A blowout that breaks the model: record $41.5B revenue, ~85% gross margin, $28.9B non-GAAP net income, and a Q4 guide of $50B. HBM is sold out beyond 2027, and new multi-year Strategic Customer Agreements are designed to take the cyclicality out of the story. The stock jumped ~15% after hours.
This is one of the most extraordinary quarters a large-cap has ever printed. Micron did $41.46B in revenue — up 74% sequentially and roughly 4.5x year-over-year — at an 84.9% non-GAAP gross margin, dropping $28.9B of non-GAAP net income and $25.11 of EPS, then guided Q4 to $50B with ~86% margins and $31.00 EPS. Memory has stopped behaving like a commodity and started behaving like a scarce, priced-for-AI strategic input.
The driver is structural AI demand colliding with constrained supply. Data-center revenue was about $25B in the quarter, HBM is fully booked, and management says industry tightness is "locked in to persist beyond calendar 2027." The new wrinkle — and the reason the stock moved on more than just the beat — is the disclosure of multi-year Strategic Customer Agreements: roughly $22B of cash and financial commitments, including ~$18B of cash deposits, explicitly designed to make Micron's earnings more durable and predictable through the cycle.
Our view: own the super-cycle, but stay honest about what it is. The fundamentals are spectacular and the SCAs genuinely de-risk the next few quarters — that supports a Constructive stance even after a ~15% pop to ~$1,160. The discipline is in the valuation: at these margins Micron is earning peak-of-cycle economics, and memory has always mean-reverted eventually. We anchor a base fair value near $1,300, but the honest range is wide ($750–1,750) because the entire debate is how long these economics last. Pay up for the structural story; size for the cycle.
| Metric | FQ3-26 | FQ2-26 / YoY |
|---|---|---|
| Revenue | $41.46B | $23.86B / $9.30B |
| Non-GAAP EPS | $25.11 | $12.20 / $1.91 · BEAT (~$20.2 est) |
| GAAP EPS | $24.67 | $12.07 / $1.68 |
| Non-GAAP gross margin | 84.9% | 74.9% / 39.0% |
| Non-GAAP net income | $28.86B | $14.02B / $2.18B |
| Operating cash flow | $25.39B | $11.90B / $4.61B |
| Adjusted free cash flow | $18.3B (record) | $6.9B / $1.9B |
| FQ4-26 revenue guide | $50.0B ± $1.0B | ~86% GM |
| FQ4-26 non-GAAP EPS guide | $31.00 ± $1.00 | vs ~$24.8 consensus |
Every line is a record, and the guide raises the bar again. Cost of goods sold barely moved (~$6.4B) while revenue exploded — the definition of pricing-led, supply-constrained operating leverage. Micron ended the quarter with $30.2B of cash and investments and declared a $0.15 quarterly dividend.
Management framed the quarter around three ideas: supply scarcity, HBM leadership, and contractual durability.
The CEO described market tightness as "locked in to persist beyond calendar 2027," with demand outpacing supply across HBM, DRAM and NAND. HBM products are fully booked, and the company expects the shortage to continue well past 2027 as the industry pivots capacity toward high-bandwidth memory.
HBM4 (on 1-beta DRAM) is in high-volume shipments for the lead customer's platform, with qualification samples out to multiple end-customers. Management said the HBM4 12-high ramp is "tracking twice as fast" as HBM3E 12-high, and Micron has already booked over $1B of HBM4 revenue. HBM4E (1-gamma) is targeted for volume production in 2027.
