Why EV/EBIT Beats EV/EBITDA for Heavy Industry

EV/EBITDA is the more forgiving multiple. That's not a compliment. For a business that has to pour billions of dollars a year back into blast furnaces and rolling mills just to keep functioning, forgiveness is exactly the wrong quality to want from a valuation ratio. EBITDA adds back depreciation and amortization before anyone divides by anything, which means the multiple never has to look at what it actually costs Nucor (NUE) to remain a steel company year after year.

That's the whole argument in one sentence, and it deserves more than one sentence. EV/EBIT and EV/EBITDA are separated by exactly one line item — depreciation and amortization — and for a capital-light software business that line item is small enough to round away. For a steel producer, it isn't. I'll build both multiples from Nucor's fiscal 2024 filings, show how far apart they land, and make the case that for heavy industry specifically, EV/EBIT is the multiple that tells you the truth and EV/EBITDA is the one being polite about it.

Price and value diverge here in a specific, mechanical way, not a mysterious one. Price is whatever the market pays for Nucor's shares on a given afternoon. Value is whether that price is cheap or expensive relative to what the business can actually generate for its owners — and answering that honestly means picking an earnings denominator that doesn't quietly excuse away a real cost. Get the multiple wrong and you can talk yourself into thinking a mediocre price is a good one.

What the Two Multiples Actually Measure

Start with the mechanics, because the entire disagreement between these two multiples lives in one accounting line.

EV/EBIT = Enterprise Value ÷ Earnings Before Interest and Taxes EV/EBITDA = Enterprise Value ÷ (Earnings Before Interest and Taxes + Depreciation and Amortization) Enterprise Value = Market Capitalization + Total Debt − Cash and Cash Equivalents (and near-cash investments)

Both multiples strip out capital structure — that's what the enterprise-value numerator is for, so a company's mix of debt and equity financing doesn't distort the comparison. Both strip out taxes, on the theory that a buyer of the whole business cares about pre-tax earning power before layering on its own tax situation. Where the two multiples part ways is depreciation and amortization. EV/EBIT treats D&A as a real expense, because in a capital-intensive business it's standing in for capital the company already spent and has to keep spending to replace. EV/EBITDA adds it back, treating it the way you'd treat a genuinely non-cash accounting artifact — the way you might treat a one-time impairment charge.

For an asset-light business, that's a defensible simplification, covered in more depth in our guide to pricing capital-light and capital-heavy businesses. D&A on a company spending 2% of revenue on fixed assets barely moves the multiple either way. Add it back or don't; the answer changes by a rounding error. Heavy industry is not that case, and Nucor is as clean an example as the market offers.

Building Nucor's Numbers From the 2024 10-K

Nucor reported $30.73 billion in net sales for fiscal 2024, down from a stronger 2023 as steel prices cooled off their post-pandemic highs. Pretax earnings came in at $2,902 million, built on $228 million of interest expense against $258 million of interest income — Nucor actually earned more on its cash in 2024 than it paid on its debt, a detail worth noting and then setting aside, since EBIT strips out both sides of that line anyway. Back out that net $30 million of interest income and EBIT for the year works out to roughly $2,872 million.

Depreciation and amortization for 2024 totaled $1,356 million: $1,094 million of depreciation on Nucor's mills and mini-mills, $262 million of amortization on intangibles built up through the company's acquisition history. Add that back to EBIT and EBITDA comes out to roughly $4,228 million. Already, before either valuation multiple gets computed, the earnings denominator has grown by 47% depending on which one you pick.

Nucor closed out fiscal 2024 with roughly 240 million diluted shares outstanding — $2.03 billion of net earnings attributable to Nucor stockholders divided by the $8.46 of diluted EPS the company reported. With shares trading near $117 in early January 2025, not far removed from the fiscal year-end close, that puts market capitalization at roughly $28.0 billion. Add back the roughly $6.2 billion of debt disclosed in the 10-K's fair-value footnote, subtract the $3.4 billion of cash and short-term investments on the balance sheet at year-end, and enterprise value lands close to $30.8 billion.

Divide that $30.8 billion enterprise value by EBIT and you get roughly 10.7x. Divide the same enterprise value by EBITDA and you get roughly 7.3x. Same company, same numerator, same afternoon's stock price. Different earnings idea entirely — and the multiple you'd quote someone depends on whether $1.356 billion a year of depreciation counts as a real cost or an accounting nuisance.

