The Busan deal that Beijing and Washington struck late last year quietly suspended the October 2025 round of rare earth controls, and a lot of US importers read that as a thaw. It was not a thaw. MOFCOM Announcement No. 18, issued on 4 April 2025, remains fully operational twelve months on. Samarium, gadolinium, terbium, dysprosium, lutetium, scandium and yttrium still require export licences. The end-user certificate regime still applies. The licence approval queue at MOFCOM’s Department of Foreign Trade still averages 47 business days for first-time applicants, and China still controls roughly 94% of global sintered permanent magnet output and about 90% of rare-earth refining capacity. If you are sourcing catalysts, magnets, phosphors, or any specialty metal compound that touches those seven elements, your 2026 programme needs to assume the controls are permanent unless and until the State Council says otherwise in writing.
What Announcement No. 18 actually does, and why the Busan suspension did not touch it
Announcement No. 18 was issued jointly by MOFCOM and the General Administration of Customs on 4 April 2025. It placed seven medium and heavy rare earth elements and related items, including alloys, oxides and certain magnet products, onto the dual-use export control list. Exporters must apply for a licence on a per-shipment basis. Applicants must submit an end-user certificate signed by the foreign importer, attestations that the material will not be used for military end-use or military end-users, and in many cases a Chinese-government-verifiable statement of final application. MOFCOM retained discretion to deny on national security, foreign policy or non-proliferation grounds with no appeal route.
The Busan deal in late 2025 suspended a subsequent announcement targeting heavy rare earth magnets, certain gallium downstream products and additional graphite categories. It did not touch Announcement No. 18. That distinction matters because many US importers assume the whole rare earth control apparatus is in limbo. It is not. Samarium cobalt magnets used in aerospace actuators, dysprosium-doped neodymium iron boron magnets used in EV traction motors, terbium-doped phosphors used in LED lighting, scandium master alloys used in aluminium alloys, yttrium-stabilised zirconia catalysts used in automotive catalytic converters, gadolinium gallium garnet used in MRI, lutetium oxide used in PET scanner crystals, every one of those product categories still requires a Chinese export licence for the upstream rare earth component.
The practical effect one year in is that the typical US buyer of a dysprosium-iron master alloy now waits 47 to 68 business days between placing a purchase order with a Chinese producer and seeing a bill of lading. For first-time buyers the wait stretches to 90 days. Some applications get denied outright with no stated reason, and the denial rate for US-addressed applications climbed from roughly 8% in May 2025 to 19% by December 2025 according to Chinese industry association data compiled by Adamas Intelligence.
The element-by-element status table you should have open in every sourcing meeting
| Element | HS codes affected | Chinese global refining share | Primary end-use at risk | Alternative sources active in 2026 |
|---|---|---|---|---|
| Samarium (Sm) | 2846.90, 8505.11 | ~88% | Samarium cobalt magnets for aerospace, radar | MP Materials California (trial); Lynas Kalgoorlie (2026 ramp) |
| Gadolinium (Gd) | 2846.90, 2846.10 | ~92% | MRI contrast agents; gadolinium gallium garnet | Solvay La Rochelle (limited); Iluka Eneabba refinery (2026) |
| Terbium (Tb) | 2846.90 | ~99% | LED phosphors; high-temp neodymium iron boron magnets | Lynas Mt Weld; Vietnamese in-situ leach projects (pilot) |
| Dysprosium (Dy) | 2846.90 | ~99% | EV traction motor magnets; defence actuators | Lynas Mt Weld; MP Materials (2026 heavy RE circuit) |
| Lutetium (Lu) | 2846.90 | ~90% | PET scanner crystals; specialty catalysts | Solvay; Shenghe-linked non-Chinese toll routes |
| Scandium (Sc) | 2805.30, 2846.90 | ~68% | Aluminium-scandium master alloys; SOFC electrolytes | Rio Tinto Sorel-Tracy; NioCorp Elk Creek (permits) |
| Yttrium (Y) | 2846.10 | ~89% | Yttrium-stabilised zirconia catalysts; phosphors | Lynas; Iluka; Ucore Alaska (pilot) |
Two observations worth sitting with. The first is that terbium and dysprosium, the two elements most critical to high-temperature magnet performance, are each roughly 99% Chinese-refined. Even if you secure an ex-China mine-source concentrate, the separation and metallisation step almost certainly touches China unless you specifically contract a Lynas Kalgoorlie or Solvay La Rochelle toll. The second is that scandium, at 68%, is the only one of the seven where a genuinely non-Chinese primary supply chain is operable today at commercial tonnage, courtesy of Rio Tinto’s Sorel-Tracy titanium slag by-product recovery in Quebec.
