Why does Gerdau produce materially different EBITDA per tonne across two regions with comparable shipment volumes?

A 3.7× EBITDA-per-tonne gap (2Q26) on near-identical volume.

OBSERVED — Brazil 2Q26: 1.352 Mt shipped, R$705m Adj. EBITDA, 10.5% margin. North America 2Q26: 1.347 Mt shipped, R$2.600bn Adj. EBITDA, 25.7% margin, backlog above 100 days. These are the reported figures.

DERIVED — EBITDA per tonne (R$521/t Brazil, R$1,930/t North America) and the resulting 3.7× ratio are calculated from the observed figures above, specifically for 2Q26. The same calculation on the 6M26 window (using a different internal audit table) yields ≈R$479/t Brazil and ≈R$1,850/t North America — ≈3.9×. Both are real, correctly computed ratios for different periods; they are not the same number, and the 2Q26 figure of 3.7× is the one used as this page's headline.

Interpretation — the gap cannot be reduced to one variable — pricing, mix, market structure, inputs and operations all plausibly contribute.

Decision relevance — establishes the scale of the question worth validating; a gap this size, on this little volume difference, is unlikely to be explained by shipment volume alone.

What would change this conclusion — a full product-mix and cost-structure reconciliation showing the gap is fully explained by known structural differences between the two markets.

North America capacity is economically scarce in the current decision environment.

OBSERVED — North America backlog stands above 100 days; 2Q26 shipments, Adj. EBITDA and margin are as reported above; Gerdau's own commentary describes record shipments under the current industrial configuration.

DERIVED — annualizing 2Q26 North America shipments against installed capacity gives a mechanical utilization figure of roughly 90%. This is our calculation from public disclosures — it is not a company-reported utilization rate, and it is a portfolio-level estimate, not a per-plant one.

INFERRED — a backlog above 100 days combined with a ~90% mechanical utilization estimate is consistent with capacity being economically scarce in the relevant decision environment. This does not establish that any individual plant is capacity-constrained — that would require a per-plant utilization disclosure this research does not have.

Decision relevance — if capacity is genuinely scarce, assigning it early to a lower-value order carries a real opportunity cost — the foundation of the optionality-pricing question this research raises.

What would change this conclusion — a disclosed per-plant utilization rate materially below 90%, or evidence that backlog length reflects scheduling choice rather than a capacity constraint.

Gerdau already runs sophisticated decision layers individually.

Evidence — public evidence shows a 3–6 month strategic forecast by segment/country/product, ~60-day order-acceptance optimization and conflict resolution, explicit portfolio/sales trade-offs between EBITDA and market share, and S&OP-guided production scheduling with corrective action on deviation. A public Gerdau/Ouro Branco digital-twin scheduling project reported a 1% monthly production increase without added infrastructure.

Interpretation — the available evidence is consistent with each of these four functions operating with real sophistication individually. That is not the same as a claim about the full internal decision architecture, or about whether the four functions are reconciled against a shared, cross-horizon objective — see Finding 04.

Decision relevance — this is the evidence base the central inference builds on; it rules out "Gerdau lacks sophistication" as an explanation.

What would change this conclusion — evidence that these four functions operate on inconsistent or conflicting objectives rather than a shared one.

The same future resource is valued on different horizons.

Evidence — a commitment moves from order acceptance (now), to near-term order conflicts (~60 days), to the market/commercial plan (3–6 months), to finance and structural decisions (6 quarters+).

Interpretation — a commitment that is locally correct inside one decision window can still change the economics of another: a profitable order can consume a scarce option — a product family, a production campaign, a downstream slot, a scarce future week — that another opportunity could have used later.

Decision relevance — combined with Finding 02 (North America capacity scarcity), this is the mechanism by which committing capacity early can carry an opportunity cost. In Brazil, comparatively available capacity poses the opposite problem — value can come from finding the best economically acceptable use of it, including domestic/export reallocation.

What would change this conclusion — evidence that order-acceptance, commercial planning and financial horizons already share a single, reconciled value function for future capacity.

A value equation for accepting an order.

Evidence — framed from the observed and inferred evidence above: the value of accepting an order is its direct contribution, minus the value of the best displaced future alternative, plus or minus system effects.

Interpretation — this reframes "accept the order" from a single-period question into a cross-horizon one — every accepted commitment changes the set of decisions that remain available later.

Decision relevance — gives a concrete, reproducible form to what "pricing future optionality" would actually mean in Gerdau's context.

What would change this conclusion — if Gerdau's own order-acceptance criteria already include an equivalent displaced-alternative term, this derivation adds no new value.

Whether lost flexibility is made visible at the point of commitment.

Evidence — public information does not show whether constrained capacity is explicitly repriced as system conditions change, whether local order-acceptance objectives remain economically coherent with 3–6 month commercial expectations and longer financial priorities, or — when a commitment consumes future flexibility — whether that loss is made visible (which feasible alternatives disappeared, their expected value, and how much optionality remains).

Decision relevance — this is the specific, bounded question this research identifies as worth validating internally — not a claim that a gap definitely exists.