Micron Earnings Show AI Memory Has Become a $34 Billion Data-Centre Business

Micron earnings for its fiscal fourth quarter of 2026 show just how central memory has become to the AI buildout. The company reported $54.23 billion in quarterly revenue on September 30, up from $11.32 billion a year earlier. Yet the more useful figure is inside its business-unit table: the Cloud Memory and Core Data Center units together generated $34.29 billion in the quarter. That is about 63% of total revenue, based on figures in Micron’s earnings release filed with the US Securities and Exchange Commission. The AI infrastructure boom is no longer merely a forecast in these accounts. It is shaping the product mix, margins and investment plan of a major memory supplier.

The result needs a careful reading. Cloud Memory and Core Data Center are company reporting segments, and neither is a pure measure of high-bandwidth memory, or HBM, sold into AI accelerator packages. Micron sells other types of memory and storage into data centres too. The combined figure shows the scale of data-centre-oriented business; it does not tell us the exact revenue from HBM, Nvidia systems or any single customer. Independent original reporting by Reuters, CNBC and Bloomberg examined the earnings and guidance, including the tension between extraordinary demand and a modest forecast margin decline.

The verified numbers at a glance

• Release date: September 30, 2026, for the fiscal quarter ended September 3.
• Fiscal Q4 revenue: $54.23 billion, compared with $41.46 billion in the previous quarter and $11.32 billion a year earlier.
• Fiscal-year revenue: $133.19 billion, compared with $37.38 billion in fiscal 2025. The $54.23 billion figure is one quarter, not the full year.
• Data-centre-oriented units: $16.28 billion of Cloud Memory revenue plus $18.00 billion of Core Data Center revenue.
• GAAP gross margin: 86.8% in Q4, compared with 44.7% a year earlier.
• Next-quarter guidance: $61.5 billion of revenue, plus or minus $1.5 billion, for fiscal Q1 2027.

Why the data-centre mix is the story

Micron’s Core Data Center business reported $18.00 billion in Q4 revenue, up from $1.58 billion a year earlier. Dividing the current figure by the prior-year figure gives roughly 11.4 times the revenue. Cloud Memory contributed another $16.28 billion, up from $4.54 billion. Taken together, these two units are far larger than the company’s $13.11 billion Mobile and Client unit and its $6.82 billion Automotive and Embedded unit. That does not mean consumers and cars have stopped needing memory. It shows that data-centre demand now dominates the company’s disclosed revenue mix.

This shift matters because an AI server is a system, not just an accelerator chip. Processors require memory close enough and fast enough to feed calculations, plus storage and data-centre infrastructure to keep models and applications running. A shortage or steep price increase at the memory layer can affect the cost and availability of complete systems even if processors are being produced. Lapaas Voice has previously explained how HBM qualification and yields can become a bottleneck: a manufacturer’s test production is not the same thing as high-volume supply accepted by customers.

Micron’s segment results cannot isolate every AI workload, but they offer a measurable contrast with last year. Core Data Center revenue grew from $1.58 billion to $18.00 billion and its reported gross margin rose from 41% to 90%. Cloud Memory revenue rose from $4.54 billion to $16.28 billion, while its gross margin rose from 59% to 83%. Those are company figures for two business units. The changes reflect product mix, pricing and demand conditions that Micron describes as AI-driven; they do not prove that every additional dollar was caused by AI alone.

There is an India angle as well. Indian data-centre developers, cloud operators, AI startups and electronics manufacturers participate in supply chains affected by memory availability. If leading suppliers allocate advanced products to high-paying data-centre buyers, pricing pressure can spread to other categories. Lapaas Voice’s earlier report on smartphone shipments and memory constraints described one possible downstream consequence. Micron’s Q4 report confirms strong supplier economics, but it does not by itself quantify Indian component prices or smartphone demand.

The margin boom — and the warning inside next quarter’s outlook

Micron reported a GAAP gross margin of 86.8% for Q4, up from 44.7% in the comparable quarter of fiscal 2025. Its non-GAAP measure was 87.0%, versus 45.7% a year earlier. These figures show how unusual the current combination of volume, pricing and product mix has been. GAAP net income was $37.70 billion in the quarter, while operating cash flow reached $43.97 billion. The company’s full-year operating cash flow was $89.68 billion. All of those values appear in the SEC-filed earnings exhibit; they are not estimates from a stock-market screen.

Looking forward, Micron guided fiscal Q1 2027 revenue to $61.5 billion, with a $1.5 billion range on either side. Its GAAP gross-margin guide is approximately 85.95%, and the non-GAAP figure is approximately 86.25%. Both are slightly below the Q4 margin. Bloomberg’s original report highlighted management’s explanation that higher employee incentive compensation is one contributor to the forecast margin dip. A lower percentage margin alongside higher forecast revenue is not the same as a forecast collapse in gross profit.