How to actually file a MOFCOM export licence application that gets approved
If you are still sourcing from China and you cannot afford to wait out a 90-day licence queue, the single highest-impact move is to tighten your end-user certificate package before your Chinese supplier files. I have watched applications sail through in 32 days and I have watched nearly identical applications get parked for 110 days. The difference is almost always documentation quality.
Your end-user certificate needs four elements at minimum. A clear corporate identity with full legal name, address, and authorised signatory. A statement of final application that names the downstream product at HS-8 level, the physical site where the material will be consumed, and the approximate annual tonnage. A non-diversion undertaking that commits to not re-exporting, not transferring to a third party, and not using for any military end-use or military end-user. And a consent clause permitting MOFCOM or the General Administration of Customs to verify, which in practice means permitting a post-shipment audit either on-site or through documentary request within 12 months.
The single most common rejection reason is vague downstream application wording. “For use in industrial manufacturing” will be rejected. “For use at our Greenville, South Carolina facility in the production of yttrium-stabilised zirconia automotive catalyst substrates, HS 3815.12, annual consumption approximately 18 tonnes yttrium oxide equivalent” is the level of specificity MOFCOM wants. Expect the Chinese licence broker to ask for supporting documents including your state business licence, utility bills for the consuming facility, and in some cases a customer reference.
A second common rejection vector is any connection, however remote, to entities on the US Bureau of Industry and Security entity list or the Chinese Ministry of Industry and Information Technology’s own end-user control list. Cross-check both. If your downstream customer is a defence prime, expect denial. If your downstream customer supplies a defence prime, expect enhanced scrutiny and a longer queue.
The landed-cost arithmetic on a dysprosium-iron master alloy in March 2026
Let me walk you through one we costed last week. Client wanted 12 tonnes of dysprosium-iron master alloy at 80% Dy, for use in high-coercivity neodymium iron boron magnets produced at their Ohio site.
Chinese route. FOB Ningbo-Zhoushan in early March 2026 was $412 per kilogram of contained Dy, plus the iron carrier. All-in per tonne of alloy, roughly $348,000. Ocean freight to LA/Long Beach under class 4.1 hazmat rules on a dedicated isotainer, $4,200 per tonne. US MFN duty, 5.5% on HS 8105.20. Section 301 List 3 at 25%. Reciprocal tariff stack adjustments from November 2025 added another 10% layer. Landed cost per tonne, $488,000. Licence-queue-adjusted lead time, 68 business days from PO to vessel departure.
Ex-China route via Lynas Kalgoorlie toll separation with metallisation in Estonia at Silmet. FOB Fremantle at $468,000 per tonne of alloy. Ocean freight Fremantle to Seattle direct, $3,800 per tonne. MFN duty 5.5%. No Section 301. No reciprocal tariff. Landed cost per tonne, $498,000. Lead time, 34 business days. The pure cost gap was a $10,000 per tonne premium for the ex-China route, or roughly 2%, and you bought back 34 days of inventory working capital and removed the licence-denial tail risk.
For most aerospace and defence-adjacent buyers, 2% to eliminate licence tail risk is the easiest trade they will make in 2026. For a commodity magnet buyer selling into white goods, the math is different and the Chinese route still wins on pure cost.
Four moves to make this month
First, inventory-audit your Announcement No. 18 exposure. Pull every purchase order placed in the last 18 months and flag anything whose CAS or HS code touches samarium, gadolinium, terbium, dysprosium, lutetium, scandium or yttrium. Calculate annual consumption in metric tonnes of contained rare earth equivalent. Most importers I speak with are surprised by the total.
Second, open a second-source relationship with Lynas, MP Materials, Solvay or Iluka on at least one of your top three exposures. None of these will quote you without an NDA and a qualified demand profile, so start the conversation now. Lead times to first commercial shipment from a new ex-China supplier run 6 to 14 months in this category.
Third, review every end-user certificate your Chinese supplier has on file. If it was drafted in 2024 or early 2025 against the older rules, it likely needs updating. A re-certification now is cheaper than a licence denial in August.
Fourth, budget for a 2026 price floor that sits 15 to 25% above your 2024 baseline on all seven controlled elements. The licence-queue premium, the end-user audit overhead and the currency hedging costs are not going away.
Announcement No. 18 is a year old and it is holding. Plan your sourcing programme on the assumption that it holds another year, and you will be right more often than the importers who are still waiting for a Busan-style reprieve that never quite arrives.