I called a beat-and-raise and a coin-flip reaction. Micron blew past both — a $41.5B record, an 84.9% margin, and a $50B next-quarter guide that broke the model. The bar was sky-high. Micron cleared it by miles.
It didn't just beat — it broke the model. Every line came in above the bull end of the range.
| Metric | FQ3 Actual | Guide | Street | Verdict |
|---|---|---|---|---|
| Revenue | $41.46B | $33.5B | ~$35.1B | BEAT by ~$6–8B |
| Non-GAAP EPS | $25.11 | $19.15 | ~$20.39 | BEAT by ~$5 |
| GAAP EPS | $24.67 | — | — | record |
| Non-GAAP gross margin | 84.9% | ~81% | ~81.8% | +3 pts |
| Non-GAAP operating margin | 81.2% | — | — | opex leverage |
| Non-GAAP net income | $28.86B | — | — | ~70% net margin |
| Operating cash flow | $25.39B | — | — | cash machine |
| Free cash flow (adj.) | $18.3B | — | — | capex $7.1B net |
| Cash & investments | $30.2B | — | — | fortress balance sheet |
There's no spin needed here — this is one of the great quarters in semiconductor history. Revenue nearly quadrupled year-over-year (from $9.30B) and grew ~74% in a single quarter (from $23.86B). A memory company posted an 84.9% non-GAAP gross margin and a ~70% net margin, converting $41.5B of revenue into $28.9B of profit and $18.3B of free cash flow in ninety days. Every figure landed above even the bullish end of the range I laid out in the preview. The "is it priced in?" debate that dominated the setup got answered the only way it could: the numbers were simply too big to be priced in.
Micron didn't just beat expectations — it shattered them. It made more revenue in three months than it used to make in a year, kept ~85 cents of gross profit on every dollar (unheard of for memory), and threw off $18 billion in spare cash. There was a big worry that all the good news was already in the price; the results were so large that worry evaporated.
The real bombshell was Q4. $50 billion next quarter — when the Street was at ~$43B.
If the quarter was great, the guide is what re-rates the stock. Micron told the Street to expect $50.0B ± $1.0B in FQ4 revenue at ~86% gross margin and ~$31.00 non-GAAP EPS (GAAP ~$30.73) — against a consensus that was sitting near ~$43B. That is not a "slight raise"; it's a ~$7B, ~16% upward reset of next quarter, with margins still expanding. Annualize that $31 quarterly EPS run-rate and you're looking at roughly ~$124 of earnings power — which is why, even after a +14% pop, the stock arguably got cheaper on forward numbers: estimates jumped more than the price did. The "sandbag and beat" pattern I flagged in the preview didn't just continue — it went vertical.
The forecast for next quarter is the real shocker: Micron expects $50 billion in sales when Wall Street guessed ~$43 billion, with profit margins going even higher. That points to roughly $124/share in annual earning power — so even though the stock jumped 14%, it's actually cheaper compared to what it's now expected to earn.
This is a data-center company now. Cloud + core data center = ~61% of revenue.
| Business unit | FQ3 revenue | Share | Read |
|---|---|---|---|
| Cloud Memory | $13.77B | ~33% | HBM & high-cap DRAM for AI |
| Core Data Center | $11.52B | ~28% | server DRAM/SSD |
| Mobile & Client | $11.52B | ~28% | recovering alongside AI PCs/phones |
| Automotive & Embedded | $4.63B | ~11% | steady, non-AI ballast |
Micron now reports along AI-era market lines, and the mix tells the story: Cloud Memory + Core Data Center together are ~$25.3B, about 61% of total revenue. The old "PC and phone memory company" is now, first and foremost, an AI data-center supplier — and management noted HBM4 is already shipping in high volume, the generation that ships with Nvidia's Rubin. That's the structural shift bulls have been underwriting: the demand isn't a cyclical bounce in commodity DRAM, it's AI infrastructure pulling the highest-margin product Micron makes. The flip side to keep honest: ~61% of the company now rides one end-market's capex budget.
Most of Micron's money now comes from AI data centers — about 61% of sales — not phones and laptops. Its newest, most advanced AI memory (HBM4) is already shipping in big volumes. Great, because that's the high-profit stuff; risky, because the company is now heavily dependent on AI companies continuing to spend.
