Critical Minerals
Supply Chain Security Strategy

The DRC has about seventy percent of the world's cobalt mine output and almost no influence over where that cobalt ends up, what it sells for, or who makes money converting it into battery-grade sulfate. Japan has no mines worth mentioning. What Japan has is JOGMEC, and JOGMEC has spent fifty years doing something that looks boring on paper and turns out to be extraordinarily effective: taking ten-to-twenty-percent equity stakes in mining projects scattered across six continents. These stakes are deliberately too small to trigger political resistance in host countries about foreign control of natural resources. They are exactly large enough to secure a board seat, see the production data, and enforce a contractual right to preferential supply when the market tightens. The gap between the DRC's position and Japan's position tells you almost everything about what supply chain security actually means versus what most policy documents think it means.

The public conversation about critical minerals is organized around individual commodities. Lithium. Cobalt. Nickel. Rare earths. Each one gets its own analysis with reserves, production data, trade flows. That framing misses where the real fragility sits. The dangerous parts of this system are in the joints, not the bones: the coupling between minerals that come out of the ground together and cannot be separated from each other's economics, the gap between where ore is extracted and where it becomes something a manufacturer can use, the fact that for many of the most strategically sensitive minerals, the price mechanism of undergraduate economics textbooks simply does not apply.

The USGS critical minerals list went from 35 entries in 2018 to 50 in 2022. The EU's 2023 Critical Raw Materials Act has 34. Those numbers keep climbing as the analytical frameworks, which score minerals on supply risk and economic importance, keep scooping more items into the net. At fifty entries the list becomes a bureaucratic artifact. Nobody can simultaneously stockpile, build processing plants for, and negotiate bilateral agreements covering fifty different minerals. The list does not tell you how to prioritize. A mineral's real vulnerability comes from three factors that multiply: supply concentration, substitution difficulty, and the degree to which downstream industries have sunk irreversible capital into technology that depends on it. When any one of those three collapses, the mineral stops being critical. That is happening right now with cobalt. Lithium iron phosphate batteries gave the EV industry a path around cobalt. DRC still has the same stranglehold on cobalt mining. It matters less every quarter. Meanwhile gallium has all three factors maxed. Overwhelmingly concentrated production. No performance-equivalent substitute for GaN in high-frequency power electronics. Billions in 5G infrastructure hardwired around gallium nitride chip designs.

Cobalt, gallium, germanium, indium, rhenium, tellurium. These minerals do not have independent supply chains. They are byproducts. Their output is determined by someone else's market.

Cobalt comes out of copper mines. Gallium comes out of the Bayer process for making alumina. Germanium comes from zinc smelting flue dust. Rhenium tags along with molybdenum. Tellurium is recovered from anode slime in copper electrolysis. When the copper industry expands, more cobalt appears whether anyone wanted it or not. When zinc demand softens, germanium supply contracts regardless of what germanium buyers are willing to pay.

The standard economics assumption that higher prices bring more supply does not work here. Gallium can triple in price and not a single new bauxite mine will open because of it. Gallium is a rounding error in alumina plant revenue. Germanium spiked in 2022. Zinc producers did not react. Germanium is less than two percent of the value in zinc concentrate. No zinc operation rethinks its mine plan for two percent.

The perverse dynamics go deeper than just price insensitivity. Say aluminum demand surges because electric vehicles need to weigh less. Bayer process lines expand. More gallium falls out as a byproduct. Gallium price drops. A small number of companies that had been running dedicated gallium extraction from coal fly ash or zinc residues find they cannot compete with byproduct gallium whose marginal production cost is essentially zero. They close. Net result: gallium output has increased, independent supply channels have been eliminated, and all gallium production is now tethered to the aluminum industry chain. More supply. Greater structural fragility. This has already played out in multiple byproduct mineral markets.

The policy response depends on whether the host metal has a growth trajectory or a declining one. Cobalt rides on copper, gallium rides on aluminum. Both host metals face strong demand, so total byproduct volumes probably hold up. The risk is processing-stage concentration. Rhenium rides on molybdenum and some indium comes from tin. If those host metals face declining demand, the byproduct supply base itself shrinks, and at that point you need dedicated extraction routes, which for rhenium or indium barely exist at commercial scale.

