Micron Guided Q4 to $50 Billion. The Street Was at $44.

The Guide Is the Story

The trailing print was a blowout on its own: $41.5 billion in revenue, up 346% year over year and 74% sequentially, an 84.9% gross margin, and adjusted EPS of $25.11 that beat consensus by $4.62. But a beat that size was largely expected — Micron went into the report at record highs with a ~17% implied move already priced. The repricing didn't come from the quarter that just closed; it came from the one being guided.

Micron guided fiscal Q4 to roughly $50 billion in revenue, plus or minus $1 billion, against a Street that was sitting near $44 billion. Gross margin is guided to ~86% and diluted EPS to $31 ± $1. That is another ~20% sequential step *on top of* a quarter that already grew 74% sequentially. The beat moved the stock; the guide moved the model.

Why the Number Holds

The reason a $50 billion guide is credible rather than promotional is that most of it is already sold. Micron is sold out of HBM for all of 2026, with no incremental high-bandwidth capacity left to offer this year. Behind that, the company disclosed 16 take-or-pay agreements worth roughly $100 billion in minimum contracted revenue through 2030, backed by about $22 billion in fixed-price commitments and deposits.

The rest of the mix supports the trajectory: a data-center run-rate now exceeding $100 billion annualized, and HBM4 already past $1 billion shipped while still in its early ramp. The same forecast strength, echoed across the supply chain, is what ignited a roughly $400 billion AI-chip rally on the print. When the next quarter is contracted at fixed prices, the guide is closer to a backlog readout than a spot-price bet — which is precisely the dynamic that historically made memory impossible to model.

What's Already Priced In

The counterweight is the chart. MU is up about 285% off its March low of $311.50 and more than 700% over the past year, and the analyst desks have split into a price-target war — bulls at $1,625 to $1,750 against one bear holding a $477 target. The reason the bear case won't die, even on numbers this good, is that memory has always mean-reverted once supply catches demand; the multiple stays compressed because the market refuses to pay a structural-growth price for a cyclical business.

The contracts push that reckoning out to 2027 and beyond — they don't delete it. My read: at $1,227 the guide justifies the move on a 12-month view, but the easy re-rate is behind it. The cheap part of this trade was buying a sold-out memory cycle the market was still pricing as cyclical. From here it's an execution-and-pricing story, not a discovery story, and the asymmetry is no longer obvious.