Micron Turned $10,000 Into About $130,000 in 5 Years. Most of It Came in the Past 12 Months.

Micron reported fiscal Q3 2026 revenue of $41.5 billion, up 346% year over year, and net income of $28.2 billion as AI-data-center demand drove memory prices sharply higher. DRAM average selling prices rose about 260%, versus shipment-volume growth of just over 20%, lifting adjusted gross margin to 84.9% from 39% a year earlier. Despite multi-year customer supply agreements, the stock trades at roughly 6x expected fiscal 2027 earnings and remains 22% below its $1,255 52-week high, reflecting investor concern that peak-cycle memory profits may not be durable.

Analysis

MU is no longer trading primarily on AI-memory unit growth; it is trading on whether exceptional pricing can remain embedded in the contract book through the next capacity additions. The key variable is the mix and enforceability of Strategic Customer Agreements: take-or-pay volumes, price-reset clauses, and qualification commitments matter more than headline duration. If agreements lock in HBM and high-end DRAM allocation while spot DRAM normalizes, MU can sustain a materially higher through-cycle margin and deserves a re-rating versus its historical commodity-memory multiple.

The near-term asymmetry is unfavorable for a directional chase despite the low headline P/E. Consensus earnings are likely extrapolating peak pricing, so a modest ASP deceleration—not an outright price decline—could trigger estimate cuts and multiple compression over the next 1-3 months. Conversely, supply is constrained by HBM packaging, wafer allocation, and customer qualification cycles; these bottlenecks make a sharp price collapse less likely before meaningful new capacity qualifies, a 6-18 month risk rather than a next-quarter base case.

Second-order beneficiaries are NVDA and hyperscaler AI platforms only if memory availability ceases to constrain accelerator system shipments; sustained elevated memory pricing instead transfers AI economics from compute buyers toward MU/Samsung/SK Hynix and pressures server OEM gross margins. The contrarian view is that the market may be treating MU as a conventional late-cycle memory peak while underestimating the strategic value of qualified HBM supply. That thesis is falsified if MU guides sequential ASPs down, if HBM bit shipments fail to outgrow the broader DRAM market, or if gross-margin guidance falls materially below the current run-rate before new industry supply is absorbed.

Key Decisions for Investors

• Do not initiate a full-size directional MU long at current levels; use a 10-15% pullback or post-earnings confirmation of stable sequential ASP guidance as entry triggers. Target a 15-25% upside over 6-12 months if contract durability is validated; exit on a sequential revenue or gross-margin guide-down that indicates pricing normalization.
• For existing MU exposure, replace part of common-stock risk with a 3-6 month collar: sell calls 15-20% above spot and buy puts 10-15% below spot. This retains exposure to a supply-duration surprise while protecting against the more likely near-term estimate-reset volatility.
• Watch MU versus NVDA as a relative-value signal rather than shorting either outright: widening MU outperformance alongside rising memory ASP commentary supports the memory-bottleneck thesis; a reversal after NVDA shipment commentary would indicate that memory pricing is beginning to impede system demand.
• Require disclosure of customer-concentration, contract price-reset mechanics, HBM mix, and capex/wafer-start plans before underwriting a structural multiple expansion. Absent those data, treat the low forward P/E as peak-earnings risk rather than a standalone value signal.