Insights › Rare-earth magnet supply chain atlas › JL MAG
China
JL MAG
Sintered NdFeB, with grain boundary diffusion as a core disclosed technology: the company states grain-boundary-diffusion products were roughly 90% of its product mix in FY2025. Documented end markets are new energy vehicles, variable-frequency air conditioning, wind power, robotics and industrial servo motors, consumer electronics, low-altitude aircraft and rail transit. Five plants across Ganzhou, Baotou and Ningbo, plus a Hong Kong research center.
At a glance
- Founded
- 2008 — Established 19 August 2008 as Jiangxi JL MAG Rare-Earth Co., Ltd. in Ganzhou, Jiangxi.
- Headquarters
- Ganzhou, Jiangxi, China
- Ownership
- Dual listed: Shenzhen ChiNext (SZSE: 300748) since September 2018 and Hong Kong main board (HKEX: 6680) since January 2022.
- Stage
- Mature. The world's largest high-performance NdFeB maker, about 40,000 t/yr at end-2025, with a 20,000 t/yr Baotou expansion under construction.
- Listing
- Shenzhen: 300748 · HKEX: 6680
- Revenue
- CNY 7.718 billion, up 14.11% year on year
Reported, FY2025, calendar year ended 31 December 2025.
Net profit attributable to shareholders CNY 705.6 million, up 142.44%. No named source published a US dollar equivalent with a conversion date, so none is given.
Source
Capacity
Each date below links to the document the figure came from. A capacity figure with no source is not on this page.
| What | Figure | Status | As of |
|---|---|---|---|
| Sintered NdFeB, all sites — company states the build was completed at the end of 2025 | 40,000 t/yr | Operating | Dec 2025 |
| Sintered NdFeB capacity in service during FY2025, at utilization above 90% | 38,000 t/yr | Operating | Dec 2025 |
| Actual FY2025 output — magnet blanks | 34,400 t, up 17.31% year on year | Operating | Dec 2025 |
| Actual FY2025 output — finished magnets | 25,300 t, up 21.25% year on year | Operating | Dec 2025 |
| Baotou Phase III high-performance magnet project | 20,000 t/yr | Under construction | Sep 2026 |
| Company target for total sintered NdFeB capacity by the end of 2027 | 60,000 t/yr | Announced | Sep 2026 |
| Rare earth recycling — approved capacity for recovery of single rare-earth compounds | 5,000 t/yr | Operating | Dec 2025 |
| Rare earth raw material recovered from recycling — actual output, not capacity | 3,681 t | Operating | Dec 2025 |
Narrative
JL MAG Rare-Earth was established in Ganzhou, Jiangxi in August 2008 and produces sintered NdFeB permanent magnets. It listed on the Shenzhen Stock Exchange's ChiNext board in September 2018 and added a Hong Kong main-board listing in January 2022. The company reports five plants across three sites, states it completed 40,000 metric tons per year of magnet capacity at the end of 2025, and produced 34,400 metric tons of blanks and 25,300 metric tons of finished magnets that year against FY2025 revenue of CNY 7.718 billion. A 20,000 metric ton per year Baotou Phase III project is under construction, with the company targeting 60,000 metric tons per year by the end of 2027.
Where public sources disagree
- Capacity and output are frequently conflated in secondary coverage. The 34,400 t figure is FY2025 magnet blank output, not capacity; at least one English wire has presented it under a capacity heading.
- The 40,000 t/yr milestone is dated differently across sources. One outlet described it as existing capacity in January 2025, while the company's own annual report gives 38,000 t in service during 2025 and a company statement in April 2026 says the 40,000 t build was completed at the end of 2025.
- An English machine translation of one Chinese report rendered 2万吨 (20,000 metric tons) as two million metric tons. That figure is a translation artifact, not a claim anyone has made.
Fact checks from the digest
Claims made about this company that the weekly digest has put against this atlas, newest first. The verdict and the working are here in full; the issue each came from is for subscribers.
-
The deal is a statement, not a document
Two things to weigh, and they pull in opposite directions.
The outcome is reported, not documented, and this digest will not print it as settled. A one-year delay renegotiated annually is what was said. Searching for the text of it returns, over and over, coverage of the Busan summit of October 2025 — which produced a structurally identical result, a one-year pause, and whose figures (tariffs from 57 to 47 percent, a suspension expiring 10 November 2026) are still circulating attached to this week's meeting. At least one account of the 2026 outcome carries 2025's numbers. Until a MOFCOM notice or a White House readout exists, the honest description is that two governments have described an understanding that neither has published.