Straight into the bull zone. +14% to ~$1,199, dragging the whole memory complex up.
The tape did exactly what a blowout-plus-raise should do. Shares jumped to ~$1,198.89, up ~14% versus the prior close (roughly +9% off the regular-session close), pushing right up against the $1,214 all-time high. The move spilled across the sector — chip names that had wobbled earlier in the week turned higher, and the Roundhill Memory ETF was up ~10% after the bell. Remember the options setup: the front-week was pricing a ~±11–12% move, so this ~14% pop slightly exceeded the implied move — meaning the result beat not just estimates but the market's own volatility expectation. For anyone who sold premium into the print, the vol-crush worked; for anyone short into it, it didn't.
The stock jumped about 14% to ~$1,199, almost to its record high, and pulled other chip stocks up with it. Traders had braced for an ~11–12% swing — the actual move was even bigger, which tells you the results surprised even the people betting on a big surprise.
How my preview call held up. Right on direction, right on the zone, wrong (low) on the size.
Honest grade: the framework worked — beat-and-raise, bull-zone reaction, record margins all called correctly — but I anchored too close to the published estimates and under-modeled just how non-linear this cycle has become. The lesson I'm taking: in a genuine supply-constrained supercycle, "sandbagged guide + sold-out capacity" can beat by far more than the historical surprise rate suggests. I'd rather log that miss plainly than pretend my range caught a $50B guide.
My preview got the big calls right (it would beat, it would raise, the stock would pop into my "bull" range) but I lowballed how huge the numbers would be. Being honest about that miss matters more than looking smart.
The print answered the quarter. It sharpened the cycle question. The better this gets, the louder the "peak" debate.
▲ The Bull Case
- Forward EPS power ~$124/yr ($31 Q4 run-rate) — stock is <10x
- $50B guide + 86% GM — cycle has another leg, not a peak
- $18.3B FCF/quarter — buybacks, debt paydown, optionality
- HBM4 shipping in volume — holds the Nvidia Rubin socket
- Estimate revisions just began — targets to $1,500–1,800 likely
- Re-rate cheaper on the pop — numbers rose faster than price
▼ The Bear Case
- Peak-earnings paradox — best print ever = closest to the top
- Memory always mean-reverts — 86% GM invites new supply
- ~61% on AI data center — one capex cycle to digest
- Samsung HBM4 ramp could crack the pricing truce in 2027
- At the ATH — chasing +14% green leaves no margin of safety
- Capex climbing ($7.1B/qtr) — supply additions seed the next glut
Bulls: profits are exploding, the stock is cheap on next year's earnings, and the boom has another leg. Bears: this industry always swings boom-to-bust, sky-high margins tempt rivals to flood the market, and the best quarter ever often comes right before the cycle turns. Both are true — that's the tension now.
My 12-month scenarios, post-print. From ~$1,199, with estimates being revised up as we speak.
The math shifted hard in the bulls' favor. With FQ4 alone guiding to $31 EPS, the full-year FY26 number now towers over the old ~$57.71 consensus, and FY27 (previously ~$97.77) is going to be revised sharply higher — at ~$1,199 the stock trades under ~10x its annualized run-rate. That's why my base case is now higher than the pre-print bull case: a stock growing earnings this fast at a sub-12x multiple has room even if the multiple doesn't expand. The bear case isn't about this quarter — it never was. It's the one constant in memory: the cycle turns, and it usually turns right after the best print. I'm not calling the top here — the $50B guide says the next two quarters are spoken for — but I'm keeping the bear weight honest at ~22%, because "this time is different" is the most expensive sentence in this industry.
Because next-quarter earnings are guided so high, the stock is cheap versus what it's about to earn — so my realistic case ($1,400–1,550) is now above what I'd have called bullish yesterday. The main risk isn't the results; it's that the memory business always eventually cools off, often right after a peak like this. I'm not calling the top, but I'm not ignoring history either.
