Micron posts better-than-expected Q2 results and Q3 guidance, but stock declines

Micron reported Q2 EPS of $12.20 and revenue of $23.86B, beating consensus EPS of $9.00 and revenue of $19.7B, and issued Q3 guidance well above estimates. Management cites AI-driven demand for HBM/DDR memory as creating supply tightness, lifting prices across consumer and enterprise devices; Micron is expanding capacity with a second Taiwan plant and a $100B New York fab. Micron shares have rallied 357% over the past 12 months (63% year-to-date), underscoring the stock’s sensitivity to AI memory demand trends and capacity announcements.

Analysis

The market is re-pricing structural memory tightness as a multi-year trade rather than a transitory beat; the non-obvious lever is allocation distortion — fabs and CMs are prioritizing HBM and enterprise-grade DRAM at the expense of low-margin consumer DDR inventory. That shift amplifies margin capture for HBM-focused vendors but feeds through to OEM BOM inflation and lower unit growth for PCs/phones, creating a feedback loop where higher device ASPs further slow shipments and sustain chip pricing.

Capex expansion plans create an asymmetric time profile: near-term pricing power persists as lead times and qualification cycles keep incremental capacity offline for 18–36 months, but announced fabs materially raise the probability of a supply inflection beyond the 2–3 year horizon. Equally important is demand concentration — a small set of hyperscalers account for most incremental HBM demand, giving them bargaining leverage that can compress spot margins if their procurement cadence shifts or model architectures become more memory-efficient.

Market structure amplifies moves: flows have rotated heavily into memory/storage, inflating multiples and leaving little room for disappointment. Key monitoring signals that would reverse the trade are visible — hyperscaler purchase cadence, HBM ASP slips, SK Hynix/Samsung inventory disclosures, or rapid yield ramp at new fabs — all of which can unwind premiums quickly. Position sizing and option-defined risk are therefore preferable to naked exposures given the asymmetric multi-year capex risk versus front-loaded demand upside.

Key Decisions for Investors

• Long SNDK (6–12 months): buy a 9–12 month call spread (long ~25–30Δ, short ~10–15Δ) to express sustained data‑center storage demand with defined cost. Target ~2–3x payoff if enterprise buildout continues; max loss = premium. Close or trim on signs of hyperscaler destock or NAND oversupply.
• Pair trade — Long MU / Short INTC (3–6 months): equal‑dollar long MU stock or 6–9 month call spread versus short INTC equity to capture GPU/HBM-led server share shift. Size 2–3% NAV, target 20–35% relative outperformance; stop-loss: cut pair if MU underperforms by 15% or INTC rallies >20%.
• Long NVDA (3–9 months) via call spread: buy a 3–6 month 50Δ call and sell a nearer OTM call to finance exposure to continued GPU demand without naked upside risk. Expect asymmetric upside if OEMs accelerate re‑orders; cap gain if model shift/policy shock compresses GPU demand.
• Tail hedge (12–36 months): buy long‑dated puts or put spreads on MU/SNDK (calendar out to 12–24 months) sized to cap portfolio downside against a faster‑than‑expected fab ramp or sudden hyperscaler contract renegotiation. Treat as insurance — cost is the price of being long a crowded memory trade.