The War Behind the Magnet

Myanmar's poisoned hills supply much of the dysprosium and terbium inside modern weapons and clean-energy machines. Losing those hills would not break China's grip. This piece explains why.

On 18 October 2024 the Kachin Independence Army took Pangwa, a town on Myanmar's border with China, and with it the last of a belt of hills that most readers have never heard of and that a good part of the world's advanced industry cannot run without. Within days the trucks stopped. By the end of the month, Radio Free Asia reported, China had shut six border gates, and towns on the Kachin side ran short of fuel and food. In the second week of November the rebels answered by sealing their own crossings at Mai Ja Yang, Lai Zar and Pang War, and for a few weeks the only cargo that mattered on that border moved in neither direction. In December there was a quiet meeting in Kunming. By 2025 the Kachin Independence Army was taxing exports at 20 percent, a fixed price of 35,000 yuan a tonne had been agreed, and the trucks were moving again, in one direction, to the same buyer they had always gone to.

The hills changed hands. The buyer did not.

It is a timely week for it. The United States Treasury Secretary met China's Vice Premier in New York on 20 September with rare earths on the agenda going in, per Reuters, and the two presidents are due to meet in Washington on 24 September 2026. Nothing on that agenda touches the hills.

Where the heavy rare earths come from. The belt at Pangwa and Chipwi, the Kachin Independence Organisation's seats, the buyer's side of the border, and the next frontier in Laos; the green dots mark provinces, not individual mines, and the inset places Jiangxi. Boundaries: Natural Earth, public domain. Map: NewsplanetAI.

What the magnet needs

Rare earths are not rare. There are seventeen of them and they are scattered through ordinary rock almost everywhere. What is rare is the ability to separate them from each other, and from the radioactive thorium and uranium that keep them company, at scale and at a price anyone will pay.

The magnets that matter are made of neodymium, iron and boron. Neodymium and its sibling praseodymium do the pulling. Two heavier elements, dysprosium and terbium, do something less visible: they keep the magnet working when it gets hot. A motor in an electric car gets hot. A generator in a wind turbine gets hot. So does, in the International Crisis Group's phrase, advanced weapons, including munitions and missiles. Take the two heavy elements out and the magnet is still a magnet, right up until the moment it is not.

Where rare earths come from is not the story. Where they go is. The International Energy Agency, in its rare-earths report revised in May 2026, puts China at 60 percent of the world's mined magnet rare earths in 2024, 91 percent of the refining, and 94 percent of the finished magnets. The chokehold is not the deposit. It is the plant that turns the deposit into something a factory can use, and almost all of those plants are in China.

Now the distinction the rest of the piece rests on. The light rare earths, neodymium among them, come from hard-rock mines: Bayan Obo in Inner Mongolia, Mountain Pass in California, Mount Weld in Australia. The heavy pair come from somewhere else. They sit, loosely held, in clay hills of a kind found across southern China and in a handful of other places, and they are got out not by digging but by soaking: drill the hill, pour in ammonium sulphate, collect what leaches out below.

Myanmar is a small rare-earth producer by the total count. The United States Geological Survey ranks it fourth in the world for 2024, at 7.1 percent of mined output, behind China, the United States and Australia. In the heavy pair it is something else. The IEA states that Myanmar accounted for over 40 percent of the world's mined dysprosium and terbium in 2024, in a sector it calls largely informal and underreported, and that Myanmar and China together hold around 83 percent of the world's mined heavy magnet supply. Global Witness reported in August 2022 that the biggest of China's licensed heavy-rare-earth miners drew 70 percent of its raw material from Myanmar, a figure it took from a Chinese customs official's research, and that all the heavy rare earths mined in Myanmar were exported to China. The trade peaked at 41,700 tonnes in 2023, worth 1.4 billion dollars, a figure Global Witness published in May 2024 and the UN human rights office repeated in August 2026. In the first nine months of 2025, according to Chinese customs data compiled by the Institute for Strategy and Policy Myanmar, Myanmar supplied 53 percent of China's rare-earth imports by volume, worth 624 million dollars against 724 million in the same months of 2024.

So: minor in total, over 40 percent of the two elements that keep the magnet working when hot, and every tonne of it to one buyer.

