How China came to own rare earths

Short versionDeng said it. He almost certainly didn't mean it the way it gets quoted.

You can't read anything about rare earths without running into this line. Deng Xiaoping, 1992. It's carved in calligraphy on stones in an industrial park in Baotou, it's in every article on the subject, and now it's in this one too.

It usually gets used as proof of a thirty-year plan: Deng saw OPEC, saw the oil embargoes, and set out to build the same weapon out of the lanthanides. It's a great story. The geographer Julie Klinger went back to Chinese newspapers and policy documents from the early 1990s to check it, and it doesn't hold up.¹ Deng popularized the line on his 1992 Southern Tour (not 1987, as a lot of sources say). The Middle East crisis on everybody's mind in January 1992 was not the 1973 embargo. It was the Gulf War, and the lesson drawn in China was closer to "oil made the Middle East a target" than "oil made the Middle East powerful." Klinger calls the master-plan reading ahistorical. The quote didn't appear in the New York Times until 2009, after China had started squeezing exports, at which point it made a very tidy explanation of something that was already happening.

The goal in raising this isn't to be contrarian about a quotation. It's that the master-plan version makes the story sound inevitable, and it wasn't. Everything below is a series of fairly ordinary industrial-policy decisions, most of them visible at the time, and most of them made easier by everybody else's.

Short versionIn about fifteen years China went from a fifth of world production to nearly all of it. Nobody banned anybody. They just couldn't compete.

From the 1960s to the mid-1980s, the rare earth industry was, to a first approximation, one mine in the California desert. Mountain Pass, near the Nevada line, was the world's main source from 1965.⁴ In 1985, the United States produced about a third of the world's rare earths and China about a fifth.²

By 1995 China had 60 percent. By 2005, about 98. None of the reasons were mysterious: a very large deposit at Bayan Obo in Inner Mongolia that comes out of the ground alongside iron ore, cheap labor, state support, and environmental standards that were, to put it gently, not California's.⁵

Mountain Pass, meanwhile, had California's standards and a wastewater problem. Leaks from its wastewater pipeline brought legal action in 1997, and with remediation costs on one side and Chinese prices on the other, US production fell from about a third of the world in 1985 to 6 percent in 2000 and stopped in 2002.²⁶ Most of the rest of the world's production went the same way.

It's worth being fair about this. While it's tempting to call it dumping, most of it was simply a lower-cost producer doing what lower-cost producers do. The West wasn't outmaneuvered so much as happy to buy cheap oxide from someone else, and happy to have the mess be somewhere else.

Short versionOnce China had the mines, it made raw rare earths expensive to export and finished goods cheap to export. The factories followed the price gap.

China capped rare earth exports with quotas from around 2000, and later added export taxes that reached 15 to 25 percent.⁷ What matters is WHAT was taxed. Oxide and metal going out: taxed and capped. A finished magnet, or a motor with a magnet in it: not.

Put yourself in the shoes of a magnet maker in Japan or Germany. Your competitor in Ningbo buys neodymium at the Chinese domestic price. You buy the same neodymium at the export price, plus the tax, plus whatever the quota does to availability that quarter. How big was the gap? At the end of 2011, dysprosium oxide cost about $450/kg inside China and about $1,400/kg for export, roughly three to one. Neodymium oxide was about two to one.⁸ The US Trade Representative later said the duties and quotas had forced American manufacturers to pay as much as 300 percent more than their Chinese competitors for the same materials.⁹ You can fight that for a while. Or you can build your next plant in China, where the input is cheap.

Lots of people built the next plant in China.

The magnet-industry version of this story has a name: Magnequench. General Motors developed one of the two original neodymium magnet processes in the early 1980s, at the same time as Sumitomo, and set up Magnequench in Anderson, Indiana, in 1986 to make it. In 1995 GM sold it for $70 million to a group that included two Chinese state-linked companies, San Huan New Material and China National Non-Ferrous Metals. Magnequench broke ground on a powder plant in Tianjin in 1998, and over the following years the US production went there.¹⁰

In practiceNobody moves a plant because of one bad quarter. They move it because the difference between their input price and their competitor's input price is structural, and it has a policy behind it rather than a market. Once a gap like that is written into a tax schedule, the question stops being WHETHER to build capacity inside China and becomes how much, and how fast. That's the part of the story that tends to get told as a failure of nerve in the West. It was mostly arithmetic.

