A beat inside a longer quarter
Micron reported revenue of $54,229M after the close on , up 379% on a year ago and $3,229M above the top of a ~$50B ± $1B guide. Non-GAAP EPS was $33.42, $1.42 above the guide's $32 top, and $1.92 above a vendor consensus of $31.50 (unverified). The first-quarter guide is $61.5B ± $1.5B and $38.15 of EPS, $2.79 above consensus (unverified). The stock closed the regular session at $1,068.95 and traded $1,067.95 that evening (unverified, market data). Nothing moved.
One reason the tape shrugged is on the calendar, not the income statement. The quarter ran from to , 98 days, a 14-week quarter against 13 in each of the three before it. Read per week, revenue grew 74.9% in the second quarter, 73.7% in the third and 21.47% in the fourth. The headline 30.81% sequential gain carries a week the first quarter will not have, so the guide's 13.41% looks like a stall and is not: over an assumed 13 weeks (unverified), the $61.5B midpoint is $4,730.8M a week, 22.13% above the fourth quarter's rate. The house rule on single periods applies here: a one-quarter move is a phase until the series says otherwise, and this series says the deceleration already happened, in the quarter the market just cheered, and the guide holds the new pace rather than slipping from it.
The numbers
Price did most of it again, at a slower rate. DRAM revenue was $39.8B, 73% of the total, on bit shipments up mid-single digits and prices up high-teens percent; in June it had been prices up in the low-60s. NAND was $14.1B, up 42% sequentially, bits up about 10%, prices up about 30%. Bit growth of that size in a quarter one week longer is a factory running flat out, which is what management says: "we do not have line of sight to when supply and demand will return to balance."
Non-GAAP gross margin reached 87.0%, up 210 basis points; GAAP was 86.76%, and 69.52% of revenue reached net income. Core Data Center grew to $18,002M at a 90% gross margin and became the largest unit, ahead of Cloud Memory at $16,283M and 83%; together the two data-center units were 63.2% of revenue. Data center SSD revenue was "nearly $10 billion." HBM "grew faster than total company revenue," and Micron has "completed agreements for the vast majority of our calendar 2027 HBM bit supply with significant price increases year over year, narrowing the gross margin gap with conventional DRAM." Operating expenses rose $1.1B, on incentive pay and a $300M community contribution; GAAP also carries a $500M patent license charge.
The rest of this dive is for paid subscribers.
The headline numbers are above. The 7 sections below carry the mechanism, the peer read, the valuation work, and the dated tests that decide it.
- The deposit is the story
- When the money goes back
- Grading June
- The strongest case against me
- What decides it
- The trade (analysis, not advice)
- Bottom line
Already a subscriber? Sign in.
The deposit is the story
Mark Murphy, the chief financial officer, gave the line the whole quarter turns on: "customer cash deposits associated with SCAs are reported within financing activities and therefore do not affect our free cash flow. Customer cash deposits received during fiscal Q4 were $12.3 billion." Decode it. Customers that signed Micron's strategic customer agreements, take-or-pay contracts on volume with a price floor under most of them, wired Micron more cash in one quarter than Micron spent on fabs. On the balance sheet, noncurrent customer contract liabilities went from $568M in May to $12,895M, a rise of $12,327M, 114.4% of the quarter's $10,774M of net capex.
The house has a model for a customer funding a supplier's build, the customer became the lender, and its reference case was Oracle, where the prepayment arrived inside operating cash flow and no leverage screen could see it. Micron's version is the visible one. The money sits in financing, adjusted free cash flow of $33,199M is clean of it, and management says outright what the money is: "SCA cash deposits are unrestricted and will be returned to customers over time, toward the latter half of each agreement's term, assuming minimum purchase requirements are met." That last clause is the mechanism. A deposit that is returned only if the customer takes its minimum volume is the enforcement behind a take-or-pay floor. The customer is not prepaying for memory. It is posting collateral against its own demand forecast.
That answers the house's standing question, who ultimately pays for the capex, more directly than any memory print before it. Commitments across the 26 agreements rose to "$32 billion, the vast majority of which are cash deposits." With $12.7B on the balance sheet, about $19.3B of committed money has not arrived yet, not all of it necessarily cash. Micron guided net capex of around $11.5B in the first quarter and about $25B for the first half, higher in the second, which puts fiscal 2027 above $50B, 82.7% above fiscal 2026. "A majority of the increase is for construction capex, most of which is to help accelerate clean room space availability in late calendar 2028 and beyond."
