Financials

Financials in One Page

Micron is a deeply cyclical memory manufacturer printing record numbers at what looks like the peak of an unprecedented AI-driven cycle. FY2025 (ended Aug 28, 2025) delivered $37.4B revenue (an all-time high, +49% YoY), 26% operating margin, 48% EBITDA margin, $8.5B net income, and $17.5B operating cash flow — but only $1.7B of free cash flow because capex hit a record $15.9B (42% of revenue). The balance sheet supports this spend: $10.3B cash against $14.6B debt, net debt of $4.3B and just 0.3x trailing EBITDA. ROIC swung from −12% in FY2023 to ~15% in FY2025, and Q2 FY2026 accelerated further with the HBM/AI price surge. The single financial metric that matters most right now is the gross-margin trajectory — from −9% in FY2023 to 56% in Q1 FY2026 to 75% (non-GAAP) in Q2 FY2026 — and the defensibility of the ~$880B equity value depends on that line not breaking.

A quick glossary for the rest of this page. Gross margin is revenue minus cost of goods sold, divided by revenue — it tells you how much pricing power Micron has after paying for silicon, gas, water, and direct fab labor. Operating margin deducts SG&A and R&D on top — Micron spends ~10% of revenue on R&D, so operating margin runs roughly 12 points below gross margin. EBITDA (earnings before interest, taxes, depreciation, amortization) strips out the ~$8B/year of depreciation from prior capex; for a fab business it is the cleanest read of cash earnings. Free cash flow is operating cash flow minus capex — the cash actually left for shareholders, debt repayment, and the next cycle. ROIC (return on invested capital) measures profit per dollar of capital tied up in fabs and equipment.

Revenue, Margins, and Earnings Power

Micron's income statement is a textbook lesson in semiconductor cyclicality. Revenue has compounded at roughly 11% over twenty years, but the path has included three full reversals: the 2008-2009 financial crisis trough, the 2016 China-CapEx oversupply, and the 2023 inventory-and-pricing collapse where operating margin went to -37%. Each trough has been followed by a sharper peak. FY2025 set a new revenue high of $37.4B (the prior peak was $30.8B in FY2022) and brought operating margin back to 26%.

Three things to notice on this chart. First, revenue has tripled since FY2014, but operating income has done a full lap on each cycle — peaks above $10B, troughs below zero. Second, FY2018's $15B of operating income on $30B of revenue (49% margin) is the absolute high-water mark for the franchise; FY2025 at 26% is good but not yet at the prior peak. Third, FY2023 was the third negative-operating-income year of the modern era — Micron's earnings power can disappear entirely in a single bad year.

The recent quarterly trajectory is the real story. From the FY2023 trough Micron has rebuilt margins step-function each quarter as HBM (high-bandwidth memory, the AI-grade DRAM stacked on top of NVIDIA GPUs) has gone from a small product to the most economically valuable thing in the company. Q4 FY2025 alone delivered $11.3B of revenue and 45% gross margin — comparable to the FY2018 cycle peak in margin terms. Q1 FY2026 then hit 56% gross margin and Q2 FY2026 (reported March 2026) printed $23.9B revenue, 75% non-GAAP gross margin (74.4% GAAP), and $12.20 non-GAAP EPS ($12.28 GAAP), blowing past the $8.42 guide midpoint.

Cash Flow and Earnings Quality

Memory accounting is unusually clean: revenue is recognized on shipment, the customers are mostly large OEMs with short DSOs, and there is very little intangible amortization. The real test of earnings quality is whether the company's enormous depreciation charge — currently $8.4B/year — is roughly matched by the actual capex needed to keep the fabs current. For Micron, the answer over a full cycle is "yes, but only barely".

Operating cash flow is consistently higher than net income — by $4-9B per year — because Micron books $8B+ of depreciation that is not a cash charge. That gap is what keeps the company solvent through troughs: in FY2023, net income was -$5.8B but operating cash flow was still positive $1.6B because depreciation more than offset the loss. The honest free-cash-flow series, however, is far less heroic. Across the last eight years FCF has averaged roughly $2B per year — about a quarter of operating cash flow — because Micron consistently spends 35-50% of revenue on capex. In FY2025, $15.9B capex absorbed 91% of operating cash flow.