Short versionIn 2010 China cut its export quota by about 40 percent and cracked down on illegal mining, and prices went vertical. Dysprosium went up 26 times in 31 months.

By the late 2000s, China was consuming most of what it produced. In 2010 it cut its export quota by about 40 percent, to roughly 30,000 metric tons.¹⁵ At the same time, it started shutting unlicensed mines and enforcing environmental rules it had not previously bothered much with, which took a lot of undocumented supply out of the market at once.

Then, in September 2010, a Chinese fishing trawler collided with Japanese coast guard vessels near the Senkaku Islands, Japan detained the captain, and rare earth shipments to Japan stopped for about two months. (Whether Beijing ordered that halt or just let it happen has been argued about ever since.¹)

The Senkaku halt is usually told as THE cause of the price spike. It wasn't, quite. Adamas Intelligence's monthly data shows dysprosium oxide was already rising, and prices barely moved on the Japan news for about three months, then took off.¹¹ What drove it was everything at once: falling quotas, rising taxes, the crackdown, the Japan episode, and then every buyer on earth deciding at the same moment to hold more inventory. From January 2009 to August 2011, dysprosium oxide went from $91 per kilogram to $2,377.

Here's the part that's easy to miss. Neodymium went up too, by about an order of magnitude, but it came back down fast. Dysprosium came down slower and stayed higher.¹⁶ That's the difference between a light rare earth that's in every major deposit and a heavy one that isn't. You can mine your way out of a neodymium shortage. There's no easy way to mine your way out of a dysprosium shortage, and that fact comes back in section 6.

Short versionExpensive rare earths did two useful things. They made magnet makers use a lot less dysprosium, and they paid for mines outside China. Then prices fell, and most of the mines went bankrupt.

Less dysprosium per magnet

The quiet winner of 2011 was a process, not a mine. Grain boundary diffusion puts the dysprosium (or terbium) where it does the most good, at the grain boundaries where demagnetization starts, instead of alloying it through the whole magnet. It gets you the same coercivity with roughly half the heavy rare earth.¹⁶ It had been around since about 2005, out of Japan. Before 2011 it was, in Stan Trout's words, "met with a shrug," and fairly, because it's an elaborate extra process, and nobody runs an elaborate extra process to save a few grams of something that costs $30 a kilogram. At $3,000 a kilogram, the arithmetic flips.

In practiceWhether you call what happened next commercialization or adoption is six of one, half a dozen of the other. The price spike made the process worth doing, and then the industry did a great deal of unglamorous work to get the cost of the process down to where it made sense at NORMAL prices too. That second part is the one people forget, and it's why the technology stuck after dysprosium came back to earth. Today most automotive magnets use it. A technology that had been sitting there for six years, and it took a 26-fold price shock to make anyone care.

New mines, and what happened to them

Lynas had been developing Mount Weld in Western Australia, and built its separation plant in Malaysia, which was producing by 2014–15.¹² Lynas came close to going under in the price collapse that followed, and survived.

Great Western Minerals Group, a Canadian company that owned Less Common Metals, the UK alloy maker, set out to reopen Steenkampskraal in South Africa (an old Anglo American thorium mine with very high-grade monazite) and feed its own alloy plant.¹⁷ That was exactly the right idea: mine to metal, outside China. It didn't survive the price collapse, and the mine never produced under its ownership.

Molycorp bought Mountain Pass in 2008 and spent heavily rebuilding it, separation plant and all. It restarted into the price collapse, filed for bankruptcy in 2015, and Mountain Pass was idled again that October.¹²¹⁸

What killed them wasn't geology. It was the price chart. In 2012 the United States, the European Union and Japan took China to the WTO over the quotas and export taxes. China lost in 2014, and by May 2015 the quotas and the taxes were gone.¹⁹ That sounds like a win, and it was, for buyers. With the export tax gone, the export price fell toward the domestic price, which is to say toward the cost of the lowest-cost producer on earth. By the end of 2015, neodymium oxide was around $40/kg and dysprosium around $220.¹² Every project outside China had been financed on 2011 prices.

So what did the West have to show for winning at the WTO? A ruling, and essentially no new separation capacity. That absence is the finding.