When the money goes back
Line the two calendars up. The Idaho ID1 fab starts wafer output in mid-calendar 2027, ID2 in late 2028, the Japan expansion in late 2028, the Singapore NAND fab in the second half of 2028 and New York in 2030. The agreements signed in June were mostly five-year terms over calendar 2026 to 2030, per the June call, and now some run into 2031. "Toward the latter half of each agreement's term" is roughly 2028 to 2030. The deposits go back as the capacity they helped pay for comes on. A commitment is an allocation claim before it is a cost: for the buyer, the deposit reserves wafers other buyers cannot have; for Micron, the refund falls due in the same window its own new supply lands in the market. That is fine at a floor price with margins that management says stay "meaningfully above any prior cycle peak margins." It matters only if the new supply arrives into a softer market, and even then the minimum-purchase condition makes the customer, not Micron, the party that forfeits.
Grading June
The June dive set four things to watch. Remaining performance obligations were guided to about $100B across 16 agreements; Murphy now puts them at "approximately $150 billion" across 26, on committed volumes at minimum prices. The 10-Q for the May quarter had recorded only about $5B, because it counted agreements in force at the quarter end and Micron was still signing others after it closed; the fiscal 2026 10-K is the first filed number on the new base. Deposits were expected at "roughly $10B" in the fourth quarter and came in at $12.3B. Price moderation arrived as flagged: DRAM prices rose high-teens percent against low-60s in June. And the margin call held: gross margin 87.0% against a ~86% guide, with the first quarter's 86.25% named "the floor for gross margins in fiscal 2027." June's read was right on all four; the part it underweighted was the cash.
The strongest case against me
Here is the other side at full strength. Micron does not need anyone's money. It ended the year with $73.5B of cash and investments, $68.3B net of debt, and produced $33,199M of adjusted free cash flow in one quarter, about three times the quarter's net capex. The deposits went straight into long-term marketable investments, which rose $25,913M. Calling them funding for the capex is a story about fungible cash: Micron would build the same fabs without a dollar of deposit, and the deposit's only job is to make the take-or-pay contract enforceable. On this reading there is no lender, only a security deposit, and the refund schedule is housekeeping.
I agree with most of that and still land the other way on what it means. The point is not that Micron needs the money; it is that the customer was willing to part with it. $12.3B in one quarter is the buyer's own balance sheet underwriting the demand behind the agreements, which is the thing memory has never had in any prior cycle. Strip the deposits and net cash is still $55.6B; there is no solvency question here. The question is whether that willingness keeps arriving at the rate capex is rising, because the $19.3B still to come is the difference between a one-quarter signing bonus and a financing model.
What decides it
One line, on one date: noncurrent customer contract liabilities on the balance sheet for the first quarter of fiscal 2027, reported around (unverified, tentative date). Above $18.0B, at least $5.1B more arrived, more than a quarter of the committed money still outstanding, in the quarter capex steps up to around $11.5B, and customers are paying into the build as it grows. No higher than $12,895M, no new money came in, the fourth quarter was a signing lump, and I'm wrong. Between the two is a partial result, and I'd call it that. The secondary checks are the RPO figure in the fiscal 2026 10-K against $150B, and first-quarter gross margin against the 86.25% floor.
The trade (analysis, not advice)
At $1,067.95 (unverified), MU trades at 14.14x fiscal 2026 non-GAAP EPS of $75.52 and 7.0x the first-quarter guide annualized, $152.60. It sits 12.2% below where it traded after the June print. The market is pricing the peak, not the floor, and a flat reaction to a guide that cleared consensus by 7.9% says the debate is no longer about the next quarter. It is about what the earnings are worth once supply catches up, and the deposits are the best evidence on that question. For a holder, the deposit line is the thing to track before the price line. For someone looking in from flat, the December balance sheet answers more than the December guide will. For the skeptic, the argument that remains is the late-2028 supply wave, and the deposits already put a date and a counterparty on it.
Bottom line
The beat was big and partly calendar, and per week the growth rate already stepped down to about 22%. What changed is the cash behind the contracts: $12.3B of customer money in one quarter, more than the capex, returned only if the customers take their minimums. Right if the deposit line is above $18.0B at the first-quarter balance sheet. Wrong if it is no higher than $12,895M.