Where ore is dug up and where it is processed into something industrially useful are two different geographies, and the way that gap was created matters for whether current efforts to close it will succeed.

The mechanism was overcapacity. Build more processing plants than the market needs. Run them at scale. Push processing fees below what anyone else can survive on. Wait for overseas competitors to shut down. Harvest the monopoly position. Australia used to separate rare earths. Canada used to refine cobalt. The US had a functioning mine-to-magnet rare earth supply chain. Those capabilities did not disappear because geology changed. They were competed out of existence through sustained price pressure. The same strategy delivered dominance in solar panels and polysilicon. At some point the pattern becomes hard to attribute to coincidence.

This happened in real time in 2023-2024 when lithium processing fees collapsed and several Western projects got shelved or cut back. Subsidies cover construction costs. They do not defend against price warfare during the ramp-up phase. And they rarely survive more than two changes of government. A smelting plant needs a decade to become self-sustaining. An electoral cycle is four or five years. Investors have to bet that successive administrations will hold the policy line. Many boardrooms will not take that bet.

Even after separation, between oxide and permanent magnet sit alloy smelting, melt spinning, jet milling, orientation pressing, sintering, surface coating. Magnet manufacturing is more concentrated than separation. Full mine-to-magnet autonomy is measured in decades of effort, not years.

Glencore. Trafigura. Mercuria. These companies barely show up in mineral security policy documents. In practice they control large fractions of physical mineral trade and operate what amounts to a private real-time intelligence system on global mineral logistics. Ship movements, smelter inventory levels, mine maintenance schedules. When a government scrambles for emergency mineral supply during a crisis, the entity across the table often has better information about global availability than the government does.

Glencore in cobalt: simultaneously a major DRC miner, the world's largest cobalt trader, a processor, and a warehouser. When supply tightens, that company's commercial decisions about inventory and customer allocation have consequences across the entire battery materials chain. Not an accusation. A description of market structure.

The market wants neodymium and dysprosium for permanent magnets. Mining rare earths produces three to five times as much lanthanum and cerium by weight as neodymium. You cannot mine neodymium selectively. The seventeen elements come out together.

Surplus lanthanum and cerium sell cheap. Cheap prices kill incentive to develop new applications. Without new demand sinks, surplus element costs get loaded onto neodymium and dysprosium. Magnet prices go up. Wind turbine and EV motor economics suffer.

Two possible exits. Build enough demand for surplus elements to absorb the extra volume: lanthanum-nickel hydrogen storage, cerium polishing compounds, cerium catalysts. Can any of these scale to the volumes needed? Genuinely unclear. Or find ore bodies with different element ratios. Deep-sea muds in parts of the Pacific show elevated heavy rare earth fractions. Deep-sea mining might reshape the balance problem more fundamentally than any demand-side fix. It also might not reach commercial scale within any planning horizon that matters. I have no confidence in anyone's predictions about deep-sea mining timelines.

The balance problem matters because permanent magnet costs feed directly into wind turbine and EV drivetrain economics. It sits inside the cost structure of the energy transition.

This one rarely gets discussed outside chemical engineering circles.

A factory that has been running on feedstock from one supplier for years has its entire process tuned to that supplier's product characteristics, not just the characteristics covered by industry specifications (purity, major impurity limits), but trace impurity profiles, particle morphology, surface area distribution. None of those appear in the published spec. A cathode precursor plant that switches lithium hydroxide suppliers may find the co-precipitation reaction behaving differently even though both products pass the standard. pH drifts, particle size distribution shifts, yield drops. Two percent. Five percent. At five-to-eight-percent profit margins, a few months of depressed yield eliminates a quarter's earnings.

So factories do not switch. This lock-in was not designed. It accumulated over years of process optimization around a single feedstock source. Breaking it requires qualifying multiple feedstock sources during initial commissioning, which costs more time and money. Almost nobody does it in advance.

