How to Price Capital-Light vs. Capital-Heavy Businesses

Put two 10-Ks side by side and strip off the cover pages. Company A and Company B, both reporting a net income margin within a point of each other, both growing revenue in the low double digits. If you stopped reading at the income statement, you'd have every reason to expect the market to price these two businesses similarly — maybe a modest premium for whichever one is growing a hair faster. You'd be wrong, and not because of anything happening above the operating-income line.

The divergence shows up two statements later, in the section most investors skim on their way to the per-share numbers: capital expenditures. Company A spends a rounding error of its revenue rebuilding what it already owns. Company B spends nearly half. Same accounting treatment on the income statement — capex doesn't touch net income directly, it flows through as depreciation over years — but a completely different claim on the cash that net income theoretically represents. And that claim is the entire difference between what a business reports and what it can actually hand to its owners.

This is the problem with comparing valuation multiples across capital intensity without adjusting for it first. A P/E ratio treats a dollar of Adobe's (ADBE) net income and a dollar of Duke Energy's (DUK) net income as interchangeable. They aren't. I'll walk through both companies' fiscal 2024 filings, build owner earnings for each, and show which multiples hold up once you account for what it actually costs each business to keep functioning — and which ones quietly lie to you.

Why Capex Intensity Breaks the Comparison

Start with the accounting mechanics, because they explain why this distortion is so persistent. Net income and capital expenditures are almost entirely disconnected in the year they occur. Capex hits the balance sheet as an asset, then depreciates over its useful life — five years for equipment, twenty or thirty for a power plant, effectively never for a decades-old brand. The cash goes out the door in year one. The income statement expense trickles out over a decade or three. A business can be spending enormous sums just to stand still, and show healthy net income the entire time.

Capex-to-revenue is the fastest diagnostic. A business spending 1–3% of revenue on capital expenditures is asset-light — the moat, if there is one, is doing its work through the P&L, not the balance sheet. A business spending 30–45% is capital-heavy, and a meaningful share of every dollar in reported earnings is spoken for before an owner ever sees it. Screening on P/E or even P/S without checking this ratio first means comparing two numbers labeled the same way. Same label. Very different things being measured underneath it.

Price and value aren't the same question here, and it's worth separating them explicitly. Price is what the market will trade the stock for today — observable, unambiguous, updated every second the exchange is open. Value is what the cash the business generates for an owner, after it pays whatever capital reinvestment the business genuinely requires, is worth today. Two businesses can trade at an identical price-to-earnings multiple and be priced completely differently relative to value, because the earnings figure in the denominator means something different in each case.

Owner Earnings = Net Income + Depreciation & Amortization − Capital Expenditures (A simplification of Buffett's original formulation, which also nets out working capital changes. For businesses where capex materially exceeds or falls short of D&A, the direction of that gap is often more informative than the owner earnings figure itself.)

Adobe (ADBE): The Capital-Light Case

Adobe's Form 10-K for fiscal year 2024 (ended November 29, 2024) reported revenue of approximately $21.5 billion and net income of approximately $5.56 billion — a net margin above 25%, itself a signal of pricing power in a subscription model with minimal marginal delivery cost. Diluted shares outstanding: roughly 449 million, putting diluted EPS at $12.36.

Now check the capex line. Adobe spent approximately $183 million on capital expenditures in fiscal 2024 — a capex-to-revenue ratio under 1%. Depreciation and amortization for the year ran to roughly $857 million, more than four and a half times what the company actually spent to maintain and expand its physical and technical infrastructure. That gap is the asset-light signature: the accounting expense running well ahead of the actual cash outlay, because most of what Adobe sells doesn't require rebuilding a factory to keep selling it.

Adobe FY2024 (10-K, fiscal year ended Nov 29, 2024) Net Income: $5.56B Depreciation & Amortization: + $0.86B Capital Expenditures: − $0.18B Owner Earnings: $6.23B Owner Earnings per share (÷449M shares): ≈ $13.88 Diluted GAAP EPS: $12.36

Owner earnings per share running above reported EPS is the tell. Adobe's accounting depreciation is overstating the actual economic cost of staying competitive, which means net income — already strong — understates what the business is generating for an owner. A P/E ratio built on GAAP earnings is, if anything, slightly conservative here.

