Micron closed fiscal 2026 with a record FQ4 (revenue $54.2B, gross margin 87%, EPS $33.42, all above the top of guidance) and guided FQ1 2027 to another record: $61.5B revenue and $38.15 EPS.
Every part of the call pointed the same way: management sees supply tighter in calendar 2027 and 2028 than in 2026, with no line of sight to balance.
Quick Summary
- Record year. FY26 revenue was $133.2B (+256% YoY), gross margin 81.1% (+40 pts), EPS $75.52 (+811%). DRAM revenue alone passed $100B for the year.
- FQ4 beat across the board. Revenue +31% QoQ and +379% YoY; DRAM prices up high teens % QoQ, NAND prices up ~30% QoQ. Operating margin reached 82.3%.
- Guide implies ~13% QoQ growth. FQ1 revenue $61.5B, but gross margin fall to ~86.25% because ~$1B of extra costs (mostly incentive compensation absorbed into inventory).
- FQ1 is called the gross-margin floor for FY27, with expected sequential revenue growth every quarter.
- Contracts are changing the business model. 26 strategic customer agreements (SCAs), estimated at over 35% of revenue through 2030 (should be 50% going forward); RPO of ~$150B at floor prices; $32B of customer financial commitments, mostly cash deposits. Over 75% of calendar 2027 output is already committed.
- Supply stays tight. Industry DRAM bits grow low-20s % in CY27–28 and NAND mid-20s %, both supply constrained; HBM outgrows conventional DRAM through CY28. CY27 HBM supply is mostly sold, at significantly higher prices.
- More CapEx, tilted to construction. FY27 CapEx goes up versus prior plans: ~$11.5B in FQ1, ~$25B in 1H, more in 2H. The extra money mainly buys cleanroom space for late CY28 and beyond.
- Capital return is the next catalyst. Net cash is $68.3B; Micron reaches its target cash level by end of FQ1, will step up buybacks from 9 December 2026, will seek a larger authorization, and aims over time to return 100% of excess cash.
Headline results
FQ4 was the sixth straight quarterly revenue record, driven far more by price than by volume.
| Metric | FQ4 2026 | QoQ | YoY | FY2026 |
|---|---|---|---|---|
| Revenue | $54.2B | +31% | +379% | $133.2B (+256%) |
| DRAM revenue | $39.8B (73% of total) | +27% | +343% | over $100B |
| NAND revenue | $14.1B (26% of total) | +42% | +526% | — |
| Gross margin | 87.0% | +210 bp | — | 81.1% (+40 pts) |
| Operating expenses | $2.6B | +$1.1B | — | — |
| Operating income | $44.6B (82.3% margin) | +110 bp margin | +47 pts margin | — |
| Tax | $6.8B (15% rate) | — | — | — |
| EPS | $33.42 | +33% | — | $75.52 (+811%) |
Price versus volume (FQ4, QoQ):
- DRAM: bit shipments up mid-single digits %, prices up high teens %.
- NAND: bit shipments up ~10%, prices up ~30%.
Reading the numbers:
- Gross margin improved on pricing and execution, partly offset by mix (more HBM, which still carries a lower margin than conventional DRAM).
- The $1.1B OpEx jump was one-off in nature: a higher FY26 incentive payout for every employee plus a $300M community investment contribution. FQ1 OpEx is guided back down to ~$2.06B.
- Data center is the engine: data center revenue rose fourfold in FY26, and data center SSDs alone were nearly $10B in FQ4 (over two-thirds of NAND revenue, more than 10x a year earlier).
Business unit performance
All four business units set revenue records; Core Data Center grew fastest and, with Mobile & Client, reached a 90% gross margin.
| Business unit | FQ4 revenue | Share of total | QoQ growth | Gross margin | Margin change QoQ | Driver named on the call |
|---|---|---|---|---|---|---|
| Core Data Center (CDBU) | $18.0B | 33% | +56% | 90% | +290 bp | Higher pricing, bit shipments, favorable mix |
| Cloud Memory (CMBU) | $16.3B | 30% | +18% | 83% | flat | Pricing and bits, offset by higher HBM mix |
| Mobile & Client (MCBU) | $13.1B | 24% | +14% | 90% | +260 bp | Pricing and mix, partly offset by lower bit shipments |
| Automotive & Embedded (AEBU) | $6.8B | 13% | +47% | 84% | +470 bp | Higher pricing and bit shipments |
What stands out:
- CMBU is the only unit with flat margins, because HBM (sold mostly under 2026 prices negotiated a year earlier) diluted it. The CY27 HBM repricing should lift this unit most.
