Micron Stock Fundamentals 2026: MU Earnings, Cash Flow and Valuation

Micron reported record fourth-quarter and full-year fiscal 2026 results after the US close on 30 September 2026, and the number that reframes the investment case is not in the earnings line. Fourth-quarter revenue was USD 54.23bn, up 379% year on year and 31% sequentially, with non-GAAP diluted earnings per share of USD 33.42 and a non-GAAP gross margin of 87%. Full-year revenue was USD 133.2bn, up 256%. Against that, management guided capital expenditure of roughly USD 11.5bn for the first quarter of fiscal 2027 and about USD 25bn across the first half — within touching distance of the USD 27.37bn Micron spent across the whole of fiscal 2026, in half the time.

Micron's stock fundamentals in 2026 therefore no longer turn on whether memory is tight. Management said plainly that it expects industry supply and demand to be tighter in 2027 and 2028 than in 2026, with DRAM the primary constraint on AI infrastructure, and the company has 26 multi-year strategic customer agreements carrying take-or-pay volume commitments to back that up. They turn instead on cash: how much of a peak-margin cycle gets reinvested in cleanroom construction before it reaches shareholders. The shares closed the regular session at USD 1,065.11, all but unchanged on the day and before the release landed; the initial after-hours reaction was slight, under 1%, and reported inconsistently across outlets. No full session has yet traded on this news.

The earnings line, on the right basis

Every headline figure here is non-GAAP, and the distinction matters more than usual in a quarter this extreme. On that basis, fourth-quarter revenue of USD 54.23bn came in ahead of a consensus near USD 51.07bn, and non-GAAP diluted EPS of USD 33.42 beat an expectation closer to USD 31.45. The company's own framing was that fourth-quarter revenue, gross margin and earnings per share all exceeded the high end of its guidance.

For the full year, non-GAAP gross margin was 81.1% of revenue and non-GAAP net income was USD 86.76bn, or USD 75.52 per diluted share. Note the shape inside that: a full-year non-GAAP gross margin of 81.1% against a fourth-quarter figure of 87% means Micron exited the year running nearly six points above its own annual average. The exit rate, not the average, is what the first quarter of fiscal 2027 is being guided off.

That guide is for revenue of about USD 61.5bn plus or minus USD 1.5bn, with non-GAAP EPS of USD 38.15. Set against the USD 54.23bn just delivered, the midpoint implies a further 13% sequential increase on top of a quarter that was itself up 31% sequentially. It arrived above the consensus the market had been carrying.

High-bandwidth memory is doing a disproportionate share of the work. HBM revenue grew faster than total company revenue in the fourth quarter as Micron ramped shipments across a widening customer base, and the company has completed agreements for the vast majority of its calendar 2027 HBM bit supply at significantly higher prices year on year — which, on management's account, narrows the gross margin gap between HBM and conventional DRAM. That is a structural point rather than a cyclical one: HBM has historically been the volume story and conventional DRAM the margin story.

The cash flow line is where the story turns

Micron spent USD 10.77bn on capital expenditure in the fourth quarter alone, which is 19.9% of the quarter's revenue. For the full year, capex of USD 27.37bn was 20.5% of revenue. Those are high ratios in absolute terms, but they are the ratios of a business whose revenue had just tripled, so they look almost restrained.

The fiscal 2027 plan is a different proposition. Roughly USD 11.5bn in the first quarter, about USD 25bn across the first half, and management has indicated higher spending in the second half still. Chief executive Sanjay Mehrotra tied the increase directly to capacity physics and to contracted demand: "Given the need for DRAM cleanroom space and supported by greater visibility from SCAs into our demand through the end of the decade and beyond, we plan to increase our capex in fiscal 2027 versus prior plans."

The composition matters as much as the number. The majority of the increase is construction, aimed at accelerating cleanroom availability in late calendar 2028 and beyond — not equipment purchases for near-term output. Micron is therefore spending peak-cycle cash today on wafer capacity that produces revenue at the end of the decade. That is a defensible decision if the tightness call is right, and it is the single largest call on shareholder cash that this company has made.

The balance sheet can carry it. Micron closed the quarter with USD 73.48bn in cash, marketable investments and restricted cash. At current margins the programme is fundable out of operations with a very large cushion behind it. The question is not whether Micron can afford USD 25bn a half; it is what return that spending earns and when.

Why the reaction was so muted

A record quarter, a double beat and an above-consensus guide produced almost no move. Coverage of the after-hours session attributed that to the capital spending outlook overwhelming the beat, and the logic holds: a record quarter is a statement about the past twelve weeks, while USD 25bn of first-half capex is a claim about 2028 to 2030. Markets discount the second much harder than they reward the first.