Why the Gap Is the Point, Not a Rounding Error

A 47% gap between two multiples measuring the same business isn't noise. Not even close. It's the exact size of the distortion EV/EBITDA introduces for a company like Nucor, and it moves in a predictable direction: EBITDA always looks cheaper than EBIT for a capital-heavy business, never the other way around, because D&A gets added to the denominator while enterprise value doesn't move. Screen a universe of stocks on EV/EBITDA without checking capital intensity first, and steel mills, railroads, and utilities will systematically screen cheaper than they actually are next to a software company carrying the same reported multiple.

But that's not a flaw in EV/EBITDA generally. It's a flaw in using it without a capital-intensity check first, and the fix is cheap: look at capex-to-revenue before you trust either multiple's answer.

I'll flag the honest limit here directly. Nucor spent $3.17 billion on capital expenditures in 2024 — more than double the $1.356 billion of depreciation running through the income statement. Some of that gap is growth capex, new mill capacity rather than replacing worn-out equipment, and Nucor's disclosures don't cleanly separate the two. If a meaningful share of that spending is optional growth rather than required maintenance, EV/EBIT actually understates how cheap Nucor's standing-still economics really are. The honest fix, at that point, isn't EBIT or EBITDA at all — it's owner earnings built from actual maintenance capex, a number neither multiple gives you on its own.

There's a second distortion worth naming before moving on, and it has nothing to do with depreciation. Steel earnings are cyclical in a way software earnings mostly aren't. Nucor's $2,872 million of trailing EBIT reflects fiscal 2024 steel prices specifically, a year the company itself described as cooler than 2023. Run the same EV/EBIT calculation at the top of a steel cycle, when EBIT might run 50% higher on the same asset base, and the multiple looks far cheaper without the business having changed at all. Neither EV/EBIT nor EV/EBITDA fixes that on its own — both are single-year snapshots — which is why cyclical businesses deserve the additional step of normalizing the earnings figure across a full cycle before trusting either multiple. Our piece on normalizing earnings for a cyclical business walks through exactly that adjustment.

That cyclicality isn't hypothetical. Nucor raised its hot-rolled coil spot price for a fourth consecutive week on August 17, 2026, a cumulative $45-per-ton increase to $1,170 per ton on tight domestic supply, and the stock itself has traded above $240 through late August 2026 — more than double the roughly $117 level this post's fiscal 2024 figures were built from. Whatever EBIT Nucor is actually running at today almost certainly sits well above the $2,872 million base used above, which means today's real EV/EBIT isn't the 10.7x calculated here. The arithmetic in this post doesn't expire. The specific multiple attached to a specific fiscal year does, and fast.

When EV/EBITDA Is the Right Tool for the Job

None of this makes EV/EBITDA a bad multiple in general — it makes it a multiple that needs a capital-intensity check before you trust it. For a business spending low single digits of revenue on capital expenditures, the gap between EV/EBIT and EV/EBITDA compresses to almost nothing, and EBITDA's other advantage — it's harder to distort through depreciation-schedule choices, and it travels better across companies using different useful-life assumptions for similar assets — starts to matter more than the small cost it's ignoring. EV/EBITDA earns its keep as a cross-company screening tool precisely in the businesses where the D&A line is small enough not to matter.

So check capex-to-revenue before you pick a multiple, not after. Above roughly 10–15% of revenue in sustained capital spending, lean on EV/EBIT, or better yet, enterprise value divided by a capex-adjusted earnings figure directly. Below that, EV/EBITDA's simplicity is a reasonable trade for the small distortion it introduces. Nucor, spending north of 10% of revenue on capex every year this decade, sits well inside the range where the multiple you choose changes the conclusion — not the direction of the conclusion, necessarily, but how much margin of safety you think you're actually getting.

None of this — not the 10.7x, not the 7.3x, not the cyclical adjustment sitting underneath both — should be read as a coordinate you can plug in and trust to the decimal. Treat any multiple, EV/EBIT included, as a compass rather than a GPS reading: it tells you the general direction a business's valuation is pointing, built on assumptions about depreciation, cycle position, and growth capex that are each individually debatable. That's a feature of honest valuation work, not a shortcoming of this particular exercise.

Key Takeaways

• EV/EBIT and EV/EBITDA differ by exactly one line item, depreciation and amortization, and that line item is large enough at a capital-heavy business to move the multiple by half or more.
• Nucor's fiscal 2024 numbers put EV/EBIT at roughly 10.7x against EV/EBITDA's roughly 7.3x: the same enterprise value, divided by two different ideas of what the business actually earns.
• Screening on EV/EBITDA without checking capital intensity first will systematically make steel mills, railroads, and utilities look cheaper than they are next to asset-light businesses carrying the same multiple.
• Even EV/EBIT isn't the last word once capex is running well above depreciation. That's when a capex-adjusted earnings figure, built from real maintenance spending, becomes the more honest number.