Is the licence leaking? Bismuth after China's Feb-2025 export-licence regime

Fourth material in the circumvention track (R72), companion to the antimony, graphite and tungsten cases. The price wedge measures scarcity (what a controlled material costs ex-China). This case measures the other half: is the control actually holding, or is Chinese-controlled material re-entering the US through a laundered origin? This is a DUAL-SCORE / alternative-track signal — never folded into any Tier-1 exposure score. Research, not investment advice; origin-relabelling is INFERRED from statistical implausibility plus public production records, never asserted as smuggling on any single shipment.

Verdict

On 4 February 2025 China's MOFCOM/GAC (Announcement No. 10 [2025]) placed bismuth — with tungsten, tellurium, molybdenum and indium — under dual-use export-licence control. Unlike the antimony case, this is a licence regime, not an outright US ban. Yet the bite on the US import line is far harder than the tungsten licence produced: China's direct bismuth line to the US collapsed, and it did so without a clean trans-shipment surger stepping into the gap. The control is biting, but the gap is being filled by a genuine ally-refiner (Korea Zinc), not by an implausible relabelled origin. That is the finding — and it is the opposite of the antimony fingerprint.

Source: UN Comtrade (reporter USA, monthly, HS 810610 + HS 810690), free public preview API; committed artifact `data/intelligence/bismuth-circumvention.json` (40 monthly periods, 2023-01 → 2026-04). Refining shares: USGS Mineral Commodity Summaries 2026 world-refinery table via `lib/iptm-material-country-production.ts`.

Two things move at once and both matter:

1. China's line near-vanishes. From 53% of US bismuth imports in 2023 to 27% in 2025 to 0.3% in the first four months of 2026 (363 kg). The licence is functioning as a de-facto stop on the US-bound direct line. 2. World volume falls 42% while world value rises 54%. Unit value went from $11.8/kg (2024) to $31.5/kg (2025) — a ~2.7× price jump. This is the price wedge showing up inside the same customs dataset: buyers imported far less metal at far higher prices. Bismuth's well-documented 2025 spike (roughly $4/lb → $20+/lb) is visible here as a value-up / volume-down scissor.

Why this is NOT the antimony fingerprint (the honest read)

The antimony case had a clean trans-shipment tell: Thailand and Mexico — countries with ≈0% of world antimony production — surged into the exact gap China vacated. Bismuth does not show that clean tell, for three reasons:

• The gap is filled by a real refiner. Korea is the USGS #2 refined-bismuth

producer (Korea Zinc's Onsan smelter). It was already the co-#1 US supplier before the control and simply held that position at crunch prices (value +126% on −28% volume). A legitimate ally-refiner absorbing demand is substitution, not laundering. Japan (a genuine ~3% refiner) behaves the same way. There is no implausible-origin country needed to explain where the metal came from.

• The implausible surgers sit in the scrap/articles line. Germany (up ~20×

from 3.3 t in 2023 to ~66–70 t in 2024–25) and Belgium (2.1 t → 14 t in 2025) are the only ≈0%-refining origins that rose — but in 2025 their volume is almost entirely HS 810690 ("bismuth, other — incl. articles, waste and scrap"), not the ≥99.99% pure-metal line (HS 810610). That is exactly where legitimate recycling and bismuth-chemical / pharma feedstock (bismuth subsalicylate, alloys) legitimately sits, and both countries host genuine multi-metal recyclers (Umicore in Belgium). The scrap-line flavour materially weakens any relabelling inference.

• The pure-metal line has no implausible surger at all. Splitting HS 810610

(≥99.99% Bi): China 149.5 t (2024) → 67.5 t (2025) → 0 (2026); Korea 332.9 t → 304.5 t (holds); Japan 106.3 t → 97.7 t. The replacement for lost Chinese pure metal is entirely Korea + Japan — both genuine. The only new pure-metal entrants are tiny (Netherlands 1,960 kg, Poland 607 kg in 2025) and immaterial.

No common-ownership tell was traced. Unlike antimony (Youngsun → Thai Unipet → Youngsun & Essen), there is no public registry/filing chain here linking a Chinese producer to a surging German or Belgian exporter-of-record. Absent that pipe, and with the surge concentrated in the scrap line, the relabelling inference is not supported. This is why the case is GATE 0.

Data caveat: several small 2025–26 partner rows in the artifact carry raw UN M49 numeric codes (M76 = Brazil, M348 = Hungary, M376 = Israel, M608 = Philippines, M699 = India) that did not map to an ISO-2 label in the pull. All are small (≤7 t) and none change the picture; they are flagged here rather than leaned on.

What it means for the buyer

1. The dependency is real and the price proves it. A risk team should not read "China's US bismuth share halved" as de-risking. World volume fell 42% and the price nearly tripled — the licence tightened the whole market, it did not open a cheap alternative. The exposure is intact; it is now expensive and Korea-concentrated. 2. Concentration migrated, it did not disperse. The US bismuth line is now disproportionately a single ally-refiner (Korea Zinc) running at crunch prices. That is a different risk shape from Chinese dependence, but it is still single-point concentration — one Onsan outage or a Korean policy shift now moves the US line. 3. A control can bite hard without leaking. Bismuth is the counter-example to antimony: same MOFCOM instrument, but because a genuine non-Chinese refiner with real capacity exists, demand rerouted legitimately rather than through a laundered flag. The presence or absence of a real ally-refiner is what decides whether a control leaks — and that is exactly what the alternative-track is for measuring, per material.

Method & honesty rails

• Trade data: UN Comtrade, US imports, monthly, HS 810610 (≥99.99% bismuth)

and HS 810690 (bismuth, other incl. articles/waste/scrap). Net weight is gross product weight, not contained Bi. 2026 is a partial year (Jan–Apr) and is used only for direction, not level. Late months lag ~2–3 months.

• Production baseline: USGS Mineral Commodity Summaries 2026 refined-bismuth

table. China ≈87.5%; Korea ≈6.25% (genuine #2); Japan ≈3.1%; Laos/Bolivia ≈0.3% each. Germany, Belgium, Netherlands, Poland: no listed refining capacity.

• Alternative-track only: never touches buyerRelativeScore or the base

exposure — it sits beside them, like the China–West price wedge.

• Inference, not accusation: the muted / negative circumvention read here is

driven by the data, not by exoneration of any party. No shipment is asserted legal or illegal beyond what the public record states.

• What would lift this to GATE 1: a traced ownership chain linking a Chinese

bismuth producer to a surging German/Belgian exporter-of-record, or a clean implausible surge in the HS 810610 pure-metal line from a ≈0%-refining origin. Neither is present in the current data.