The unit of the mine

Tungsten has become nine times more expensive in eighteen months, and Europe still cannot finance a mine. This paper takes that sentence apart. The price is the wrong unit to decide in, the mine has a unit of its own, decades, and the only instrument that has reopened a Western mine since the shock reckons in the unit of the mine rather than in the unit of the market. It is the first Keystone paper on the matter axis, and it finds there the same error as on the time axis, seen from the other side.

The problem, concrete

On the edge of Dartmoor in Devon lies one of the largest known tungsten deposits in the world. Wolf Minerals bought the Hemerdon project in 2007, completed its feasibility study in 2011 and started production in 2015, with an open pit designed for an initial ten years. In October 2018 the company could not meet an interest payment of 2.1 million pounds, its lender would not extend again, and it went into liquidation. Some 250 people were sent home from a pit that had run for three years, on ore that turned out more friable than the study had assumed, at a price that averaged 245 dollars per metric ton unit in 2017, the middle of the mine’s three years. Nothing was wrong with the deposit. The clock was wrong.

In 2026 the same pit is being readied to restart. The European Commission designated Hemerdon a strategic project under the Critical Raw Materials Act in June 2025, commissioning began in July 2026 with the first processing circuit due in the third quarter and full commissioning in early 2027, and the new owner is assembling private money, a bridging loan of 25 million dollars from its largest shareholder and a debt package of up to 85 million dollars, to get there. The rock has not changed. The price has, by a factor of nine.

The number that carries the debate

The figure behind that factor is a quotation: ammonium paratungstate, 88.5 percent tungsten trioxide, delivered Rotterdam or Baltimore, in dollars per metric ton unit, assessed weekly by a price-reporting agency. In June 2025 it stood at 430 to 475 dollars. In January 2026 at 900 to 940. In mid-February 2026 at 1,650 to 1,900, in March at 2,500 to 2,800, in April at 2,800 to 3,280. On 19 June 2026 it was assessed at 2,900 to 3,210 dollars, 575 percent above the same week a year earlier, and at the end of July at 3,075. In the weeks after the controls the European assessment stood at 370 to 380 dollars; measured from there the rise is eightfold, and measured from before the controls it is the ninefold rise the researchers who sounded the alarm mean.

What moved it is not geology. On 4 February 2025 the Chinese Ministry of Commerce and the customs administration placed tungsten, together with tellurium, bismuth, molybdenum and indium, under export licensing, 41 tariff lines, with immediate effect, minutes after an American tariff of ten percent on Chinese goods took effect. Since January 2026 only fifteen companies, named by the Ministry of Commerce in December 2025, may export tungsten at all, for 2026 and 2027. China mined roughly 67,000 tonnes of tungsten in 2025, close to eighty percent of world output, and accounts for 86 percent at the processing stage. A licence desk in Beijing therefore decides what reaches Rotterdam.

Then the market split. By June 2026 the domestic Chinese price of the same compound had fallen to roughly 1,200 to 1,300 dollars, more than a thousand below its spring high, while the seaborne quotation held near 3,000; the concentrate that feeds it had lost half its value since March. The Rotterdam number measures one thing with great precision: the cost of getting tungsten past a licence. It does not measure what tungsten costs to produce, and it does not measure what it will cost in 2030. A mine financed on that number is financed on a policy decision in another capital, not on rock.

The United States read the same number and answered in kind. A temporary rule of the Bureau of Industry and Security, published on 4 August 2026 under the Defense Production Act, requires from 27 August that all sales of tungsten waste and scrap, and of the black mass from battery recycling, go to American buyers for one year. No tungsten has been mined in the United States since 2015; scrap is the only domestic source it has, and it is being kept at home. A Dutch newspaper wrote this week that the tungsten market is now read as a predictor of war, because the metal that makes armour-piercing rounds and cutting tools moves before armies do. A Dutch processor put the industrial version more plainly: the worry is not the price, but whether there will be anything to deliver.

Two clocks

Set the mine’s clock next to the market’s. The International Energy Agency found that the major mines that opened between 2010 and 2019 took more than sixteen years on average from discovery to first production, over twelve of them in exploration and feasibility, four to five in construction. Hemerdon took eight years from purchase to first ore and then three to die. Sangdong, in South Korea, once one of the largest tungsten mines in the world, closed in 1994 when cheaper Chinese tungsten reached the market, was bought by the Canadian miner Almonty in 2015, and started processing ore on 1 July 2026: thirty-two years shut, eleven years to reopen.

The market’s clock runs in months. A quotation moves on a tariff announcement, a licence decision, a stockpile purchase. And it runs in one direction the mine cannot follow. The director of the Centre for Materials and Resilience in Delft put it in one sentence: if Europe decides to mine more tungsten, China can simply open the tap again. The price falls, the lender will not roll the debt, and the pit closes, as it did in 1994 and in 2018.

This is the mechanism that the first Keystone paper named horizon compression, in its financial form. The object reckons in decades; the decision about it is taken in the unit of a price cycle, by a party that will have marked its book and moved on long before the object has paid for itself. Every single refusal to finance is defensible on the quotation of the day, and the sum of those refusals is a continent without a mine. The structure is the same as the owners’ association that keeps its reserve low year after year; only the sign is reversed. There the benefit comes now and the bill later. Here the bill comes now, the benefit over thirty years, and the one who decides is still not the one who lives with the outcome.