MP Materials. Out of Molycorp's bankruptcy, a group led by JHL Capital and QVT, with China's Shenghe Resources as a minority holder, bought Mountain Pass in 2017 and brought it back into production by 2018.¹⁸²⁰ For its first several years MP mined in California and shipped its concentrate to China for separation.¹⁸ Separation came back to Mountain Pass later, and magnet making in Texas later still.

Short versionChina now mines about 70 percent of the world's rare earths but refines about 90 percent, and makes most of the magnets. That's what makes export controls work. It has used them several times, and it has picked its targets carefully.

Mining share is the number everyone quotes, and it's the least important one. China's share of mining has actually fallen, to about 69 percent in 2025.³ What didn't fall is the middle of the chain: China produces around 90 percent of the world's refined supply.²¹ An oxide separated in China, reduced to metal in China and made into a magnet in China has passed through three chokepoints, and the mine comes before all of them.

Beyond the 2010 Japan episode, the ledger looks like this:

• December 2023. China banned the export of rare earth extraction, separation and refining TECHNOLOGY: know-how, not material. That one has never been suspended, and it may be the most durable of the lot, because it's aimed squarely at everybody else's attempt to build a separation industry quickly.²²
• April 2025. During the US–China tariff escalation, China put seven medium and heavy rare earths under export license (samarium, gadolinium, terbium, dysprosium, lutetium, scandium and yttrium), along with their alloys, compounds, metals and oxides.³
• October–November 2025. Five more elements (europium, holmium, erbium, thulium and ytterbium) were added in October. In November, after President Trump and President Xi met in Busan, China suspended the October measures for one year. The April controls stayed.³ The suspension runs to 10 November 2026.
• December 2025 onward. Japan again. After Prime Minister Takaichi said Japan could help defend Taiwan, China sharply reduced magnet shipments to Japan and has come close to suspending some elements outright, dysprosium among them.²³ Fifteen years after the fishing trawler: same country, same element.

Notice what's NOT on that list: neodymium and praseodymium. China hasn't restricted either one. The controls are on the heavy and specialty elements, the ones the rest of the world can't easily replace, which also happen to be the ones China can restrict without doing much damage to its own magnet exporters. That's section 4 again. You can mine your way out of a neodymium problem. Dysprosium, not so much.

The two-price system is back

Look at the right-hand end of the price chart and something seems backward: the USGS average dysprosium price FELL in 2025, the year of the controls, to about $239/kg.³ That isn't the controls failing. It's the controls working. When exports need a license, Chinese supply has fewer places to go and the domestic price sags, while buyers outside China pay whatever they have to for what gets out. At the end of August 2026, Fastmarkets had dysprosium oxide at $270–330/kg FOB China and $1,250–1,675/kg in Rotterdam, with European dysprosium trading at about 4.9 times the Chinese domestic price.¹³

At the end of 2011 the same gap was about three to one. It's the 2000s export-tax gap, rebuilt with a licensing desk instead of a tariff schedule, and it has the same effect: the cheapest place in the world to make a dysprosium-bearing magnet is inside China.

In practiceWhen somebody quotes you "the price of dysprosium" today, the first question is WHICH one. The Chinese domestic price, the FOB China export price and the delivered price in Europe are three different numbers, and in 2026 the biggest of them is several times the smallest. A cost model built on the number in the headline is usually built on the one you can't buy at.

Short versionChina won rare earths by being cheap, then by moving the value-add home, then by being patient. Everyone else kept losing the same way: financing projects at peak prices and watching them die at trough prices.

If there's one lesson in the price chart, it's that a price spike is a terrible foundation for an industry. Every project outside China from 2011 was financed on the top of the curve and killed by the bottom of it, and China didn't need to do anything clever to make that happen. It just had to stop propping the price up.

Which is why the most interesting recent development in this story isn't an export control. It's the Defense Department's July 2025 agreement with MP Materials, which set a minimum price of $110/kg for NdPr oxide.²⁴ Whatever you think of the government underwriting a mining company (and there are respectable arguments on both sides), a price floor attacks the exact failure mode that killed Molycorp and Great Western. In early September 2026 the Chinese NdPr oxide benchmark was about $97/kg, below that floor.²⁵ That's the floor doing its job, and it's also a fair measure of what it costs to do this without one.

What would I watch? Three things. Whether the October 2025 controls come back on 10 November. Whether anyone outside China gets heavy rare earth separation running at scale, which is the only thing that closes the dysprosium gap for real. And whether the price floors last longer than the price spikes that made everyone want them.