The war the hills sit in

Myanmar has been at war with itself since independence in 1948. The Kachin Independence Organisation was founded in 1961 and its army has fought the central government, with pauses, for most of the years since. The coup of 1 February 2021 restarted the war nationally, and in 2025 and 2026 the junta has been clawing ground back with conscription, an air war, and Chinese backing, as Al Jazeera reported on 30 August 2026. This month the resistance began striking the junta's airbases with drones, per Asia Times on 16 and 17 September 2026.

The toll depends on who is counting and what. The UN humanitarian office plans for displacement rising from 3.6 million to 4 million people; the World Food Programme puts more than 12 million people in acute hunger in 2026; the Assistance Association for Political Prisoners had documented 8,414 killings by the junta and allied groups as of 21 September 2026, a narrow count of a particular kind of death. Broader conflict-fatality series run far higher.

The rare-earth belt is a corner of this. Pangwa and Chipwi, hard against the Chinese border, were held until late 2024 by a militia that had broken from the Kachin Independence Organisation decades earlier and later became a Border Guard Force under the junta's flag; it began mining there in 2014, according to a Shanan Foundation study published in February 2025. The Kachin Independence Army ran its own, smaller operations around Mai Ja Yang from 2011. Both sides of the Kachin war have mined. Both have sold to the same customer.

The mine changed hands. The buyer did not.

The chronology is short and it is the argument.

On 29 September 2024 the Kachin Independence Army and allied resistance forces took Chipwi, per Burma News International. On 15 October they took the Border Guard Force headquarters at Pangwa, per Radio Free Asia. On 18 October they held the town. The belt that had supplied China's heavy-rare-earth refiners for a decade was now in the hands of the junta's enemy.

China's answer was the border. In late October it shut six gates, RFA reported on 14 November 2024, and the Kachin towns that live on Chinese goods began to run short. The rebels sealed their own crossings in reply. Then, in December, representatives of the Kachin Independence Army met Chinese interlocutors in Kunming, according to the Wilson Center's New Security Beat on 4 August 2025. Out of that came the arrangement described above: a 20 percent levy on exported concentrate, a fixed price of 35,000 yuan a tonne, about 4,830 dollars by that account's own conversion. Shipments, which had collapsed sharply after the capture, resumed through 2025, and by late October 2025 the trade had fully reopened, according to the International Crisis Group, which notes that Beijing has since pressed the Kachin Independence Organisation far less to stop fighting than it has pressed other border groups. The IEA's own summary of the episode is dry: local governance issues led to border closures at the end of 2024, followed by gradual reopening in 2025. The customs data above, 53 percent of China's imports in the first nine months of 2025, is what reopening looks like in a table.

Read that as an experiment, because it nearly is one. An armed group captured a mining district from its rival. The buyer squeezed the new owner. The new owner squeezed back with the only weapon it had, which was to withhold ore from the only customer it had. A price was set. The ore moved again, to the same refineries, under a new landlord. The political authority over the hills changed completely. The industrial destination did not change at all. New Security Beat estimates the arrangement could bring the Kachin Independence Army over 200 million dollars a year, an estimate modelled from public tonnage and the agreed price, not a disclosed budget.

The lesson is not that China is pro-junta. China backs the junta and it deals with whoever holds the ground it needs, and in northern Myanmar that is several people at once. In Shan State the United Wa State Army, a different armed group with its own long arrangement with Beijing, had leased at least 26 rare-earth sites to Chinese firms as of February 2025, according to Shan Human Rights Foundation satellite imagery reported by the Jamestown Foundation in December 2025, and the UN's August 2026 report counts Shan's mining sites rising from 12 before the coup to 85 after it. Whoever wins the hills sells to China. There is nowhere else to sell.

The Kachin side is not a pit that ore comes out of. The Shanan Foundation's researchers, who conducted 23 interviews inside Kachin Independence Organisation territory between November 2024 and January 2025, record that villagers near Mai Ja Yang petitioned the organisation against the mining in December 2022, marched in January 2023, and in March 2023 destroyed Chinese mining facilities; that Kachin civil society groups condemned the organisation for pressuring communities in April 2023; and that on 15 April 2023 the organisation's chairman told 1,500 villagers the mining had been meant to fund development and buy weapons for the war, and halted it. Eighteen months later the same organisation took Pangwa. One of its own officials told the researchers that investors pay taxes to the organisation and pocket money and cars to its officials, and that departments had been instructed not to raise issues about rare-earth mining. A worker put it more simply: the Chinese knew the Kachin army was on their side. The rebels chose. They chose the way governments choose, with the money in view.