- MCBU grew on price alone while volumes fell. Management expects PC and smartphone unit declines possibly in the double digits this year, offset by premium devices with more memory. Nearly half of MCBU revenue already comes from 1-gamma products.
- AEBU had the largest margin gain (+470 bp). Management frames autonomous vehicles as the first wave of physical AI: Level 4+ vehicles carry over 200GB of memory and multiple TB of storage, more than 10x Level 2+/3 vehicles, and humanoid robots are expected to need similar amounts.
Cash flow, balance sheet and capital return
FQ4 free cash flow was $33B, and management said FQ1 free cash flow will be “significantly higher”; that cash, not the P&L, is now the main debate.
| Item | FQ4 2026 |
|---|---|
| Operating cash flow | $44.0B |
| CapEx (net of government incentives) | $10.8B |
| Free cash flow | $33.2B |
| Customer SCA deposits received in quarter | $12.3B (in financing, not in FCF) |
| Cash and investments | $73.5B (record) |
| Debt | $5.2B (~$500M repaid; avg. maturity ~9 years) |
| Net cash | $68.3B |
| SCA deposits on balance sheet | $12.7B |
| Inventory | $10.4B; 129 days (+9 days QoQ) |
| Credit rating | BBB+ or equivalent at all three agencies (two upgrades) |
Points worth noting:
- Deposits are not owned cash. The $12.7B of SCA deposits is unrestricted but is returned to customers in the latter half of each agreement, if minimum purchases are met. Netting them out, net cash is about $55.6B. Total customer commitments are $32B, so more deposits may still arrive.
- Inventory days rose for known reasons: build-ahead before node end-of-life and incentive pay capitalized into inventory. Management called supply “extremely tight” and expects DIO to fall.
- Capital return plan: return rises from 9 December 2026 (two years after the CHIPS agreements were signed). Micron expects to hit its target cash level by end of FQ1, will return excess cash mainly through buybacks, will seek a larger authorization soon (current: $2.2B), and aims over time to return 100% of excess cash. No target cash level or dollar amount was disclosed.
Guidance: FQ1 2027 and FY2027
FQ1 guides revenue up ~13% QoQ to $61.5B and EPS up ~14% to $38.15, with a small, explained gross-margin step down.
| Metric | FQ1 2027 guide | vs. FQ4 2026 |
|---|---|---|
| Revenue | $61.5B ± $1.5B | +13% at midpoint |
| Gross margin | ~86.25% | –75 bp |
| Operating expenses | ~$2.06B | –$0.54B (FQ4 one-offs drop out) |
| Tax rate | ~15.5% | +0.5 pt |
| Share count | ~1.15B | — |
| EPS | $38.15 ± $1.00 | +14% at midpoint |
| CapEx | ~$11.5B | +$0.7B |
Why margin dips in FQ1. About $1B of extra cost lands in FQ1: mainly FY26 incentive pay that was capitalized into inventory in FQ4 and is now sold through, plus start-up and other costs. That is roughly 1.6 points of FQ1 revenue, so the underlying margin would be near 88%.
FY2027 framework:
- Another record year, with sequential revenue growth in every quarter.
- FQ1 is the gross-margin floor; margins rise after it, with “a more moderate rate of price increases”.
- FQ2 gets less of the FY26 pay effect but absorbs the higher FY27 incentive plan, so it is not a clean rebound. About $1B of ongoing extra cost (incentive pay, start-up, other) persists through FY27; management calls part of it variable or temporary.
- OpEx rises ~$2.5B for the year, mainly R&D and higher incentive compensation.
- CapEx: ~$25B in 1H, higher in 2H, with construction growing much faster than equipment. Analysts on the call inferred a full year of $50–55B or more; management did not confirm a number.