Stock-based compensation is non-trivial — $972M in FY2025 versus $328M in FY2020 — and dilutes shareholders by roughly 1% per year. Combined with paused buybacks, share count has crept up from ~1.09B in FY2017 to ~1.12B in FY2025. Not abusive, but not accretive either. The investor question on cash conversion is therefore simple: with depreciation running at $8B and capex at $16B, the next $8B of "growth capex" each year only converts to free cash if HBM-grade pricing holds. FY2025 had to drop capex relative to OCF to get even a positive FCF figure.

Balance Sheet and Financial Resilience

Micron entered the AI cycle with a much sturdier balance sheet than it had at the start of the FY2018 cycle. Net debt peaked at $5.9B in FY2024 — that is 0.6x trailing EBITDA, well inside any prudent rating-agency band — and has stayed roughly flat in dollar terms while EBITDA more than doubled. Interest coverage is now 38x EBITDA / interest, versus 5x in the FY2023 trough.

The shape of the balance sheet matters as much as the size. Property, plant and equipment grew from $39B at the start of FY2023 to $47B at FY2025 close — a $9B build during a period when the company lost $5B and earned $8.5B respectively. That growth is funded by a combination of cash flow, $7B of new debt issued in FY2023 (when the company had to borrow through the trough), and $4.4B of fresh debt in FY2025 with $4.6B of repayments — essentially a refinancing rather than a leverage-up. Goodwill is a trivial $1.15B; this is a hard-asset, organic-growth balance sheet.

Inventory at 139 days remains elevated. In a memory company, days inventory is the single best leading indicator of cycle direction: it climbs to ~160 days at the top of supply-glut periods (it was 162 in FY2023 and FY2024), then collapses back to ~100 days as demand absorbs the build. The FY2025 print is moving in the right direction but is not yet at "clean" levels. Watch this.

Returns, Reinvestment, and Capital Allocation

Returns on capital are the most volatile line on the page for a memory maker. Across the last decade Micron has earned ROIC anywhere from -12% to +18%, with no clear central tendency. FY2025's 14.8% ROIC is good — meaningfully above the company's roughly 9% weighted average cost of capital — but it depends on margins staying anywhere close to FY2025 levels. The FY2023 print of -12% ROIC tells you what a single bad year does to compounded value.

The capital-allocation mix tells the underwriting story. From FY2014 through FY2022 Micron spent roughly $4-12B per year on capex, returned $5B to shareholders via buybacks (cumulatively), and paid down ~$10B of net debt in the FY2018 cycle peak. From FY2023 onward, however, capex has been the only meaningful use of cash; buybacks have been paused since the small $300M repurchase in FY2024, and the new $0.46/share annual dividend (started in FY2022) is a token $520M/year. Almost every dollar of operating cash flow is being plowed into HBM-grade fab capacity in Idaho and New York.

The share count chart is benign — diluted shares are roughly flat at 1.12-1.13B for eight straight years — but the underlying story is that ~$1B/year of SBC is being absorbed by intermittent buybacks. With buybacks paused, expect modest creep (~1%/year) until management resumes repurchases. The reasoned argument for the pause is correct: at FY2025 close the market cap was $137B and trailing FCF was $1.7B; new HBM capacity at 20%+ projected returns is a better use of dollars than repurchasing the stock. That argument starts to fail at $880B market cap and a different forward FCF math — see Valuation below.

Segment and Unit Economics

Segment-level financials were not staged in this run. From the FY2025 10-K, the company reports four business units: Compute and Networking (the largest, includes HBM/data-center DRAM), Mobile, Embedded (auto/industrial), and Storage (NAND). DRAM is approximately 70% of consolidated revenue and the dominant profit pool; NAND is roughly 25% and structurally lower-margin. Within DRAM, HBM is the swing factor: the FY2025 10-K guides to HBM revenue scaling from a small base in FY2024 to mid-teens-percent of revenue by late FY2026, with HBM ASPs running multiples of standard DRAM.