Duke Energy (DUK): The Capital-Heavy Case

Duke Energy's Form 10-K for fiscal year 2024 reported total operating revenues of approximately $30.4 billion and net income attributable to Duke Energy Corporation of approximately $4.39 billion — GAAP diluted EPS of $5.71, on roughly 769 million diluted shares. On a standalone basis, that net margin (about 14%) looks unremarkable next to Adobe's. The real gap doesn't show up until the cash flow statement.

Duke's 10-K shows net cash used in investing activities of approximately $12.5 billion for 2024, the large majority of which is capital spending on the regulated grid and generation fleet — utilities typically have little other investing activity to speak of. Depreciation and amortization for the year ran approximately $5.8 billion. Run the same owner earnings formula:

Duke Energy FY2024 (10-K, fiscal year ended Dec 31, 2024) Net Income (attributable to DUK): $4.39B Depreciation & Amortization: + $5.79B Capital Expenditures (approx.): − $12.50B Owner Earnings: ≈ −$2.32B GAAP Diluted EPS: $5.71

Same formula, same fiscal year discipline, and the sign flips. Duke reported $4.39 billion of positive net income and, on the owner earnings math, generated roughly negative $2.3 billion for an owner who wanted to pull cash out of the business rather than let it keep reinvesting. That's not a business in trouble — regulated utilities earn a contractual return on the capital they deploy into rate base, and Duke's growth capex is largely pre-approved spending that will earn back a regulated return over decades. But it does mean the net income figure and the cash-in-your-pocket figure have almost nothing to do with each other in a given year, and a P/E ratio computed on the former tells you very little about the latter.

I'm genuinely unsettled about how to treat this cleanly. Strict owner earnings math treats a regulated utility's growth capex the same way it treats a steel mill's — as cash that isn't available to the owner, full stop. But growth capex on a rate base that's contractually allowed to earn something like 9–10% isn't destroying value the way discretionary growth spending at an unregulated capital-heavy business might. I don't think the formula is wrong, exactly. I think it's incomplete for regulated businesses, and the honest fix is to weigh rate base growth and allowed ROE alongside owner earnings, not to trust either number alone.

Which Multiple Actually Applies

Once you've run the owner earnings math, the choice of multiple mostly makes itself. For capital-light businesses like Adobe, P/E, FCF yield, and owner-earnings multiples converge closely enough that any of them will get you in the right neighborhood — the fundamental analysis is doing most of the work in the earnings quality, not the multiple selection.

For capital-heavy businesses, P/E is close to useless as a standalone signal, and EV/EBITDA is worse — EBITDA adds back depreciation without asking whether that depreciation reflects real, unavoidable reinvestment. A regulated utility's EBITDA looks robust precisely because the metric ignores the capex line doing the damage. Better fits: regulated rate base growth and allowed return on equity, both disclosed in the 10-K's regulatory matters section; dividend coverage against regulated cash flows; and price-to-book against rate base, which more directly captures what you're actually buying — a claim on a regulated earnings stream, not a claim on free cash today.

This is also where asset-light businesses with genuine moats earn their premium multiples honestly rather than through screener noise. A moat that costs almost nothing to defend converts a larger share of every reported dollar into cash an owner can actually use. A moat — even a real, durable one — sitting on top of a capital-heavy asset base has to fund its own defense before anything reaches the owner. Same moat quality, in principle. Wildly different claim on the cash.

None of this produces one clean number you can trust without a second look. A fair value estimate built on owner earnings is a compass, not a GPS coordinate — true of any business, but especially true once capital intensity is doing this much work behind the scenes to separate what a company reports from what it can actually pay out.

Key Takeaways

• Capex-to-revenue is the fastest screen for whether a P/E or FCF-based multiple will mislead you. Under 3% of revenue is asset-light; above 20–25% requires switching frameworks entirely.
• Adobe's fiscal 2024 owner earnings (~$6.23B, or ~$13.88/share) ran above its GAAP diluted EPS ($12.36) — depreciation is overstating the real cost of staying competitive, so P/E is a reasonable starting multiple.
• Duke Energy's fiscal 2024 owner earnings were negative (~−$2.3B) despite $4.39B in reported net income — capex of roughly $12.5B against $5.8B of depreciation means P/E and EV/EBITDA both understate how much of net income is already spoken for.
• For capital-heavy, regulated businesses, rate base growth, allowed ROE, and price-to-book are more honest starting points than earnings-based multiples alone.