The distinction matters more than a one-day share-price reaction. Gross margin is revenue left after the costs of producing goods; it excludes some expenses that affect operating or net profit. A modest sequential percentage decline may be compatible with greater absolute profit if sales rise, but actual future results will depend on shipments, pricing, mix, manufacturing costs and compensation. Company guidance describes expectations, not guaranteed sales. Reuters separately reported that Micron’s customers had increased long-term supply commitments, which supports demand visibility while leaving volumes, delivery timing and pricing conditions subject to real contracts.

The full-year comparison also requires perspective. Fiscal 2026 revenue of $133.19 billion is more than three and a half times fiscal 2025’s $37.38 billion. It would be careless to extrapolate that growth rate into every future year. Memory is historically cyclical, and Micron’s own forward-looking-statement section cautions that results can diverge from management’s expectations. The report captures an extraordinary moment in AI infrastructure spending; it is not proof that the same pricing and margins can persist indefinitely.

Capacity expansion is real spending, but not instant new supply

Micron said net capital-expenditure investment was $10.77 billion in Q4 and $27.37 billion for fiscal 2026. It ended the year with $73.48 billion in cash, marketable investments and restricted cash. Those figures show both the scale of the manufacturing programme and the financial resources generated during the strong cycle. Capex is not simply a bill for more machines: a memory fab or advanced packaging line takes time to build, equip, qualify and ramp. Customers then have to validate products for demanding uses.

The company’s release lists product milestones, including samples of dense 512GB DDR5 server modules, more customer qualifications for high-speed server memory, and SSD products shipping for AI-related cache applications. Sampling and qualification are different stages from volume shipment; a claim that every listed product is already broadly available would overstate the announcement. Micron also said it had begun sampling LPDDR6 products for multiple physical-AI markets. The commercial effect of those samples will depend on customer adoption and production scale.

Its earlier $10 billion Micron Research Labs plan is a separate, longer-term research commitment. The earnings release’s annual capex number concerns investment in the fiscal year just ended. Mixing those two figures would make both the spending timetable and the purpose look more certain than they are. Research, equipment installation, qualified output and billed sales are distinct milestones.

The revenue and margin surge may encourage competitors and customers to invest in alternative memory capacity and supply arrangements. But a production response cannot instantly create qualified HBM stacks or next-generation server modules. If Micron’s demand outlook proves accurate, buyers of AI servers will still need to plan around memory supply and total system costs. If supply expands faster than demand, the pricing environment could change. The filing does not settle which path will dominate beyond its near-term guidance.

How to read the AI-memory signal without overclaiming

Three distinctions keep this report grounded. First, Micron’s record $54.23 billion is quarterly revenue, while $133.19 billion is full-year revenue. Second, the combined $34.29 billion from its two data-centre-oriented units is a segment sum, not a disclosed HBM-only figure. Third, the $61.5 billion for fiscal Q1 2027 is a management forecast with a stated range, not revenue already earned. These separations are essential because fast-growing technology businesses are often described with figures from different periods or categories as if they are interchangeable.

It is also important to attribute forward-looking claims. Micron calls AI demand a driver of its results and expects an even stronger fiscal 2027. Those are management’s assessments. The reported sales, margins, cash flow and business-unit values are concrete historical outcomes in the filed release. Reuters, CNBC and Bloomberg add independent reporting and context, but no news report can turn a forecast into a completed result.

For Indian AI builders and technology buyers, the practical takeaway is that memory should be treated as a strategic input. Model performance, inference economics and server availability depend on a complete hardware stack. Micron’s results show that suppliers at the memory layer are capturing a large share of the current infrastructure wave. Whether that improves access for buyers or makes systems more expensive will depend on future output, contracts and competition, not solely on one record quarter.

Frequently asked questions

Did Micron make $54.23 billion in its full fiscal year?

No. That is fiscal Q4 2026 revenue. Full-year fiscal 2026 revenue was $133.19 billion, according to Micron’s SEC-filed release.

Does $34.29 billion equal Micron’s HBM sales?

No. It is the sum of Micron’s Cloud Memory and Core Data Center business-unit revenue. Both include products beyond HBM, and the release does not provide an HBM-only revenue total.

Why did the margin forecast edge down despite stronger sales guidance?

Micron guided to a slightly lower gross-margin percentage for fiscal Q1 2027 while forecasting higher revenue. Bloomberg reported management’s explanation that increased incentive compensation was one contributor. The ultimate margin depends on actual costs and mix.

When was the result released?

Micron released it on September 30, 2026, after the US market close. In India it arrived on October 1. The fiscal quarter itself ended on September 3.

Sources and verification

Primary: Micron fiscal Q4 and full-year 2026 earnings release, filed with the SEC, September 30. Independent original publisher reporting: Reuters, September 30; CNBC, September 30; Bloomberg, Dina Bass, September 30. The SVG revenue and segment shares are calculated directly from Micron’s filed values. One Reuters story should be counted once even if it appears on multiple licensed sites.