In February 2024 a Belgian mining company signed a roughly $350 million long-term supply agreement with a Chinese manufacturer in Inner Mongolia. The material was high-purity quartz sand. The mine was in Spruce Pine, North Carolina. The buyer makes the fused quartz crucibles that Chinese semiconductor and solar producers use to grow silicon ingots.
No export licence was required, because none exists. No CFIUS review was triggered, because the transaction did not involve an acquisition. No agency objected. The United States maintains no export control regime of any kind on high-purity quartz.
This is worth sitting with, because for two years Washington has been running an industrial policy premised on the idea that upstream material chokepoints are instruments of national power. It has spent billions to break Chinese control of rare earth separation. It has built a strategic minerals stockpile. It has placed export controls on lithography systems, on electronic design automation software, on advanced logic. And it has left completely open the one upstream node in the semiconductor chain where American territory holds something close to a global monopoly.
What the deposit actually is
Quartz is the second most abundant mineral in the Earth’s crust. Quartz that can be refined to the purity a semiconductor process requires is not abundant at all. Most deposits carry iron, aluminium and titanium at concentrations that are trivial by any ordinary industrial standard and disqualifying here, where contamination measured in tens of parts per billion ruins finished product.
The Spruce Pine district in Mitchell County supplies somewhere between 70 and 90 percent of world output. Two operators work it: Covia, a subsidiary of the Belgian group Sibelco, and The Quartz Corp, Norwegian owned. Between them they ship an estimated 180,000 to 200,000 tonnes a year, up sharply over the past decade as monocrystalline solar displaced older cell architectures.
The material does not go into the chip. It goes into the crucible that holds the molten silicon during Czochralski growth. Above 1,400 degrees, molten silicon attacks nearly every containment material available; fused quartz holds its structure and transfers almost nothing into the melt. Each crucible costs several thousand dollars, survives roughly 400 hours of operation, and represents on the order of 30 percent of ingot production cost. It is consumable, it is unavoidable, and it comes from one valley in Appalachia.
Deposits exist in Brazil, Australia, Madagascar and Russia. None has demonstrated consistent five-nines material at industrial volume. Qualification cycles in this industry run years, because the cost of a bad melt is measured in scrapped wafers rather than in raw material price.
The asymmetry nobody has closed
Beijing has been methodical about the class of instrument this represents. Gallium and germanium went under licence in 2023. Graphite followed. Rare earth processing technology and equipment came next, with named American firms added to control lists in 2026 specifically to reach the construction phase of the domestic magnet buildout. The pattern is consistent: identify the node where global capacity is concentrated in your territory, then convert commercial position into licensing authority.
Washington has run the same play in the other direction on lithography and design software, where the concentration sits with allies. It has not run it on the one material node it physically holds. The contracts flow the other way instead, under multi-year offtake, to firms that supply the Chinese wafer and solar industries.
Whether that gap should be closed is a separate question from whether anyone has noticed it. The evidence suggests nobody has.
Why it is still the smaller risk
The dramatic version of this story overstates itself, and the correction matters more than the alarm.
Hurricane Helene ran the live experiment in September 2024. Both Spruce Pine operations went offline. The world’s primary source of semiconductor-grade quartz stopped for weeks. Analysts put buffer inventory in the chain at three months or better, production resumed, prices moved and settled. A single point of failure holding 80 percent of world supply turned out to be survivable across a quarter.
The deeper reason is the escape hatch. Synthetic quartz reaches higher purity than natural material and is already sold commercially as crucibles, at five to ten times the cost. For a wafer maker that premium is absorbable, because the ingot is a small fraction of finished chip value. For photovoltaics, which consume 25 to 30 tonnes of quartz per gigawatt of wafers, it is not. Concentration converts into leverage only where substitution is priced out. On quartz, the chip industry can buy its way out and the solar industry cannot.
Where the chip-relevant exposure actually sits
Two corrections follow.
The first concerns polysilicon. The widely quoted figure of roughly 94 percent Chinese capacity share is a photovoltaic statistic and is routinely presented as a semiconductor one. Electronic-grade polysilicon at eleven-nines purity is a separate qualification market with different suppliers, and Western capacity across Hemlock, Wacker and Tokuyama is adequate for chip demand precisely because chips consume a rounding error of global tonnage. The 94 percent number describes a solar vulnerability.
The second concerns metallurgical silicon, which is where the genuine structural dependency sits and which almost nobody writes about. China produces around 72 percent of global output at roughly 2.7 million tonnes, against Brazil near 262,000 and Norway near 203,000. The constraint is not technology. Submerged arc furnace smelting is century-old metallurgy. The constraint is electricity price, and the United States has no cost position at all. Sustained Chinese overproduction has pushed prices below the cash costs of Western producers, which turns onshoring from an engineering problem into a permanent subsidy commitment.
That is the unglamorous shape of the problem. The photogenic chokepoint is a mine in North Carolina that the country already owns and does not defend. The consequential one is a furnace bill it cannot pay.
Watch for a licensing regime on high-purity quartz exports. Its absence is the clearest available measure of how seriously the material layer is actually being treated.
