Better Memory Stock to Buy: Micron or SanDisk?

Micron reported Q1 DRAM revenue of $10.8B (+69% YoY, +20% QoQ) and NAND revenue of $2.7B (+22% YoY), is sold out of HBM for 2026 and expects HBM TAM to grow ~40% CAGR to $100B by 2028; shares trade at ~11.5x forward EPS. SanDisk posted fiscal Q2 revenue +61% YoY and +31% QoQ, expects Q3 revenue to rise >50% sequentially at midpoint, is partnering with SK Hynix on High Bandwidth Flash, and trades at ~13.4x forward EPS. Analyst view in article favors Micron for lower HBM contract risk, but both firms benefit materially from AI-driven demand and current supply tightness, with cyclical risk remaining a valuation discount factor.

Analysis

Micron is the higher-conviction way to own memory exposure to the AI compute cycle because HBM creates a multi-year structural revenue stream that is sticky via long-term contracts and multi-quarter lead times. The key second-order effect is that HBM’s performance/power delta forces system architects to prioritize supply continuity over spot pricing, which lengthens pricing stickiness and improves margin visibility for a supplier with contract footprints. Expect that to translate into asymmetric returns if HBM supply remains constrained for the next 6–18 months while wafer fabs scale only slowly.

SanDisk’s sweet spot is hyperscaler storage — a different bottleneck than HBM — and that specialization both concentrates upside when LLM dataset buildouts accelerate and concentrates downside when large customers over-provision or switch architectures. The SK Hynix collaboration to standardize HBF is a two-edged sword: it can expand the market faster but also accelerates commoditization and cross-supplier pricing competition once ecosystem validation occurs. Operationally, NAND inventory churn at hyperscalers will produce lumpy quarter-to-quarter results and sudden ASP compression if capex or procurement cadence shifts.

Major near-term catalysts that could flip the trade are binary: Nvidia and hyperscaler platform choices for HBM4/HBF over the next 3–12 months, and incremental capacity announcements or faster-than-expected yield gains from Samsung/SK Hynix/Chinese entrants that can create supply inflection within 6–18 months. Tail risks include a macro-led enterprise capex pullback (1–3 quarters) or geopolitical restrictions on cross-border supply that could materially reroute demand and strain contractual coverages. Position sizing should therefore treat memory as a cycle with asymmetric upside but non-trivial downside and concentrate hedges around the next 2–4 major earnings/guidance updates.

Key Decisions for Investors

• Overweight MU (6–12 month horizon): buy MU equity sized to 3–5% of fund NAV as core exposure to HBM. Hedge with a 6–12 month out-of-the-money (OTM) put (strike ~25–30% below entry) to cap cycle reversal risk. Target relative total return +50–80% if HBM market remains tight; downside capped to ~-25% with the hedge.
• Directional pair (12 months): long MU / short SNDK to express HBM outperformance vs commoditized NAND. Size 1:0.6 (MU:SNDK) so long exposure dominates; expect MU to outperform SNDK by 30–50% if HBM adoption continues. Exit or rebalance on NVDA/Hyperscaler announcements that materially change supplier mix.
• Options momentum trade on SNDK (3 months): buy a near-term straddle or asymmetric call-heavy structure around the next earnings/guidance date to capture high guidance dispersion from hyperscaler commitments. Keep position small (<=1% NAV) — implied vol often misprices guidance-driven moves in cyclical memory names.
• Event tail hedge (12–18 months): buy protection on broad memory beta via an index/ETF put or buy deep OTM puts on MU sized to ~0.5–1% NAV to protect against rapid capacity-induced ASP collapse from Samsung/SK Hynix/Chinese fab ramps. This is cheap insurance against a 6–18 month supply surge that would compress multiples across the sector.