Strategic customer agreements (SCAs) and RPO
Micron has signed 26 multiyear take-or-pay SCAs, now estimated at over 35% of revenue through 2030, with a goal of about 50%.
| Element | What was disclosed |
|---|---|
| Agreements signed | 26 (up from 16 at the prior call, per an analyst) |
| Revenue coverage | Over 35% of estimated revenue through 2030; target ~50% |
| Pricing structure | 3/4 has a defined pricing framework (most with floor and ceiling bands); 1/4 repriced periodically to market |
| Duration | Some new SCAs run into 2031; two existing ones extended by 1 year to 2031 |
| Customer financial commitments | $32B, vast majority cash deposits |
| RPO | ~$150B: committed volumes × minimum prices, only for SCAs with a defined pricing framework |
| Margin at floor | “Meaningfully above any prior cycle peak margins” |
| New negotiations | Priced higher, on current market conditions and outlook |
Why it matters. Management presents SCAs as a structural change for a historically volatile industry: they give multiyear demand visibility to plan fabs, and they give customers supply assurance. Mark Murphy called the ~$150B RPO “inherently conservative” and expects revenue to well exceed it over the agreements’ terms.
Caveats from the call itself:
- RPO excludes the quarter of SCA revenue that is repriced to market, so it understates contract value but also says nothing about that part’s price.
- Floors and ceilings cap the upside as well as the downside on most of the covered volume.
- Management wants to keep flexibility across customers and new entrants, so SCA coverage may stay below 50% if the rest of the business grows faster.
Market outlook: DRAM, NAND, HBM and end markets
Management expects both DRAM and NAND to stay supply constrained through calendar 2028, with demand strengthening since the last call.
| Industry bit shipment growth | CY2026 | CY2027 | CY2028 | Micron vs. industry (CY26) |
|---|---|---|---|---|
| DRAM | Mid-20s % | Low-20s %, supply constrained | Low-20s %, supply constrained | About in line |
| NAND | Low-20s % (slightly above prior view) | Mid-20s %, supply constrained | Mid-20s %, supply constrained | Below industry |
| HBM | Grows faster than conventional DRAM through CY2028 |
Why supply can’t catch up quickly, in Sanjay Mehrotra’s words on the call:
- Cleanrooms take a long time to build, and output ramps only gradually after first wafers.
- The HBM mix shifts from HBM3E to HBM4 and HBM4E, which consume more wafers per bit (the “trade ratio”).
- Future node transitions yield less bit growth per wafer.
- Even with the new industry cleanroom plans, there is no line of sight to supply-demand balance.
Demand drivers:
- AI data center. Larger models, longer context windows, higher concurrency and more agents all need more and faster memory. Server units should grow high teens % in both CY26 and CY27, with slightly lower memory content per server than expected, because customers trim content to ship more units in a tight market.
- HBM. CY27 HBM bit supply is mostly contracted at significantly higher prices, narrowing the margin gap with conventional DRAM. HBM4 ramp is on track; Micron is working with NVIDIA on the first custom HBM4E (“NVHBM”) for next-generation GPUs and NVLink Fusion.
- Data center SSD. KV-cache offload for AI inference and HDD replacement expand the market. Micron expects a fifth straight year of record data center SSD share in CY26.
- PC and smartphone. Industry revenue grows this year despite possible double-digit unit declines, carried by premium AI devices with more DRAM and NAND.
- Physical AI. Autonomous vehicles first, then humanoid robots; several customers are sampling next-generation products, and Micron sees it as a significant demand driver by the end of the decade.
Capacity expansion and CapEx roadmap
Micron is raising FY27 CapEx above prior plans, mostly for construction, to pull cleanroom space forward to late calendar 2028 and beyond.
Technology: 1-gamma DRAM and G9 NAND are now Micron’s largest nodes and are heading to the highest volumes in its history. U.S. fabs in Virginia, Idaho and New York anchor U.S.-made DDR, LPDRAM and HBM supply. Management says equipment will be added in step with demand, and some equipment spending is being pulled forward to get more out of existing cleanrooms.
Highlights From Q&A
- Mark Murphy on how the cash comes back: “we plan to return excess primarily done through share repurchase.” On the authorization: “our current authorization… stands at $2.2 billion. And you can assume that we will seek additional authorization in the near term.”
- “we have the ability and the intent to increase our capital return and… you can expect us to seek to increase our authorization and commence stronger capital return from December 9 in accordance with the agreements we have with — on CHIPS.”(We should expect announcements soon!)
