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

NdPr blended oxide priced at USD 126.16/kg on 1 April 2026 per Shanghai Metals Market and traded into the USD 135–140/kg range by late April, with Northeast Asian regional benchmarks running materially higher (Neodymium oxide hit USD 147.61/kg in March 2026 in Northeast Asia). For perspective, NdPr opened 2026 at roughly USD 53/kg, crossed USD 100/kg in early March, jumped 37% in the first week of April, and has continued grinding higher since. That puts year-to-date appreciation at approximately 155% as of this writing. Dysprosium oxide held at roughly USD 191/kg ex-works China in early March 2026 (SMM); international landed prices for Dy carried 3–4x premiums over the Shanghai ex-works figure across 2025-2026 as the MOFCOM licensing layer compressed availability outside China. Terbium oxide remains the tightest of the three benchmarks, with Shanghai quotations gated behind paid feeds and approval-only access — directional moves track Dy but at higher absolute price points and steeper licensing friction. The structural read: NdPr is in its second consecutive year of supply deficit, Chinese mining and smelting quotas have not been raised enough to clear the deficit, and Western and Australian production additions are still measured in single-digit percent of global supply.

• •NdPr oxide (Shanghai blended): ~USD 126/kg on 1 April, ~USD 135–140/kg by 30 April
• •NdPr oxide YTD 2026: opened ~USD 53/kg → ~USD 135/kg = +155%
• •Neodymium oxide (Northeast Asia regional): USD 147.61/kg in March 2026
• •Dysprosium oxide ex-works China: ~USD 191/kg early March 2026; international landed running 3–4x
• •Terbium oxide: behind paid feeds, directionally tracking Dy with higher absolute price and tighter licensing

Two MOFCOM regimes are operating in parallel right now, and the distinction matters for every order placed in the next two quarters. The April 2025 controls (MOFCOM Announcement No. 18) cover seven medium and heavy rare earths — samarium, gadolinium, terbium, dysprosium, lutetium, scandium, yttrium — plus their oxides, metals, alloys, compounds, mixtures, and any finished product (including NdFeB grades) carrying them above defined thresholds. These rules are fully in force. The October 9, 2025 expanded controls — which would have swept additional downstream products and dual-use technologies into scope — were suspended via Announcements No. 70 and 72 from 7 November 2025 to 10 November 2026. That suspension expires in roughly six months. As of April 2026, license processing under the April 2025 regime continued to run 'slow, unpredictable, tightly scrutinized' — Arnold Magnetic Technologies' phrase, and one borne out by the working pattern: best case ~45 working days at MOFCOM (10 weeks application-to-shipment total), with many applications sitting 60–120 days without explicit decision. On the EU side, the Council adopted its Critical Raw Materials Act position on 4 March 2026; the magnet-label implementing act is due 24 November 2026, with the 0.2 kg disclosure threshold applying from 24 May 2027. US Section 232 and IRA-aligned procurement continue to favor non-Chinese supply for defense, semiconductor, and federal R&D programs.

China produced roughly 270,000 metric tons of rare-earth-oxide-equivalent in 2025 (USGS, February 2026 report), and continues to account for ~85% of global neodymium mine production and ~90% of global NdFeB magnet manufacturing. Quota policy this year has been notable for what it has not done: MIIT has not announced the staged increases that markets had penciled in for early 2026. Instead, MIIT published draft administrative-penalty rules covering production-quota breaches in mining and smelting and unauthorized separation activity — in effect, tightening enforcement on the existing quota envelope rather than expanding it. The combined effect is straightforward: the supply curve out of China is being held flatter than 2024 consensus assumed, exactly as EV, wind, and humanoid demand pulls harder on it. Domestic Chinese magnet manufacturers are still receiving feedstock, and standard N-grade NdFeB without Dy or Tb continues to flow without licensing friction for export. The friction is concentrated on the H/SH/UH/EH grades and on any application using SmCo, where every shipment needs an export license and every license needs end-use documentation that survives MOFCOM review.

March-April 2026 produced more concrete Western magnet milestones than any month in the past five years. MP Materials announced its first rare earth magnets produced on commercial-scale machinery in late March 2026, sets up Q1 2026 earnings for 7 May 2026, and is moving from its operating Independence facility in Fort Worth into the $1.25 billion 'Ten X' campus in Northlake, Texas — 1,500 jobs, ~$200 million in state and local incentives, ground-breaking imminent, commissioning targeted for 2028, planned total NdFeB capacity ~10,000 metric tons per year. USA Rare Earth commissioned Phase 1a of its Stillwater, Oklahoma sintered NdFeB line in April 2026 and began Q2 2026 customer fulfillment, targeting 600 tpa by Q4 2026 and 1,200 tpa by Q1 2027. The company also closed a $2.8 billion Brazil mining acquisition and took an equity stake in France's Carester. Neo Performance Materials commissioned its HREE solvent-extraction small-scale line at Silmet in Narva, Estonia in April 2026, complementing its Estonian magnet plant which is running PPAP samples through H1 2026 toward mass production later in 2026 (initial target 2,000 tpa). Lynas added first samarium production from Mt Weld feedstock in April 2026, on top of its existing Dy production (since 2025) and Tb (since June 2025), heading toward 1,500–2,000 tonnes Dy-oxide-equivalent annually with potential to reach 3,000+ by FY28. France's Carester Caremag — the recycling-and-refining facility in Lacq — remains scheduled for late 2026 operations, with 2,000 tpa of magnet recycling and 600 tpa Dy/Tb oxide output planned (~15% of global Dy/Tb).