Who pays

The method is the cost. A study in the Lancet Regional Health Southeast Asia, published 25 May 2026, puts it at an estimated 75 cubic metres of acidic wastewater and one tonne of radioactive residue for each tonne of rare earth produced, and states that no environmental impact assessment has ever been conducted for any rare-earth operation in Myanmar. The reagent is ammonium sulphate, and China's exports of it to Myanmar rose from 93,000 tonnes in 2015 to 1.5 million tonnes in 2023, per the same study. Nearly 400 sites now operate in Kachin State alone, the Lancet authors count, a 194 percent increase since the coup. The UN human rights office, using satellite imagery over a different baseline, counts 302 new sites in Kachin since 2021, a 149 percent rise. The two counts use different baselines.

What the soaking does to people is reported, not yet measured. The UN report of 25 August 2026 states that local documentation efforts recorded sharp surges in miscarriages, severe bleeding in pregnancy and childbirth, and pregnancy complications attributed to heavy-metal and chemical exposure. The Lancet authors are explicit that no epidemiological study exists from Myanmar's mining zones and that the health picture is inferred from comparable Chinese regions. That inference gained a data point this week: a Chinese cohort study of 4,891 women who gave birth in Wuhan between 2012 and 2014, reported by the South China Morning Post on 20 September 2026, found prenatal exposure to rare-earth elements raised the risk of low birthweight, for some elements more than threefold. The country that moved the mining across its border has the science on what it does.

The hills also fall on people. A collapse at a Pang War mine on 19 June 2024 killed 15 people, RFA reported, and a slide on 4 June had killed more than 20 by RFA's later count, after a first-day count of five dead and 20 missing; other outlets count the same slides differently, and the counts should not be added. A former site supervisor told Grist, in a piece published 23 July 2026, that when a Myanmar worker dies the family is paid between 80,000 and 100,000 yuan, roughly 12,000 to 14,000 dollars, and that when a Chinese national dies the payment can exceed one or two million yuan. The Shanan researchers found workers, many of them displaced people, earning 2,000 to 4,000 yuan in wages and landowners paid 4,000 yuan an acre, and found that the rule requiring investors to refill their leaching pits went unenforced; a villager described cattle falling into the holes and dying there.

China moved the mine, not the monopoly

Here is the turn.

The heavy-rare-earth clays that Myanmar mines are the same kind of clays that made Jiangxi province the centre of China's own heavy-rare-earth industry for decades. In 2016, according to the International Crisis Group, Jiangxi shut down much of that mining, along with smaller deposits in Yunnan and other southern provinces, and extraction shifted across the border into Myanmar, where the armed group in control imposed few safeguards. The Shanan study dates the crackdown on illegal domestic mining to 2015 and says plainly that Chinese businessmen outsourced the mining to Myanmar and imported the product for refining. Crisis Group's own words for the arrangement are that China is outsourcing environmental damage outside its borders.

Notice what moved and what stayed. The soaking moved. The wastewater and the radioactive residue and the miscarriages moved. The refineries did not move. The separation plants that turn Kachin's leachate into dysprosium oxide, and the magnet works that turn the oxide into a motor, stayed in China, and the 91 percent and the 94 percent above are theirs. The war in Kachin cannot touch them. The Kachin Independence Army can capture every hill in the belt and the only thing it will have captured is the right to sell to them.

Nor did China wait for Myanmar to fail. In Laos, the Stimson Center's Mekong monitoring found at least 26 rare-earth mines operating, the first opened in 2022, shortly after the Myanmar coup and two years before the Kachin army took Pangwa; its wider census, published in November 2025, identified 517 unregulated mines along Laotian rivers, by Stimson's own count. The deposits in Xieng Khouang and Houaphanh are the same ion-adsorption clays, and the IEA lists Laos among the countries where they occur. The Wilson Center's July 2026 piece on Laos is titled for spiralling rare-earth pollution. The same harms, a different border. The sacrifice zone is not migrating when one site fails. It is spreading in parallel.

