JL MAG Rare-Earth (300748) Zen Horizon Framework Deep Dive: A Global Rare-Earth Permanent Magnet Leader, With a V-Shaped Earnings Rebound Meeting a Rich Valuation and a Robotics Narrative

Hunting ten-year five-baggers among great growth stocks — pressing the upside question: "Can it get much bigger?"

• QHow high is its market ceiling? Is it expanding an existing pie, or creating an entirely new market?4/10

  A

  Conclusion: JL MAG is "making an existing pie bigger," not "creating an entirely new market": electrification is a real and long runway with enough size, but JL MAG's role is to penetrate existing motor demand and take an upstream position as major customers expand capacity. It is not a creator of end markets.

  The pie is large enough, but it is existing demand scaling up. High-performance NdFeB is a "picks and shovels" business for electrification. Under the ChinaIRN / Frost & Sullivan framework, global high-performance demand rose from about 74,000 tons in 2020 to about 226,000 tons, with a CAGR of about 25%, and the global NdFeB market is in the tens of billions of U.S. dollars. The report uses "about 130,000 tons/year of total demand" as its anchor. The definition of "high-performance" varies by framework, and narrower external estimates such as 58,000 tons also coexist; this answer follows the report's framework. The drivers are motor penetration in NEVs, inverter air conditioners, wind power, and industrial / energy-saving motors. These are real incremental volumes, but the essence is existing demand scaling up, not a new market appearing from nowhere.

  JL MAG is "expanding capacity alongside major customers' motor demand." The report states explicitly that it is "good at expanding capacity alongside major customers' motor demand and making magnetic materials more refined and larger in scale, rather than defining the end market itself": its customers are Tesla, BYD, Midea, and Goldwind; capacity is 40,000 tons/year, with actual output of 38,000 tons, and a 2027 target of 60,000 tons. This is a follower-style upstream position in magnetic materials, not the creation of terminal demand.

  Looking at the three downstream segments one by one: NEVs, which account for about 56% of main business, are the largest base, but China's EV penetration rate had already exceeded 55% in 2025, and 2026Q1 sales had already declined year over year. This is a mature pie being expanded. Energy-saving motors are a real source of incremental demand, but their absolute scale is still being cultivated. The only area with a "new market" imagination is humanoid robots, yet current incremental demand is only at the hundred-ton level, about 104/229/458 tons in 2025/26/27. Even if annual humanoid sales reach 1 million units, the incremental contribution to total NdFeB demand would still be less than 3%. This is a long-term TAM after 2030, not the current ceiling.

  Dimension tilt: medium to weak. TAM is large enough and the runway is long enough, but JL MAG is a follower penetrating an existing pie rather than a market creator, and the robot "new market" has not yet materialized.

  Jun 5, 2026
• QCan its revenue at least double over the next five years? Will growth be driven mainly by volume, price, or new businesses?5/10

  A

  Conclusion: revenue can double over five years, but the quality is poor. It is a "volume-price mix" that relies on cyclical price leverage to fill the gap. Sustainable volume growth alone cannot support a doubling, new businesses are negligible, and the profit base has not doubled in parallel.

  Sell-side forecasts have already put the doubling inside three years: iFinD consensus estimates project 2026/27/28E revenue of RMB 10.342 billion (+34%) / RMB 13.170 billion (+27%) / RMB 15.602 billion (+18%). That means revenue nearly doubles from RMB 7.718 billion in 2025, according to the report's +14.1%, by 2028E. A five-year doubling is therefore the base case.

  Breaking down the drivers: sustainable volume growth exists, but is not enough. Capacity rises from 40,000 tons/year to a 2027 target of 60,000 tons, or +50%, with an EV base and energy-saving motor penetration. That can support only about +50% revenue; volume alone does not double the top line. The real factor filling the gap is price, and price is an unsustainable cyclical lever: praseodymium-neodymium oxide rebounded from the 2024 trough of RMB 360,000 to about RMB 700,000 on the reference date, near a three-year high, and became the main engine of this cycle. The report provides hard evidence: in 2024, magnetic-material sales volume rose +37.88%, but revenue increased only +1.1%, because prices were falling. Once price reverses, even large volume cannot hold up revenue. Consensus 2026E growth of +34% is far above the actual +16% in 2026Q1, implying a strong assumption that "prices keep standing at a high level," which is fragile by itself.

