Despite Rising More Than 300% in 1 Year, Micron Stock Looks Cheap. But Is It a Buy?

Micron reported fiscal Q2 revenue of $23.86B, up 196% YoY and 75% sequentially, with adjusted EPS of $12.20 (up 682% YoY). Management guided fiscal Q3 revenue of roughly $33.5B, an adjusted gross margin of ~81%, and EPS of $18.90, reflecting extreme pricing power from AI-driven demand for HBM and datacenter storage. The stock trades at ~19x trailing EPS and just ~8x forward P/E despite a >300% 12-month gain, as the market prices in potential cyclicality and the risk that capacity additions could rapidly compress prices and margins.

Analysis

Hyperscaler procurement behavior is now the dominant driver — not spot markets — which creates a time-limited window where a handful of memory suppliers can sustain structurally higher prices because customers are willing to pre-pay for scarcity. That creates two non-obvious effects: (1) higher bargaining power for suppliers who have locked design wins (increasing switching costs for customers) and (2) a disincentive for new entrants to immediately expand HBM capacity (capex economics look worse vs. DRAM), prolonging tightness for 12–24 months. Tool lead times (EUV/advanced lithography, backend HBM assembly) and capital intensity mean capacity additions are lumpy and slow — the market cannot be instantaneously arbitraged away by incremental fab builds.

Primary downside catalysts are conventional but fast-acting: a hyperscaler capex pause that reduces contracted purchases (weeks–months) and a coordinated capacity push from Samsung/SK Hynix that can depress pricing within 12–24 months. Policy/trade risk is asymmetric too — export controls or procurement nationalism could either tighten supply (positive) or restrict addressable demand (negative), creating binary 30–50% swings in revenue for vendors exposed to specific customers. Also monitor model-level changes from large AI customers (HBM per GPU reductions, increased compression offload) which can materially reduce demand intensity in 6–18 months.

Consensus is implicitly pricing a near-term reversion; the contrarian gap is that the market may be underestimating how long a durable HBM oligopoly could persist because of technical barriers and hyperscaler lock-ups. That argues for asymmetric exposure (option structures or hedged equity) rather than naked leverage: if you’re right, payoffs are multiples in 6–24 months; if wrong, downside is capped and reversible once capex-driven oversupply materializes beyond ~18 months.

Key Decisions for Investors

• Buy MU equity with a protective collar (size 2–4% portfolio): buy MU stock and simultaneously buy a 6–9 month 20% OTM put and sell a 6–9 month 10% OTM call. Timeframe 3–9 months; objective +40–60% if demand persists; downside limited to ~20% net (after collar). Exit/roll if hyperscaler capex guidance turns negative within two consecutive quarters.
• Directional long via LEAPS (size 1–2% portfolio): buy MU Jan-2028 LEAP calls and finance by selling near-term (3–6 month) calls. Time horizon 12–30 months; risk limited to premium paid, reward asymmetric if HBM tightness persists for multiple build cycles. Close or hedge if industry-wide capacity guidance increases by >20% year/year.
• Pair trade: long MU / short INTC (equal-dollar, size 1–3% net long exposure) over 6–12 months — rationale is exposure to HBM-driven margin expansion vs. Intel’s heavier exposure to legacy CPU server cycles and margin compression. Target pair outperformance 25–50%; stop-loss: reallocate if MU underperforms Intel by >25% over a rolling 3-month window or if broad semiconductor indices rally >20% on cyclical recovery.
• Small tactical long NVDA options (size 0.5–1% portfolio) as leveraged exposure to continued AI infrastructure spend: buy 3–6 month calls (or call spreads) to capture upside from continued GPU/HBM demand. Use this as a hedge for MU long exposure — NVDA pullbacks tied to hyperscaler pauses would signal reassessment; target 2–4x return on option premium if AI capex persists, loss limited to premium.