- On Capital Intensity: “the capital intensity is low on historic levels, and that is reflective of the strategic asset that memory and storage has become and the industry is structurally reset.” He added that capacity will be added “in a very disciplined manner, ensuring that we’re getting adequate return on that capacity investment.” Murphy confirmed the shift toward construction spending: “that mix of capital spend is shifting to more construction… And we would expect that trend to continue here next few years.”
- Sanjay Mehrotra: “a large part of the volume is already sold out for 2027 for HBM and the prices are much higher than 2026 prices. And of course, that is helping us narrow the margin gap with the non-HBM memory.” On timing: “with the price increases starting in calendar year 2027.” On growth: “we expect HBM to be outgrowing the DRAM in terms of the demand growth.”
- Mehrotra on supply increases and price: “supply-demand environment is only getting tighter… cleanrooms take a long while to build, even after they are built, even after first wafer output, production ramps up only gradually.” He listed further headwinds to supply: the HBM trade ratio, and node transitions that “give less productivity gain per wafer.” On despec: “when customers reduce their content growth… that really is to enable them to ship more units.” On visibility: “more than 75% of our output is already committed for 2027. And a majority of discussions with our customers today are already around 2028.” He closed: “we do not have a line of sight to when supply and demand will get in balance.”
- Mehrotra on SCAs: The 75%+ figure is “across our customer base, SCAs as well as non-SCAs.” On the ceiling: “that basically is still around 50% of our revenue through the 2030 time frame… Of course, that could be less too.” On why he wants room to move: “It is important for us to be able to maintain flexibility in terms of managing our supply across our customer base.” On why SCAs matter: “that really is a fundamental change to the nature of this industry… that is really good for the health of the industry versus the past when you would experience more volatility.”
Analysis: strengths, open questions and what to watch
The call makes a strong case that this upcycle is longer and better contracted than past ones; the open questions are about how much of that cash reaches shareholders and how long pricing keeps rising.
Strengths shown on the call:
- Pricing power, not volume. FQ4 growth came mostly from price (DRAM high teens %, NAND ~30% QoQ) on modest bit growth, and margins of 87% gross and 82% operating are far above anything in prior cycles.
- Visibility the industry never had. Take-or-pay SCAs, ~$150B RPO at floor prices, $32B in customer commitments and over 75% of 2027 output committed reduce the classic memory risk of a sudden demand drop.
- Customers are paying to secure supply. Cash deposits, extensions to 2031 and “higher pricing” on new SCAs are evidence of real scarcity, not just management optimism.
- Broad-based strength. Every business unit set a record; data center, mobile premium, auto and physical AI all point up.
- Fortress balance sheet. $68.3B net cash, BBB+ ratings, and a stated aim to return 100% of excess cash over time.
Open questions and risks raised by the call itself:
- Rate of price increases slows. Management guides “a more moderate rate of price increases” after FQ1. Growth in FY27 depends more on bits and HBM repricing than on blanket price hikes.
- Content trimming. Management acknowledged lower memory content per server than expected and did not directly address the reported HBM despec at one large customer.
- Capacity coming in 2027–2028. Micron’s own new fabs (ID1, Tongluo, ID2, Japan, Singapore NAND) and “additional industry DRAM cleanroom space” add supply; management argues ramps are slow, but this is the main long-term pricing risk analysts raised.
- Pricing caps. Most SCA revenue sits inside floor-and-ceiling bands, so part of the upside in a very tight market is given up in exchange for stability.
- Deposits are liabilities. $12.7B of SCA deposits must be returned later (but does not impact FCF); headline net cash flatters the owned cash position.
- Cost creep. ~$1B of ongoing extra cost, OpEx up ~$2.5B and rising CapEx mean margins depend on prices holding.
- Capital return unquantified. Two analysts pushed and got no number; the size of the new authorization will be the first real signal.
What to watch next:
- Size of the new buyback authorization and the pace of repurchases after 9 December 2026.
- FQ2 gross margin versus the “FQ1 is the floor” claim.
- CY2027 HBM pricing flowing into Cloud Memory margins.
- New SCAs: count, coverage toward ~50%, and extensions beyond 2030.
- Full-year FY27 CapEx figure and the construction-versus-equipment split.
- Execution on 2027 milestones: Singapore HBM packaging, ID1, Tongluo and next-generation DRAM/NAND nodes.
- Any sign of customers despeccing memory beyond trimming content growth.
Thank you!
Core ValueResearch Team
