Three Ways China's Rare Earth Export Suspension Can End Before December 2026

Two administrative clocks are running down inside the critical-minerals system, and both expire before the end of this year. Neither is a mine, a shipping lane, or a price. Both are dates in Chinese regulatory announcements, and how they resolve will do more to set magnet availability in 2027 than any tonnage figure published between now and then.

The first is 10 November 2026. On 7 November 2025, China's Ministry of Commerce issued Announcement No. 70, suspending six export-control announcements dated 9 October 2025 — Nos. 55, 56, 57, 58, 61 and 62 — through that date. The second is 27 November 2026, the expiry attached to Announcement No. 72 of 9 November 2025, which suspended Article 2 of Announcement No. 46 of 2024, the provision barring U.S.-bound shipments of gallium, germanium, antimony, graphite and superhard materials.

A suspension is not a repeal. Nothing was struck from the control list; the instruments were parked with a return date attached. That distinction is the subject here: a parked rule behaves differently from a dead rule in every system that plans around it — procurement calendars, inventory policy, second-source qualification.

What Is Suspended, and What Never Was

Reporting on this file tends to collapse a layered regime into one on-off switch. As of late August 2026, four layers operate in different states.

Layer	Instrument	Status
Seven medium and heavy rare earths on the dual-use list	MOFCOM Announcement No. 18 of 2025 (April 2025)	In force. Never suspended
Extraterritorial reach over foreign-made goods	Announcement No. 61 of 2025	Suspended to 10 Nov 2026
Technology and equipment controls	Announcements Nos. 56 and 62 of 2025	Suspended to 10 Nov 2026
U.S.-bound ban on gallium, germanium, antimony	Article 2, Announcement No. 46 of 2024	Suspended to 27 Nov 2026

The April 2025 measure is the load-bearing one and has never been paused. It placed samarium, gadolinium, terbium, dysprosium, lutetium, scandium and yttrium — with related alloys, oxides and finished magnet materials — under case-by-case export licensing. A sintered magnet containing terbium or dysprosium still needs a permit for every shipment. What lapsed in November 2025 was the tier built on top of that.

The suspended October instruments were far more aggressive. Announcement No. 61 asserted jurisdiction over items made outside China where Chinese-origin controlled rare earths make up 0.1 percent or more of the value of the finished item, and over goods made abroad using Chinese extraction, smelting-separation, metal-smelting, magnetic-material or recycling technology. That is the architecture Western export-control regimes have used for decades, pointed the other way. Parts of it were due to bind on 1 December 2025; it was parked before most of it ever did.

The Binding Constraint Is the Licence Queue, Not the Ore Body

Most treatment of this subject reaches for mine-production share, the least concentrated part of the chain and therefore the least informative. The U.S. Geological Survey's 2026 Mineral Commodity Summaries put 2025 world rare earth mine production at 390,000 tonnes of oxide equivalent, of which China accounted for 270,000 tonnes — roughly 69 percent — with the United States at 51,000 tonnes. World reserves exceed 85 million tonnes; on mining alone the position looks contestable within a decade.

One step downstream, concentration tightens sharply. The International Energy Agency puts China's share of global rare earth refining and separation near 91 percent and its share of sintered permanent magnet production at about 94 percent, up from roughly 50 percent two decades ago. The chokepoint is separation chemistry and sintering — and above both sits an administrative layer that can be tightened without moving a tonne of capacity.

That layer has measurable parameters. Under the Regulations on Export Control of Dual-Use Items, in force since 1 December 2024, MOFCOM is to grant or deny a licence within 45 working days, extendable. That is roughly nine calendar weeks before any extension, before document preparation and end-user certification. For a buyer running six weeks of magnet inventory, the statutory ceiling alone is a stockout.

The queue shows up in trade data more clearly than in price. Chinese rare earth magnet exports fell to 1,239 tonnes in May 2025, down 74.3 percent year on year, in the first full month after the April controls landed. By December 2025 monthly exports had recovered to 5,952 tonnes — unevenly and directionally. In November 2025, U.S. magnet imports were down about 11 percent year on year while European shipments rose about 60 percent. Yttrium shipments to the United States totalled roughly 17 tonnes from April to December 2025, against 333 tonnes in the preceding eight months.

