Steel Dynamics' record quarter was a spread story, not a volume story — and that distinction matters

Steel Dynamics reported Q3 2026 net income of $756 million, up 87% year on year on record steel shipments of 3.78 million net tons. Headlines will focus on the record tonnage. The more consequential number is buried two lines lower: the average external selling price rose $102/ton quarter on quarter while the average ferrous scrap cost per ton melted fell $8/ton. That combination widened the company's metal spread by roughly $110/ton, or about 12%, and explains nearly all of the earnings jump.

This is a classic EAF margin quarter. It is also a quarter that tells us relatively little about underlying demand, and quite a lot about how much of SDI's profitability now hinges on the relationship between two prices it only partly controls.

What the numbers actually say

Prior-period sales back-solved from the reported growth rates; shipments and scrap cost derived the same way. Analysis, not company-reported figures. Margins below are my calculations on reported numbers.

• Operating margin: ~15.5% in Q3 2026, versus ~11.5% in Q2 and ~10.5% in Q3 2025.
• Adjusted EBITDA margin: ~18.3%.
• Metal spread: $996/ton, against ~$886/ton in Q2.

The mechanics: why $110/ton did so much work

On 3.78 million tons, a $110/ton spread gain is worth roughly $416 million in gross metal margin versus Q2. Operating income rose about $310 million. The gap between those two figures is the tell: some of the spread gain was absorbed by higher conversion costs, alloy and coating expenses, or weaker results elsewhere in the portfolio (fabrication, metals recycling, the aluminum ramp).

That asymmetry is important. In a quarter where shipments grew only 1.2%, the operating leverage came from price, not volume. SDI sold marginally more steel into a market where its realised price rose nearly 8% sequentially — a combination that is excellent for a quarter but inherently self-limiting. Either price holds and volumes flatten as buyers balk, or price retreats and the spread compresses faster than tonnage can compensate.

The demand picture is narrower than the headline suggests

SDI cites energy, non-residential construction, automotive and industrial as the leading demand sectors. That list is worth reading carefully. It is a list of cyclical, rate-sensitive and policy-sensitive end markets, and it notably omits the two categories that dominated the 2021–22 upcycle: residential construction and white-goods/appliance consumption tied to housing turnover.

Non-residential construction strength is real but concentrated in specific verticals — data centres, warehouses, energy infrastructure, semiconductor fabs. These are large, lumpy projects with long lead times and heavy structural and plate content. They support order books for 12–24 months but do not create the broad-based, high-frequency demand that sustains pricing through a cyclical downturn.

The automotive component deserves scrutiny. US light-vehicle production has been rangebound, and automotive is the most tariff-exposed steel-consuming sector in the country. Steel sold into autos carries a price premium, so its presence in the demand mix flatters the average selling price — but it is also the most vulnerable to a demand shock.

Millett's trade-policy argument is the crux — and the risk

The CEO's outlook leans explicitly on a "supportive U.S. trade environment and reduced unfairly traded import competition." Translated: Section 232 tariffs on steel and aluminium, and the current enforcement posture, are doing structural work in this earnings model.

That is a legitimate and material point. It is also the single largest exogenous variable in SDI's 2027 outlook, and it is a policy variable, not a market one. A $1,400/ton average selling price is high by historical standards for a mill with SDI's product mix (long products, flat-rolled, and a growing value-added coating book). Part of that premium is domestic scarcity created by trade barriers.

For industry readers, the practical implication is this: SDI's earnings quality now depends on the persistence of an import barrier as much as on the persistence of demand. Investors and counterparties should treat the current spread as policy-supported rather than structurally earned, and stress-test accordingly.

What to watch in Q4

Three items will determine whether Q3 was a peak or a plateau:

1. Scrap direction. Scrap settled up $20–30/gross ton in most US locations in October, per trade reporting. If that flows through to SDI's Q4 cost per ton melted, the spread compresses even if steel prices hold. Every $10/ton of scrap inflation costs roughly $38 million on 3.78 million tons.
2. Aluminium Dynamics. Millett explicitly groups aluminium with steel in the outlook. The flat-rolled aluminium operation is still in ramp and is a capital and opex drag before it is a margin contributor. Watch for any disclosure on volume, qualification and cost absorption.
3. The price/volume trade-off. If realised prices hold near $1,400 and shipments stay at record levels, the quarter was a genuine cyclical expansion. If prices slip while shipments hold, it was a spread peak.

The bottom line

SDI executed well and delivered a genuinely strong quarter — an 18% EBITDA margin is a good outcome by any standard. But the record shipment headline obscures the real driver: a $110/ton spread expansion driven by rising steel prices against falling scrap costs, in a market protected by trade policy. That is a favourable configuration, not a durable moat. The company's own demand narrative — data centres, infrastructure, energy — is credible and real, but it is narrower and more project-driven than the 2021–22 cycle, and it cannot by itself offset a spread reversal. The question for Q4 is not whether SDI can ship more steel. It is whether the price-scrap relationship that made this quarter can survive an October scrap increase and a policy environment that is supportive today but not guaranteed tomorrow.

All company-reported figures are as reported by SteelOrbis, citing SDI's Q3 2026 release. The primary source — SEC Form 8-K accession 0001104659-26-117217, Exhibit 99.1 — should be consulted for segment detail, the non-GAAP reconciliation and the income statement before these figures are used for valuation. Derived ratios, spread calculations and prior-period estimates are the author's analysis.