The one thing China cannot do is take the heavy pair from its own great mine. Bayan Obo is the largest rare-earth deposit on earth and it is a light-rare-earth deposit: medium and heavy elements make up about 0.2 percent of its ore by weight, about 3 percent of its rare-earth content, according to a 2025 paper in the journal Minerals. It cannot substitute for the clays at volume. That is why the clays are mined at all, and why they are mined where the rules are weakest.

Armies can capture the mine. China still owns the destination.

What would change it

Not the end of the war. A peace in Kachin, of any shape, leaves the refineries where they are and the clays where they are, and whoever governs the hills afterward will face the same single buyer.

Not buying mines. That is where the money has gone this year. In the August 2026 package announced in Washington, the largest single award, 1.4 billion dollars, went to a battery-materials company, 400 million to a scandium project in Australia, 150 million to a company making magnets without rare earths, and nothing to rare-earth separation, per the White House fact sheet of 7 August 2026 and the reporting around it; the money for separation came earlier and separately, and in smaller sums. A mine outside China produces concentrate. Concentrate has to go somewhere, and 91 percent of somewhere is the country the mine was meant to escape.

Only refining, outside China, at scale, changes the destination. The IEA's May 2026 report measures how far that is. Production from current capacity in regions outside China and Myanmar meets 45 percent of those regions' own mining demand, 20 percent of their refining demand, and 14 percent of their magnet demand. Counting every planned expansion, the IEA projects those shares at about 50 percent, 25 percent, and well below 20 percent in 2035. That is the honest close and it is also our falsifier. If the regions outside China and Myanmar are refining more than a quarter of their own magnet rare-earth demand by 2035, the grip described here has loosened and this argument was wrong. The IEA publishes the series. We will publish either way.

What that quarter costs is the part the spending announcements do not say. The separation is the same chemistry whoever does it, and the wastewater and the residue in the Lancet's figures have to sit somewhere with a name on it. The largest heavy-rare-earth separation plant outside China is Lynas's in Kuantan, Malaysia, which shipped its first separated dysprosium and terbium under contract in late 2025 and whose four-year supply deal with the Pentagon became a Malaysian domestic political fight in July 2026. The owner of Mountain Pass, the American mine, went bankrupt in 2015 when prices collapsed after China eased its export restrictions, and the mine's heavy-rare-earth separation circuit is still being commissioned. Independence means paying to separate, paying to hold the waste, paying to keep a plant alive through the years China can price it under water, and paying to make magnets before anyone has promised to buy them. Every one of those costs is a cost China's customers have spent a decade displacing onto hills where no one asks.

And there is no clean counterparty. The junta stands accused by the UN of unrelenting violations. The Kachin Independence Organisation runs the poisoning mines and, by its own official's account, takes the pocket money. We saw the same shape in Sudan this month, where the buyer's willingness to deal with whoever holds the ground is the mechanism, not a failure of it.

The war is where the cost lives, not where the power lives. The world can end the war, or refuse the ore, or buy the hills. It will still be sending the trucks to the same gate.

Where this comes from

Our own work, cited above

• Why Doesn't Sudan's War End?, parts one and two, 6 and 13 September 2026: the buyer who deals with whoever holds the ground.

Primary documents and data

• International Energy Agency, Rare Earth Elements: Pathways to Secure and Diversified Supply Chains, revised May 2026: China's 60, 91 and 94 percent shares for 2024; Myanmar over 40 percent of mined dysprosium and terbium; Myanmar and China around 83 percent of mined heavy magnet supply; the 2035 projections for regions outside China and Myanmar.
• United States Geological Survey, Mineral Commodity Summaries, February 2026, and its Burma country page: Myanmar fourth-ranked, 7.1 percent of world mine production, 2024.
• Office of the UN High Commissioner for Human Rights, report on Myanmar covering 1 June 2025 to 31 May 2026, released 25 August 2026: 302 new Kachin sites and 149 percent; Shan 12 to 85; the locally documented reproductive-health findings; the 1.4 billion dollar 2023 peak.
• Hein Thu and Minn Khant Kyaw Linn, Lancet Regional Health Southeast Asia, 25 May 2026: 75 cubic metres and one tonne per tonne; no environmental impact assessment; ammonium sulphate 93,000 to 1.5 million tonnes; nearly 400 sites and 194 percent; the absence of epidemiology in Myanmar.
• General Administration of Customs of China, compiled by the Institute for Strategy and Policy Myanmar, 27 November 2025: 53 percent of China's rare-earth imports January to September 2025; 624 million against 724 million dollars.

