical cycles. The bank notes that memory has evolved into a critical chokepoint for AI builds and for agentic CPU configurations, changing the implications of the usual price and investment signals.
While the analysts concede that a second derivative deceleration is unavoidable - meaning growth rates will slow after a strong period - they emphasize that the duration of favorable conditions becomes the dominant factor given current valuation levels, which the bank notes sit at approximately four times earnings. Key indicators, the firm says, continue to point in a positive direction.
The bank also addressed recent debate over KV cache optimization, which had pressured investor sentiment in the sector. Morgan Stanley characterized such optimization as a routine step of productivity improvement rather than a structural reduction in memory demand.
Overall, the firm interprets the recent market selloff as the market pricing in concerns about durability. Morgan Stanley maintains that the underlying strength of the memory sector - supported by constrained supply and AI-related demand patterns - is more sustainable than prevailing sentiment suggests.
Implications
This view matters for participants in the memory and broader semiconductor supply chains and for investors tracking AI infrastructure spend. It also bears on capital allocation decisions inside firms that build AI compute systems, where memory availability can constrain build schedules and contract structures.