Micron Is 1.6 Points From Taking the No. 2 Spot in DRAM

Micron's Q2 2026 DRAM revenue rose 66% sequentially to $36.0 billion, lifting its global market share to 23.3% and narrowing its gap with SK Hynix to 1.6 percentage points from 7.5 points in Q3 2025. The gain was driven primarily by conventional DRAM price increases and scarce inventory rather than HBM, where Micron's share fell to 18% from 21%. Micron generated $41.5 billion of fiscal Q3 revenue, up more than fourfold year over year, and guided for roughly 86% gross margin, but conventional DRAM price growth is expected to slow to 13%-18% in Q3 and HBM4 competition remains a key risk.

Analysis

MU’s earnings torque is currently more exposed to commodity-like DRAM pricing than to a durable technology-share gain. That supports estimate revisions over the next 1-3 months while contract repricing remains positive, but it also means the incremental revenue carries a lower terminal multiple than HBM-led revenue: conventional-memory profits historically mean-revert quickly once customers rebuild inventory or supply additions arrive. The key near-term question is whether management can convert current spot/contract tightness into fiscal-2027 pricing commitments rather than merely report a peak-margin quarter.

The more consequential competitive variable is HBM4 qualification and yield, not the headline DRAM ranking. SK Hynix’s entrenched packaging/customer position and Samsung’s manufacturing scale create a risk that MU’s HBM mix remains below what its valuation requires, even if conventional DRAM stays tight; a weaker HBM mix would make MU more vulnerable to a conventional-price normalization. Conversely, successful multi-customer HBM4 ramps could cause the market to re-rate MU from a cyclical memory supplier toward a structurally AI-linked earnings compounder over 6-18 months.

Second-order pressure should emerge at server OEMs and hyperscaler capex budgets if memory input costs persist: NVDA’s accelerator demand is unlikely to be impaired immediately, but system-level BOM inflation can delay non-GPU server refreshes and squeeze ODM/OEM gross margins. The contrarian view is that the low earnings multiple may already discount a sharp memory downcycle; therefore, evidence that pricing decelerates without turning negative could be sufficient for MU to outperform despite slowing sequential growth.

Key Decisions for Investors

• Tactically long MU for the next earnings/contract-pricing update, sized as a 1-3 month cyclical trade rather than a core AI holding. Target upside is 15-20% if forward EPS revisions continue; exit or hedge if next-quarter conventional DRAM contract guidance turns flat-to-down or gross-margin guidance misses by more than 300 bps.
• Use a market-neutral pair: long MU / short SK Hynix exposure via 000660.KS or an available liquid proxy, with a 2-3 month horizon. The trade monetizes MU’s greater sensitivity to conventional DRAM tightness; cap risk at a 10% adverse spread move because a confirmed HBM4 qualification win at SK Hynix could rapidly reverse relative performance.
• Do not add structural MU exposure until HBM4 customer qualification, yield, and revenue-mix disclosures are independently confirmed. Set an alert for HBM revenue mix and HBM gross-margin commentary at the next two earnings calls; sustained mix expansion is the required catalyst for a 6-18 month multiple re-rating.
• Watch NVDA, Dell, and SMCI for evidence of memory-cost pass-through. If server-system pricing rises while enterprise demand weakens, prefer memory suppliers over OEMs; if OEM backlogs remain resilient despite higher BOM costs, the broader AI infrastructure trade can absorb tighter DRAM pricing.