DRAM: We May Be At The Start Of The Second Memory Rally

Roundhill initiated coverage of the Memory ETF with a strong buy, citing unprecedented AI-driven memory demand alongside compressed valuations in leading memory suppliers. The DRAM exposure is concentrated in Samsung, SK hynix, and Micron, positioning investors for access to dominant global memory makers. Management argues the recent DRAM selloff reflects over-leveraged Korean retail rather than weak fundamentals, noting earnings and guidance are exceeding expectations.

Analysis

The setup is less about a single price move and more about a multi-quarter earnings revision cycle: concentrated memory supply plus AI-linked bit demand can turn a cyclical commodity into a scarcity asset. That matters because these names still trade with a discount to the broader AI complex; if utilization stays tight, the upside comes from both multiple re-rating and operating leverage, not just spot pricing.

Second-order winners are the tool and materials vendors that benefit once producers feel confident enough to add capacity, but the timing is lagged by 2-4 quarters. The near-term losers are server OEMs and other hardware assemblers with the least pricing power, since memory is one of the few inputs that can move faster than end-market ASPs. That creates a potential relative-value short in names like DELL/HPE versus memory exposure if pricing stays firm into the next earnings season.

The contrarian risk is that the market may be extrapolating AI demand before confirming sustained order growth outside a narrow set of customers. If inventory rebuilds rather than true end-demand is doing the heavy lifting, the trade will fade quickly once channel data normalizes. What falsifies the thesis is a pause in sequential DRAM/HBM pricing or guidance that stops improving over the next two reporting cycles; if that happens, the rally should be treated as flow-driven rather than fundamental.

Key Decisions for Investors

• Build a starter long in SKHYV now, but size it as a 3-6 month cyclical trade rather than a structural hold; target 15-20% upside if memory pricing and guidance keep inflecting, with a hard risk cut if the ETF breaks 8-10% below entry on no new positive pricing data.
• Pair long SKHYV / short DELL or HPE for 1-3 months to isolate memory inflation versus OEM margin compression; this is the cleaner expression if you want to avoid broad semis beta.
• Add AMAT or LRCX on any pullback only after the next memory producer earnings calls confirm capex follow-through; the better entry is when equipment orders start to re-accelerate, not on the initial news flow.
• If options liquidity is workable, use 3-6 month call spreads on SKHYV or MU instead of outright stock to express upside with defined downside; this is attractive only if you expect another positive earnings revision within the next two quarters.