  New businesses, namely robots, are negligible: robots and industrial servo combined contribute only about 4% of revenue, with 2026Q1 at RMB 118 million and +82%, but mostly from traditional servo. Humanoid incremental demand is only at the hundred-ton level, about 104/229/458 tons in 2025/26/27, and cannot support a second curve.

  One framework needs to be made explicit: revenue growth has in fact stayed around ~15%, with FY25 +14% and Q1 +16%. What surged was profit, +142%, before falling back to +20% in Q1. The report notes that 2025 net profit attributable to the parent of RMB 706 million is roughly equal to RMB 703 million in 2022; the "normalized profit center has completed a three-year round trip without moving higher." The revenue line can double, but the profit center may not.

  Tilt: medium. The revenue-doubling threshold is achievable, and is even the consensus base case, but growth relies on cyclical price leverage, new businesses are absent, and the profit base does not double with revenue. It does not meet Baillie Gifford LTGG's standard for high-quality, self-driven compounding.

  Jun 5, 2026
• QAfter five years, what will take over as the next growth engine? Does this "second curve" exist today?3/10

  A

  Conclusion: there is still no "second curve" that can be counted in the financial statements today. The humanoid robot magnetic components most anticipated by the market are still at the "sample delivery / small-batch" stage. They are a real long-term option, not a growth engine already being realized.

  Scale: financial reports never disclose humanoid revenue separately. There is only the combined line item "robots and industrial servo": RMB 196 million for full-year 2024, RMB 133 million for 2025H1, and RMB 118 million for 2026Q1 (+82%). But this accounts for only about 4% of revenue, and the vast majority is still traditional industrial servo. Humanoids are even smaller and only in small-batch sample delivery. The company has also never acknowledged being Tesla's "exclusive supplier."

  Realization: each Optimus uses about 3.5kg of high-performance NdFeB, about 1.75 times an EV, but the absolute incremental magnetic-material demand is only about 104/229/458 tons in 2025/26/27. Even if annual humanoid sales reach 1 million units, the incremental contribution to about 130,000 tons/year of total demand would still be less than 3%. This is a long-term TAM after 2030. Downstream timing is also being pushed out: the Optimus V3 release has been delayed again, Musk has admitted the progress is disappointing, and external expectations generally put real volume ramp in 2027.

  Rare-earth elimination: the current Optimus still uses about 3.5kg of NdFeB per unit, and frameless torque motors cannot avoid high-performance NdFeB. Musk's 2023 statement about "no rare earths" pointed to next-generation / EV drive motors. The near-term assumption has not yet been falsified, which is somewhat mitigating for JL MAG, but over the long term it remains an overhang.

  The company does have real forward-looking investment: in 2025 it upgraded the humanoid magnetic components R&D department into a business unit, with the CEO personally coordinating it, and invested in dedicated production lines. Energy-saving motors are also seen by brokers as a high-elasticity penetration direction, but absolute scale is still being cultivated, and low-altitude economy receives very little attention in JL MAG's case. Overall, the second curve has not yet taken over, while the first curve, EV at about 56%, already saw total volume decline year over year in 2026Q1. That is exactly why this dimension is weak.

  Jun 5, 2026
• QWhat is its core competitive advantage? Will this moat widen or narrow over the next three to five years?4/10

  A

  Conclusion: the core advantage is the triple barrier of "grain boundary diffusion (GBD) + scale + tier-one customers." It is real and the widest among A-share peers. But as the report says, this is a "moat in peer competition," not a "moat against cyclicality," and over the next three to five years the direction is toward narrowing.

  The hard evidence for the moat is solid: GBD diffuses terbium and dysprosium along grain boundaries, reducing heavy rare-earth usage by 50–70%. GBD products account for more than 90%, the company has accumulated about 127 patents, and it has invention patents in China, the U.S., Europe, and Japan. 2025 R&D investment was RMB 506 million, or 6.55% of revenue. This technology + scale barrier gave JL MAG a FY2025 gross margin of 21.18%, firmly the highest among peers: Zhenghai 13.05%, Ningbo Yunsheng 17.81%, and Zhong Ke San Huan 11.20%. Capacity rises from 40,000 tons to a 2027 target of 60,000 tons, making it the largest pure NdFeB player among A shares. But the report is honest: when praseodymium-neodymium prices fell in 2024, its net profit was still cut in half and gross margin was compressed to 11.13%. The moat can fend off peers, but not the cycle.

