Rare Earth Market Update: June 2026 — NdPr Pricing, MOFCOM Licensing, and Western Supply

The headline this month is a correction, not a collapse. NdPr blended oxide on the Shanghai Metals Market domestic benchmark fell from USD 126.16/kg on 1 April 2026 to approximately USD 99.61/kg in early May — about a 21% month-over-month drop, the sharpest monthly decline since the Q1 rally began. The move reflects profit-taking and inventory adjustment after a roughly 105% run from the January open, not a structural reversal: even at the early-May trough, NdPr was still up about 88% year-to-date from its ~USD 53/kg January start. Prices firmed again through late May, with regional Northeast Asia neodymium-oxide quotations reaching approximately USD 144/kg. International reference points held up better than the domestic benchmark wobbled — FOB China neodymium oxide averaged around USD 183/kg (range ~USD 167–199) in late April, and CIF Rotterdam held near USD 255/kg, a sign that Western consumers were not panic-selling into the dip. Treat the SMM domestic number as the most volatile of these gauges and the FOB/CIF figures as the better proxy for what a Western buyer actually pays.

• •NdPr oxide (SMM domestic): USD 126.16/kg (1 Apr) → ~USD 99.61/kg (early May), about -21% MoM
• •NdPr oxide YTD: ~USD 53/kg January open → ~USD 100/kg trough = roughly +88%, recovering late May
• •Neodymium oxide (Northeast Asia regional): ~USD 144/kg in May
• •Neodymium oxide FOB China: ~USD 183/kg (range ~USD 167–199), late April
• •Neodymium oxide CIF Rotterdam: ~USD 255/kg, broadly unchanged

The most important thing in this month's data is what did not happen. While NdPr corrected, the heavy rare earths kept climbing. Dysprosium metal traded near USD 930/kg in late May (roughly +105% YTD) and terbium metal near USD 4,030/kg (roughly +103% YTD); on the oxide side, Dy was around USD 191/kg ex-works China and ~USD 317/kg FOB earlier in the quarter, with Tb oxide quoted in the USD 730–760/kg range. This divergence is not noise — it is structural. NdPr is a light rare earth that flows out of China without an export license for standard N-grade magnets, so its price is driven mostly by Chinese mining quotas, smelter inventory, and speculative positioning, all of which can swing on profit-taking. Dy and Tb sit behind the MOFCOM Announcement No. 18 licensing layer, so their availability outside China is gated by a slow, discretionary approval process that profit-taking cannot relieve. The second-order effect for procurement is concrete: the cost stack of high-temperature grades that require dysprosium or terbium — the H, SH, UH, and EH series used in EV traction and humanoid actuators — kept rising even as the NdPr headline fell. A buyer who watches only the NdPr number would have concluded magnet prices were softening in May. A buyer running an SH or UH grade saw the opposite, because the heavy rare earth content in that grade is where the cost pressure actually lives.

Two MOFCOM regimes still run in parallel, and the distinction governs every HREE-bearing order. The April 2025 controls (Announcement No. 18) cover samarium, gadolinium, terbium, dysprosium, lutetium, scandium, and yttrium plus their oxides, metals, alloys, compounds, and any finished product carrying them above threshold — including the relevant NdFeB grades. These are fully in force. The October 9, 2025 expanded controls — which would have swept additional downstream products and dual-use technologies into scope — remain suspended through 10 November 2026, now roughly five months out. The practical texture of licensing eased slightly but did not normalize. General licenses issued since December 2025 to JL MAG, San Huan, and Yunsheng let those approved exporters ship specified products to pre-cleared customers under multi-shipment, year-long permits rather than filing per shipment, which has improved throughput for established civilian buyers already on an approved customer list. New or unlisted end-uses still face the slow, case-by-case path. Geopolitical friction added a wrinkle: following January 2026 dual-use controls aimed at Japan, Chinese customs data showed permanent-magnet exports to Japan up just 2.5% in April after a 17.3% contraction in March — a reminder that licensing can be applied unevenly by destination. On the Western side, the EU Critical Raw Materials Act implementing act for magnet labeling remains due 24 November 2026, with the 0.2 kg disclosure threshold applying from 24 May 2027.

China continues to account for roughly 85% of global neodymium mine production and about 90% of NdFeB magnet manufacturing, and the supply posture this year has been defined by restraint. MIIT has not announced the staged 2026 quota increases that some in the market had penciled in, and has instead leaned on enforcement — penalty rules covering production-quota breaches and unauthorized separation — to hold the existing envelope rather than expand it. The early-May NdPr correction is best read against that backdrop: it was a demand-side and inventory event (profit-taking, destocking) layered on top of a supply curve that Beijing is deliberately keeping flat, which is precisely why prices firmed again within weeks rather than entering a sustained decline. Standard N-grade NdFeB without Dy or Tb continues to export without licensing friction; the friction stays concentrated on the heavy-rare-earth-bearing H/SH/UH/EH grades and on SmCo, where every shipment needs a license and every license needs end-use documentation that survives MOFCOM review. The general-license mechanism is gradually widening the channel for pre-cleared civilian buyers, but it does not change the underlying quota math.

