Tearing apart the Pentagon's antimony gap

Antimony.

Two weeks ago, the U.S. Army cut a ribbon at Idaho National Laboratory. A modular pilot plant. The press release called it a "new processing capability for military-grade antimony trisulfide." The defense reporters filed it under feel-good industrial policy. CNBC didn't mention it. Bloomberg ran a brief.

Here's what none of them said: the United States mined zero tonnes of antimony from 2001 through 2024. Not low production. Not declining output. Zero — for more than two decades. Small-scale output resumed last year, but it barely registers. The national defense stockpile is, in the Pentagon's own filings, a rounding error. And the metal they just celebrated processing? It hardens bullet cores, features in ammunition primers, and goes into the pyrotechnic mixture of tracer rounds the U.S. military fires.

The mainstream narrative treats antimony like a curiosity — a niche element overshadowed by the rare earths panic and the lithium drama. But rare earths go into magnets. Lithium goes into batteries. Antimony goes into the things that keep soldiers alive and keep buildings from burning down. Flame retardants account for 49% of U.S. demand. Ammunition and lead-acid batteries take another 40%. It's the metal nobody can name that sits underneath everything nobody wants to catch fire.

And Beijing just proved — conclusively, in real time — that it can switch the supply off whenever it wants.

Numbers. No narrative.

The timeline is what matters. August 2024: Beijing imposed export licensing on antimony. By October, overall Chinese antimony exports had collapsed 97%. December 2024: outright ban on antimony shipments to the United States. Prices, which had traded in a $3 to $15 per kilo band for four decades, went vertical — touching $59,750 per tonne by July 2025. A move of roughly 345% in eleven months.

Then the truce. The November 2025 suspension of the ban — notionally through November 27, 2026. Prices eased. The financial press called it normalization. Rotterdam spot is around $38,000–50,000 per tonne today, depending on grade. Still triple the pre-2024 baseline.

But here's the diagram nobody is drawing. The ban was suspended. The licensing regime was not. Approval rates run below 25%, with processing times of two to three months. The valve was turned from "off" to "partially open." The physical flow is still running well below historical volumes.

The new supply everyone is celebrating? The Stibnite mine in Idaho — Perpetua Resources' $1.3 billion project — broke ground last year. Commercial production: 2028 or 2029. That's not a typo. Two to three years from now. The Hillgrove mine in Australia, Larvotto's flagship, would supply about 7% of global demand. It has only just begun mining, with first development ore reaching the surface earlier this year.

There is no quick fix. There is a pilot plant processing about a tonne of ore per day in Idaho Falls and a 40,000-tonne annual deficit that is being covered by reserves China can cut off with a memo.

November 27, 2026. That's the date.

The suspension of China's antimony export ban to the United States expires in 105 days. There has been no confirmation of extension. No framework for renegotiation. No trial balloon from Beijing or Washington. The market is trading as if the truce is permanent. I've seen that before — with rare earths this spring — and it wasn't fun for anyone positioned that way.

The deeper problem is what happens even if the suspension holds. The licensing regime remains in place. Sub-25% approval rates with multi-month processing delays don't fill a 40,000-tonne hole. The Defense Logistics Agency is trying to rebuild the national stockpile from scratch — antimony metal ingots — but you can't stockpile what isn't available. They're buying in a market where the dominant seller has its hand on the valve.

Meanwhile, the demand side is quietly expanding. Solar glass — a new and growing consumer of antimony trioxide — has become one of the fastest-growing end markets globally. The military drawdown from Ukraine replenishment hasn't been backfilled. Europe is ramping defense spending with procurement lists that all assume antimony arrives from somewhere. They haven't, to my knowledge, checked where "somewhere" actually is.

The weak link isn't production. It isn't demand. It's time. The Western response is real — Stibnite, Hillgrove, the DLA stockpile program — but every piece of it is two to three years from producing a single commercial tonne. The next 105 days are unhedged.

The sequence from here is mechanical. You don't need to guess. You just need to watch the calendar.

If the ban snaps back, the first casualty isn't the military — they'll pay whatever it costs and invoke DPA priority. The first casualty is the civilian flame retardant supply chain. That's plastics, textiles, electronics, building materials. Companies that budgeted antimony trioxide at $25,000 per tonne would be buying at $50,000 or more. Margins don't absorb that. They pass it through or they stop production.

The second-order effect is the one the market consistently underprices. Antimony is a co-product and byproduct of gold, lead, and silver mining. You can't just "turn on" antimony supply the way you ramp a copper mine. There are maybe four or five companies on Earth with near-term antimony production outside the China-Russia-Tajikistan bloc. Perpetua, Larvotto, Mandalay, US Antimony, a couple of early-stage explorers. That's the entire Western bench.

Where does the money go? Not into broad critical minerals ETFs — those are diluted with dozens of elements and stages of development that don't move on antimony. The asymmetric exposure sits with pure-play or near-pure-play producers that have permitted projects, secured offtakes, and production timelines inside a two-to-four-year window. Companies whose revenue inflects directly on spot antimony. If you're looking, you already know the names. There are about five of them.

The market is treating the truce as the new normal. The physical supply chain is telling you it's a ceasefire with an expiration date printed on it. In my experience, when those two disagree, the calendar wins. It just takes longer than you'd like.

One hundred and five days.