  Why the judgment is "narrowing": ① the technology gap is being erased. Zhenghai's heavy rare-earth diffusion capacity has exceeded 11,000 tons/year, with coverage above 70%, and Zhong Ke San Huan's grain boundary diffusion has been widely used in mass production. GBD is moving from JL MAG's patent wall to an industry standard. ② Listed rare-earth permanent-magnet companies are in an expansion wave, with planned capacity far above real demand, about 400,000 tons vs about 200,000 tons of actual demand, and utilization below 70%. Once high-end products enter a price war, the premium margin is the first thing to be squeezed. ③ Value capture is weak: the top ten customers account for more than 70% of NEV revenue, EV business gross margin is only about 14%, and receivable turnover is 128 days and lengthening. Scale has not translated into pricing power. Export controls on heavy rare earths raise the resource value of GBD, and total-quota controls benefit leaders as a reverse support, but these are cyclical tailwinds rather than structural widening. By Baillie Gifford's standard of "sustained widening over ten years," this moat does not qualify.

  Jun 5, 2026
• QIf its core business is disrupted, does it have the DNA to reinvent itself? How does it treat mistakes and bad news?3/10

  A

  Conclusion: weak. JL MAG is an "optimizer / follower within the cycle," not the kind of "self-reinventor" LTGG prefers. But its restrained and candid disclosure is a real positive in this dimension.

  DNA for reinvention against disruption: weak. The report's framing is accurate: JL MAG is good at "expanding capacity alongside major customers' motor demand and making magnetic materials more refined and larger in scale, rather than defining the end market itself." Every leap in its capital-market history, from wind power to NEVs to robots, followed customer volume, rather than creating its own track. Facing a core-business disruption such as "rare-earth elimination," its response is reduction within the existing paradigm: GBD reduces expensive heavy rare-earth usage by about 50%–70%, with GBD products already accounting for about 90% and 2025 R&D investment of RMB 506 million / 6.55% of revenue. This is real adaptive technology, but it is still essentially "using less rare earth," not "using no rare earth." I verified that all six core technologies of the company are NdFeB optimizations, with no self-developed rare-earth-free permanent-magnet product. In other words, if mainstream motors truly shift toward ferrite / rare-earth-free routes, JL MAG has no ready second paradigm to switch into.

  But this hard-disruption assumption has not materialized so far: Musk's 2023-03 claim that "next-generation permanent-magnet motors will use no rare earths at all" is now more than three years old, and by 2026 Optimus still relies on rare-earth magnets. He has also complained that China's magnet controls slowed Optimus mass production (The Hill). Moreover, switching high-performance motors to ferrite magnets would require a 10–12 times increase in volume and fail lightweighting requirements (NetEase), so hard disruption is difficult in the short term. Organizationally, JL MAG has indeed moved closer to robots: in 2025 it focused on developing humanoid robot magnetic components and completed a magnetic components production line (Sina Finance), and according to the 2025 annual report it established an embodied-robot motor rotor business unit and made small-batch deliveries. But it is still making "magnetic components / rotors" as parts and following customer motor demand.

  Candor: strong. When the robot theme was at its hottest, the company never acknowledged being Tesla's "exclusive supplier," did not name Optimus, and clearly stated that "scaled mass production and profit contribution will gradually be reflected with downstream customers' mass-production progress," actively cooling overheated narratives. This restraint is rare among A-share theme stocks and represents real disclosure integrity. Overall: reinvention DNA is weak, candor is strong, and the combined result is weak.

  Jun 5, 2026
• QDoes management, especially the founder, have a long-term view and deep alignment with the company? Is it willing to sacrifice current profits for five to ten years out?4/10

  A

  Conclusion: management "alignment of interests" is real, but the part Baillie Gifford truly values, namely "sacrificing current profits for five to ten years out and reinvesting all cash," is not present. JL MAG looks more like a steady, shareholder-return-oriented founder company than an LTGG-style long-term reinvestment company. This dimension is medium.

