Micron Guided to $50 Billion and the Memory Glut Trade Died With It

The Guide, Not the Beat

The headline numbers were already strong: fiscal Q3 revenue of $41.46 billion against roughly $35.25B expected, non-GAAP EPS of $25.11, and a record 84.9% gross margin — Micron's fifth straight quarterly revenue record. But a beat on a stock that had already tripled into the print needed more than a clean quarter to justify a 20%-plus move. The thing that actually repriced MU was the fiscal Q4 guide: about $50 billion in revenue (±$1B), some $7 billion above the ~$43B consensus, with gross margin guided to ~86% and non-GAAP EPS near $31.00. That is not a company climbing the next rung of a normal up-cycle — it is a step-function in both volume and margin, guided a single quarter out.

The Cycle Argument Just Got Harder to Make

Memory has historically traded at single-digit multiples because of the glut reflex: every shortage pulls in capacity that eventually craters pricing. Micron spent this print attacking that reflex head-on. It disclosed 16 strategic customer agreements, 14 of which carry roughly $100 billion in minimum contracted revenue, mostly on five-year terms running calendar 2026 through 2030. HBM is fully booked, with management saying it expects the supply shortage to persist well beyond 2027 and HBM4 12-high ramping roughly twice as fast as HBM3E. Layer on the Anthropic agreement signed June 22 — multi-year HBM, DRAM and SSD supply plus a stake in Anthropic's $65B Series H — and the demand side stops looking like a spot market and starts looking like a backlog. Contracted minimums don't repeal cyclicality, but they put a floor under it that the old model never priced in.

What the Street Is Actually Underwriting

The analyst response makes the re-rate explicit. Needham went from $500 to $1,550, valuing MU at 10x its fiscal 2028 non-GAAP EPS estimate of $155 — a bet on 2028 earnings, not 2026. BofA reset to $1,500 from $950, and UBS ($1,625), Stifel and TD Cowen ($1,500), and Bernstein and Wedbush ($1,300) are all clustered far above where the stock traded a month ago. The counterpoint is the obvious one: targets anchored to 2028 estimates assume the shortage holds and that Samsung and SK hynix don't flood DRAM and NAND on the way up, and a name that has run this hard already discounts a lot of that. The contracted-revenue disclosure is the bull's direct answer to that risk — the open question is whether minimums underwritten today survive the next capacity wave.