EV/EBITDA vs. P/E: what AT&T's $163B in debt hides
AT&T's P/E of 8.4 looks cheap on its own, but EV/EBITDA of 5.9 tells a fuller story once $163B of debt enters the calculation.
The number P/E doesn't see
AT&T trades at a trailing P/E of about 8.4x — a number that, on its own, reads as cheap next to a market where the median stock sits well above that. AT&T's EV/EBITDA, the other valuation multiple that shows up on most stock screeners, is 5.9x. Both are "cheap" in isolation, but they aren't measuring the same thing, and the gap between them is not noise. It's $162.9 billion of debt, sitting on AT&T's balance sheet, that P/E has no mechanism to show.
P/E divides share price by earnings per share. Earnings per share is what's left over after interest on that debt has already been paid. So a company that carries a lot of debt can post a low, flattering P/E for a reason that has nothing to do with how cheap the underlying business is — the earnings number has already absorbed the cost of that leverage, and the multiple only prices what's left for equity holders. EV/EBITDA is built to strip that distortion back out.
What EV/EBITDA actually adds up
Enterprise value is market capitalization, plus total debt, minus cash — in effect, what it would cost to buy the whole company and retire its capital structure, not just its stock. EBITDA is earnings before interest, taxes, depreciation and amortization — operating cash generation before either the tax authority or the lenders take their cut. Divide one by the other and you get a multiple that prices the operating business the same way whether it was built on debt or equity, because both the debt (in the numerator) and the interest it generates (excluded from the denominator) are accounted for the same way for every company.
That's the whole idea: EV/EBITDA is capital-structure-neutral in a way P/E isn't. Two companies with identical factories, identical revenue and identical operating margins can post very different P/E ratios purely because one financed its factories with debt and the other with equity — and EV/EBITDA is largely indifferent to that choice.
Same company, two different stories
AT&T's numbers make the mechanism concrete. Its enterprise value runs 83% above its market cap, entirely because of debt sitting on top of the equity value:
| Metric | AT&T (T) | Carnival (CCL) | Alphabet (GOOGL) |
|---|---|---|---|
| P/E (trailing) | ~8.4x | ~9.5x | ~17.1x |
| EV/EBITDA | ~5.9x | ~7.5x | ~13.1x |
| Market cap | ~$174.2B | ~$29.9B | ~$4.18T |
| Enterprise value | ~$319.5B | ~$53.8B | ~$4.24T |
| EV vs. market cap | +83% | +80% | +1.4% |
Carnival shows the same pattern for the same reason — a heavily debt-financed balance sheet post-pandemic pushes its enterprise value 80% above its market cap, and its EV/EBITDA sits meaningfully below its P/E as a result. Alphabet is the control case: with debt largely offset by cash and short-term investments, its enterprise value is barely above its market cap, so leverage isn't doing much to separate the two multiples. They still aren't identical — taxes, depreciation intensity and other items P/E and EBITDA treat differently see to that — but the capital-structure gap that dominates AT&T's and Carnival's numbers is mostly absent here.
Where the metric still needs guardrails
None of this makes EV/EBITDA a clean substitute for P/E in every case. Zorro's screener — which runs EV/EBITDA as one of 20 metrics across roughly 3,445 stocks — excludes financial companies from it entirely, because a bank's business is lending money, not running operations with an EBITDA line that means anything comparable to an industrial company's. The metric is also clamped to a 0–200x range before it's used in ranking, because a handful of near-zero-EBITDA companies would otherwise produce multiples in the thousands that distort everything they're compared against. A newly listed company is a common source of that noise: fresh off an IPO, EBITDA is often depressed or negative from one-time listing costs and early-stage spending, which is one more reason the recent-IPO cohort tends to get judged on different terms — first-day pop, lockup timing — before multiples like this one are stable enough to mean much.
Cheap on EV/EBITDA isn't always a bargain
There's a sharper trap than noisy data: a very low EV/EBITDA can mean distress rather than a bargain. A company priced at 2–3x EBITDA is often being priced that way because the market expects the debt behind that low multiple to not get fully repaid — distressed shippers and airlines coming out of a down cycle are the textbook case. That's exactly why a growth-oriented screen puts a floor under EV/EBITDA rather than rewarding the lowest number available: Zorro's GARP screen requires EV/EBITDA of at least 5x precisely so that a distressed, cheaply-priced-for-a-reason company can't rank ahead of a business growing on a sound balance sheet. The same multiple that flags a bargain in a value screen flags a red flag in a growth screen when it drops too far — context, not the raw number, is what tells you which one you're looking at.
Read on its own, a single EV/EBITDA number is just a ratio. Read next to P/E, on the same company, it's a way of asking a specific question: how much of what looks cheap is really the business, and how much is just the debt sitting underneath it.