  Positive alignment, and real. According to the report, the company was founded in 2008 by Chairman Cai Baogui together with Hu Zhibin and Li Xinnong, and the three jointly control it. The founders remain in place. Through the controlling shareholder Ruide Venture Capital, or Jiangxi Ruide, the three contributed 40%/30%/30% and collectively control about 28%, with clear and stable control. Together with the 2025 A-share employee stock ownership plan of 8.0158 million shares, including 7 directors, supervisors, and senior executives, and H-share restricted shares, company-level assessments require 2025–2027 net profit to grow 20%/45%/80% versus 2024, linking management with three-year performance. The skin-in-the-game test is passed.

  The fundamental tension with Baillie Gifford. LTGG wants founders willing to burn current profits and reinvest internally with full force. JL MAG is the opposite: a high-dividend return-oriented company. The report notes that in the 2024 trough year, when net profit attributable to the parent was only RMB 291 million, it still paid cash dividends of about RMB 271 million, about 93% of attributable net profit. In 2025, total dividends rose to RMB 550 million, +103% year over year, with a payout ratio of 77.98%. Capacity expansion from 40,000 to 60,000 tons is mainly funded by external financing: H-share proceeds of RMB 5.056 billion, convertible bonds, and a 2021 private placement of RMB 521 million under the report's framework. This is not internally generated cash and runs opposite to the LTGG logic of "reinvestment compounding." One more thing needs monitoring: in 2025H2, concert party Ganzhou Xinsheng reduced its stake by 1% and cashed out about RMB 500 million, and 5 directors, supervisors, and senior executives planned to reduce holdings by about 2.1061 million shares. The control circle trimmed holdings at a cyclical / thematic high.

  Trade-off: founder alignment is real, and multi-year equity incentives also align the team. But the capital-allocation philosophy is "high dividends + externally financed expansion," definitely not Baillie Gifford's "sacrifice current profits and reinvest cash with full force," and it is compounded by high-level selling from the control circle. Overall, this dimension is medium: alignment is present, but the LTGG-style long-term reinvestment gene is clearly lacking.

  Jun 5, 2026
• QIf it disappeared tomorrow, how much would customers miss it? Is its growth model sustainable, without depending on harm to society or regulatory arbitrage?4/10

  A

  Conclusion: Question 7 is medium to weak. ① Indispensability is weak: customers would miss it, but others can replace it. ② Social and regulatory sustainability is relatively strong: growth is clean and favored by policy. Under Baillie Gifford's heavy weighting on "irreplaceability," the binding constraint is ①.

  ① Indispensability: customers would miss it, but it is not irreplaceable. JL MAG is the A-share leader in pure NdFeB scale and technology, with GBD accounting for about 90% and gross margin of 21.18%, the highest among peers. Certification cycles are long and switching costs are high, so downstream customers would indeed feel pain if it disappeared tomorrow. But the report is honest about its chain position: this is a "to large-B, weak-bargaining-power" business. JL MAG sits upstream on the cost sheets of Tesla / BYD / Midea / Goldwind, with gross margin and payment terms squeezed: EV magnetic-material gross margin is only about 14%, and 2024 receivable turnover is about 128 days and lengthening. Customer concentration is high: during the IPO period, the top five accounted for 68%/73%/68%; in 2023 this fell to 46.54%; the top ten account for more than 70% of NEV revenue. This shows stickiness, but also exposes that JL MAG depends on major customers more than major customers depend on it. Nor is it exclusive: Zhenghai Magnetic Material, Zhong Ke San Huan, and Ningbo Yunsheng can all take orders, and GBD is not exclusive. This is "replaceable by someone" rather than a TSMC-like position. JL MAG does not meet Baillie Gifford's most valued standard of "irreplaceability."

