Micron: the quarter the cost line did not move

Micron reported $41.46B of revenue for the three months to 28 May 2026, against $9.30B in the same quarter a year earlier and $37.38B for the whole of fiscal 2025. Over that same year its cost of revenue went from $5.79B to $6.40B. Almost the entire difference fell through to gross profit, which is how a company that makes memory chips printed an 84.56% gross margin. This is five readings of that, and none of them is a view on what happens next.

Micron files a 10-Q every quarter like everybody else, and the one it filed on 25 June 2026 contains a number that stops you: revenue of $41.46B for the three months to 28 May 2026, against $9.30B for the same quarter a year before.

A figure that size is usually a mistake. It is the shape a filing takes when a year-to-date total has been read as a quarterly one, so it is worth saying at the top that this one is not. The company reported $78.96B for the nine months and $41.46B for the three, separately, in the same document. The quarter really was bigger than the whole of the year before it, which Micron closed at $37.38B.

What follows is five readings of that, taken from the company’s own filings on one afternoon. None of them is about what happens next.

1. 01

   The cost line that did not move

   The earnings tab prints every quarter Micron has filed, which is 68 of them, and puts the latest one beside the average of all of them. Revenue for the May 2026 quarter was $41.46B against a 68-quarter average of $5.76B. Diluted EPS was $24.67 against an average of $1.28.

   

   Close

   The interesting half is not on this page, or on any page here, so it comes straight out of the 10-Q. Over the same year the revenue line quadrupled, the cost of revenue went from $5.79B to $6.40B. That is a rise of 10.5% against a rise of 345.7%.

   This is what a memory cycle is, stated as arithmetic rather than as a story. The cost of producing a bit of DRAM is set by fabs that were built years ago and does not move much from quarter to quarter. The price a bit sells for moves enormously. So revenue is very nearly a price index with a fixed cost base underneath it, and when the price triples almost the whole of the increase arrives as gross profit.

   The financials tab carries the fiscal-year version of that line back to 2009, and it is the part worth looking at twice, because the shape either side of the peak is the point.

   

   Close

   • Fiscal 2018 was the previous peak, at a 58.87% gross margin on $30.39B of revenue.
   • Fiscal 2023 was the trough, at a gross margin of negative 9.11% and an operating margin of negative 36.97%. Micron sold $15.54B of product that year for more than it received.
   • Fiscal 2025 closed at 39.79%, which was already a recovery, and is less than half the margin the May 2026 quarter carried.

   Thirteen quarters separate the worst gross margin in the series from the best one. That is the fact to keep hold of through the four sections below, because every one of them reads differently depending on which end of that span you think you are standing at.

   The read: Reassuring

   Micron's revenue rose 345.7% over the year to May 2026 while its cost of revenue rose 10.5%, from $5.79B to $6.40B.

   Almost the entire difference fell through to gross profit, which is how the quarter produced an 84.56% gross margin against a previous best of 58.87% in fiscal 2018. Both figures are the company's own, and the second is on the page.

2. 02

   A trillion dollars at a P/E of 22

   Somewhere in the middle of all this Micron became a trillion-dollar company. The overview page prices it at $977.41, delayed, from the close on 10 September 2026, and puts the market cap at $1.1T.

   The P/E ratio beside it is 22.17, which for a company whose earnings grew the way section one describes looks unremarkable. The line underneath is the one that does the work.

   

   Close

   Against its own record the company is 45.56% more expensive than usual, on a median P/E of 15.2. That inverts the impression the bare multiple gives, and it is worth being precise about why the two disagree. A P/E is earnings in the denominator, and Micron’s earnings are at the top of the range they have ever occupied. The same denominator was negative in fiscal 2023 and produced no usable ratio at all, which is why the median behind that sentence is drawn from four filed years rather than seventeen.

   So the page is comparing a multiple built on the best earnings in the company’s history against a median built from the handful of years in which there were earnings to divide by. Both halves of that are on the panel, and neither of them says which way the cycle turns next.

   The read: Worth watching

   Micron's market cap is $1.1T and its P/E ratio is 22.17, which the page reads as 45.56% more expensive than its own median of 15.2.

   A modest multiple on peak-cycle earnings is a different fact from a modest multiple, and the comparison that catches it is against the company's own history rather than the market's. The median behind it rests on four filed years.

3. 03

   What Micron did with the cash

   A company earning at this rate has to do something with the proceeds, and the cash flow panel is a plain account of what Micron did. In fiscal 2025 it generated $17.53B of operating cash flow, spent $15.86B of it on capital expenditure, paid $522M in dividends and bought back nothing at all.

   

   Close

   $0 of buybacks against $17.53B of operating cash flow is the row that stands out, because a zero in that column is rare for a company this size and it is a choice rather than a constraint. Free cash flow for the year was $1.67B, the residual after capital expenditure took nine tenths of the total.

