Every cycle produces a wave of peak gold commentary, and every cycle it is treated as new. The supply side does matter — just not in the way the headlines imply. This edition sets out what mine output actually determines and what it does not.
The arithmetic that defuses most supply headlines
There is an estimated 216,000 tonnes of gold above ground. Annual mine supply runs roughly 3,000 to 3,700 tonnes. New production therefore adds something on the order of 1.5 to 2 per cent to the existing pool each year.
That single ratio explains why a major mine outage rarely moves the price for long, and why the behaviour of existing holders dominates. Gold is a stock market masquerading as a flow market. Almost every ounce that will trade this year has already been mined.
What the cost curve does set
All-in sustaining cost is the industry's attempt to state what it really costs to keep producing an ounce, including sustaining capital and site overheads. It is imperfect and inconsistently defined between producers, but the aggregate curve is still informative: prices that sit below the upper cost quartile for long enough cause the highest-cost operations to curtail, and curtailment is a slow, sticky decision.
This produces a soft floor rather than a hard one. Producers hedge, high-grade their deposits to survive a weak patch, and defer stripping — all of which can keep uneconomic output flowing for quarters. Anyone treating the cost curve as a guaranteed price floor is reading a tendency as a rule.
The lag nobody prices
The gap between a discovery and a producing ounce is routinely a decade or more: resource definition, feasibility work, permitting, financing, construction and ramp-up. Grades in most major districts have been declining for years, which means more rock moved and more energy consumed per ounce, and energy is now a first-order input cost rather than a footnote.
The consequence is that supply cannot answer a demand shock inside the horizon that shock plays out over. A sustained high price does eventually produce more metal. It does not produce it this year, or next.
Recycling is the real shock absorber
Scrap has historically supplied roughly a quarter to a third of annual supply, and unlike mine output it responds within weeks. Price spikes pull jewellery and industrial scrap out of drawers and into refineries, which is why sharp rallies tend to be self-limiting without any change in mine plans.
Watch this closely during a fast move: if refinery throughput and scrap flows are surging while the price climbs, the rally is meeting elastic supply. If the price is climbing and scrap is not appearing, holders are choosing not to sell, which is a materially more durable move.
How we use the supply side
Not as a price signal. As a constraint: it tells us how far the price can fall before output responds, how long any response will take, and how quickly a spike is likely to be capped by recycling. Those are boundary conditions on the other drivers, and boundary conditions are more useful than forecasts.
Where the metal comes from now
Production is far less concentrated than it was a generation ago. China, Australia, Russia, Canada, the United States, Kazakhstan, Mexico, Peru, Ghana and Indonesia between them account for the majority of annual output, and no single country holds a share large enough to make its domestic politics a systemic supply event. Artisanal and small-scale mining adds a further meaningful slice that is poorly measured, informally financed and largely invisible in producer reporting.
That dispersion is the reason country-level supply headlines rarely justify the attention they receive. A permitting freeze in one jurisdiction removes tonnes from a 3,000-plus-tonne annual flow that itself only adds a couple of per cent to the above-ground stock. The arithmetic is unkind to the narrative.
The reserve replacement problem
The more interesting structural story is not annual output but reserve replacement. Exploration budgets have been directed disproportionately at brownfield extensions rather than new districts, discovery rates for large high-grade deposits have fallen, and average head grades at major operations have drifted lower for two decades. Producers have responded by buying reserves through consolidation rather than finding them, which reshuffles ownership without adding metal.
If that persists, the consequence is not a shortage — recycling and the existing stock make a shortage nearly impossible — but a slowly rising real cost of the marginal ounce. That is a decade-scale tailwind under the cost curve, not a trade.
What a producer's numbers actually tell you
Reading a producer report properly means separating three things that are routinely conflated. Reserves are the economically extractable subset of a resource at a stated price assumption — raise the assumed price and reserves grow without a single new discovery. All-in sustaining cost excludes growth capital, so a company can report falling costs while quietly underinvesting in its own future output. And grade guidance matters more than tonnage guidance, because mining a high-grade zone early flatters this year's numbers at next year's expense.
Anyone using equities as a proxy for metal exposure should read those three items before the headline earnings line. They explain most of the cases where a producer underperforms the metal during a rising price.
Energy, water and permitting as the real constraints
Modern gold mining is an energy business with a precious-metal output. Diesel, grid power, explosives and freight are the dominant cost inputs, which means energy prices transmit into the cost curve with a short lag. Water access and tailings management increasingly determine whether a permitted project can actually be built, and community consent processes have lengthened timelines in several major jurisdictions.
None of this changes the two per cent arithmetic. It does mean the response lag we described above is more likely to lengthen than shorten from here.
What would change our view
Three things would make the supply side genuinely price-relevant: annual mine output growing materially faster than the low single digits for several consecutive years, scrap flows failing to respond to a sharp price rise, or a sustained period where prices sat below the aggregate cost curve without curtailment. Each is measurable in public data, and each would tell us that the framework in this edition needs revising rather than repeating.
Grade decline is the quiet story
Average head grades at large open-pit gold operations have drifted lower for two decades. The consequence is mechanical: to hold ounces flat, a mine must move and process more rock every year, which raises energy, reagent and haulage costs per ounce even when nothing about the deposit has changed. When you read that a producer "missed guidance on throughput", check whether the tonnes moved actually rose while the ounces fell — that pattern is grade, not mismanagement, and it does not reverse.
This is also why cost inflation in mining is stickier than in most industries. A rising gold price improves margins immediately, but it also makes lower-grade material economic, which pulls average grade down again and re-absorbs part of the margin. The industry's own success at higher prices is a partial brake on its profitability.
Streaming, hedging and why reported output is not free ounces
A share of published production is already sold. Streaming and royalty agreements assign a fixed percentage of output to a financier at a pre-agreed price, and some producers still run forward sales on part of their book. When you compare a company's ounces to the spot price, you are frequently pricing metal it will never receive at spot. The disclosure is in the notes to the accounts, not the production headline: look for stream percentages, delivery obligations and any hedge book maturity table before drawing a conclusion about leverage to price.
On the programme
The supply block on 10 October covers capex cycles, grade decline, permitting timelines and what producers see that spot does not. Educational content only — not investment advice.