  ② Social and regulatory sustainability: clean and favored. Growth does not cross social / regulatory red lines. JL MAG is a downstream magnetic-material company and is not in the environmentally heavy mining, selection, and smelting stage. GBD reduces heavy rare-earth usage by about 50%–70%, saving resources and fitting China's rare-earth strategy. Regulation is generally a tailwind: total-quota indicators are issued only to China Rare Earth Group and China Northern Rare Earth, raising industry concentration, and export licenses since 2025-04 for heavy rare earths and permanent magnets containing terbium and dysprosium are favorable to domestic leaders, strengthening bargaining power and tightening supply to support prices. The only overhang is geopolitics rather than the company's own behavior: the expanded FDP controls have been suspended for one year and expire on 2026-11-10. The whole board is tied to rare earths, a geopolitically sensitive product, and whether the suspension can be extended is an exogenous variable.

  Jun 5, 2026
• QWhat are the unit economics of this business, including gross margin and incremental returns? Do they improve or deteriorate as scale grows? Where does the money it earns go?3/10

  A

  Conclusion: JL MAG's unit economics are "capped by the cycle and bargaining power," not "the bigger the scale, the better." It lacks the high incremental-return compounding attribute Baillie Gifford wants, so it is weak.

  The gross-margin end is determined by exogenous rare-earth prices, not scale. Gross margin moved from 16.07% (2023) to 11.13% (2024) to 21.18% (2025), and the key was praseodymium-neodymium prices rather than capacity utilization. Even the cyclical high of 21% is only medium within manufacturing. The structure is even less favorable: the largest downstream segment, EV magnetic materials, accounts for about 56% of main business and has gross margin of only about 14%, the lowest in the company. That means the market where JL MAG scales fastest dilutes, rather than lifts, consolidated gross margin. Add planned industry capacity above 400,000 tons vs demand of about 200,000 tons and utilization below 70%, and "the more it expands, the more it fights a price war" further caps gross margin.

  Incremental return is the hardest falsification in this question. Even when 2025 net profit returned to the cyclical top at RMB 706 million, roughly equal to RMB 703 million in 2022, current weighted ROE and ROIC were only about 7.4%, and 2024 trough ROIC was about 4.95%. Even under a broader framework, ROE is still only single digit. Over these three years, the company raised about RMB 5.0 billion from H shares plus convertible bonds and private placements, and capacity doubled from about 20,000 tons to 40,000 tons. After a huge amount of incremental capital was deployed, normalized profit was almost standing still, from RMB 703 million to RMB 706 million. Incremental capital did not produce incremental profit, which is exactly the low incremental return growth investors most dislike. Working capital is also continuing to bleed cash: 2024 inventory was about RMB 2.178 billion, or about 32% of revenue, amplifying inventory gains / impairments in both directions when prices rise or fall. Receivable turnover is about 128 days and lengthening. Operating cash flow swings with the cycle: in 2024 it fell year over year by −32% to −66%; in 2025 it not only failed to recover but turned negative. In the first three quarters of 2025, net operating cash flow was an outflow of about RMB 199 million, down about 196% year over year, and the interim report showed a net outflow of about RMB 549 million. The main reason was that rising volume and prices drove a large increase in raw-material purchases and tied up cash. It earned accounting profit but could not collect cash, while working capital continued to drain cash.

  Where the money goes: high dividends, with RMB 271 million paid even in the 2024 trough, equal to about 93% of attributable net profit, plus RMB 506 million of R&D, or 6.55% of revenue, to support the GBD moat, which is an efficient spend. But capacity expansion relies mainly on external financing rather than internal cash. Paying a 93% dividend with one hand and dilutive financing for expansion with the other means capital is in some sense spinning in place, pinning incremental ROIC in the single digits. Unit economics are weak.

  Jun 5, 2026
• QWhat conditions must all be true for it to rise fivefold over ten years? Are these conditions realistic? What expectations are implied in today's share price?2/10

  A

  Conclusion: weak. The "earnings × valuation" combination required for a 5x gain over ten years is almost impossible to hold simultaneously for a highly cyclical stock like JL MAG. The current price has already prepaid the dual optimism of "rare earths keep rising + robots materialize," leaving insufficient margin of safety.

  Quantitative reverse-engineering: at a current price of about RMB 31 and a total market cap of about RMB 42.8 billion, a 5x gain over ten years implies a market cap of about RMB 214.0 billion.

