Map exposure by process step, not by address
The most common error in magnet risk assessment is treating the finishing factory's country as the answer. A magnet sintered and coated outside China may still have been made from Chinese-separated oxide, Chinese-produced metal, or alloy from a single strip-casting source — and may depend on production technology that is itself subject to export controls.
Rare earth material must be mined, separated, refined, converted to metal, alloyed, milled to powder, pressed, sintered, heat treated, machined, coated, magnetized and inspected. Each step is a potential constraint. Diversification at one step tells you nothing about the others.
Mining has diversified meaningfully. The steps immediately after it have not, which is why material mined in one region is still routinely shipped to Asia for processing — the downstream capability exists there and is still being built elsewhere. A country can mine rare earths without being able to manufacture a competitive permanent magnet.
Suppliers vary in how much of this they will disclose, and unwillingness is itself information. A supplier with multiple qualified production routes usually has an incentive to tell you, because it is a genuine competitive advantage. Fold the answers into your supplier audit record rather than keeping them in a separate document that ages independently.
The export-control landscape
Rare-earth export controls moved from background condition to operating constraint over 2025 and 2026. The specifics change; the structural point does not, which is that a licensing decision can now interrupt supply independently of price or capacity.
| Date | Development | Practical effect |
|---|---|---|
| April 2025 | MOFCOM Announcement No. 18 adds seven medium and heavy rare earth elements and derived products to the restricted list | SmCo and Dy/Tb-bearing NdFeB brought into scope; license required per shipment |
| Mid 2025 | Review timelines extend well beyond informal expectations | Lead times become licensing-dependent and difficult to forecast |
| October 2025 | Scope expanded to 13 rare-earth input materials, four permanent-magnet materials and seven target materials | Broader documentation demands, including end-use disclosure |
| December 2025 | Extraterritorial provisions take effect for goods made outside China using Chinese-origin material or technology | Non-Chinese producers can fall within scope |
| Late 2025 | Following high-level trade discussions, some measures suspended; general licences introduced | Approved producers gain year-long, customer-specific permits |
| Dec 2025 onward | First general licences issued to major magnet manufacturers | Faster flow for named customers; others remain on per-shipment approval |
| 2026 | Further licensing requirements for high-performance magnets and related manufacturing technology | Technology transfer and dual-use applications under closer scrutiny |
Export-control status changes on short notice and with retroactive practical effect. Treat the table above as orientation, not as a compliance position. Confirm current requirements through official publications and your own trade-compliance counsel before making a sourcing commitment. This page is general reference material and is not legal advice.
Three consequences matter for procurement. First, being on a general licence is a real commercial advantage for a supplier, and worth asking about directly. Second, applications tied to aerospace, medical, sensing or anything potentially dual-use attract more scrutiny and longer review — so end-use documentation quality directly affects your lead time. Third, a supplier can be entirely capable of making your part and still be unable to ship it, which is a failure mode that capacity planning does not capture.
Build the risk register
A magnet risk register is a part-level document, not a supplier-level one. Two parts from the same supplier can carry entirely different exposure because of grade, heavy rare-earth content and end application.
The requalification field is the one most often filled in optimistically. Ask the question concretely: if this supplier stopped shipping tomorrow, what is the sequence, who performs each step, and how long does the slowest one take? For a part requiring customer approval or a regulatory submission, the honest answer is frequently six to twelve months — which changes the inventory decision entirely.
Design flexibility belongs in the register because it is often the fastest genuine mitigation available. A part specified at a high-coercivity grade for a duty cycle that never approaches the limit carries unnecessary cost and unnecessary licensing exposure. Reviewing those specifications against real thermal data is usually the highest-return risk action in the portfolio, and it is engineering work rather than procurement work — which is why it needs a named owner.
