Report contents
At a glance
- Business model: Intel earns most of its money selling PC and server processors, where the x86 software base supports the installed base but caps pricing against AMD and Arm. It funds a contract foundry with few confirmed outside customers, so fab depreciation arrives before external wafer revenue. US leading-edge manufacturing is the strength; converting 18A yields into binding volume is the obstacle.
- Financial quality: Operating cash flow of $9.70 billion in FY2025 fell $4.95 billion short of capex, a fourth negative free-cash-flow year in five. The $23 billion equity offering lifted cash to $21.85 billion and left net debt near $28.3 billion, buying time not capacity. A $13 billion gap between operating income and net income in June 2026 leaves normalized earnings unclear.
Business model
Business model: the fab bill comes before foundry customers
Intel sells processors by the unit and, increasingly, manufacturing by the wafer. Core and Core Ultra chips go to PC makers such as Dell, HP and Lenovo; Xeon goes to server vendors and cloud operators; Mobileye sells a chip-and-software package per vehicle. Fiscal 2025 revenue was $52.9 billion, down 0.5% from $53.1 billion, the company reported in its 10-K for the year ended December 27, 2025. That makes Intel an integrated device manufacturer (IDM: a company that designs and fabricates its own chips) attempting to become a contract foundry as well, so the right label is a turnaround grafted onto a cyclical franchise.
The cost base is the business model
Gross profit was $18.4 billion, or 34.8% of revenue, against 55.4% in fiscal 2021. Research and development of $13.8 billion and sales, marketing and administrative costs of $4.6 billion then left a $2.2 billion operating loss. Intel reports three segments: Client Computing Group, Data Center and AI, and Intel Foundry. The last of these charges Intel's own product groups for their wafers, so its result shows what manufacturing costs before any outside customer pays a markup. That matters to the business judgment: every external wafer sold covers depreciation that Intel's own volume would otherwise carry alone.
Client processors have historically supplied the largest share of revenue and the cash that funds each process transition. Data center carries the higher margin per unit but is the line most exposed to accelerator substitution, and foundry is where the capital goes. The awkward shape of the company is that the segment generating the cash has the weakest pricing, while the segment absorbing the capital has the fewest confirmed outside customers.
Demand is not the swing factor
PC units track replacement cycles and have been flat to shrinking, while data center budgets are growing but moving toward accelerators and custom chips. Intel's revenue slipped 0.5% in a year when AI infrastructure spending climbed steeply. Position inside those markets, not their size, is what has to change before the margin does; a demand recovery alone would not repair it.
Buyer power sits with a small, well-informed group. PC processors go to a handful of OEMs that run thin margins and qualify more than one supplier; Xeon goes to cloud operators that increasingly design their own silicon. The x86 instruction set is the real switching cost, because decades of Windows, Linux and business software are compiled for it. That protects the installed base but not the price. AMD ships competitive parts in every tier and Arm-based designs now target PC sockets, which caps what Intel can charge at a given level of performance. On the foundry side the buyers are chip designers comparing Intel with TSMC on yields, design kits and packaging. Intel's genuine differentiator is geography: it is the only company doing leading-edge research and high-volume manufacturing in the United States, which counts with governments and with customers that have sovereign-supply requirements. Whether it is worth a premium is unsettled.
The foundry bet changes unit economics
Selling wafers on contract is a different business from selling a finished processor for hundreds of dollars. Contract prices must cover depreciation, yield loss and the cost of a fab running below capacity, and they are negotiated years ahead. That makes 18A, in high-volume production since 2025, the variable that binds both halves of the company: yields set the cost of Intel's own chips and decide whether outside designers commit. Low utilization hurts either way; the difference is whether someone else's revenue helps pay for it.
The fragile part is timing. Intel's fabs in Oregon, Arizona, New Mexico, Ireland and Israel carry fixed costs set by the IDM business, and the headline segment figures do not isolate how much foundry revenue comes from outside customers. If external volume stays immaterial, Intel is a product company carrying a foundry's cost structure, and the useful frame becomes share defense in client and data center. What would overturn that reading is specific: named external customers in volume production, prepayments or take-or-pay terms that shift utilization risk to the buyer, and a second node committed by an outside designer. Partnerships announced at a process launch do not qualify, and neither does internal wafer demand presented as foundry growth.
Questions worth following
- Can Intel win committed external foundry customers for 18A at meaningful volume?
External wafer commitments are what would turn the foundry from a cost center into a second revenue engine and justify the capital already spent. Without named customers, fab utilization depends almost entirely on Intel's own product roadmaps.
- Will Arm-based PC chips take premium notebook sockets from Intel?
Client processors are Intel's largest revenue line and its most price-sensitive. If buyers can substitute an Arm or AMD part without rewriting their software, Intel must defend volume with price, compressing the gross margin that the fab base depends on to cover fixed costs.
- Can Intel hold data center CPU volume and pricing as AI budgets shift to GPUs and in-house silicon?