The takeaway for an investor without segment detail: roughly 70-75% of incremental profit in FY2025-FY2026 is HBM-related, even though HBM is much less than half of revenue. That concentration is the real risk hiding inside a "diversified memory portfolio" headline.

Valuation and Market Expectations

This is where the underwriting question lives. At the FY2025 close (Aug 28, 2025, $122/share) Micron looked statistically cheap on FY2025 numbers: 16x P/E, 7.8x EV/EBITDA, 3.7x sales, 2.5x book. Since then the stock has run from $122 to roughly $782 — a 6.4x move in eight months — driven by the Q1/Q2 FY2026 HBM-margin shock and a re-rating of the franchise as an AI infrastructure name. The market cap now sits near $880B, with TTM EBITDA of roughly $37B and TTM EPS of roughly $21.

The key tension: on trailing numbers, MU is extraordinarily expensive — 36x P/E, 24x EBITDA, 15x sales for a cyclical memory maker. On forward consensus EPS (street estimates around $95-100 for FY2026 based on the run-rate after Q2 FY2026), the forward P/E collapses to about 8x — cheap, or very cheap, for a non-cyclical. Whether to underwrite the forward number requires three things to hold: (a) HBM ASPs within ~30% of the Q2 FY2026 print, (b) DRAM supply discipline through Samsung's and SK hynix's competing capacity adds, and (c) AI training and inference demand not cooling through CY2026.

The street's median 12-month price target of $600 (vs current ~$782) implies the consensus base case is closer to "Bear-to-Base" — analysts already see normalization risk. The bull case requires both HBM share leadership and a permanent valuation re-rating from "fab cyclical" to "AI infrastructure". Neither is impossible; both are unproven over a full cycle.

Peer Financial Comparison

Direct merchant-memory peers are concentrated in Korea (Samsung, SK hynix), Japan (Kioxia), and the post-spin US market (Sandisk, Western Digital). The 10-K-named competitor list also includes two private Chinese players (CXMT, YMTC) that we can characterize qualitatively but not in a financial table. The table below uses each peer's most recent reported full-year results. US-listed peers (MU, SNDK, WDC) use FY2025 calendar; SK hynix is FY2025 (calendar 2025); Samsung is CY2025; Kioxia is FY ended March 2025. Currency-converted at the staged FX in the snapshots.

Two reads from this table. First, on FY2025-reported numbers, SK hynix has the best margins and ROIC in the group (49% gross margin, 30% ROIC) because it is more DRAM-skewed and more HBM-leveraged than Micron. Sandisk is the worst (negative op margin, negative ROIC) because pure-play NAND is in a much worse pricing position. Micron sits in the middle on returns but with a stronger balance sheet than Korean or Japanese peers. Second, multiples are now scrambled. The peer table here reflects the late-FY2025 close; in the months since, the AI-memory re-rate has lifted Micron's EV/EBITDA from 7.8x to roughly 24x while SK hynix has also re-rated. A current-day apples-to-apples valuation comparison would require fresh FX-adjusted market caps for the Korean and Japanese peers, which were not retrieved in this run — flagged for follow-up in the queries file.

The peer-gap that matters: SK hynix is Micron's most-comparable peer (pure-play memory, similar scale on DRAM, leading HBM customer at NVIDIA), and on FY2025 it earned 49% gross margin vs Micron's 40%. That nine-point gap is the most important benchmark on this page — closing it is the entire HBM-leadership thesis Micron is selling.

What to Watch in the Financials

The first financial metric to watch is gross margin in Q3 FY2026. Q2 FY2026's 75% non-GAAP is partly one-time; Q1 FY2026's 56% is the truer underlying level. A Q3 print in the high-40s to mid-50s would keep the cycle in extension and leave the forward multiple intact. A slip below 45% would force FY2026 EPS consensus revisions and remove that multiple support.