A unit without an instrument

Europe has, in fact, set the unit right. The Critical Raw Materials Act, in force since May 2024, counts in shares of supply and in years: by 2030 the Union should extract ten percent of what it consumes, process forty percent, recycle twenty-five percent, and take no more than 65 percent of any strategic material from a single third country. Tungsten is on both lists, critical and strategic. In March 2025 the Commission designated 47 strategic projects inside the Union, three of them for tungsten, among them El Moto and the integrated P6 Metals project in Spain, explicitly for the resilience of the defence industry; in June 2025 it added thirteen outside the Union, Hemerdon among them. Strategic status brings a permitting deadline, 27 months for extraction and fifteen for processing, and a place in the queue of a financing hub. A second call closed in January 2026 with more than 160 applications.

That is a goal expressed in the right unit, and it is not an instrument. A permit that arrives in 27 months rather than in five years shortens one stretch of the mine’s sixteen; it does nothing about the question that closed Hemerdon, which is who carries the loss when the price halves in the mine’s fourth year. The benchmarks describe 2030. Nobody signs for it.

The move from goal to instrument has begun, late. In December 2025 the Commission adopted RESourceEU: three billion euro to be mobilised within a year, a European Critical Raw Materials Centre to be set up in 2026 to monitor needs, purchase jointly on behalf of member states and hold stockpiles, a joint purchasing platform from March 2026, a stockpiling pilot in the first months of the year, and an export ban on battery black mass from September. In May 2026 tungsten, rare earths and gallium were reported to be shortlisted for the first joint reserve. Each of these reckons in a longer unit than the week. But a stockpile buffers a shock and finances nothing, and a purchasing platform finances nothing either unless it signs for years. The instrument that matters is not the existence of a buyer. It is the length of the contract.

The instrument that exists

It exists, and it reopened Sangdong. The Korean mine was financed with a project loan of 75 million dollars from KfW IPEX-Bank, an arm of the German state development bank, and with an offtake agreement with Global Tungsten & Powders, a Western processor that supplies the American defence chain. On 14 July 2026 that agreement was extended to twenty-one years from first delivery, covering about ninety percent of the mine’s first-phase output, 4.41 million metric ton units in total and at least 210,000 a year once ramped up, worth some 490 million dollars a year at the prices of that month. A mine that closed in 1994 because the price fell reopened in 2026 because somebody signed for twenty-one years.

That contract is the instrument, and it has exactly the shape of the one described in the first Keystone paper. The statutory multi-year maintenance plan forces the owners of a building to reckon in the unit of the building, decades, rather than in the unit of the annual meeting. A twenty-one-year offtake with a state bank behind it forces the financing of a mine into the unit of the mine rather than into the unit of the quotation. Both replace a decision that is retaken every cycle, and therefore always deferred, with a commitment on the lifespan of the object. Both need a party with a balance sheet long enough to hold that commitment when the cycle turns: a statutory reserve in the one case, a treasury in the other.

The Delft proposal is the same move in European form: long-term offtake guarantees, floor prices, and a buy-European line in the public procurement of critical materials, together with the observation that Europe must get over the idea that there always has to be a business case. A business case is a statement in the unit of the cycle. A guarantee is a statement in the unit of the mine. Hemerdon makes the gap visible: it holds a European strategic designation and is still raising its money from private lenders, on a price nobody will promise for 2030. The line the Union has not yet written is the cheapest one in its toolkit, a European counterparty prepared to sign for twenty years.

Why this belongs to Keystone

The first Keystone paper made a claim about measurement: a tractable proxy crowds out the physical reality beneath it, and what went uncounted turns out to be constitutive. The tonne of CO2 is governed; the material chain that produces every avoided tonne is not. Tungsten is the cleanest test of that claim, for two reasons. It is not a transition mineral, so none of the comforts of the transition argument apply: no learning curve thins the demand for cutting tools and armour, and no substitute waits in the wings. And its concentration, four-fifths of mining and a comparable share of refining in one country, is as sharp as anything in the rare-earth debate.

But look closely and the matter error turns out to be the time error seen from the other side. The proxy that crowds out the object here is the quotation, the one number that is daily, global and comparable, and what it crowds out is a deposit with a thirty-year life. The decision is then taken in the unit of the proxy. Two axes, one dossier, one mistake: we decide in the administrator’s unit about things that have a unit of their own. Thinking in the right units, the theme that holds this pillar together, is not a metaphor here. It is the difference between a mine and a hole.

What follows

What Europe has to decide is not whether to mine tungsten. It is in which unit to decide it. Three things follow, none of them revolutionary and all of them against current practice.

The Critical Raw Materials Centre should be a counterparty, not only a warehouse. Its first task is to sign offtakes on the lifespan of projects, with the Sangdong structure as the template: a state bank’s loan and a processor’s contract, twenty years long. A stockpile is the right answer to next year’s shortage. A contract is the only answer to the one after 2030.

