EV/EBITDA explained: valuing the whole business, not just the shares

The P/E ratio values a company's shares. EV/EBITDA values the whole business — the shares and the debt together — against the profit the operations produce before financing, tax and accounting charges for wear and tear. Professional investors and acquirers lean on it because it puts companies with very different balance sheets on the same footing. It is also the first multiple in our valuation factor. This guide explains how it is built, why it so often disagrees with the P/E, and where it goes blind.

The two halves

Enterprise value (EV) is what you would have to pay to own the entire business: the market value of the shares, plus the company's debt, minus its cash. Debt is added because a buyer would inherit it; cash is subtracted because it comes with the business and could pay some of that debt off.

EBITDA stands for earnings before interest, tax, depreciation and amortisation. We build it as operating income plus depreciation and amortisation, both read from the company's filings. Leaving out interest makes EBITDA independent of how the company is financed; leaving out depreciation and amortisation removes charges that reflect past spending rather than this year's cash.

Divide one by the other and you get the number of years of current operating profit the whole business costs. Coca-Cola's EV/EBITDA was 27.6 at the 14 October 2026 close; PepsiCo's was 11.95. On this measure, a buyer of PepsiCo pays well under half as much for each dollar of operating profit. Margins cannot explain that gap, because the multiple is already per dollar of profit: it is the extra the market will pay for each dollar of Coca-Cola's earnings, and our Coca-Cola vs PepsiCo comparison sets out the growth and returns behind it.

Why it is often more useful than P/E

Imagine two companies with identical operations. One has borrowed heavily and bought back half its shares; the other has no debt. The indebted one will show a lower P/E — fewer shares, so higher earnings per share — even though the businesses are the same. EV/EBITDA sees through this, because the debt is added back into the price. For the same reason it is the multiple acquirers use: they buy the whole company, debt and all.

It also sidesteps two kinds of noise in net income: interest costs that depend on financing choices, and gains or losses from outside the operating business.

When P/E and EV/EBITDA disagree

For most companies EV/EBITDA comes out below the P/E, because EBITDA is a bigger number than net income. When the two diverge sharply, the gap itself is information:

EV/EBITDA above the P/E is unusual, and it nearly always means part of net income came from outside the operations. Alphabet earned $244.21bn of net income over the past twelve months against operating income of $147.63bn. The P/E of 17.44 is flattered by that extra profit; the EV/EBITDA of 25.01, built from operations only, is the steadier read of what the business costs.

EV/EBITDA well below the P/E usually means large depreciation — common for companies that are spending heavily on physical assets. Meta and Amazon are both building data centres at enormous scale. Their depreciation charges, added back in EBITDA, are real costs that will recur for as long as they keep building. Micron adds a second effect: it holds more cash ($38.36bn) than debt ($5.18bn), so its enterprise value is below its market value.

A dash in the P/E column with a number beside it is the loss-maker case. Intel lost money after tax, so it has no meaningful P/E, but its operations still generate positive EBITDA once depreciation is added back. EV/EBITDA is one of the few valuation multiples that stays usable here — with the caveat in the next section.

The blind spots

EV/EBITDA is popular partly because it flatters. Keep four weaknesses in mind:

1. Depreciation is a real cost. Equipment wears out and must be replaced. A company that has to spend heavily every year just to stand still can look cheap on EBITDA while generating little free cash. Always look at capital spending, or at the free cash flow yield, alongside it.
2. Interest and tax are real too. EBITDA ignores them, so a heavily indebted company can look affordable on EV/EBITDA and still struggle to pay its lenders.
3. "Adjusted" EBITDA is a different number. Many companies publish an adjusted EBITDA that adds back stock-based pay and one-off costs. Ours is built from reported operating income, which already deducts both, so it is usually lower than the figure a company quotes — and the multiple correspondingly higher.
4. Cash is not all cash. We subtract cash and cash equivalents only. Companies such as Apple, Alphabet, Microsoft and Nvidia hold much of their liquidity in marketable securities, which our figure leaves out, so their enterprise values — and their EV/EBITDA multiples — are somewhat overstated on our pages. We would rather overstate a multiple than subtract an investment portfolio we cannot verify as liquid.

Where EV/EBITDA does not work at all

For banks it is meaningless. A bank's "debt" is mostly its customers' deposits — the raw material of its business — and it has no operating-income line to build EBITDA from. That is why our JPMorgan and Bank of America pages show no EV/EBITDA; how to compare bank stocks explains what to use instead. It is meaningless when EBITDA is zero or negative, and we withhold it then; when EBITDA is merely small, as at Palantir, Tesla and AMD, the multiple is published but swings sharply with small changes in profit — see the numbers we refuse to print.

How to use it

Compare EV/EBITDA within an industry: capital-hungry businesses and asset-light software companies settle at very different levels. Put it next to the P/E, and treat a large gap as a question to answer before trusting either. And remember that it is one of four votes in our valuation factor — alongside price-to-sales, price-to-book and free-cash-flow yield — rather than a verdict on its own, as the six-factor score explained sets out. To see all four at work on two very different businesses, try Alphabet vs Amazon.