What form to stockpile. Ore is stable and cheap to store. It is useless without smelting capacity to process it, and if smelting is the bottleneck, you have warehouses full of expensive rock. Intermediates go directly into production lines, but lithium hydroxide carbonates in air, lithium carbonate clumps in humidity, metal sputtering targets oxidize. Japan stockpiles ore, intermediates, and finished metals depending on the mineral. The differentiation reflects genuine thinking about what each material's downstream chain requires.

How to release. If markets read a release as a signal that the government considers things bad enough to tap strategic reserves, the release can trigger hoarding. The SPR release history shows exactly this happening. Mineral reserves are harder to release than petroleum because most critical minerals have no liquid market to absorb the flow.

Whether to publish quantities. JOGMEC discloses which metals Japan reserves, not how much. Strategic ambiguity. China discloses nearly nothing. Oil reserve transparency works because the oil market is deep and liquid. Critical mineral markets are thin. Full transparency may be destabilizing rather than stabilizing.

China's export controls on gallium, germanium, antimony, graphite. Indonesia's nickel ore ban. Zimbabwe's lithium restrictions. Chilean and Mexican lithium nationalization debates.

These are not the same thing. Indonesia wants foreign companies to build smelters in Indonesia. That is industrial development policy. Respond with investment negotiation, not confrontation. Zimbabwe follows the same logic. China's gallium and germanium moves are retaliation in the semiconductor technology contest. Different motivations, different appropriate responses. Treating them all as "resource weaponization" leads to reaching for the wrong tools.

DRC artisanal cobalt fails social criteria. The same region's industrial mines can have lower carbon footprints than Australian sulfide nickel operations requiring high-temperature roasting. Argentine lithium brine has minimal carbon emissions and maximum indigenous water rights conflict. The three letters point in different directions at the project level.

The hardest version: mining critical minerals damages local environments. Those minerals are needed to build clean energy systems. You cannot square this circle with better language. It requires case-by-case tradeoffs.

Permitting reform that actually works without weakening protections: parallel review, standardized templates, early engagement mechanisms. Canada and Australia have demonstrated thirty to forty percent time reduction. Procedural changes, not substantive ones.

Oil has IEA monthly reports, EIA weekly inventories, live futures, satellite production monitoring. Most critical minerals have none of that. Rare earth pricing depends on survey-based assessments from reporting agencies. Gallium and germanium trade data is scattered across national customs databases with inconsistent codes and months of lag.

Satellite remote sensing is beginning to fill gaps. Commercial imagery estimates stockpile volumes, tracks mine advance rates, infers smelter utilization from thermal signatures. Sentinel-2 open data plus machine learning can monitor hundreds of mines without requiring anyone's cooperation. Satellites see surfaces, not warehouses. Where refined product goes after leaving the plant gate is invisible from orbit. A working early warning system needs to fuse remote sensing, customs flows, shipping AIS data, price reporting, and human intelligence. That integrated system does not exist yet.

The MSP's fourteen members share an interest in supply security and compete in downstream battery and EV manufacturing. Joint smelter investment is cooperative until someone asks whose territory, whose companies get priority supply, whose taxpayers cover overruns.

Witwatersrand gold tailings hold uranium and rare earths. Montana copper tailings contain cobalt and germanium. Chilean copper tailings contain rhenium and tellurium. The material is already mined, crushed, ground. No new land disturbance. Processing reduces legacy environmental liabilities. Every tailings facility needs custom process development because the mineralogy has been altered by prior processing and residual reagents interfere with new extraction chemistry. The US Superfund list includes over a hundred mining sites, many with tailings never assessed using modern techniques.

Lithium in 2000: psychiatric medication ingredient, glass additive, lubricating grease component. Annual production in the tens of thousands of tons. It would have appeared on no critical minerals list. Technology shifts rewrote its strategic position within fifteen years.

Antimony. Small market. Flame retardants, ammunition, lead-acid alloys. Almost no public visibility. Showing up in liquid metal battery research and III-V semiconductor work. Extremely concentrated production. If any of those applications crosses from lab to commercial within five to eight years, this market breaks overnight.