$50.9bn in a 14-Week Quarter (Micron Earnings Preview, 30 September 2026): What to Expect and Why the Next Guide Will Look Smaller Than It Is

Micron reports fiscal fourth-quarter 2026 results after the US close on Wednesday 30 September, and LSEG consensus expects $50.9 billion of revenue and $31.49 of adjusted earnings per share against the company's own guide of $50.0 billion ±$1.0 billion and $31.00 ±$1.00. That puts the Street within $0.1 billion of the top of Micron's range before a single number is printed. The detail most previews skip is on page one of the quarterly filing: this quarter runs 14 weeks, not 13. That flatters the print, and it means the next guide will look smaller than the underlying business it describes.

When Micron reports, and what the numbers are

Micron confirmed on 26 August that it will report after the close on Wednesday 30 September, with the call at 2:30 p.m. Mountain time. The guidance it has to meet is the set it issued with third-quarter results on 24 June. The consensus is the LSEG figure reported by CNBC on 25 September.

The per-week row is this note's arithmetic, not a company figure; it divides each revenue number by the weeks in its quarter. A year earlier the same quarter produced $11.3 billion in 13 weeks, so the consensus implies a quarter about 4.5 times larger. The stock has carried that. CNBC reported on 25 September that the shares were up nearly 280% this year but still more than 10% below the record close of about $1,213 reached on 25 June, the day after the last report. That report sent the stock up 15% in extended trading because revenue beat LSEG consensus by $5.6 billion and the guide beat by more than $6 billion. The setup this time is different. Consensus has moved up to sit on top of the guide rather than below it, so there is less room for the printed quarter to surprise.

The mechanism: the margin is almost all price now

Memory margins normally tell you about the cycle. At these levels they mostly tell you about pricing. Micron's non-GAAP gross profit was $35.2bn on $41.46bn of revenue in the third quarter. The June guide of about 86% on $50.0bn implies gross profit of about $43.0bn. That is $7.8bn more gross profit on $8.5bn more revenue, an implied incremental margin of roughly 91%. Cost of goods sold was $6.40bn over 13 weeks, about $0.49bn a week. The guide implies about $7.0bn over 14 weeks, about $0.50bn a week. So revenue per week is guided up 12%, while cost per week is almost flat.

The implication for Wednesday is specific. When cost is only 14 cents in the revenue dollar, the percentage margin has very little room to move, while gross profit in dollars keeps moving one-for-one with price. Each point of gross margin at $50bn of revenue is about $0.5bn of gross profit. Micron's June figures imply a conversion of roughly 86% from operating income to net income on about 1.15bn diluted shares, which puts one margin point at roughly $0.37 of EPS. That is about a third of the width of the ±$1.00 guidance band. The margin line will therefore look almost boring, in the mid-to-high 80s. The information is in the revenue per week and in what management says about contract pricing for the next year.

That chain is why Micron's own business units moved together in the third quarter even though they sell into very different markets. Core data centre revenue rose more than sevenfold year on year to $11.52bn, but mobile and client also reached an 87% gross margin, and automotive and embedded reached 79%. When capacity is pulled toward the AI products, every other product becomes scarce as well. The same dynamic sits behind the laptop makers turning to Chinese DRAM this summer.

The three scenarios

Consensus has already absorbed the quarter, so the scenarios turn on the guide and the call, not on the print.

1. The per-week rate rises into December. Revenue lands at or above $51bn, and fiscal Q1 2027 guidance implies more revenue per week than the fourth quarter, which means roughly $47bn or more for 13 weeks. This is the version where the shortage is still tightening. It would reach the semiconductor complex through the pricing read-across, and it would support the argument, made in CNBC's 25 September week-ahead piece, that higher yields are coming from strong growth, with AI spending as one of the engines.

2. The per-week rate holds, but the headline falls. The print is in range and the December guide comes in somewhere around the mid-$40bn range, flat or slightly lower per week, and below the fourth quarter as reported. The fundamental information is neutral; the headline looks like a sequential decline. The first after-hours move may be driven by the headline comparison before the per-week comparison catches up. That gap is the mechanism worth understanding, whichever way the stock moves.

