Material procurement: delivered cost, usable output and the cash tied up between them

Analysis

Six August–September disclosures frame a procurement problem as of October 9, 2026. Portland’s working pots concern operating equipment; Hydro’s Norwegian power contract concerns future energy; Equus gas concerns a conditional development pathway; RevoCast concerns product conversion and regional distribution; the IEA coal review concerns fuel markets; Glencore concerns inventory funding. These are different boundaries, periods and evidence types. Combining them into one measured metal-cost index would be unjustified. Their common economic question is how a material reaches an accepted specification, becomes usable output and consumes cash before the buyer collects payment.

Begin with the receiving boundary. Delivered material cost includes the purchase reference, conversion charge, transport and agreed delivery expenses. Inspection, rejection, handling losses and yield then determine cost per accepted tonne. Finance has a time denominator, while production has a physical denominator. Record who owns the cargo in transit and when payment starts. Otherwise the same material can enter both transit stock and warehouse stock, or financing can begin before the buyer actually funds it. A cash-flow model should reconcile physical batches and payment dates before introducing a commodity-price scenario.

The Portland disclosure establishes a restart milestone, not saleable metal output. More operating equipment can support supply, but quality, stable production and subsequent casting remain necessary. RevoCast’s announced conversion route reinforces this product distinction: billet capacity is not a guaranteed delivery of the alloy and dimensions a customer needs. The procurement review should ask for qualified specification, acceptable batch size and a dated dispatch plan. Local production can reduce distance without proving an observed lead-time saving. If receiving inspection delays release, nearby material may still tie up capital rather than immediately enter production.

Energy sourcing needs its own boundary. Hydro’s contracted 876 GWh per year corresponds arithmetically to 100 MW averaged across 8,760 hours; this is an average energy equivalent, not a guarantee of 100 MW in every hour. Ten years give 8.76 TWh, consistent with the disclosed approximately 8.8 TWh. The tariff is undisclosed. Smelting cost cannot be inferred without actual power consumption, hourly terms and the remaining supply portfolio. Moreover, the contract starts in 2031, so it is not a measured saving on electricity purchased in 2026.

Alcoa’s approximately 50 TJ per day prospective Equus supply is likewise an energy quantity, conditional on project delivery in the early-to-mid-2030s. It cannot establish current refinery cost or erase the need for interim supply. Development conditions, demand coverage and backup arrangements belong in the review. Converting gas energy to volume requires a stated heating value and reference conditions. Comparing gas with electricity also requires the useful service: process heat at a specified temperature differs from a generic unit of final energy, and a technically unsuitable fuel is not made suitable by a low price.

Coal makes the quality denominator especially visible. Consider a fictional delivered fuel at US$120 per tonne purchase plus US$30 per tonne logistics, a lower heating value of 24 GJ per tonne and 80% useful-heat efficiency. The calculated heat cost is 150 / (24 × 0.80) = US$7.8125 per useful GJ, or US$28.125 per useful MWh using 3.6 GJ/MWh. If assumed logistics rise to US$50 per tonne, it becomes 170 / 19.2 = US$8.8542/GJ, or US$31.875/MWh. These are hypothetical inputs, not coal quotations from the IEA. Moisture, ash, losses and plant performance can change the comparison; metallurgical coal is not an interchangeable heating input.

Now separate financing from the invoice. Assume an accepted metal lot of 1,000 tonnes with US$2,000/t metal, US$150/t conversion, US$100/t logistics and US$50/t other receiving expenses. The mutually exclusive components give US$2,300/t and US$2.3 million for the lot. These prices and quantities are invented examples, not offers from Alcoa, Hydro, Rio Tinto or Glencore. Assume the entire lot is funded for 30 days at 8% simple annual interest using a 365-day basis. Carrying finance is 2,300,000 × 0.08 × 30 / 365 = US$15,123.29, or US$15.1233 per tonne for that period.

The native chart isolates four financing cases. A 10% assumed increase in the whole landed invoice gives US$2,530/t and 30-day finance of US$16.6356/t. A separate increase in funding time from 30 to 60 days, with price unchanged, gives US$30.2466/t. Applying both assumptions gives US$33.2712/t. The difference from baseline is about US$18.1479/t, or US$18,147.95 for the 1,000-tonne lot. Price and days multiply; their percentage effects on finance are not simply additive. These scenarios neither predict shipment delays nor estimate the probability of future prices.

The holding period must represent actual funding exposure. Supplier credit can postpone cash outflow; a deposit can start exposure before delivery; customer collection can continue it after processing. For a steady operating business, inventory days, receivable days and payable days can describe a cash-conversion cycle, but each component needs its own cost or revenue basis. Applying a single stock price indiscriminately to receivables and payables would distort the amount funded. An advance payment is also not automatically the full shipment value. Use actual cash flows and avoid adding a separate safety-stock expense when that stock already appears in the funded balance.

Glencore’s net-funding disclosure illustrates a reported inventory-price channel, but its company-defined net funding is different from net debt. It cannot calibrate the fictional buyer’s rate, credit days or balance. Higher replacement values can increase cash requirements without increasing physical stock. A price hedge may reduce market exposure while requiring collateral; physical delivery can still be delayed. Funding availability, counterparty terms and product qualification therefore sit beside commodity reference prices, rather than being inferred from them. A low purchase price is insufficient if the material cannot be used or the buyer cannot finance the payment schedule.

Comparing procurement options should preserve these separate mechanisms. A longer contract can improve supply visibility but expose a different price formula; a nearer caster can shorten a route but require qualification; a restart can increase equipment availability without guaranteeing a finished product; a fuel switch can lower a modeled energy cost while needing capital and maintenance changes. Assess each option against the same accepted product or useful service, time period and cash boundary. Do not double-count a logistics saving that is already embedded in a delivered quotation, or label a financing saving as a measured reduction in smelting cost.

The next evidence is contract-specific: dated quotes, energy and quality bases, usable output, shipment and acceptance dates, ownership, deposits, credit terms and funding rates. This analysis uses evidence available through October 9 and examines procurement planning for the remainder of 2026–2027, with later contract starts separately identified. Its arithmetic is deterministic and has no calibrated probability distribution or confidence interval. Taxes, foreign exchange, hedges, storage, equipment investment and losses are excluded from the numerical cases unless stated. The purpose is a reproducible cost decomposition, preserving observed disclosures, conditional plans and analytical assumptions as distinct evidence.