Two scores, updated. The business: 9.5/10 · chasing it at the ATH: ~6.5/10.
| Dimension | Grade | Why |
|---|---|---|
| Business quality | A+ | 84.9% GM, $18.3B FCF/qtr, three-supplier HBM moat |
| Demand visibility | A+ | $50B guide; HBM sold out; HBM4 shipping in volume |
| Forward valuation | A− | <10x annualized run-rate — cheaper after the pop |
| Trend / momentum | A | +14% to the ATH on record numbers |
| Cyclical risk | C | Peak margins always invite the next supply wave |
| Entry timing | B− | At the ATH after +14% — better to scale than chase |
The preview rated the business 9 and the entry 6. The print earns a bump on both: the business to 9.5 (this was extraordinary, not just good), and the entry to 6.5 — because while the multiple is now lower and the support is real, you'd be paying up green at the all-time high. The framing doesn't change: a 9.5-business / 6.5-entry is a name you want to own and don't need to chase. The forward earnings power is so large that patience is cheap; let the next pullback come to you rather than buying the after-hours candle.
The company just scored a 9.5/10. Buying it the instant it spikes to a record high is more like a 6.5 — the value is real, but you're paying top dollar in the heat of the moment. It's a stock to own; you don't have to chase the pop.
John's read. Spectacular business, vindicated thesis, one nagging discipline.
- Credit where it's due — this was a monster. $41.5B revenue, 84.9% margins, $18.3B of free cash flow, and a $50B guide that humiliated the Street's ~$43B. The "strategic value of memory in the AI era" line from Mehrotra isn't a slogan anymore; the income statement is the proof.
- My preview call worked, and I'll own the part that didn't. Beat-and-raise: right. Reaction in the bull zone (~$1,200–1,260): right, +14% to ~$1,199. But I modeled ~$35–36B / ~$21 — I was too conservative by a mile. In a real supercycle, a sandbagged guide on sold-out capacity can blow past the historical surprise math.
- The "priced in" fear is dead — for now. The numbers were too big to be in the price, which is why it gapped through the implied move. That debate is settled for this quarter. The next one always comes back.
- The only risk that matters from here is the cycle, not the quarter. Watch FQ4 HBM4 share, the margin trajectory toward 86%, and any whiff that Samsung's HBM4 yields are fixed. Memory has made fortunes and vaporized them; an 86% gross margin is the kind of number that, historically, plants the seed of the next glut. I'm long the trend, eyes open.
- How I'd play it: own quality, scale don't chase. The stock is genuinely cheap on its new run-rate (<10x), so I'm not bearish — but I refuse to pay up green at the ATH on an after-hours candle. I'd build on pullbacks, key the thesis to HBM4 share + margins + the FQ4 print, and respect that the better this gets, the closer we drift to the moment this industry always, eventually, reminds you what it is. Great company. Stay disciplined on price.
Want my live read as the FQ4 print and HBM4 share data roll in?
Join the Discord to find out! →Primary — Micron Technology, Inc. — "Reports Record Results for the Third Quarter of Fiscal 2026" (official press release, June 24, 2026): revenue $41.46B (+73.8% QoQ, +345.7% YoY); GAAP gross margin 84.6% / non-GAAP 84.9%; GAAP operating margin 80.4% / non-GAAP 81.2%; GAAP net income $28.24B / non-GAAP $28.86B; GAAP EPS $24.67 / non-GAAP EPS $25.11; operating cash flow $25.39B; adjusted FCF $18.3B; capex $7.1B net; cash & investments $30.2B; business units Cloud Memory $13.77B, Core Data Center $11.52B, Mobile & Client $11.52B, Automotive & Embedded $4.63B; HBM4 shipping in high volume; FQ4 FY26 guide revenue $50.0B ±$1.0B, GM ~86%, non-GAAP opex ~$1.65B, non-GAAP EPS $31.00 ±$1.00 (GAAP $30.73 ±$1.00); CEO Sanjay Mehrotra quote. Corroboration via StockTitan and TradingView News. Market data — live MU quote, range, options & implied vol via Interactive Brokers (after-hours ~$1,198.89, +13.99% vs prior close $1,051.77; intraday +4.2% pre-print; 52W $103.27–$1,213.56; front-week IV ~185% → implied move ~±11–12%). Context — consensus/preview & analyst targets (UBS $1,625, Aletheia $1,600, BofA/TD Cowen $1,500, Raymond James/HSBC/Melius $1,100, Citi $840) and HBM share/HBM4 allocation via TradingKey, TheStreet, Yahoo Finance, MarketBeat, TrendForce & Counterpoint.