Score and rank exposure
Scoring converts a register into a priority list. Keep it simple enough to be maintained and transparent enough to be argued with.
| Factor | Score 1 | Score 3 | Score 5 |
|---|---|---|---|
| Qualified sources | Three or more, all active | Two, one warm | One |
| Origin concentration | Independent to alloy level | Shared upstream at one step | Single upstream chain |
| Heavy rare-earth content | None | Moderate, H or SH grade | High, UH or EH grade |
| Licensing exposure | Outside controlled categories | Controlled, general licence in place | Controlled, per-shipment approval |
| Requalification time | Under 8 weeks | 8 to 26 weeks | Over 26 weeks |
| Inventory cover | Exceeds requalification time | Half of it | Under a quarter of it |
// rank descending; the top ten is the working list for the year
// re-score quarterly — scores drift as alternates go cold and cover is consumed
Multiplying by revenue rather than by spend is deliberate. It surfaces the low-spend, high-consequence parts that a conventional procurement analysis sorted by purchase value will never show — and those are consistently where the unmanaged risk lives. It also produces a number that means something to a finance audience, which matters when you need budget for inventory or qualification work.
Leading indicators worth watching
By the time a shortage appears in your delivery performance, your options have narrowed. These indicators move earlier, and most cost nothing to track.
Quoted lead time deserves particular attention because it is free, it is in your own data, and it moves several months before availability problems become visible. A supplier whose standard quote drifts from six weeks to ten over two quarters is telling you something they may not say directly.
The heavy rare-earth spread is worth tracking separately from the headline index. Constraints on dysprosium and terbium hit high-coercivity grades specifically, and a portfolio weighted toward UH and EH parts can be badly exposed even when general magnet availability looks unremarkable. That is the same signal your indexed contracts should already be decomposing.
The mitigation playbook
Different risks need different responses, and matching them correctly matters more than the size of the response. Inventory does not solve a licensing problem, and qualification does not solve a price problem.
| Risk | Primary mitigation | Secondary | Typical time to effect |
|---|---|---|---|
| Single source | Qualify an alternate with independent upstream chain | Safety stock covering the qualification window | 3–12 months |
| Licensing interruption | Source with a general licence or outside controlled scope | Inventory ahead of known review cycles | 1–6 months |
| Heavy rare-earth exposure | Validate a lower grade against real duty cycle | Grain boundary diffusion variant at equal coercivity | 2–6 months |
| Price volatility | Index-linked contract with a collar | Fixed-rate stocking program for standard parts | Next contract cycle |
| Allocation risk | Term agreement with priority clause and forecast sharing | Warm second source carrying real volume | 1–2 quarters |
| Upstream concentration | Require alloy-level origin disclosure and diversity | Regional supplier despite premium | 6–18 months |
| Design rigidity | Approve two grades or geometries on the drawing | Ferrite or SmCo alternate where performance allows | Next design cycle |
The last row is undervalued. Approving a second acceptable grade or a small dimensional variation on the drawing costs almost nothing at design time and dramatically widens the sourcing options later. Doing it retroactively, under pressure, with a line down, is expensive and slow. Push it into the design release checklist rather than treating it as a procurement problem.
Forecast sharing is similarly underrated as a mitigation. Suppliers reserve material and capacity for customers whose demand they can see. A realistic rolling forecast, delivered consistently, buys allocation priority that no contractual clause reliably delivers — and it costs nothing but discipline.
Reporting risk upward
Magnet risk competes for attention against every other category, and it usually loses on spend. It wins on consequence — but only if you present it that way.
Lead with revenue, not with spend
“Magnets are 0.3% of purchase value” ends the conversation. “$140 million of revenue depends on eleven single-sourced magnet part numbers, six of which need a heavy rare earth subject to export licensing” starts a different one. Both sentences describe the same portfolio.
Show the trend, not the snapshot
A quarterly exposure number moving in the right direction demonstrates that the strategy works. A single point-in-time assessment is a report; a trend is a program, and programs get funded.
Be explicit about what money buys
Qualification work, insurance volume with alternates and safety stock all cost real money and all reduce a quantified exposure. Present them as priced options against the revenue at risk rather than as budget requests, and let the decision be made explicitly rather than by default.
Name what you are choosing to accept
Some exposure is not economically worth removing. Say so, in writing, with the reasoning. A risk that has been consciously accepted and documented is a managed risk. The same risk unmentioned is the one that generates a post-incident review asking why nobody flagged it.
The register feeds directly into category positioning, the licensing and origin fields feed your country-of-origin position, and the requalification estimates size both inventory cover and the second-source program. Kept as one document rather than four, it stays current.