Expectations are the other half of it. A stock at roughly USD 1,065 after a year in which revenue rose 256% has already absorbed a great deal of good news. On the trailing full-year non-GAAP figure of USD 75.52 per diluted share, that price is about 14.1 times earnings. Annualise the guided first-quarter non-GAAP EPS of USD 38.15 and you get roughly USD 152 and a multiple near seven times. That second number is a scale marker, not a forecast — nobody should extrapolate four quarters from one guide in a cyclical industry — but it illustrates the central difficulty in valuing memory. The multiple is at its lowest precisely where the earnings are hardest to extrapolate.

The Openbook read

Growth is as strong as this factor gets. Revenue up 379% year on year, up 31% sequentially, and guided up a further 13% sequentially. Full-year revenue of USD 133.2bn against roughly USD 37.4bn the year before is not a growth rate so much as a change of scale. The factor reads at the top of the range, and the only honest caveat is that a comparison base this distorted will mechanically deflate the score within a few quarters.

Profitability reads near maximum and the forward question is duration rather than level. An 87% non-GAAP gross margin in the fourth quarter is not a normal memory margin by any historical standard. What is genuinely new is the contracting: 26 strategic customer agreements with take-or-pay commitments, estimated by management to represent more than 35% of revenue through 2030, with around three-quarters of that expected revenue sitting under defined pricing frameworks. That is the first substantial evidence that DRAM revenue can be contracted rather than left to spot, and if it holds it changes the amplitude of the next downcycle rather than abolishing it.

Solvency is strong and deliberately being consumed. USD 73.48bn of liquidity against a capex programme running near USD 25bn per half means coverage is comfortable today, but the direction of travel is a steadily smaller cushion for as long as construction spending runs ahead of the capacity it creates.

Momentum is the weak factor, and it is weak for an informative reason. The stock was unchanged in the regular session and moved under 1% after hours on a record print. A result this strong that generates no move tells you the marginal buyer is already positioned. That is a very different kind of weak momentum from a stock falling on bad news, and it is the one most likely to resolve on the next catalyst rather than on the last one.

Reward/Risk has changed shape rather than magnitude. Until this print, the principal risk in Micron was demand. After it, with tightness contracted into 2030 and cleanrooms being poured, the principal risk is reinvestment: the possibility that capacity arriving in late calendar 2028 lands into a looser market than the one that justified building it. The take-or-pay agreements are the mitigation, and they are unusually specific. On the other side, the trailing multiple is low and the contracted revenue base is long. The asymmetry now sits on construction timing and on cycle duration, not on whether AI needs memory.

The read-across

For the semiconductor equipment makers the detail of this guide matters more than the total. Micron is raising capex, which reads as positive for Lam Research and Applied Materials — but the majority of the increase is construction rather than tools. A capex line that rises on concrete does not convert into equipment orders at the same rate or on the same timetable. The equipment spend attached to cleanroom space that comes available in late calendar 2028 is a fiscal 2028 event, not a fiscal 2027 one. The read-across is real but shifted right.

For the memory complex more broadly, management's tightness call through 2027 and 2028 applies to the industry, not to Micron alone, and the pricing achieved on calendar 2027 HBM bit supply sets a reference point every supplier will be measured against.

For the buyers of AI infrastructure the most consequential sentence in the release is that DRAM, not compute, is the primary constraint. If memory availability is the gate on data-centre build-outs, then memory supply agreements become a competitive variable for hyperscalers and systems builders rather than a procurement detail. Investors tracking where in the AI chain pricing power actually sits can compare the margin profiles directly on our screener.

What to watch next

• The first-quarter fiscal 2027 report, due around the turn of the calendar year, against the guided USD 61.5bn plus or minus USD 1.5bn of revenue and USD 38.15 of non-GAAP EPS.
• First-quarter capex against the USD 11.5bn figure, and whether the roughly USD 25bn first-half number holds once two quarters are on the board.
• The full fiscal 2027 capex total. Management has said it will rise versus prior plans and that the second half will be heavier than the first, but has not framed a year number. That is the gap in the disclosure.
• HBM pricing and the margin gap. Calendar 2027 bit supply is largely contracted at significantly higher prices; whether the gross margin gap with conventional DRAM keeps narrowing is the test of whether HBM has become a margin business.
• Cleanroom availability timing. Late calendar 2028 is the stated target. Slippage there is a direct hit to the return on this spending, and it is the one variable the take-or-pay agreements do not protect.

Micron has finished a year in which it earned USD 75.52 per diluted share on a non-GAAP basis and told the market it intends to spend USD 25bn in six months. Both of those are facts about the same business, and reconciling them is what a fundamental view of this stock now consists of.