   The comparison that gives it meaning is the last time Micron was here. Fiscal 2018 carried the 58.87% margin from section one. In fiscal 2019, with the gross margin already falling to 45.72%, the company spent $2.73B buying its own shares. The filings record both decisions and explain neither, which is the ordinary condition of reading them: a cash flow statement says where money went and never why somebody sent it there.

   The same tab scores Micron 7 of 9 on the answerable Piotroski checks for fiscal 2025, a published method from 2000 applied to the company’s own filed figures rather than this product’s opinion. The two it fails are a current ratio that fell from 2.64 to 2.52 and a diluted share count that rose from 1,118,000,000 to 1,125,000,000. The second of those is the buyback row again, arriving from a different direction.

   The read: Both

   Micron spent $0 on buybacks in fiscal 2025 and put 90.48% of its operating cash flow into capital expenditure.

   Fiscal 2019, the year after the last margin peak, went the other way. Which of those is the better decision is not in any filing, but the difference between them is.

4. 04

   Who was holding, and who was not

   The slowest of the documents is the 13F, which reports what a manager held on the last day of a quarter. The overview prints how old the data is before it prints any of the data, and at the time of writing that was 79 days.

   

   Close

   • Coatue Management disclosed $3.63B, which is 7.46% of a $48.63B book, after adding 1,793.72% to the position over the quarter.
   • Appaloosa disclosed $1.13B and trimmed 41.44%. At 15.06% of its book it is the largest Micron position here as a share of the manager holding it.
   • Gotham Asset Management trimmed 47.99% to $147M, and the Sequoia Fund did not move at all: $2.5M, 0.04% of its book, the same 2,162 shares it reported in its Q3 2024 filing.

   The absence is worth as much as the presence. Aquamarine Capital last disclosed a Micron holding in its Q3 2025 filing; the three 13Fs it has filed since list seven positions each and Micron is in none of them. A 13F says nothing about why a manager left, or when inside the quarter, or at what price, and I wrote a longer walkthrough of what a 13F leaves out that applies to every line of this section.

   The read: Both

   Six of the 47 tracked managers disclosed a Micron position in their Q2 2026 13F, and the list changed shape rather than size.

   Coatue Management added 1,793.72% to reach $3.63B while Appaloosa trimmed 41.44%. Every figure here describes 30 June 2026 and was published up to 45 days later, so any of it may already be wrong.

5. 05

   The one sale that was not scheduled

   Micron has 508 Form 4 filings on file. Against a year in which the stock did what section two describes, the headline available here is that insiders sold 355 times and bought three times, and that headline is close to worthless without the next panel.

   

   Close

   Since 2023 a Form 4 carries a box saying whether the trade ran under a Rule 10b5-1 plan, adopted in advance at a time the insider was permitted to trade and executing on a schedule afterwards. $336M of the selling here ticked it. $56.88M did not. Not one of the filings is unlabelled, which is unusual and means nothing on this page has to be guessed at.

   The per-person view is where the split stops being an aggregate. The three insiders whose holdings fell furthest over the twelve months are the Chief People Officer at 41.51%, the Chief Financial Officer at 36.45% and the Chief Legal Officer at 36.44%, and every dollar of all three was sold on a plan.

   

   Close

   Which leaves the discretionary column, and inside it one person worth naming. The Executive Vice President and Chief Business Officer sold $34.87M across the year with no plan behind any of it, taking his stake from 273,021 shares to 191,021, a fall of 30.03%. The largest single filing in that run was 15,000 shares at $934.29 on 18 August 2026, about $14.0M, with the 10b5-1 box not ticked.

   The last row of the same table is the one that stops this becoming a story. A director finished the twelve months with 25,910 shares against 2,710 at the start, up 856.09%, having sold nothing. The table is not unanimous, and a panel that only counted sellers would not have shown you that.

   So the readable part is one executive selling on his own initiative and one director holding a great deal more than he did. Both are decisions rather than schedules, they point opposite ways, and neither is a reason, because a Form 4 records a trade and never why. The stance here is neutral for the plainest possible cause: the large numbers are uninformative by construction and the informative numbers disagree.

   The read: Not a signal

   Of the 40 open-market sale filings by Micron insiders in the year to 21 August 2026, 26 ran under a Rule 10b5-1 plan and 14 did not.

   The planned ones carry $336M and say almost nothing about what anybody thinks now. The $56.88M decided without a plan is the readable part, and a sale is still not a reason.

The four tabs this post used are the quarterly earnings table, the filed statements, the insider transactions and the holder list on the company overview. The managers in section four have their own pages under the investor list. Everything above describes documents that were already public when it was written, and the dates are on every figure for the reason that matters most here: a cycle is only ever read from somewhere inside it.