Dysprosium at 16x China, Neodymium Below the U.S. Floor: The Rare-Earth Spread Just Split in Two

The gap between rare-earth prices inside and outside China was flagged on November 10, 2026 as a supply-chain warning ahead of the Trump-Xi summit. It is one gap in name only. Neodymium-praseodymium oxide and dysprosium oxide are moving in opposite directions on opposite sides of the border, and the two halves now carry opposite investment implications.

The inversion is the part that breaks the consensus framing. Normally ex-China material trades at a premium because export controls strand supply outside China. Through the third quarter of 2026 the opposite happened on light rare earths: Western separation capacity arrived faster than ex-China magnet demand, and the ex-China assessment fell below the floor the Pentagon had set. On September averages, China's own domestic neodymium-praseodymium was 2.1% more expensive than the ex-China price.

The November 9 extension covers six October 2025 notices — production equipment, superhard materials, battery and anode materials, licences for rare-earth mining and separation technology, the addition of five medium and heavy rare earths to the licensing list, and the 0.1% foreign-content rule extending Chinese jurisdiction to foreign goods containing Chinese-origin material. It does not cover the April 2025 regime, which requires a licence for every shipment of samarium, gadolinium, terbium, dysprosium, lutetium, scandium and yttrium, plus their metals, alloys, oxides and permanent magnets. That regime was never suspended and still gates every shipment.

Prices for the seven April 2025 controlled elements are set by how many export licences Beijing actually grants, so the spread between the domestic assessment and the ex-China price is a running scoreboard on policy intensity rather than on geology. It widened from 6.8 times at the end of June to 16.3 times at the end of September, and it moved because the licence queue did, not because the ore moved.

Volume-based forecasts and price-based assessments disagree, and the price is the one that clears. S&P Global's August 2026 neodymium-praseodymium market report counted announced 2026 ex-China oxide production of 12 kt against an ex-China demand gap of roughly 32 kt, which reads like a shortage. What buyers actually paid shows ex-China oxide below the Pentagon's floor in August and 2% below China's own domestic price in September. Announced capacity has not turned into delivered tons, and the near-term glut and the 2030 separation deficit are the same story at different horizons.

MP Materials is the cleanest listed expression of the ex-China thesis, and its own numbers show why the split matters. In the third quarter of 2026 it produced 1,018 tonnes of neodymium-praseodymium and sold 1,020 tonnes, up 41% and 94% year over year, at a realized price of about $100 a kilogram. That is roughly 7% below China's domestic assessment of $107.50 a kilogram, and below the $110 a kilogram floor the Department of Defense set in the July 2025 partnership. MP realizes about 7% less than China's own domestic price.

Because the April 2025 regime was never suspended, the beneficiaries of the status quo are the Chinese producers and magnet makers sitting inside it, and the evidence is in their reported margins rather than in the headlines. China Northern Rare Earth lifted gross margin from 10.4% to 16.9% year over year in the third quarter and grew net income 161%, while JL MAG lifted first-half gross margin from 15.4% to 21.7% and grew net income 52%. Chinese quota holders have already converted the spread into margin.

Not every Chinese producer is capturing the premium, and that difference is the second-order signal. Shenghe Resources grew revenue 15% in the third quarter but saw gross margin slip to 10.1% from 11.4% and net income fall 42%; Ningbo Yunsheng grew revenue 38% while gross margin fell to 16.5% from 21.1% and net income fell 34%. Volume without pricing is what happens outside the quota. Only the quota holders are positioned for the re-opening to fail.

Meanwhile the equity market has started pricing the opposite. China Northern Rare Earth's market value fell from CNY173.7 billion at the end of June to CNY132.2 billion by November 10, 2026, and Ningbo Yunsheng's from CNY16.0 billion to CNY11.1 billion, tracking a domestic neodymium-praseodymium price that peaked at $123 a kilogram in February 2026 and has since given back about 13%. The de-rating has priced a normalization the spread does not show.

Below the mining layer, the buyers of magnets are where the 16.3x premium actually lands. High-grade traction and wind magnets need dysprosium and terbium to hold coercivity at temperature, and the ex-China price for those elements now embeds a licence premium more than an order of magnitude above China's domestic level. Downstream users are paying a scarcity rent set by Beijing's licence queue.

Near term, the spread moves before the headlines do. The dysprosium premium widened from 6.8 times at the end of June to 16.3 times at the end of September, and it is the first place a licensing concession would show: a narrowing back toward the June level would be the earliest hard evidence that Beijing is granting heavy-rare-earth licences, and it would arrive days before any communiqué. The spread, not the summit schedule, dates the deal.

Over one to three years the same data points somewhere else. Forecasts cited in current market reviews put non-Chinese neodymium-praseodymium output rising 4.4 times between 2024 and 2030 while the world still runs a 36% deficit in separation and refining by 2030, and the International Energy Agency projects demand for magnet rare earths outside China rising 50% by 2035 after doubling since 2015. Near-term glut and long-term deficit are the same story at different horizons: today's oversupply is the separation build-out arriving before the magnet demand it is meant to serve. MP's stock capitalises a 2030 separation deficit, not a 2026 price.

The conclusion is narrow, and it is the one the spread supports. Beijing has not re-opened heavy rare earths, and the two-month extension bought time without changing the constraint, so Chinese quota holders keep their premium while Western producers keep working under a floor that now sits above the market they sell into. What the data does not support is the popular version of the story, that rare earths are uniformly tightening: ex-China neodymium-praseodymium has fallen to $95 a kilogram and now prints below China's own domestic price, which means the Western supply build-out is being absorbed without scarcity rent. Only the heavy-rare-earth spread carries investable information.