Data center CPUs have funded much of Intel's roadmap. If hyperscalers treat the processor as an accessory to their own accelerators, Intel loses pricing power precisely when it needs cash for fabrication capacity, and the loss compounds with each platform generation.
Financial quality
Financial quality: cash still comes from the balance sheet
Weak. Intel has reported negative free cash flow in four of the last five fiscal years, and the positive year, FY2021, predates the foundry build. Operating cash flow of $9.70 billion in FY2025 fell $4.95 billion short of capital spending; the FY2024 gap was $15.66 billion. From FY2022 through FY2025 the cumulative shortfall was about $44.5 billion, and financing activities supplied $32.6 billion over the same span. The capital program is still funded by claims on the balance sheet, not by the plants it has already built.
The five-year record is a compression of returns, not a normal cycle. Operating margin fell from 24.6% in FY2021 to -4.2% in FY2025 as revenue dropped 33%, and ROE went from 22.5% to -0.3%, ROA from 12.4% to -0.1%. Gross margin bottomed at 32.7% in FY2024 against 55.4% in FY2021, then climbed to 43.0% in the September 2026 quarter. Revenue of $16.51 billion was up 21% from a year earlier, with data center and AI at $7.3 billion, up 78%, while client computing fell 5% to $8.1 billion, Intel said in its third-quarter release. Mix explains part of the gain and utilization explains the rest; Intel does not split the two, so it is unclear how much of the recovery survives a softer PC market.
Reported earnings are not yet a series anyone can project. Net income was -$11.03 billion in the June 2026 quarter and +$3.32 billion in September, against operating income of $1.98 billion and $2.52 billion. Even FY2025's small net loss of $0.27 billion sat $1.94 billion above its operating loss. The June quarter is the one that matters: roughly $13 billion separated operating income from net income, alongside a $19.3 billion increase in total liabilities and a $23.9 billion fall in equity. The disclosed debt lines account for only about $5.5 billion of that liability increase.
Cash conversion looks adequate only because the charges are non-cash. Operating cash flow covered net income 1.5 times in FY2021 and 1.9 times in FY2022; in the loss years the ratio means nothing, so the useful test is operating cash flow against capex. Capital spending fell from $25.75 billion in FY2023 to $14.65 billion in FY2025 and averaged about $3.4 billion a quarter in the first three quarters of FY2026. Free cash flow improved because spending fell. Asset turnover went from 0.49 to 0.26 over the same years, so the smaller outlay has not produced more revenue per dollar of assets.
Leverage is not the immediate pressure point. Liabilities were 42.4% of assets at September 2026, the current ratio was 2.21, and cash of $21.85 billion offset $50.14 billion of debt, leaving net debt near $28.3 billion. Quarterly EBITDA of $7.54 billion covers roughly $1.3 billion of annualized interest several times. The August equity offering of about $23 billion is what repaired those numbers: equity rose $27.4 billion in the quarter and assets rose $26.1 billion. It bought time, not capacity, and the Arizona and Ohio schedules imply spending returning toward earlier levels before operating cash flow can carry it.
Working capital is currently a comfort rather than a warning. Receivables of $4.09 billion equal about 23 days of revenue, and inventory turnover held near 0.74 times a quarter, roughly 120 days, on inventory of $13.07 billion. That could turn quickly. Global PC shipments fell 4% year over year in the second quarter, Counterpoint Research estimates, and Intel said the PC slowdown should continue at least through the first half of next year. An October explosion at a fab in Israel is expected to cost about $600 million of revenue across two quarters, which cuts shipments without cutting the depreciation those fabs carry.
Annual metrics, USD bn unless noted
| Metric | FY2021 | FY2022 | FY2023 | FY2024 | FY2025 | Q3 FY2026 (quarter) | Assessment |
|---|---|---|---|---|---|---|---|
| Revenue | 79.02 | 63.05 | 54.23 | 53.10 | 52.85 | 16.51 | Weak: down 33% over four years; quarter up 21% y/y |
| Gross margin % | 55.4 | 42.6 | 40.0 | 32.7 | 34.8 | 43.0 | Adequate: recovering, still below FY2021 |
| Operating margin % | 24.6 | 3.7 | 0.2 | -22.0 | -4.2 | 15.3 | Very weak: near breakeven since FY2022 |
| Net margin % | 25.1 | 12.7 | 3.1 | -35.3 | -0.5 | 20.1 | Very weak: whipsawed by non-operating items |
| ROE % | 22.5 | 8.1 | 1.6 | -18.3 | -0.3 | 3.3 | Very weak |
| ROA % | 12.4 | 4.6 | 0.9 | -9.7 | -0.1 | 1.5 | Very weak |
| Liabilities/assets % | 43.4 | 43.3 | 42.6 | 46.5 | 40.2 | 42.4 | Adequate |
| Current ratio (x) | 2.10 | 1.57 | 1.54 | 1.33 | 2.02 | 2.21 | Adequate |
| Operating cash flow | 29.99 | 15.43 | 11.47 | 8.29 | 9.70 | 5.08 | Weak: one-third of FY2021 |
| Capex (OCF − FCF) | 20.33 | 25.05 | 25.75 | 23.95 | 14.65 | 4.02 | Weak: consuming most operating cash |
| Free cash flow | 9.66 | -9.62 | -14.28 | -15.66 | -4.95 | 1.06 | Very weak |
| OCF/net income (x) | 1.5 | 1.9 | 6.8 | Not meaningful (loss) | Not meaningful (loss) | 1.5 | Not meaningful in loss years |
| Asset turnover (x) | 0.49 | 0.36 | 0.29 | 0.27 | 0.26 | 0.08 | Very weak |
| Inventory turnover (x) | 3.67 | 3.02 | 2.67 | 3.07 | 2.90 | 0.74 | Adequate |
| Receivables turnover (x) | 9.73 | 9.28 | 14.39 | 15.44 | 14.45 | 4.06 | Strong: about 25 days |
The final column is the standalone quarter ended September 30, 2026; its flows are not comparable with full-year figures.
