Micron: The Peak Memory Myth

Micron remains a Buy as analysts expect nearly 600% EPS growth this fiscal year and ~70% in fiscal 2027, driven by AI-related memory demand and a significant backlog. The note deems fears of peak memory demand, higher capex, and cyclical risk overstated despite Google's TurboQuant announcement, implying materially stronger earnings power and upside to consensus estimates.

Analysis

Micron sits at the intersection of structural AI demand and a cyclical memory market; the subtle winner from current dynamics is the company that converts design wins into higher HBM/DRAM content per rack rather than simply selling more commodity bits. Second-order beneficiaries include EUV and inspection equipment vendors (tightening tool cadence raises barriers to rapid capacity adds) and specialized module/substrate suppliers that capture mix-upside as designs shift to higher-bandwidth packaging.

Main risks are technological and timing mismatches: software-led efficiency gains (quantization, pruning, sparsity-aware runtimes) can compress memory needs per inference while the hardware replacement cycle for data centers is multi-quarter to multi-year, creating a timing pitfall where bookings can lead or lag real ASP recovery. Macro and inventory shocks—OEM destocking or a competitor accelerating capacity—can flip margins quickly; watch server OEM billings and spot DRAM/NAND pricing as high-frequency signals over the next 4-12 weeks.

Trade implementation should separate beta from idiosyncratic execution: play conviction in Micron’s product mix with asymmetric structures that protect against near-term cyclical snaps while keeping long-term upside. Use pair structures to isolate share-gain narratives vs broad semicap or consumer-NAND weakness. Keep catalysts-driven time buckets: near-term (earnings, 0-90 days), medium (capex cadence, 3-12 months), and secular (architecture shifts, 1-3 years).

Contrarian read: the market may be under-discounting the probability that software efficiency materially slows unit growth while simultaneously over-penalizing capex increases as purely dilutive; the sweet spot to exploit is where Micron’s roadmap for HBM and packaging compounds ASPs even if overall bit growth decelerates. Key reversal triggers: OEM order cadence, spot ASP momentum, and any surprise design-win reversals—monitor daily/weekly for rapid position adjustments.

Key Decisions for Investors

• Long MU equity + protective puts: Buy MU (size = 1-2% NAV) and purchase 3-month 5% OTM puts to cap left-tail risk (cost ~2-4% of position). R/R: asymmetric — retains full upside into secular AI cycle while limiting drawdown from a cyclical ASP reset over the next 90 days.
• LEAP call for secular upside: Buy MU Jan 2028 LEAP calls (delta ~0.35) to capture multi-year HBM/DRAM content gains with limited capital outlay. Timeframe 12–36 months; stress case: loss limited to premium, upside >2x if Micron wins sustained design share.
• Pair trade to isolate share gain: Go long MU / short equal-dollar SMH (or short 000660.KS for direct competitor exposure) to express Micron-specific outperformance versus broader semiconductor cyclicality. Timeframe 3–12 months; risk if sector rallies broadly, neutralize with beta-adjusted sizing.
• Tactical hedge against model-efficiency shock: Buy short-dated GOOGL puts (30–90 days) as a hedge if you own a large MU position, since outsized software efficiency rollouts could compress hardware reorder cycles—puts protect portfolio-level exposure to demand deflation within 1–3 months.