3. Pricing or contracts disappoint. Gross margin prints below 86%, the guide implies a lower per-week rate, or commentary suggests the contract agreements are locking in prices lower than spot. Because the incremental margin is almost all price, even a small pricing change moves earnings much more than it moves revenue. This is the scenario the memory cycle has historically delivered at the top, and it is what the contract programme was designed to soften.

The contract question: $22bn committed, $5bn audited

The newest variable in Micron's model is the one the market understands least. The June filing describes the strategic customer agreements as take-or-pay, with binding volume commitments over multi-year terms. Pricing is fixed or banded for most agreements and floats with the market for a minority. Micron said on the June call, as CNBC reported, that it had signed 16 of them, lasting three to five years, and expected $22 billion of cash deposits and related financial commitments, about $18 billion of that in cash. Mehrotra said that when the programme is complete he expects approximately half or more of company revenue to sit under these agreements.

The filing also gives the conservative version of the same number. As of 28 May, remaining performance obligations were approximately $5 billion, measured on minimum committed volumes at minimum prices. About a third of that was expected to be recognised within twelve months. The measure excludes agreements with no price band and agreements signed after quarter end, and the filing warns that it is not expected to indicate future revenue. The gap between the headline commitments and the audited floor is the right frame for Wednesday. More agreements, more committed volume and a larger floor would make the next downturn shallower by contract. A programme that stalls at 16 would leave the business exposed to spot pricing, which is the part of memory that has always reversed.

What the Street is watching lines up with this. In the same CNBC preview, UBS said its checks showed the supply gap continuing to widen into calendar 2027, with DRAM fulfilment at about 60%, and Wells Fargo put the share of revenue expected to be supported by minimum-pricing commitments at roughly 40%. Both are analyst views, not company disclosures. They show that the market's question is how long the shortage lasts, not whether Micron has a good quarter.

The channel to what you actually trade

For equity indices the channel is direct. Micron's market value passed $1 trillion in June, so its after-hours move shows up in NAS100 and US500 futures in the overnight session. The wider effect is the read-across. Micron reports in US hours under US securities law, while Samsung and SK hynix report later on Korea's calendar, so Wednesday night's guide is the first priced memory data point for Seoul's open on Thursday. The KOSPI's dependence on those two names, and the leveraged products built on them, is traced in the note on the August memory selloff. The same guide also tests the September argument that AI demand is slowing. Micron is the first supplier to issue guidance under securities law since that debate, laid out in the AI slowdown note, began.

For currencies this is not a first-order event, and treating it as one would be dishonest. The backdrop is a bond market that CNBC reported took the 10-year Treasury yield briefly to 5.23% this week, the highest since June 2007, with the Nasdaq and semiconductors advancing while cyclicals lagged. The dollar's direction is set by the August PCE release at 8:30 a.m. on the same Wednesday and by Friday's payrolls, so the dollar page and the five factors behind it are the right place to read that leg. The one real crossover is risk sentiment, which reaches the yen through its safe-haven role; that is on the yen page. A large miss in the market's biggest AI supplier story could tighten risk appetite quickly enough to show up there.

What would change the picture

There are six things to watch on the call. First, the December-quarter guide measured per week, since the 13-week comparison is the only fair one. Second, gross margin against the 86% guide, given that a point is worth about $0.37 of EPS. Third, whether the number of strategic customer agreements has grown past 16, and whether the annual report that follows shows the audited floor rising from $5 billion. Fourth, any capital-return commentary, which UBS told clients it would be listening for. Fifth, capital spending, which the June filing shows had already reached $19.60 billion over nine months. That spending is where the eventual supply response starts. Sixth, any change to management's view that industry supply improves only gradually in 2028, because that date is currently the shortage's stated end date. Semicon West runs in San Francisco from 13 to 15 October, and equipment makers there will give a second read on how quickly that capacity is coming.

The quarter itself is already known in outline. The calendar makes its successor look smaller than it is, and the contracts will show how much of the current pricing will still be in place when supply catches up.

More on the method behind these notes is on the about page.

Educational macro context only — not investment advice.