Micron Guided to an 86% Gross Margin. No Memory Maker Has Ever Held a Number Like That for Long.

Micron guided for fiscal Q4 gross margin of about 86% on $50 billion, plus or minus $1 billion, of revenue, following an 84.6% margin in fiscal Q3 driven by AI data-center memory demand and constrained supply. Management said price increases are moderating, but cited 16 three-to-five-year take-or-pay agreements with pricing floors and ceilings as support for margins above prior-cycle peaks. The article cautions that Micron's previous margin peaks—61% in fiscal 2018 and 47.3% in 2021—reversed sharply, while consensus fiscal 2027 EPS of about $158 implies a roughly 7x forward P/E at a $1,070 share price.

Analysis

The key underwriting issue is not whether peak profitability is achievable, but how much of the earnings stream is genuinely contracted versus merely protected by nominal price floors. Take-or-pay terms can reduce volume downside, yet their economic value depends on product mix, qualification specificity, termination provisions, and whether floors cover HBM only or broader DRAM/NAND output. If hyperscaler AI capex pauses, customers may honor minimum commitments while shifting mix toward lower-value configurations, leaving realized gross margin materially below the headline floor implied by management.

The near-term setup into results is asymmetric: a small beat is unlikely to expand the multiple if investors already capitalize peak-cycle earnings at a discount, while any evidence of slower sequential pricing, elevated customer inventory, or incremental supply plans can trigger a sharp de-rating. Over the next 1-3 months, watch FY2027 capex, bit-supply growth, HBM yield/qualification commentary, and the proportion of revenue under enforceable multi-year commitments. The structural 6-18 month risk is that unusually high returns pull forward capacity additions from Samsung and SK Hynix, turning a supply-constrained HBM market into a conventional memory downcycle sooner than consensus expects.

A less obvious beneficiary of a sustained high-margin regime is SK Hynix, whose HBM positioning provides similar pricing exposure but potentially less downside from commodity NAND. Conversely, NVDA is not a clean sympathy long: higher memory content and tighter HBM availability raise system BOM costs, potentially pressuring gross margin or constraining accelerator shipments if supply—not GPU demand—becomes the bottleneck. Consensus may be too focused on the durability of AI demand and insufficiently focused on the reflexivity of supplier capex once margins remain extraordinary for several quarters.

Key Decisions for Investors

• Maintain a tactical underweight in MU into and immediately after earnings; add only if management fails to quantify contracted revenue, floor-price coverage, or FY2027 supply discipline. Thesis horizon: 1-3 months. Falsifier: sustained gross-margin guidance above 80% alongside explicit evidence that contracted HBM revenue covers a majority of FY2027 sales.
• Express downside through a 3-month MU put spread initiated after the earnings event if implied volatility compresses and the stock rallies on an in-line print. Target a structure with roughly 2:1 payoff-to-premium; avoid pre-event naked puts given the risk of another upward guidance reset.
• Consider a relative-value long SK Hynix / short MU position only after comparing HBM revenue mix, valuation, and capacity guidance at both companies' next reports. The trade works if MU's premium embeds a broader and more durable contract shield; close if MU demonstrates superior HBM yield or customer commitments.
• Set an alert for announced DRAM/HBM capacity expansions by Samsung, SK Hynix, or MU and for a sequential decline in DRAM contract pricing. Either signal would shift the risk from peak-margin skepticism to a more actionable 6-18 month memory-cycle short thesis.