The pricing consequence of a queue is a geographic spread, not a global spike: the IEA noted European prices reaching up to six times Chinese domestic levels in 2025. A licence regime prices certainty of delivery, not the element.

On 2 December 2025, MOFCOM began issuing general licences to a first batch of three Chinese magnet producers, with one-year validity and scope tied to named downstream customers rather than product categories. That is not deregulation. It converts a per-shipment queue into a per-relationship queue, makes the customer list itself an object of policy, and puts the first cohort's renewal in the same window as the November expiries.

Scenario A — Quiet Rollover, Scope Unchanged

The November dates pass with a further suspension announcement. The April 2025 list stays in force, general licences are renewed and extended to a second batch, and the 0.1 percent rule stays parked. Nothing is repealed; the pause is re-dated.

Triggers: a fresh suspension announcement two to four weeks before 10 November; early renewal of the first general-licence cohort rather than renewal at the wire; expansion of the cohort to additional producers; monthly magnet exports holding at 5,000–6,000 tonnes without a widening U.S.–Europe gap.

Rough weight: 45–55 percent. A suspension is cheaper than either a repeal or a re-imposition; it preserves the option without paying to exercise it. What it would not fix is the queue: the April regime, the 45-working-day window and the customer-tied licence structure all survive intact.

Scenario B — Selective Lapse With a Widened Carve-Out

The two expiries are treated differently. Part of the October stack lapses back into force — most plausibly the technology and equipment items rather than the 0.1 percent content rule — while the general-licence channel widens enough that the largest qualified buyers barely feel it and everyone outside that channel feels it considerably.

This branch produces the most confusing data: aggregate tonnage can look stable while the distribution underneath shifts hard. The November 2025 divergence — U.S. imports down 11 percent, European up 60 percent, same month — is the shape to watch for.

Triggers: an announcement naming specific items rather than the whole 2025 package; a widened licence cohort alongside a partial reinstatement; end-use certification attached to technology transfer while magnet shipments are left alone; renewed volatility in yttrium, lutetium or scandium while neodymium-praseodymium flows stay orderly. Rough weight: 25–35 percent. Selective instruments are deniable and reversible, and the December 2025 general-licence architecture makes selectivity administratively easy for the first time.

Scenario C — The October Stack Returns Substantially Intact

The full October 2025 package resumes, including Announcement No. 61's 0.1 percent threshold and its reach over foreign-made goods produced with Chinese process technology. This branch changes the compliance obligation for firms that have never bought from China directly.

A 0.1 percent-of-value threshold is extraordinarily low. In an assembly containing a small sintered magnet, rare earth content can clear that bar without the assembler having any visibility into it. The burden is not the licence application; it is bill-of-materials traceability to the element level across several supplier tiers, which most manufacturers outside aerospace and defence do not maintain.

Triggers: silence through the first week of November with no re-suspension; a statement recharacterising the November 2025 suspensions as conditional; implementing guidance for Announcement No. 61 issued before its expiry, signalling preparation to administer rather than shelve it; non-renewal of the first licence cohort at its one-year mark. Rough weight: 15–25 percent. The counter-case deserves equal statement: reinstatement most directly accelerates substitution and non-Chinese capacity — a permanent cost paid for a temporary lever. Using it twice in eighteen months would test whether it degrades with use.

The 2010–2015 Episode Rhymes, but Only Halfway

There is one prior full cycle of this instrument, and it ran to completion. From 2010, Chinese rare earth exports were capped at roughly 35,000 tonnes per year under a quota system. The United States, European Union and Japan brought a joint WTO complaint; panel reports in DS431, DS432 and DS433 circulated on 26 March 2014 and Appellate Body reports on 7 August 2014, finding the quotas inconsistent with China's obligations. The quotas were abolished in January 2015 and the export taxes removed later that year.

The instructive part of that cycle is not that the restriction was defeated. It is what the restriction had already caused by the time it was lifted — engineering changes that did not reverse when prices fell.

Magnet users engineered heavy rare earths out of their designs wherever they could. One Japanese automaker cut dysprosium content in its electric-motor magnets by roughly 40 percent, and lighting demand for europium and terbium collapsed as fluorescent tubes gave way to LEDs. When quotas ended, the demand that returned was structurally smaller than the demand that had left.

