Eighty-Seven Percent and the Stock Went Down

Eighty-seven percent. Read it again. A memory company, a business that spent most of its life selling a commodity to buyers who could swap suppliers over lunch, just reported an 87% gross margin and an 82.3% operating margin, and the stock shrugged. Down about 1% in the premarket. Barely moved, in the polite language of the wires.

Let me lay out what it shrugged at. Revenue of $54.23 billion against roughly $51.5 billion expected. The same quarter a year ago was $11.32 billion, so go ahead and sit with a 379% jump for a second. EPS of $33.42 against a Street number around $31.60. Fiscal 2026 revenue closed at $133.2 billion. The guide for the next quarter: $61.5 billion and $38.15 a share, gross margin around 86.25%, which management describes as the low point of fiscal 2027. And the shares, already up 273% this year, responded like a man who was told his lottery ticket won again.

So here is what the market said, since nobody else will say it out loud. It said: we saw the number, we have the number, we are not paying for the number. Annualize that $38.15 guide (my arithmetic, check it yourself) and you land near $150 a share. The stock trades around $1,050. Roughly seven times. Seven times earnings for a company printing 87% margins with most of next year already sold. When a market charges seven times for a growth story, it has already booked the funeral and is haggling over the flowers.

Memory has done this to everyone before. Every upcycle arrives with the same sermon: discipline, structural demand, a new regime, this time the suppliers learned. Then prices rise, every fab owner on earth pours concrete, and three years later the supply lands on a demand curve that stopped climbing the quarter before. The cycle never needed to be repealed. It only needed time and a construction schedule.

The tell lives in the guidance, not the beat. Micron is raising capex. Roughly $25 billion in the first half of fiscal 2027 alone, more in the second half, with the growth tilted toward construction (cleanroom shells for DRAM) rather than tools. Management says supply and demand stay tight through fiscal 2027 and 2028, and I believe them. Cleanrooms poured this year become wafers in 2028. The company telling you the shortage has two more years to run is, in the same breath, building the end of it. Seoul has been voting on this for weeks: as of Monday's rout, SK Hynix was down about a third and Samsung about a fifth over three months. Assume everybody with a fab has read the same price sheet.

Now the part I love. Long-term supply agreements rose to $32 billion from $22 billion in a single quarter, largely backed by customer cash deposits, and management is nearly giddy about the visibility. Fine. Look at who is wiring the deposits. Customers prepaying to lock in chips they cannot otherwise get, in a week when the ten-year Treasury touched 5.34%, the highest since 2002, and the AI build-out was out competing for the same long-dated money as every sovereign on the planet. A prepayment is a loan from the customer no matter what the contract calls it, and a contract is only as strong as the balance sheet signing it. Visibility is a lovely word until you ask who is carrying the paper at the other end.

And here is the cruelty of the rate backdrop. The bull case says through-cycle earnings power sits far above anything memory has seen, so the multiple on "peak" profits deserves to expand. I have heard that sentence at three prior tops. At a 5.3% risk-free rate the sentence gets expensive to believe. Every dollar of those earnings is discounted against a Treasury that now pays you handsomely to own nothing at all, and the stock price already embeds the possibility that nobody ever sees profits like these again.

Friday made it weirder. The Nasdaq 100 closed at a record, 30,807.93, up 1%, and the index partied through a weak jobs report as if yields had not just printed a 24-year high a day earlier. Against that backdrop, the one company that actually showed the receipts for the AI trade got a collective shrug on Thursday. The index will pay up for a story. Apparently it gets pickier when handed an income statement. Read that twice if you hold the group.

I am not telling you to short it. A stock that falls on a miss is information about the miss. A stock that fails to rise on a monster beat and a raised guide is information about the buyers, and the information is that the buyers have already bought. Everyone who wanted exposure to this cycle has it. Everyone who has it is waiting for a reason to leave. A beat of this size was supposed to be the reason to stay, and it wasn't enough to make the chart blink.

Eighty-seven percent margins and a seven multiple. One of those numbers is lying about the other, and I know which one history says will mean-revert.