Industry Synthesis 001 · Steel & Metals
What four steel producers reveal about capital architecture, operating flexibility and the structure of the next decision.
Disclosure
Independent research. Not commissioned by, and not endorsed by, any company discussed. This document is not an investment recommendation and contains no buy/sell, valuation or price-target conclusion.
Research question
The cluster suggests that current profitability alone is a weak description of decision capacity. A more useful discriminator is the structure of the next decision: what the next unit of capital must accomplish, how reversible the commitment remains, and which parts of the operating system must absorb the consequences.
Executive map
| Company | FY2025 revenue | EBITDA / Adj. EBITDA | Shipments | EBITDA/shipment proxy | CAPEX | Capital / operating state |
|---|---|---|---|---|---|---|
| Gerdau | $12.507bn | $1.804bn | 11.630m t | ~$155/t | $1.092bn | Hybrid routes; Brazil competitiveness reset / NA earnings strength |
| Nucor | $32.494bn | $4.174bn | 26.615m t | ~$157/t | $3.422bn | EAF/downstream; concurrent greenfields + liquidity + distributions |
| Steel Dynamics | $18.2bn | $2.151bn | 13.749m t | ~$156/t | $0.948bn | High-utilization steel; marginal growth capital shifts to aluminum |
| Cleveland-Cliffs | $18.6bn | $0.037bn | 16.2m t | ~$2/t | $0.561bn | Integrated footprint; deleveraging/idling/repricing after 2025 compression |
Gerdau figures are converted from BRL to USD at Gerdau's disclosed FY2025 average rate of R$5.5855/US$1, for single-currency comparability only. Original reported figures: R$69.859bn revenue, R$10.074bn EBITDA, R$6.1bn CAPEX.
The ~$155–157/t clustering across Gerdau, Nucor and Steel Dynamics is treated below as an analytical coincidence worth investigating — not as causal similarity.
Opening finding
Similar current metrics do not imply similar future action space.
Companies
Capital state: low leverage (0.76x net debt/EBITDA); capital pulled toward the structurally weaker node (Brazil), not the strongest current earner (North America).
Strongest observed fact: 80% of 2Q26 CAPEX directed to Brazil, 56% of it classified as competitiveness spend.
Central inference: group-level geographic optionality coexists with materially different local capital requirements.
Strongest unknown: no public hurdle rate or cross-regional capital ranking.
Full research object →Capital state: FY2025 FCF of -$188m alongside $2.699bn cash, an undrawn $2.25bn revolver and $1.212bn returned to shareholders.
Strongest observed fact: $3.422bn CAPEX committed to concurrent greenfields (incl. West Virginia sheet mill) while distributions continued.
Central inference: negative FCF reads as the cash signature of an investment cycle, not capital stress; group optionality is converting into project-specific rigidity.
Strongest unknown: project-level hurdle rates and minimum utilization thresholds for new capacity.
See findings on this page →Capital state: 86% steel-mill utilization; largest segment CAPEX ($593m) went to an aluminum platform, not additional steel capacity.
Strongest observed fact: relocated a planned satellite aluminum slab center (Arizona → Mississippi) in 2026 before full construction lock-in.
Central inference: the relevant decision layer has moved from steel-only optimization to portfolio construction across metals.
Strongest unknown: no comparative hurdle rates across steel, aluminum and shareholder distributions.
See findings on this page →Capital state: $18.6bn revenue produced only $37m adjusted EBITDA; issued 75m shares ($951m) to repay ABL borrowings.
Strongest observed fact: an explicitly value-destructive legacy slab contract combined with weak auto demand and adverse Canadian conditions.
Central inference: physical throughput and economic value separated; integration likely amplified — but does not solely explain — the downside.
Strongest unknown: no official FY2025 utilization KPI, since configured capacity changed mid-year.
See findings on this page →Evidence
Gerdau, Nucor and Steel Dynamics cluster near ~$155–157/t on the chosen FY2025 proxy, yet Gerdau is funding competitiveness in Brazil, Nucor is carrying multiple growth projects, and Steel Dynamics is building an aluminum platform.
Caveat — proxy denominators and business mix differ across issuers; project-level returns remain UNKNOWN. Contradicted / bounded by: Cleveland-Cliffs, which does not fit the cluster at all.
Steel Dynamics and Cleveland-Cliffs both generated roughly $18bn of FY2025 revenue (~$18.2bn vs. ~$18.6bn) but occupied radically different decision worlds: Steel Dynamics built a new metals platform and returned ~$1.19bn to shareholders; Cleveland-Cliffs idled assets, raised equity and deleveraged.
Caveat — part of the gap is 2025-specific (slab contract, auto demand, Stelco integration effects).
