A chipmaker's cost of capital runs more than double a utility's

The number Intel's spending didn't clear

Intel spent $14.6 billion in fiscal 2025 building factories and buying equipment.¹ The company still closed the year with a net loss, $(0.06) of earnings per share, while carrying $46.6 billion of debt.¹ A company posting a net loss did not clear any positive return requirement, let alone the specific one its lenders and shareholders were actually charging it. That gap, between what a company earns and what its money is priced to require, is the hurdle this lesson builds: the cost of capital.

Aswath Damodaran, the NYU Stern professor, arrives at that hurdle by naming three questions before answering them with one number. What does it cost a company to raise money? What is an investor giving up by investing here instead of an equally risky alternative? What rate should a future dollar be discounted at, to price what it is worth today? For an outside investor, the answer to all three is an opportunity cost, "the rate of return that they would expect to make in other investments of equivalent risk."² For the company raising it, that same number is what the money costs. For a project inside the company, it is the hurdle rate the project has to beat.

The return-on-capital lesson in this course built the other half of that comparison: after-tax profit, divided by the money it took to produce it. It asked whether that return was good by comparing it to a hurdle this course had not yet built. Cost of capital is that hurdle, priced from two sources: what a company pays its lenders, and what it must return to its shareholders.

Why borrowed money is cheaper than it looks

A dollar of interest a company pays a lender is, in the United States, a dollar it can deduct before computing the tax it owes: "there shall be allowed as a deduction all interest paid or accrued within the taxable year on indebtedness."³ A separate cap limits the deduction to roughly 30 percent of taxable income for a highly indebted company. Most large, profitable companies never reach it.⁴

That deduction is why the cost of debt is computed after tax: the long-term borrowing rate, times one minus the tax rate. Use the marginal, statutory rate the tax code sets, not a company's effective or book rate for the year, which can swing with one-time items unrelated to how the company actually finances itself.⁵ Say you can borrow today at 6 percent, and the statutory rate you owe on the next dollar of income is 25 percent. Every dollar of interest you pay saves you 25 cents of tax. Your after-tax cost of debt is not 6 percent. It is 6 percent times 0.75, or 4.5 percent.

The tax shield narrows the gap between industries that price very differently on the equity side. Aswath Damodaran's live industry dataset, fetched August 2, 2026, puts the after-tax cost of debt at 3.55 percent for the general utility industry and 3.97 percent for semiconductor companies, close together.⁶ Lenders mostly care whether they get paid back, not how much upside a business has, so debt prices closer to a floor even where equity does not.

A return nobody can actually observe

Debt is a contract: a company promises a lender a fixed payment. Equity has no such contract. A shareholder is owed nothing beyond what the business generates after every other claim is paid. So the cost of equity is the required return given that risk, inclusive of both dividend and price appreciation, explicitly not the dividend yield alone or the earnings-to-price ratio.⁵ Unlike interest, that return is never paid out, so it has to be estimated. The standard tool for estimating it is the capital asset pricing model, CAPM: a stock's required return sits above the risk-free rate by beta, its sensitivity to the overall market, scaled by the extra return investors demand for holding the market at all.⁷

CAPM is standard, but it is not accurate, and the economists who established that are honest about the gap. Eugene Fama and Kenneth French tested CAPM's central prediction, that required return rises with beta in a straight line, against decades of US stock returns. The actual relationship is flatter: high-beta stocks earn less than CAPM says they should, low-beta stocks earn more.⁷ The inputs are uncertain enough to swamp the answer on their own. The average annual US equity risk premium is 8.3 percent, but its standard error is wide enough that a two-standard-error range runs from 3.5 to 13.1 percent, "sufficient to make most projects appear either profitable or unprofitable."⁷ Fama and French still teach CAPM, "but we also warn students that despite its seductive simplicity, the CAPM's empirical problems probably invalidate its use in applications."⁷ It survives because nothing has replaced it, not because its problems were resolved.

A real regulator setting a real required return shows what that means. The Federal Energy Regulatory Commission sets utilities' allowed return on equity, and for years leaned mainly on one model, discounted cash flow. In 2018, doubting that model amid unusually low bond yields, FERC proposed averaging four instead: DCF, CAPM, an "expected earnings" model, and a risk-premium model.⁸ Four standard models, applied to the same utilities in the same proceeding, produced four different numbers. FERC's response was not to pick the "true" one. It was to average them. That standard traces back further than any of the four models. In 1944 the Supreme Court held that a utility's allowed return has to be set in opportunity-cost terms:

Hope Natural Gas sets an opportunity-cost standard without specifying how to estimate it, and as FERC's own four-model proceeding shows, that "how" is still contested more than eighty years later. "What a company's own regulators call fair" is this standard: a return set by comparison, not produced by a formula.

One hurdle built from two different prices

A company is financed by lenders and shareholders at once, so its cost of capital blends the two prices, weighted by how much of the company each one actually finances:

Market value, not book value, belongs in those weights. Book equity is an accounting residual, small, negative, or stale, and weighting by it is "not only irrelevant when it comes to cost of capital but come with problems that can be insurmountable."² A low-risk business and a high-risk one end up with very different hurdles, not only because their costs of debt and equity differ but because they finance themselves in different proportions. The general utility industry funds nearly half of itself with cheap debt. Semiconductor companies finance themselves almost entirely with pricier equity.

A semiconductor company's cost of capital, 10.55 percent, runs more than double a utility's, 4.36 percent. That gap is why the same return on invested capital means something different depending on what financed it.

The return-on-capital lesson stated the rule this hurdle enforces: a business earning more than its cost of capital creates value as it grows, and one earning less destroys it.¹⁰ That rule is not a claim about one company. Across roughly 4,000 of the world's largest public companies by revenue, from 2005 to 2009 the ten biggest averaged an 8-point gap between return on invested capital and cost of capital. The next 90 largest averaged 4 points, the smallest 3,500 barely more than 1. By the following decade the smallest group's gap had approached zero, and its share of global economic profit fell from 19 percent to 3 percent.¹¹ The size of that gap, not the return alone, tracks how much economic profit a company actually earns. Intel's fiscal 2025 result, a net loss on $14.6 billion of spending, is what the gap looks like when it goes negative: a return that did not clear zero.