The marquee disclosure: ~$22B of cash and financial commitments under multi-year SCAs, including roughly $18B of cash deposits from customers. This is the structural pivot — pre-committed, pre-funded demand that is meant to smooth the historic boom-bust of memory. Enterprise SSD revenue was ~$5B (about 20% of the ~$25B data-center business), evidence the strength extends beyond HBM into high-capacity NAND.
| Business Unit | FQ3-26 Rev | Gross Margin |
|---|---|---|
| Cloud Memory | $13.77B | 83% |
| Core Data Center | $11.52B | 87% |
| Mobile & Client | $11.52B | 87% |
| Automotive & Embedded | $4.63B | 79% |
Data center (Cloud Memory + Core Data Center) was ~$25B — over 60% of revenue — and the highest-margin part of the mix. But note the breadth: Mobile & Client also printed ~$11.5B at 87% margin, and even Automotive & Embedded margins reached 79%. This isn't a single-product spike; it's an across-the-board memory price reset.
| After-hours price (6/24, web-sourced) | ~$1,160 |
| Prior close (6/24) | $1,009 |
| After-hours move | ≈ +15% |
| FQ3 revenue | $41.46B (+74% QoQ, ~4.5x YoY) |
| FQ3 non-GAAP EPS | $25.11 (beat ~$20.2) |
| FQ3 non-GAAP gross margin | 84.9% (record) |
| Adjusted free cash flow | $18.3B (record) |
| Cash & investments | $30.2B |
| 9-month GAAP EPS (diluted) | $41.40 |
| Implied FY26 GAAP EPS (9M + Q4 guide) | ~$72 |
| FQ4-26 revenue guide | $50.0B ± $1.0B |
| FQ4-26 non-GAAP EPS guide | $31.00 ± $1.00 |
| Strategic Customer Agreements | ~$22B commitments (~$18B cash deposits) |
| HBM4 revenue to date | >$1B; 12-high ramp ~2x HBM3E pace |
| Dividend | $0.15/quarter |
| Street bull targets | up to $1,625–1,750 |
Valuing a memory company at peak-cycle margins is the hardest job in the sector — straight-lining 85% gross margins is how investors get hurt at tops. The bull/bear gap here is entirely about durability: do the SCAs and the AI build-out hold these economics for years, or does memory mean-revert as supply catches up? We frame three states rather than pretend to precision.
| Scenario | Basis | Implied Value |
|---|---|---|
| Bear | Memory cycle turns in 2027 as supply floods in; margins normalize toward mid-cycle; multiple compresses on peak earnings | $750–900 |
| Base | SCAs + HBM scarcity sustain elevated earnings into 2027; a disciplined ~10–12x on still-elevated forward EPS | $1,250–1,350 |
| Bull | Super-cycle extends "beyond 2027" per guidance; HBM4 ramp + pricing power; the Street re-rates toward top targets | $1,600–1,750 |
We center the base near $1,300 — modest upside to the ~$1,160 after-hours level. On implied FY26 GAAP EPS of ~$72 the stock trades ~16x; annualizing the Q4 run-rate (~$31/quarter) implies a >$120 EPS pace and ~9x, but that is precisely the number you should not straight-line. The SCAs are the swing factor: to the extent ~$18B of customer cash deposits genuinely pre-funds multi-year demand, Micron deserves a higher trough multiple than memory has ever earned. That is the bull case — and the thing to verify quarter by quarter.
| FQ4 print (guide $50B / $31.00 EPS) | Confirms the super-cycle trajectory |
| HBM4 / HBM4E ramp & qualifications | Share and pricing read |
| Strategic Customer Agreement disclosures | Durability of pre-funded demand |
| Industry capacity additions (MU, SK Hynix, Samsung) | The eventual supply response |
| DRAM / NAND spot & contract pricing | Earliest sign of a cycle turn |
| Capex pace vs. demand signals | Downside risk if demand softens |
Micron just printed one of the great quarters in semiconductor history — $41.5B revenue at ~85% margins, a $50B guide, HBM sold out beyond 2027 — and backed it with ~$22B of pre-funded customer agreements built to tame the cycle. We're Constructive with a ~$1,300 base even after the ~15% pop, because the structural AI-memory story is real and the SCAs genuinely change the durability math. The discipline: you're paying for peak-cycle economics, so let the SCAs prove out and watch DRAM/NAND pricing for the first crack. Own the super-cycle; respect the cycle. Confirm the live tick on your Schwab/TOS feed — the after-hours quote here is web-sourced.