Cleveland-Cliffs combined a value-destructive legacy slab contract with high physical throughput and simultaneous footprint actions (mine idlings, facility closures).
Caveat — this cluster does not compute an official Cliffs utilization rate from 18.8m tons of output against 20.0m tons of year-end configured capacity, because configured capacity changed within the year. No precise utilization percentage should be inferred.
Cleveland-Cliffs' and Gerdau's captive upstream assets (mines, pellets, energy) reduce some purchased-input exposure but create owned assets whose utilization must stay coordinated with downstream demand.
Bounded by — different forms and degrees of vertical integration at Nucor and Steel Dynamics, where internal scrap/downstream links may support through-cycle flexibility. Causality between integration and the 2025 downside is not established.
Steel Dynamics built working capital (+$1.1bn) tied to the aluminum ramp. Gerdau released working capital (-9% YoY) through capacity and inventory optimization in Brazil. Cleveland-Cliffs reduced inventory amid footprint and balance-sheet adjustment.
Caveat — the underlying facts are OBSERVED and HIGH confidence; classifying each move by "strategic function" is INFERRED, and FX/commodity price effects also influence balances.
Steel Dynamics relocated a satellite aluminum project before full lock-in; Nucor retains corporate liquidity while multiple greenfields harden into fixed commitments; Gerdau and Cleveland-Cliffs carry capital embedded in less modular, more integrated asset systems.
UNKNOWN — no company discloses a standardized option-value measure. This is not put forward as a universal law.
Same symptom, different cause
Simple derived FCF = operating cash flow − CAPEX
Negative free cash flow represented radically different decision states: growth commitment with substantial liquidity at Nucor versus earnings compression, footprint adjustment and balance-sheet repair at Cleveland-Cliffs.
Different symptoms, one recurring pattern
Nucor's CAPEX, Steel Dynamics' aluminum working-capital build and Gerdau's Brazil competitiveness spending are different observable expressions of one possible analytical theme: liquid corporate optionality becoming more specific production capability. A single causal mechanism across all three is not established — this is a pattern worth tracking, not a proven law.
Alternative explanations
Differences across the four companies could also reflect:
Architecture is treated as a possible mechanism, not a standalone causal proof.
Unknowns / limits
A definitive causal decomposition of these findings would require data that is not publicly disclosed by any of the four companies:
Extension case
CMC is analytically interesting as a scrap/EAF/long-steel producer with a growing downstream-construction architecture. It is not included in the core normalization above because of an August fiscal year-end and a growing Construction Solutions / precast mix, both of which reduce clean comparability to the calendar-year core four. CMC is retained as a robustness candidate for future work — a future test of whether these findings generalize to another flexible long-steel architecture, not a substitute for the core four.
Sources
Primary sources only — SEC filings, company IR releases and official annual reports. Numbers used in the Executive Map and Findings above trace to the entries below.
Gerdau
Nucor
Steel Dynamics
Corrections
| Version | Date | Change |
|---|---|---|
| 0.1 | Sep 12, 2026 | Internal research build: core four-company comparative dataset, individual company research objects, first industry synthesis draft. Not published. |
| 1.0 | Sep 12, 2026 | Public release. Renamed headline from "Same Steel Cycle" to "Same Industry, Different Decision Space" to avoid implying a unified market state. Renamed comparison table to "Directional FY2025 comparison" and added a prominent proxy-comparability warning. Renamed Finding 03 to "Physical throughput and economic value can diverge" and removed any implied official Cliffs utilization calculation. Softened Finding 04 from contradiction to "bounded by," with causality explicitly not established. Downgraded the Steel Dynamics causal hypothesis from STRENGTHENED to UNRESOLVED and narrowed its optionality claim to the one project that actually relocated. Removed "temporary" from the Nucor rigidity finding, since duration was not evidenced. Normalized the Nucor/Cleveland-Cliffs cash-flow comparison to one explicit formula (operating cash flow − CAPEX). Renamed "Different symptoms, one structural condition" to "Different symptoms, one recurring pattern." Made the Unknowns/Limits block prominent on the page rather than an appendix. Separated company research (linked) from this synthesis visually via compact company cards. Added direct primary-source links throughout. Removed the internal LinkedIn/media-thesis distribution block from the public page. Kept Commercial Metals Company as an extension case only. |
| 1.1 | Sep 12, 2026 | Converted Gerdau's revenue, EBITDA and CAPEX from BRL to USD (at Gerdau's disclosed FY2025 average rate of R$5.5855/US$1) so the Executive Map uses a single currency throughout; original BRL figures disclosed in a footnote. No change to findings or conclusions. |
Download
The PDF is a direct rendering of this page (v1.1) — same findings, same evidence labels, same information cutoff.
Research lead
Founder & Research Lead, aiclavis
LinkedIn