Critical Materials Atlas
Critical material · profile · CN 7202 93 00

Ferro-niobium

Brazil leads both the mining and the export of ferro-niobium — a genuine, not an accounting, concentration.

64/100
supply-risk index
215 yr
reserve life (reserves ÷ mining)
100%
US import reliance
🇧🇷 Brazil
lead reserves · 90%
🇧🇷 Brazil
lead miner · 90%
🇧🇷 Brazil
lead refiner · 88%
🇧🇷 Brazil
top exporter · 76%
0.60
export concentration (HHI)

Primary uses. high-strength low-alloy steel for pipelines; superalloys; superconducting magnets.

Brazil’s share of world ferro-niobium exports, 2002–2024: 75% → 76%

The chain — from the ground to the buyer

A critical material passes through distinct stages, and the country that leads each stage is often different — that gap is what this atlas exists to show. Each layer comes from a different public source with its own vintage, labelled below; shares are % of the world total.

● Reserves — where it could come from

Reserves are the deposits known to exist and economically worth mining — the long-run ceiling on supply, distinct from what is actually produced today. Ferro-niobium's reserves are highly concentrated: 🇧🇷 Brazil holds the most (90%), ahead of 🇨🇦 Canada (9%) and 🇺🇸 United States (1%). At today's extraction rate the known reserves would last roughly 215 years, so scarcity is not the binding constraint — access, capital and processing capacity are. USGS 2024 · reserves as of 2023

🇧🇷 Brazil90%
🇨🇦 Canada9%
🌍 Others (not detailed)1%

● Mined — where it is dug up today

Mining is where ore leaves the ground — the stage most people equate with “the source”, though it is rarely where the supply risk actually concentrates. Output is highly concentrated: 🇧🇷 Brazil supplies 90%, with 🇨🇦 Canada (8%) next. That is effectively a single-country dependence — one government's policy or one region's disruption can move the entire upstream. USGS 2024 · 2023 production

year2023
🇧🇷 Brazil90%
🇨🇦 Canada8%
🌍 Others (not detailed)2%

● Refined / processed — where it becomes usable metal

Refining / processing converts ore into the metal or compound buyers actually purchase; it concentrates in fewer hands than mining and sits at the buyer's doorstep, which is why it is usually the true chokepoint. 🇧🇷 Brazil processes 88% — highly concentrated. Brazil leads both mining and refining, so its grip is structural rather than a pure processing gate. On the buffers: 0% of supply is recovered from end-of-life recycling and substitutes are only partial, while the US imports 100% of what it consumes. leading refiner only — fuller breakdown not publicly reported

🇧🇷 Brazil88%
🌍 Rest — not separately reported12%

● Recycling & substitutability — the mitigants (EU CRM)

0% of supply comes from recycling end-of-life products — there is essentially no end-of-life recycling, so a disruption has no secondary cushion. Substitutability is medium — partial substitutes exist.

● Traded — who ships it

Actual bilateral trade of the traded form, reconciled from UN Comtrade / CEPII BACI. Pick a year below — 2018–2024 measured, 2025* nowcast, 2026** directional scenario. The full 2002–2026 range is on the interactive atlas.

year2024
Top exporters — reconciled trade, 2024
Countrysharevaluetonnes$/t
🇧🇷 Brazil76%$2.6B100.4 kt$25,718/t
🇨🇦 Canada9%$315M10.5 kt$29,908/t
🇳🇱 Netherlands8%$267M9.0 kt$29,802/t
🇨🇳 China2%$68M5.5 kt$12,268/t
🇸🇬 Singapore2%$57M2.2 kt$26,010/t
🇲🇾 Malaysia0%$17M4.8 kt$3,448/t
Top importers — reconciled trade, 2024
Countrysharevaluetonnes$/t
🇨🇳 China32%$1.1B42.9 kt$25,655/t
🇳🇱 Netherlands15%$531M18.8 kt$28,271/t
🇺🇸 United States8%$277M10.3 kt$26,822/t
🇰🇷 South Korea6%$215M8.3 kt$25,792/t
🇸🇬 Singapore5%$177M7.1 kt$25,085/t
🇯🇵 Japan5%$175M5.9 kt$29,464/t

Context

A near-monopoly, but on Brazil, not China: Brazil supplies ~84% of EU ferro-niobium (the additive for high-strength steel), with Canada the rest. The naive view points at the Netherlands (~49%, a Rotterdam clearing artefact); the origin is Brazil.

Explore ferro-niobium in the atlas → The origin gap Methodology

Every figure on this page is computed from out/data.json and out/flows_2024.json by build_profiles.py — no hand-entered numbers.