• •MP Materials: first commercial-scale NdFeB late March 2026; Q1 earnings 7 May; Ten X Northlake breaking ground, 10,000 tpa target by 2028
• •USA Rare Earth: Stillwater Phase 1a commissioned April 2026, 600 tpa by Q4 2026 → 1,200 tpa Q1 2027; Carester equity, Brazil acquisition
• •Neo Performance: HREE solvent-extraction line commissioned at Silmet (Estonia) April 2026; magnet plant PPAP H1 2026, 2,000 tpa initial
• •Lynas: first Sm from Mt Weld April 2026; existing Dy/Tb production, scaling toward 1,500–2,000 tpa Dy-oxide-equivalent
• •Carester Caremag (Lacq, France): late-2026 start; 2,000 tpa magnet recycling + 600 tpa Dy/Tb oxides + 800 tpa NdPr oxides

The demand-side change since the last update is that humanoid robots have stopped being a forward narrative and started showing up as specific unit volumes. Tesla's Optimus V3 is on track for mid-2026 reveal and Q3 2026 production at Fremont, where Model S/X lines are being repurposed; the first-generation Fremont line is engineered for ~1 million units annually, and Giga Texas is breaking ground on a follow-on facility targeting 10 million units long-term. 2026 calendar-year output is realistically in the hundreds-to-low-thousands range, but the BOM impact is locked: each humanoid carries roughly 0.4–1.5 kg of NdFeB across actuators. Figure AI opened BotQ with a 12,000-unit annual capacity scaling to 100,000, and Figure 02 has already produced 30,000+ BMW X3s at Spartanburg. Agility Robotics' Digit moved 100,000+ totes at GXO Flowery Branch under a multi-year RaaS agreement. BMW Plant Leipzig confirmed deployment of Hexagon's AEON humanoid for summer 2026 pilot. EV demand has not paused — first-quarter 2026 European and Chinese registrations were ahead of model — and offshore wind direct-drive build-outs continue to specify 600–1,200 kg of NdFeB per MW. The cumulative effect: the deficit case for NdPr is no longer a function of one demand vector. Three independent demand surfaces (EV traction, humanoid actuators, wind) are pulling on the same supply curve, and Western capacity additions cannot close the gap before 2028.

Three concrete actions are worth taking this month. First, check the price-lock language on every open PO. NdPr is +155% YTD and continued upward through April; any quote written in Q1 2026 with a 90-day lock is roughly 30–40% under the current market, and suppliers will be seeking renegotiation triggers, force-majeure invocations, or shorter lock windows on renewals. Treat 30-day price-lock as the new default; insist on indexation language for anything longer. Second, plan Q4 2026 deliveries now. The MOFCOM October 9, 2025 suspension expires 10 November 2026 — six months out. If the rules snap back into force, license processing for any HREE-bearing grade gets harder, not easier, exactly as Q4 inventory builds peak. Buyers should book Q4 deliveries by August/September 2026 at the latest and add a 45-day buffer above normal lead time on any H/SH/UH/EH grade. Third, qualify a non-Chinese source for at least one application now. With MP Materials, USA Rare Earth, Neo Estonia, and Lynas all running commercial product in 2026, the qualification queue is forming this year, not in 2027. Engaging early — design input, sample submission, customer-list registration — buys allocation priority and better pricing once capacity matures. The premium for US-domestic NdFeB is still 30–60% over comparable Chinese product, so the move is not 'replace Chinese supply'; it is 'add a 10–20% strategic-volume Western tranche' that gives you continuity if China tightens.

Five items will move the market between now and the next update. (1) MP Materials Q1 2026 earnings on 7 May 2026 — the first detailed read on NdFeB ramp economics, Apple and GM contract volumes, and capex pacing on Northlake. (2) Any signal from MOFCOM on the disposition of the November 10, 2026 suspension — extensions are typically telegraphed 60–90 days in advance, so August-September 2026 is when the market will price in expectations. (3) MIIT's mid-year quota review — if the 2026 mining and smelting quotas are raised in the June-July window, NdPr could correct downward; if held flat or tightened, the deficit case extends through year-end. (4) Tesla's Optimus V3 reveal date and any explicit FY26 unit guidance, which will feed directly into 2027-2028 NdFeB demand modelling. (5) Neo Performance's PPAP-to-mass-production transition in Estonia — a successful crossover proves Western magnet capacity can serve automotive PPAP standards, which is the actual qualification gate that has held back European OEM uptake of non-Chinese supply for two decades. Watch these five, and the rest of the noise stays noise.