Investigations and research

• Global Witness, Myanmar's Poisoned Mountains, 9 August 2022, and Fuelling the Future, Poisoning the Present, 23 May 2024: China Southern's quota and 70 percent sourcing; all heavy rare earths exported to China; 41,700 tonnes and 1.4 billion dollars in 2023.
• International Crisis Group, The Critical Minerals Linking Myanmar's Civil War to the World, visual explainer, rare earths section, published 10 September 2026: the Jiangxi shutdown of 2016 and the shift across the border; outsourcing environmental damage; magnets in munitions and missiles; the border's full reopening by late October 2025.
• Shanan Foundation, Governance of Rare Earth Mining by the Kachin Independence Organization/Army, Research Report 1, February 2025, published by the Heinrich Böll Foundation: 23 interviews, November 2024 to January 2025; the 2022 to 2023 protests and the April 2023 halt; wages, land rents, unenforced refilling; the official's and the worker's accounts.
• Stimson Center, Unregulated Mining Along Rivers in Mainland Southeast Asia, November 2025: at least 26 rare-earth mines in Laos since 2022; 517 unregulated mines identified. Stimson's published sub-totals of 145 heap-leach and 346 alluvial sites sum to 491; we report its stated total as stated.
• Minerals (journal), 2025: medium and heavy rare earths at about 0.2 percent by weight of Bayan Obo ore, about 3 percent of its rare-earth content.
• Wilson Center, New Security Beat, Amara Thiha, 4 August 2025: the Kunming meeting, the 20 percent levy, the 35,000 yuan price, the 200 million dollar estimate; and its July 2026 piece on rare-earth pollution in Laos.
• Jamestown Foundation, Terrorism Monitor, 4 December 2025, crediting Shan Human Rights Foundation satellite imagery: the United Wa State Army's leased sites as of February 2025.

Reporting

• Burma News International, 2 October 2024: Chipwi taken 29 September. Radio Free Asia, 15 October 2024: the Border Guard Force headquarters at Pangwa; 14 November 2024: six gates shut, the Kachin counter-seal; 5 and 21 June 2024: the landslides of 4 and 19 June.
• Grist, Emily Fishbein and Jauman Naw, 23 July 2026: the death-compensation account of a former site supervisor.
• South China Morning Post, 20 September 2026: the Wuhan cohort study.
• Reuters via the Straits Times, 18 September 2026, and Nikkei Asia, 21 September 2026: the Bessent and He Lifeng meeting of 20 September; the 24 September summit.
• Al Jazeera, 30 August 2026: the junta's 2025 to 2026 offensive. Asia Times, 16 and 17 September 2026: the drone war.
• White House fact sheet, 7 August 2026, with the Straits Times, Bloomberg and CNBC, 7 and 8 August 2026: the United States minerals package.
• UN OCHA, Humanitarian Needs and Response Plan 2026; World Food Programme, 2026; Assistance Association for Political Prisoners, daily briefing of 21 September 2026: the displacement, hunger and documented-killing figures.

Interested parties, reported as such

• The Kachin Independence Organisation: its official and its chairman, via the Shanan Foundation and Frontier Myanmar.
• The former site supervisor "Kaung", via Grist: one witness.
• The Institute for Strategy and Policy Myanmar: a specialist tracker compiling Chinese customs data.
• The United States government: the August 2026 package as announced.

Image

• Locator map drawn by NewsplanetAI from Natural Earth public-domain boundaries; place positions approximate; Laos markers are provinces named by the Stimson Center, not mine sites.

A note on method

• This newsletter is written with AI assistance, including Anthropic's Claude. Every figure above was checked against the document that carries it where that document could be reached; where a figure rests on one witness, a modeled estimate, or local documentation rather than measurement, the text says so.