The compression that does not stand. The Times grouped lanthanum, cerium, neodymium and praseodymium as “widely used in oil refining, glass polishing and low-end magnets.” Split the list: lanthanum and cerium are the refining and polishing elements. Neodymium and praseodymium are the basis of NdFeB, the strongest commercial permanent magnet there is. The arithmetic settles it — a sintered NdFeB magnet is roughly 30 percent rare earth by weight, nearly all of it neodymium and praseodymium, with dysprosium or terbium added in low single digits and only where the grade must survive heat. By mass, the elements China has left unrestricted are almost the entire magnet. Read as written, the sentence says Beijing declined to restrict the unimportant stuff. What it says is that Beijing declined to restrict the feedstock for every traction motor, wind turbine and hard drive on the planet.
-
About a third of what, exactly?
Two claims, and they land differently.
The capacity comparison is narrower than stated. From this atlas, on JL MAG's own filings: 40,000 t/yr of sintered NdFeB across all sites with the build completed at the end of 2025, 38,000 t/yr in service through FY2025 at above 90 percent utilization, and FY2025 output of 34,400 t of blanks against 25,300 t of finished magnets. Ten thousand metric tons is a third of one of those and two fifths of the other. The atlas profile for JL MAG already carries a note saying capacity and output are conflated in secondary coverage; this is that note happening in real time.
But the MP figure itself stands, and we should say so. Bloomberg puts Northlake at bringing MP "to about 10,000 tons a year" — company total, not the plant alone. That is right. Issue 004 was built on correcting a version of this exact number elsewhere, so a clean bill here is worth printing: the error is in the denominator, not the numerator.
On the dates: the article says both "by 2028" and "by next year" for the same 60,000 t target. The filing says end of 2027. Use the filing.
-
A capacity figure with a decimal point, and two years of tables that do not match
The capacity figure is the smaller half of this. The output figures are worse, and they are worse in the same way the Bloomberg comparison earlier in this issue is worse.
The FY2025 annual report of 27 April 2026 states production of 9,189.02 t, down 11.26 percent. That percentage only works against 10,355 t — which is the FY2024 report's blank output, not the 6,183.88 t line that the same FY2024 report labels finished NdFeB. The two years' tables are not on the same basis, and the FY2025 row carries no qualifier saying so.
The 98.81 percent utilization rate in the same document is that identical blank tonnage divided into a capacity the company words as high-performance rare-earth permanent magnet material: 9,300 × 0.9881 = 9,189.0. Two consequences follow that no source explains. Blank output fell in a year reported at 98.81 percent utilization. And 2024 blank output of 10,355 t exceeded the 9,300 t capacity it is being measured against.
The sales line reconciles to nothing at all: 8,971.30 t, down 10.63 percent, implies a prior-year base of about 10,038 t — matching neither the 6,238.98 t in the FY2024 table nor the 8,116.47 t in that report's own text.
None of this is a translation problem and none of it is a reporter in a hurry. It is in the audited filings.
-
JL MAG's first half, from the filing rather than from the press release
Sintered NdFeB capacity on our atlas, largest first: JL MAG 40,000 t/yr, Yantai Zhenghai 30,000, Ningbo Yunsheng 26,000, Zhongke Sanhuan 25,000. The claim holds without needing the word independent to carry it.
One figure that looks larger and is not comparable: China Northern Rare Earth is on the chart at roughly 100,000 t/yr — of NdFeB alloy, with only about 5,000 t/yr of sintered magnets through a joint venture. Alloy and magnets are different products at different points in the chain, and conflating them is how China's magnet capacity gets overstated.
Sources
- FY2025 annual report summary (Chinese) — JL MAG Rare-Earth / Shenzhen Stock Exchange disclosure (2026-03-30)
- Company statement on the 40,000 t/yr build (Chinese) — Sina Finance (2026-04-09)
- Baotou Phase III progress and the 60,000 t/yr target (Chinese) — Tencent News (2026-09-08)
- Baotou Phase III announcement — Yicai Global (2025-01-21)
- Company establishment date and registered address (Chinese) — Tonghuashun 10jqka (accessed 2026-09-18)
How this page is sourced
Everything here comes from public sources — company filings, annual reports, exchange announcements, government releases and trade press. Every figure is linked to the document it came from and carries the date that document was published, because a capacity number without a date is not a fact, it is a rumor with a decimal point.
What is deliberately absent: material nonpublic information. I have spent more than two decades in this industry and a good deal of what I know about these companies I learned under NDA. Those obligations do not expire and they do not bend for a website. Nothing I have been told in confidence appears here in any form, and nothing on this page — or missing from it — should be read as a signal about what I know privately. I hold NDAs absolutely. It is the reason clients tell me things.
Judgment is the other thing left out on purpose. This page reports what has been published and who published it. Whether a target is credible, whether a ramp will hold, whether a capacity figure survives contact with the plant behind it — that is the work I do for clients, and it does not belong in a free reference page.
Where public sources disagree, I say so rather than picking a winner. Where a figure has not been published, the field says so rather than carrying an estimate. An empty field is information too.
Public filings tell you what a company has said. Whether it holds up is a different question, and the one I get hired to answer.
Get in touch