How many copper refineries does the United States have?

Two primary electrolytic refineries, alongside two primary smelters, fourteen electrowon refineries and three secondary fire refineries, as recorded for 2024 by the United States Geological Survey.

Does the United States refine all the copper it mines?

No. In 2024 the United States mined an estimated 1,100 thousand tonnes of copper and refined 850 thousand tonnes from ore, exporting 320 thousand tonnes as ore and concentrate.

How much refined copper does the United States import?

810 thousand tonnes in 2024. Net import reliance was 45 per cent of apparent consumption, against 32 per cent in 2015.

Where is the world's copper actually refined?

World refinery production was an estimated 27,000 thousand tonnes in 2024. China accounted for 12,000 thousand tonnes. The United States accounted for 890 thousand tonnes including secondary production.

Research

The 2026 copper forecasts agreed on the direction. The price went the other way.

Two independent forecasters called a small surplus. Copper traded well above the range that surplus was expected to enforce. The lesson is about what a balance forecast can and cannot tell you.

Author: Bbabsal Research
Published: 23 August 2026
Subject: Market
Evidence: Goldman Sachs Research,
International Copper Study Group

Key findings

Goldman Sachs Research, December 2025, forecast a 2026 global surplus of 160 thousand tonnes and an LME price range of 10,000 to 11,000 dollars a tonne.

The International Copper Study Group, April 2026, forecast a 2026 surplus of 96 thousand tonnes, reversing an earlier forecast of a 150 thousand tonne deficit.

On the day that ICSG figure was reported, benchmark three month copper on the LME was 13,290 dollars a tonne.

A surplus of 96 thousand tonnes is a fraction of a percent of a refined market measured in tens of millions of tonnes.

What the forecasters said

Two forecasters, different methods, four months apart, same answer. The price disagreed.

In December 2025, Goldman Sachs Research published its 2026 copper outlook. It expected the global market to run a surplus of 160 thousand tonnes, revised down from a 500 thousand tonne surplus in 2025, and it expected that surplus to hold the London Metal Exchange price inside a range of 10,000 to 11,000 dollars a tonne. The team was explicit about the implication: the market was moving closer to balance, but a shortage was not coming soon. It forecast a first half 2026 average of 10,710 dollars.

In April 2026, the International Copper Study Group published a forecast of its own. It projected a surplus of 96 thousand tonnes for 2026, reversing its earlier forecast of a 150 thousand tonne deficit, and attributed the reversal to slower demand growth and increased secondary production. It put refined usage growth at 1.6 per cent for 2026, revised down from 2.1 per cent, and refined production growth at 0.4 per cent, constrained by limited concentrate availability. It expected a larger surplus, 377 thousand tonnes, in 2027.

On the day the ICSG figure was reported, benchmark three month copper on the LME was 13,290 dollars a tonne. That is roughly 2,300 dollars above the top of the range Goldman had expected the surplus to enforce, and above the 11,771 dollar record the same December report had described.

Why the price did not follow

There is a temptation to read this as forecasters being bad at their jobs. That is the wrong lesson, and it is worth resisting for a specific reason: a balance forecast and a price forecast answer different questions.

A surplus of 96 thousand tonnes on a refined market measured in tens of millions of tonnes is a rounding error, well inside the error bars of the production and consumption estimates that produce it. A market can be nominally oversupplied by a fraction of a per cent and still price like a market that is short, because price responds to inventory location, disruption risk, trade policy and the cost of the marginal tonne, none of which appear in a balance line.

Several of those factors were visible in 2026. Goldman's own report flagged the prospect of a United States tariff on refined copper imports, with a recommendation to the White House expected by mid year, and noted that importers building stock ahead of any tariff would pull metal into one jurisdiction. Metal in the wrong warehouse is not available to the market that needs it, whatever the global balance says. The ICSG, for its part, was explicit that its figures carried geopolitical and trade risk.

So the useful reading is not that the forecasts failed. It is that the surplus was too small to be load bearing. When a forecast balance is a fraction of a per cent of the market, it cannot tell you where the price goes, and treating it as though it can is a category error rather than a bad estimate.

The number underneath

Our own reading, held as a reading rather than a demonstrated result.

The interesting number in a copper forecast is rarely the balance. It is the concentrate line underneath it. The ICSG put refined production growth at 0.4 per cent for 2026 and attributed the constraint to concentrate availability, then expected 3 per cent growth in 2027 as concentrate improves and new capacity arrives. That is a statement about whether there is enough material to feed refineries, and it is the part of the forecast that a change in mine supply or processing capacity actually moves. A market balance can be flipped from deficit to surplus by a demand assumption shifting half a percentage point. The constraint underneath does not move that quickly.

Declared interest. Bbabsal is developing copper processing capacity, and a view that emphasises the concentrate and refining constraint is a view that flatters our own position. That is exactly why every figure above is third party, dated and attributable, and why we have quoted the forecasters in their own terms rather than assembling a case from the parts that suit us.

Method and sources

Method. Figures are quoted from the publishing institution in the edition in which they appeared, with the publication date attached. No figure has been adjusted, rebased or interpolated. All values are in United States dollars.

A note on secondary sources. In preparing this study we found widely circulated secondary figures for 2026 copper balances that did not match the primary publications they claimed to summarise, including a surplus figure attributed to Goldman Sachs several times larger than the one in the bank's own published research. We have used only figures traceable to the publishing institution. The dispersion between forecasters is a real and interesting signal. The dispersion between what forecasters said and what is reported about what they said is noise, and it is worth learning to tell them apart.

Limitations. A balance forecast is an estimate of production minus consumption, both of which carry their own revisions. Price on a single day is not a trend. This study compares published forecasts against an observed price at a point in time and draws a methodological conclusion from the gap, not a price prediction.

Goldman Sachs Research, Copper Prices Are Forecast to Decline Somewhat from Record Highs in 2026, 11 December 2025. Analyst Eoin Dinsmore.

International Copper Study Group forecast, as reported in ADM Investor Services market commentary, 24 April 2026, which also carries the LME three month price quoted here.

Corrections to this study will be published here with the date attached. Bbabsal has no declared mineral resource or reserve, and nothing in this study relates to ground the company holds. Nothing here is an offer of securities or investment advice.