Two differences make the analogy incomplete. The 2010 instrument was a quantitative quota, straightforward to challenge under trade law; the 2025 instrument is a dual-use licensing regime framed on national-security grounds, far harder to litigate. And the heavy rare earths controlled in April 2025 exist to preserve magnet coercivity at high temperatures — the property hardest to engineer around in traction motors.

The supply-side response is subsidised differently this time. A U.S. Department of Defense public-private partnership announced in July 2025 set a ten-year floor price of $110 per kilogram for NdPr oxide against a market benchmark then near $60, with full offtake on 7,000 tonnes per year of new magnet capacity and a $140 million annual earnings guarantee. Floors of that construction are built to survive the post-squeeze collapse that killed the 2010-vintage projects — a question for 2030, not November.

Where This Frame Would Fail

This frame treats the licensing layer as the dominant variable. Several conditions would make that wrong.

• If inventory is deeper than assumed. Buyers spent 2025 building stock. If the average buffer is nine to twelve months rather than six to eight weeks, a November reinstatement produces almost no observable effect until well into 2027 and the scenario tree resolves into noise on this timescale.
• If the queue is no longer binding. The 45-working-day figure is a statutory ceiling, not an observed average, and 2026 approval times are not published in any verifiable source. If processing has compressed to two or three weeks under the general-licence system, refining capacity reasserts itself as the real constraint.
• If the expiry dates are not the decision points. Suspensions can be extended, shortened or superseded at any time. Treating 10 and 27 November as the moments of resolution imposes a calendar on a process that has repeatedly moved outside its own schedule.
• If demand falls faster than supply is restricted. A slowdown in electric-vehicle and wind-turbine build rates would slacken magnet demand independently of policy, and a slack market makes a licence queue far less consequential.
• If substitution has already moved further than trade data shows. Design changes appear in trade statistics with a long lag; motor architectures adopted in 2024 and 2025 would not yet be visible in 2026 import volumes.

What to Watch Next Week

• MOFCOM's export-control portal, for any announcement referencing Nos. 55–58, 61, 62 or 70. Early re-suspension points to Scenario A; item-by-item language to B; guidance for No. 61 to C.
• Monthly Chinese customs data for rare earth permanent magnets, by destination. The level matters less than the U.S.–Europe spread; a repeat of the November 2025 divergence is the leading marker for selective treatment.
• Narrow-element flows — yttrium, lutetium, scandium, terbium — rather than headline NdPr. In 2025, small-volume elements moved first and recovered last.
• General-licence cohort news ahead of the first batch's December anniversary. Additions signal widening; silence into November is a warning on renewal.
• Ex-China separation commissioning. Each new heavy-rare-earth line outside China lowers the ceiling on what a future licence action can achieve.

Concrete Framework

A monitoring checklist for the next ninety days, cheapest to most costly to maintain.

1. Set an alert for 27 October 2026, two weeks before the first expiry. If no re-suspension has appeared by then, raise the weight on Scenarios B and C and lower it on A.
2. Track two series monthly: Chinese customs exports of rare earth permanent magnets in tonnes, and the same series split by U.S. versus EU destination. A widening destination gap with a flat total is the Scenario B signature.
3. Maintain a four-row state table covering Announcements Nos. 18, 61, 56 and 62 of 2025 and Article 2 of No. 46 of 2024. Update the status column only from MOFCOM portal text — secondary reporting has repeatedly conflated the October and November dates.
4. Answer the exposure question early. What share of finished goods, by value, contains a sintered rare earth magnet anywhere in the bill of materials? Under a 0.1 percent threshold the answer is usually higher than procurement expects, and establishing it takes months.
5. Keep the two clocks separate. 10 November governs the extraterritorial and technology layers; 27 November governs the U.S.-bound gallium, germanium and antimony provision. They can resolve in opposite directions, and treating them as one event gives the wrong read on both.
6. Log the base rate. The one completed cycle ran from 2010 to 2015 and ended with the restriction removed and demand permanently smaller. A single observation is not a distribution, but it is the only one available, and it argues for weighting the durability of substitution above the durability of the control.

The honest position is that the November outcome is not knowable in advance. The useful work is not guessing the branch but holding a trigger list that identifies which branch is being taken while it is still early. The instrument is administrative, the schedule is published, and the signals arrive in documents rather than prices — an unusually observable setup.

Disclaimer: This article is analysis of policy and market structure for general information. It is not investment or financial advice.