Guarantees must be designed for the tap being reopened. The risk the Delft researchers name is not that prices stay high but that they fall, on purpose, as they did in 1994 and in 2018. A floor price or a contract for difference on a mine’s output is what lets a European mine survive a Chinese decision to flood the market. Without it, every strategic designation is a list of future liquidations.

And report in the object’s unit. The Act’s benchmarks are measured in 2030 shares, and progress is reported in projects designated. Report instead how many years of European consumption are covered by signed contracts. That number would today be close to zero, and a number close to zero is where a policy begins.

The open question is the one the owners of the building faced before the law forced their hand: who carries the loss in the years a guarantee pays out, and whether a democracy elected on a four-year clock will hold a twenty-year line against a falling price. The maintenance plan needed a statute to make owners save. A mine needs a treasury willing to be wrong for a decade. In Devon commissioning began in July, at a price nobody will promise for 2030. That is the unit problem, whole.

Sources

• Liza van Lonkhuyzen, NRC, 17 August 2026: the tungsten market read as a predictor of war; a ninefold price rise in just over a year; Europe too passive (Centre for Materials and Resilience, TU Delft; Benjamin Sprecher). Quotations via NOS, Ingrepen China en VS maken gewild metaal wolfraam schaars, Europa betaalt hoofdprijs (August 2026), and TPO, 20 August 2026.
• Argus Metals, Europe minor metals (February–March 2025): APT Europe assessed at 370–380 $/mtu duty unpaid in the weeks after the 4 February controls, 400 $/mtu by spring, an 18 percent rise since February.
• Fastmarkets, tungsten APT 88.5% WO3 cif Rotterdam/Baltimore, $/mtu: 430–475 (20 June 2025); 900–940 (January 2026); 1,650–1,900 (mid-February 2026); 2,900–3,210 (19 June 2026, +575% year on year); 3,075 (30 July 2026). Fastmarkets, Tungsten 2026: geopolitics sets global tone (18 March 2026) and Tungsten markets fragmenting as domestic Chinese APT market diverges from exports (29 June 2026).
• Ministry of Commerce and General Administration of Customs of the People’s Republic of China, Announcement No. 10 [2025], 4 February 2025: export controls on items related to tungsten, tellurium, bismuth, molybdenum and indium. Ministry of Commerce, 26 December 2025: fifteen enterprises designated to export tungsten in 2026–2027 (reported by Asian Metal and Investing.com).
• IISS, Critical Raw Materials and European Defence (March 2025): China accounts for 86 percent of tungsten at the processing stage.
• USGS, Mineral Commodity Summaries 2025 and 2026, tungsten: production outside China around twenty percent of world output; no commercial tungsten mining in the United States since 2015; China’s 2025 output of roughly 67,000 tonnes as reported by NOS. USGS, Minerals Yearbook 2018, tungsten: annual average APT price of 245 $/mtu in 2017.
• US Bureau of Industry and Security, temporary final rule of 4 August 2026 (Defense Production Act, DPAS directive allocation order): tungsten waste and scrap and black mass to be sold to US persons only, effective 27 August 2026, for one year.
• IEA, The Role of Critical Minerals in Clean Energy Transitions (2021): more than sixteen years on average from discovery to first production for major mines commissioned in 2010–2019.
• Wolf Minerals, Hemerdon: acquisition 2007, feasibility study 2011, production 2015, liquidation October 2018 (Mining Journal, October 2018). Tungsten West: Hemerdon designated a Strategic Project under the Critical Raw Materials Act (June 2025); Q1 2026 project funding and debt update and July 2026 operational update (commissioning begun, fines circuit Q3 2026, full commissioning Q1 2027; 25 million dollar bridging loan, debt package of up to 85 million dollars).
• Almonty Industries, 14 July 2026: Sangdong offtake with Global Tungsten & Powders extended to 21 years, 4.41 million mtu, minimum 210,000 mtu a year, about 490 million dollars a year at current APT pricing; KfW IPEX-Bank project financing of 75.1 million dollars; processing start 1 July 2026. Closure in 1994: Korea Economic Institute of America, Reopening the Sangdong Mine.
• Regulation (EU) 2024/1252, the Critical Raw Materials Act: 2030 benchmarks of ten percent extraction, forty percent processing, twenty-five percent recycling, at most 65 percent from one third country; permitting deadlines of 27 and 15 months. European Commission, 25 March 2025 (47 strategic projects, three for tungsten, among them El Moto and P6 Metals in Spain; IEA policy database) and 4 June 2025 (13 projects in third countries); second call closed 15 January 2026 with more than 160 applications.
• European Commission, RESourceEU action plan, 3 December 2025: European Critical Raw Materials Centre and joint purchasing platform (2026), stockpiling pilot, black-mass export ban (September 2026). Reuters, May 2026, on the shortlist of tungsten, rare earths and gallium for a first joint reserve.
• Keystone, Discounting the future (August 2026), for the mechanism and the instrument; on the same dossier from the institutional side, Statecraft, Reeks IV Nº 06, De ongetelde keten (June 2026).