Metal margins depend on contracts and throughput, not the energy index alone

Analysis

Three different disclosures frame the mining and metals question as of October 9, 2026. The World Bank's October 6 commentary describes September benchmark movements. BHP's August 18 results describe historical copper output and pre-commitment capital. Hydro's October 5 Alunorte update describes gas procurement and an uncertain future-quarter cost estimate. They should be connected through mechanisms, not combined into a single measured industry cost series. Their reporting periods and units differ, and none supplies a complete operating model for a representative mine or smelter.

Start with a cost identity: total operating cost per saleable tonne equals variable cost per tonne plus fixed period cost divided by saleable output. Keep sustaining and expansion capital, finance and tax separately visible. A saleable-tonne denominator differs from tonnes of ore treated. For copper, ore volume multiplied by copper grade and metallurgical recovery gives recovered copper before other adjustments. Lower grade can raise energy consumed per tonne of copper even if energy per tonne of ore is stable. A cost comparison that omits grade, recovery and scope can therefore mistake geology or accounting boundaries for procurement efficiency.

The benchmark channel depends on exposure. The World Bank's September energy increase is not the percentage change in a producer's total budget. The relevant calculation applies a price change only to an exposed energy component. Diesel, electricity and gas should ideally be modeled separately with realized prices and contracts. Fixed prices, escalation clauses, hedges, local currency and settlement dates can change or delay transmission. Revenues have their own metal-price settlement rules; a higher selling benchmark does not establish a larger cash margin when input purchases, hedges and payable metal differ.

Consider a fictional operation solely to make this arithmetic auditable. Assume annual saleable output of 100,000 tonnes, variable cost of US$1,200 per tonne and annual fixed operating cost of US$80 million. Baseline operating cost is US$2,000 per tonne: 1,200 + 80,000,000 / 100,000. Within variable cost, assume energy is US$300 per tonne, with half exposed to repricing. An assumed 20% increase on that exposed half adds US$30 per tonne: 300 × 0.50 × 0.20. Cost becomes US$2,030 per tonne with output unchanged. The 20% shock is an analytical assumption; it is neither the World Bank's measured September change nor a forecast of contract prices.

Now assume a separate 10% output reduction, keeping annual fixed cost unchanged and variable cost proportional to actual output. Production becomes 90,000 tonnes. Fixed cost rises from US$800 to about US$888.89 per tonne, making operating cost about US$2,088.89 before any energy shock. Applying both assumptions gives US$2,118.89 per tonne. The chart compares these calculated unit costs. It does not imply that total annual spending increases in every case: variable spending falls when output falls. It also excludes restart expenses, lost sales, inventories, capital, financing and tax. A margin calculation would additionally need a separately stated realized selling price.

Alunorte illustrates why continuity and price are separate engineering controls. Replacement gas can preserve heat availability while changing procurement cost. Hydro's fourth-quarter US$90–110 million range addresses the contract-to-spot gas difference; its earlier third-quarter range also included lost production. Neither can be divided by an invented output denominator to derive a refinery unit cost. The public update does not provide the required quarterly gas consumption, full price formula and realized saleable production. A resilient supply arrangement should therefore be assessed for physical delivery, price exposure, storage or backup options and the cash needed before customers pay.

BHP's Escondida pre-commitment concerns a different constraint: future processing capacity. The disclosed US$0.5 billion is not the complete construction budget or already commissioned throughput. An investment case needs incremental saleable copper, remaining capital, commissioning dates, grade and recovery, water and power availability, maintenance and contingency. Enlarging the plant can reduce some fixed cost per tonne while requiring more energy or additional infrastructure. Compare discounted cash flows against a baseline with the same metal-price and operating assumptions, rather than dividing early expenditure by current group production.

For downstream cable and equipment budgets, metal content, fabrication and financing should also remain separate. A hypothetical 1,000-tonne order would change its unhedged metal bill by US$100,000 under an assumed US$100-per-tonne price change, before premiums, scrap credits and processing. This identity is not a reported copper or aluminium price. Contract timing determines whether that change reaches a bid, an invoice or inventory. Working capital rises when more cash is tied up in inputs, even if final sales eventually pass through the cost.

The next useful evidence is matched rather than merely newer: energy volumes and realized prices for the same production period, contract exposure, ore and metal output, inventory, and a consistent fixed-versus-variable cost definition. Company guidance should remain labeled until results arrive. The scenario chart is deterministic arithmetic without a probability model or confidence interval; it predicts neither prices nor output. Its practical conclusion is to test both procurement and throughput when reviewing metal economics, and to preserve the distinction between historical observations, company estimates and assumptions.