  • If PE falls back to a reasonable cyclical-stock level of 15–20×, net profit attributable to the parent would need to reach RMB 10.7–14.3 billion, or 15–20 times 2025's RMB 706 million;
  • Even if a generous permanent growth premium of 30× is assigned, already expensive for a highly cyclical stock, net profit would still need to be about RMB 7.1 billion, or 10 times;
  • Only by stubbornly holding the current roughly 58× multiple, corresponding to EPS-TTM of about 0.54 and absolutely unsustainable over 10 years, would net profit "only" need to be about RMB 3.7 billion, or 5.2 times.

  How unrealistic is this? Even if net margin expands from 9.1% in 2025 to a cyclical peak of 12–15%, RMB 10.7–14.3 billion of net profit would correspond to revenue of about RMB 70.0–120.0 billion, 9–15 times the current RMB 7.7 billion. The report points out that JL MAG's "normalized earnings center" has not moved higher in three years: 2025's RMB 706 million basically just returned to the 2022 high of RMB 703 million, and this is EPS with praseodymium-neodymium standing at a three-year high of about RMB 700,000/ton, near the cyclical top, not a depressed trough base. To achieve another 15–20 times earnings from a near-cyclical top, multiple demanding conditions must all hold: "rare earths move up another level + robots ramp from about 3–4% of combined revenue and small-batch sample delivery into real volume + 60,000 tons of capacity are fully absorbed while industry utilization is below 70%." The realism is low.

  What the share price implies: PE-TTM is about 58×, the most expensive among peers, and PB is about 4–5.8×, more than double Zhenghai / Ningbo at 2.2–2.3× despite their 2025 net profits also doubling. The combined sell-side target price is only RMB 39–41, implying only 25–30% upside. Even bullish sell-side analysts give only 20–30% one-year upside, not multi-bagger growth. The price has already paid for the dual positives of "rare-earth prices rising + robots materializing." A reversal in rare-earth prices alone could trigger a Davis double kill of lower earnings and lower valuation, which 2024 already demonstrated once with a full net-profit halving.

  Jun 5, 2026
• QWhy has the market not realized all this yet? Is it too hard to understand, too disliked, or too long term? What will become the "narrative inflection point"?3/10

  A

  Conclusion: the direction of JL MAG's perception gap is "the opposite." Baillie Gifford's usual assumption is that "the market underestimates great growth stocks," but JL MAG is not too hard to understand, not disliked, and not too long term. It is instead "understood too clearly and priced too fully," with almost no neglected perception gap.

  It simultaneously occupies two of the hottest A-share themes, rare-earth price increases and humanoid robots, and coverage is extremely crowded: in the past 90 days, about 8 institutions have concentrated on giving "Buy/Overweight" ratings, with a combined target price of about RMB 41 (iFinD earnings forecasts), only about 25–30% above the current CNY~31. PE-TTM is ~58× and PB is the highest among peers. Zhenghai and Ningbo also doubled net profit in 2025, yet trade at only 38–42× PE and PB of 2.2–2.3×, about half JL MAG's level. The robot option and cyclical reversal have already been prepaid into the price together, as discussed in sections seven and nine of the report. This is a premium, not a discount.

  The real perception gap is likely in the opposite direction: the market overestimates cycle persistence and robot realization. Praseodymium-neodymium has already stood at a near three-year high, once breaking through RMB 800,000/ton in spring 2026 and then retreating to about RMB 700,000 by the reference date. The cheapest part of the cyclical trade has already been captured by the market. Robot combined revenue is only about 4%, only in small-batch sample delivery, and the company has never officially acknowledged being Tesla's "exclusive supplier." It also still faces the falsification overhang from Musk's 2023-03 statement that next-generation motors will "use no rare earths at all".

  The narrative inflection points are two-sided. Upside: robots move from sample delivery to batch production and disclose meaningful revenue as a separate financial line item, or praseodymium-neodymium holds above RMB 800,000/ton and moves up another level. Downside: praseodymium-neodymium falls below about RMB 500,000/ton and fails to stabilize, robots are falsified, or EVs weaken. Any one of these could trigger a Davis double kill of "earnings moving down + valuation compressing from 58× back to the peer range of 35–40×," which 2024 already demonstrated once with a full halving. The current CNY~31 sits slightly above the upper end of a reasonable range and lacks margin of safety. Only a pullback below CNY 24 would make it possible to discuss a perception-gap buying point.

  Jun 5, 2026