Last month the Western story was milestones; this month it is financial actuals. MP Materials reported Q1 2026 results in May with its first quarter of commercial magnet revenue: record NdPr production of 917 metric tons (up 63% year-over-year), record NdPr sales of 1,006 metric tons (up 117% year-over-year), and record Q1 REO production of 12,983 metric tons. The Magnetics segment booked USD 21.1 million in revenue and USD 9.6 million in adjusted EBITDA — small in absolute terms, but the number now exists, which is the point. The Materials segment added USD 72.2 million in revenue plus USD 42.3 million in production-price-adjustment income, and the company finished the quarter with USD 1.7 billion in cash and short-term investments. Crucially, MP broke ground on its 10X magnetics campus in Northlake, Texas during Q1, with more than USD 200 million in combined state and local incentives including a Texas Semiconductor Innovation Fund grant; commissioning is targeted for 2028 toward roughly 10,000 tpa. Lynas, the only scaled producer of separated heavy rare earths outside China, more than doubled quarterly revenue to about A$265 million with REO output up roughly 69% to ~3,230 tonnes, and now runs three separated heavy products (dysprosium, terbium, and newly added samarium) from Mt Weld feedstock at its Malaysia facility. In Europe, Neo Performance Materials continues running PPAP samples from its Narva, Estonia magnet plant (Phase 1A targeting 2,000 tpa, infrastructure to scale beyond 5,000 tpa) toward mass production later in 2026, while France's Carester Caremag recycling-and-refining project in Lacq — backed by about EUR 216 million in financing — remains on track for late-2026 startup with planned Dy/Tb and NdPr oxide output.

• •MP Materials Q1 2026: 917 t NdPr produced (+63% YoY), 1,006 t sold (+117% YoY); Magnetics revenue USD 21.1M, EBITDA USD 9.6M; USD 1.7B cash; 10X Northlake groundbreaking
• •Lynas: revenue ~A$265M (more than doubled), REO output ~3,230 t (+69%); Dy + Tb + samarium now produced in Malaysia
• •Neo Performance: Narva (Estonia) PPAP samples shipping; 2,000 tpa Phase 1A, mass production later in 2026
• •Carester Caremag (Lacq, France): ~EUR 216M financed, late-2026 startup for Dy/Tb and NdPr oxides

The demand vectors are intact, but two of them moved their dates this month. Tesla pushed the Fremont Optimus line start to late July or August 2026 (converting Model S/X assembly capacity engineered for up to one million units annually), with several thousand units targeted on its own lines by year-end and external sales now framed for 2027 — so 2026 calendar-year output remains in the hundreds-to-low-thousands range, with the per-unit NdFeB content (roughly 0.4–1.5 kg across actuators) locked regardless of exact volume. The quieter data point is Unitree, which shipped more than 5,500 humanoids in 2025 — more than the combined US output of Tesla, Figure, and Agility — and targets 20,000 units in 2026. That matters for the supply picture in a non-obvious way: the current humanoid volume leader is a Chinese company with domestic magnet access, which means a meaningful slice of near-term humanoid magnet demand is being satisfied inside China's quota system rather than pulling on the exportable supply Western buyers compete for. Figure AI continues toward warehouse and manufacturing pilots, and Agility's Digit remains deployed in commercial logistics. Underneath the robotics narrative, EV traction and offshore wind direct-drive (still specifying roughly 600–1,200 kg of NdFeB per MW) keep the structural deficit case for NdPr intact. The net: three independent demand surfaces remain on the same supply curve, and the May price correction did nothing to change that.

Three actions for June and July. First, do not over-read the NdPr dip. The early-May correction took the headline number down about 21%, but it firmed within weeks and your high-temperature grades never followed it down — Dy and Tb kept climbing. If you run SH/UH/EH grades, your input cost did not fall in May; budget accordingly and resist the temptation to treat a softer NdPr print as broad relief. Second, if you do have standard N-grade volume to place, the early-May window was the kind of correction worth acting on opportunistically — pair any spot buying with indexed contract coverage so you are not fully exposed if the recovery extends. Third, the MOFCOM math has not changed and the calendar is shorter. The October 2025 suspension expires 10 November 2026, now about five months out, and the procurement window for Q4 deliveries closes in August/September. If you need H/SH/UH/EH grades for Q4, book by August at the latest and add a 45-day buffer above normal lead time. If you are an established civilian buyer, ask your supplier whether they ship under a general license — the multi-shipment permits granted to JL MAG, San Huan, and Yunsheng can materially shorten lead time versus per-shipment filings. And keep qualifying a non-Chinese source for at least one application: with MP Materials now booking magnet revenue and Lynas, Neo, and Carester all scaling, the allocation queue is forming this year.

Five items will move the market before the next update. (1) Whether the NdPr recovery holds or the early-May correction resumes — watch the SMM domestic benchmark against the firmer FOB/CIF references for the real signal. (2) The MIIT mid-year quota review: if 2026 mining and smelting quotas are raised in the June–July window, NdPr could soften again; if held flat or tightened, the deficit case extends through year-end. (3) Any early telegraphing on the 10 November 2026 suspension — extensions are typically signaled 60–90 days ahead, so the August–September window is when the market will start pricing expectations, and a hint either way will move HREE-bearing grade pricing first. (4) Whether the general-license channel widens beyond the initial three exporters, which would be the clearest sign that civilian throughput is genuinely normalizing. (5) The next read on Dy/Tb — if heavy rare earths keep diverging upward while NdPr stalls, the cost gap between standard and high-temperature grades widens further, and that, not the NdPr headline, is what will show up in Q4 high-temp budgets. Watch those five; the rest stays noise.