Quarterly metrics, USD bn unless noted
| Metric | Q1 FY25 | Q2 FY25 | Q3 FY25 | Q4 FY25 | Q1 FY26 | Q2 FY26 | Q3 FY26 | Assessment | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Revenue | 12.67 | 12.86 | 13.65 | 13.67 | 13.58 | 16.13 | 16.51 | Scale still modest | Rising since Q1 FY26 |
| Gross margin % | 36.9 | 27.5 | 38.2 | 36.1 | 39.4 | 41.5 | 43.0 | Recovering | Up six straight |
| Net income | -0.82 | -2.92 | 4.06 | -0.59 | -3.73 | -11.03 | 3.32 | Erratic | No trend |
| Total assets | 192.24 | 192.52 | 204.51 | 211.43 | 205.33 | 202.44 | 228.57 | Lifted by offering | Up sharply |
| Total liabilities | 85.83 | 86.77 | 87.78 | 85.07 | 80.34 | 99.30 | 97.02 | Step-up in Q2 FY26 | Elevated |
| Net PP&E | 109.76 | 109.51 | 105.05 | 105.41 | 104.46 | 105.74 | 105.31 | Stable | Flat |
| Cash | 8.95 | 9.64 | 11.14 | 14.27 | 17.25 | 12.87 | 21.85 | Rebuilt | Up |
| Inventory | 12.28 | 11.38 | 11.49 | 11.62 | 12.43 | 12.49 | 13.07 | Highest in series | Rising |
| Goodwill | 24.69 | 23.91 | 23.91 | 23.91 | 20.46 | 20.46 | 20.46 | Written down $4.2bn | Lower |
| Operating cash flow | 0.81 | 2.05 | 2.55 | 4.29 | 1.10 | 7.01 | 5.08 | Positive | Improving |
| Capex (OCF − FCF) | 5.18 | 3.55 | 2.43 | 3.49 | 3.64 | 2.56 | 4.02 | Below build years | Edging up |
| Free cash flow | -4.37 | -1.50 | 0.12 | 0.80 | -2.54 | 4.45 | 1.06 | Positive in four of seven | Improving |
| Current ratio (x) | 1.31 | 1.24 | 1.60 | 2.02 | 2.31 | 1.60 | 2.21 | Healthy | Recovered |
Seven quarters are shown because that is the extent of the reported quarterly series.
What to check
- The June 2026 charge. About $13 billion separated operating income from net income, and the disclosed debt lines explain only a fraction of the $19.3 billion liability increase. Whether the item recurs decides what normalized earnings are.
- Capex against operating cash flow into FY2027. If construction resumes toward the FY2023–24 pace of $24–26 billion while operating cash flow stays near $10 billion, the funding gap reopens and another round of borrowing or share issuance follows.
- Goodwill and inventory together. Goodwill has already fallen $4.2 billion to $20.46 billion; a slower foundry ramp invites further write-downs. Inventory of $13.07 billion heading into a PC downturn is where a demand miss would hit gross margin and cash in the same quarter.
Questions worth following
- Can Intel keep inventory from building as PC demand slows?
Inventory of $13.07 billion is the highest in the reported series and turns only about 0.74 times a quarter. A build into falling PC shipments would cut gross margin and operating cash flow at once, weakening the cash conversion the equity offering bought time to fix.
Generated by DeepVal AI. Data and models may contain errors or be outdated. For research only; not personalized investment advice.
Continue reading the full report
- Key risksWhat could break the investment case
- Valuation and margin of safetyWhat the current price assumes, and how much room for error it leaves
- SummaryOur overall judgement and what to watch next
Free to sign up · 20 research points · unlocking this report uses 10
20 points cover 2 full reports. No card needed.
Research reflects its generation date. Preview the first two modules.
More on Intel Corporation
Intel Corporation
Business, moat and key questions
Open company page