Morgan Stanley is urging market participants to view the latest weakness in memory shares as a corrective realignment rather than evidence of a broken cycle. In a note, analyst Shawn Kim described the recent price action as "a healthy reset," and said earnings revisions for memory companies remain "robust and more sustainable than most believe."
The firm highlighted that consensus profit expectations for memory manufacturers have climbed in step with share prices, providing an underpinning for the earlier rally. Rather than signaling an inflection in the industry cycle, Morgan Stanley sees the pullback as a consolidation that leaves intact the drivers of future returns.
Central to the bank's outlook is DRAM, which it identifies as the primary bottleneck in the AI buildout. Morgan Stanley pointed to substantial moves in the memory market: pricing for memory has nearly doubled since February, and lead times have extended materially. Those dynamics, the firm said, help explain the tightness in supply and the strong pricing trajectory observed to date.
Looking ahead, Morgan Stanley expects meaningful price gains to continue near term. The firm estimated more than 20% to 30% DRAM price hikes in the third quarter of 2026 - a magnitude it described as "enough to keep the YoY rate of change accelerating." That projected step-up in pricing is central to the bank's view that earnings momentum can persist.
The bank also highlighted the growing role of contractual commitments in supply: it estimated long-term agreements could account for 70% or more of total DRAM supply over the next three to five years. That concentration, Morgan Stanley argued, could support a valuation re-rating for DRAM-exposed stocks - "could re-rate to at least 8-10x PE vs. 5x PE now."
In updating downside scenarios, Morgan Stanley raised its bear-case values for two major suppliers. The firm revised its bear case for SK Hynix higher by 175% and for Samsung higher by 58%.
On the broader cycle, Morgan Stanley acknowledged that memory prices will eventually soften as planned capacity investments come online from late 2027. Still, the bank said the AI-driven demand profile changes the pricing equation: lower DRAM costs reduce the expense of running AI inference, which, the firm said, "create new demand, rather than simply reducing costs for a fixed number of consumer devices."