Micron's 'Stunning' Q2: Record Margins, Tight Supply And 5‑Year AI Deals

Q2 was a "significant beat and raise" with revenue and EPS well ahead of estimates; analysts raised price targets—Rosenblatt to $600 (+$100), Needham to $500 (+$50), BofA to $500 (+$100) and Goldman to $400 (+$40). Firms cite sharply higher DRAM, NAND and HBM pricing and record revenue across business units, expect supply to remain constrained through at least 2026 (Goldman sees DRAM/NAND bit growth just over 20% in 2026), while BofA warns Micron's guided Q3 gross margin of 81% may mark a near‑term peak. Shares traded down 2.56% at $449.89 on publication despite the beat; multi‑year strategic customer agreements and a projected HBM TAM expansion to >$100B by 2028 are structural positives, though some analysts caution HBM pricing could normalize as new supply arrives in 2027.

Analysis

The recent re-rating of memory expectations amplifies a multi-year capital-cycle story where lead times, customer contracting behavior, and node-transition economics interact. If large buyers increasingly secure multi-year supply, visibility into revenue will improve but spot-market liquidity and price discovery can deteriorate — creating episodic price spikes followed by abrupt normalization when committed volumes roll or new cleanrooms come online. Equipment and materials suppliers will earn the tail-end benefit of any sustained re-investment cycle, but their revenue cadence will lag semiconductor memory makers by 18–36 months, creating a window to play upstream exposure separately from memory equity risk. Finally, downside scenarios are concentrated in three mechanisms: faster-than-expected capacity additions by incumbents or China-enabled entrants, demand elasticity where software/hardware teams substitute away from high-bandwidth memory, and any accounting/contractual mismatch that defers downside into later quarters while compressing spot pricing sooner.

Key Decisions for Investors

• Directional equity: Buy MU on intraday weakness into $380–420 with a 6–12 month horizon; size to 2–3% net exposure, target +30–40% upside and hard stop at -15%. Rationale: asymmetric upside from cycle re-rating but respect high volatility and policy/capex tail risks.
• Options risk-defined: Construct a 12-month bull call spread (buy MU 2027-Jan 350C, sell MU 2027-Jan 650C) sized to 1% portfolio P&L risk. This captures sustained memory-tightness upside while capping premium spend if HBM pricing normalizes in 2027.
• Relative-value pair: Long MU / short a broad AI-chip beneficiary (e.g., NVDA) in a 60/40 notional ratio for 3–9 months. This isolates memory-specific cyclicality — rewards outperformance from memory pricing vs. pure GPU-for-compute demand, with expected payoff if memory pricing outpaces GPU-driven expectations.
• Play the lag to capex: Buy wafer-equipment exposure (LRCX, AMAT) on pullbacks with a 18–36 month horizon; initial position size 1–2% each. Expect revenue acceleration to lag memory margin expansion; this is a lower-volatility way to capture the capex reacceleration once OEM bookings convert.
• Hedging/Vol trade: If implied vol compresses after earnings, sell a 30–45 day strangle around near-term print (size to 0.5–1% portfolio risk) and buy longer-dated put protection (LEAP) as tail insurance. This monetizes elevated event vol while preserving crash protection from geopolitical/capex shocks.