Haplo

🚨 Caremag is designed to make 500 t of dysprosium oxide and 90–100 t of terbium oxide a year. Jack Lifton has now run the feed arithmetic: the 2,000 t of scrap magnets gets you about 67 t. The rest has to come from mines dated 2029–2031, or not dated at all. 🧵 Jack Lifton, "Jack-in-the-Stox: Heavy Rare Earths, the French Connection", InvestorNews, 12 Oct 2026: 🔗investornews.com/market-opin… 📅On 20 April I asked on X what actually feeds Lacq in 2026, 2027 and 2028 when every named primary supplier was a future mine (link below). Lifton's 12 October piece, "Heavy Rare Earths, the French Connection", does the full mass balance and lands in the same place, with the numbers attached. He is not hostile: he calls Carester "a company addressing a real industrial bottleneck" and says the absence of a public feed reconciliation "is not proof that the business plan fails". He just wants the feed schedule published. ⚗️ The nameplate 🏭Carester's plant page: 650 t NdPr oxide, 150 t Nd oxide, 500 t Dy oxide, 90 t Tb oxide a year. USA Rare Earth's July release rounds it to 800/500/100 "when fully ramped", commissioning late 2026 🏗️Feed: 2,000 t/yr scrap magnets + 5,000 t/yr mining concentrates. The plant page labels the concentrate line "REO equivalent": 5,000 t of contained oxide, not 5,000 t over the weighbridge 💰€216m financed: €106m French state, up to €110m JOGMEC and Iwatani 🧲 What the magnets can carry 📍Measured end-of-life magnet assay 2.92% Dy, 0.33% Tb, 90% recovery → 60.3 t Dy oxide + 7.0 t Tb oxide = 67.3 t, 11.4% of a 590 t target 📍His deliberately generous case, 10% Dy and 1% Tb, which no collection system averages → 228 t. Still short 📍Lifton: recycling "cannot be assumed to supply the heavy rare earth nameplate" ⛏️ What the concentrate must be 📍Working backwards from the products, the primary feed must bring in ~581 t/yr of Dy/Tb oxide equivalent, 10.6–12.6% of a 5,000 t REO basket 📍Ordinary monazite or bastnäsite baskets carry Dy+Tb at well under 1%. His Kangankunde-type illustration yields ~4 t/yr after losses 📍So the 5,000 t has to be heavy-enriched intermediates with the lights already stripped out somewhere else. Every contract Carester has signed is for exactly that 📅 The feed, by date 🇧🇷 Brazilian Rare Earths: binding 10-yr offtake capped at 150 t/yr contained Dy+Tb ≈ 156 t finished oxide, 26% of target. Scoping study (±40%): FID mid-2029, first production 2031 🇨🇦 Torngat / Strange Lake: 10-yr deal for a partially separated Dy/Tb stream "once operational"; project targets 540 t Dy + 80 t Tb oxide; operation ~2029–30; Caremag's allocation undisclosed 🇿🇦 Frontier / Zandkopsdrift (not in Lifton's piece, but in my April post): 7-yr MHREC offtake signed Feb 2026, 114 t Dy + 25 t Tb oxide a year expected; DFS due H1 2027, first production targeted 2030 🇧🇷🇺🇸Serra Verde via USA Rare Earth: Carester gets "access to" USAR feed, but USAR itself describes Serra Verde as under a 15-year, 100% offtake with price floors through a US Department of War vehicle. Lifton's all-in scenario (~155 t Dy, ~28 t Tb) is "an allocation scenario, not a disclosed Caremag delivery commitment". USAR's 13.6% Carester stake was still pending at 1 Oct 🇦🇺🇲🇼🇰🇿 Lindian / Kazakhstan: 70% of SEGH heavy carbonate, no Dy/Tb tonnes specified; Kangankunde monazite is a light feed 🇲🇾 Malaysia / Malaco: ~13,000 t/yr raw-feed plant in Perak, no cost, no timeline, approvals pending. "An important unquantified variable" One operating source already contracted elsewhere, three dated 2029–2031, two with no quantities. 🇯🇵 The buyer is already signed 📍JOGMEC and Iwatani's €110m came with 50% of Caremag's heavy output for Japan 📍A JOGMEC representative told Argus that means ~250 t/yr Dy and 45 t/yr Tb, ~20% of Japan's potential demand, from as early as 2027 📍Half the output is sold. The heavy-enriched feed behind it is, Serra Verde aside, not in production anywhere ⚖️ The terbium problem inside the dysprosium problem 📍Caremag's product ratio is 5.6 Dy : 1 Tb. End-of-life magnets ~8.6:1, Strange Lake 6.75:1, Serra Verde ~5.7:1 ♻️Blend recycling + the Serra Verde scenario + enough Strange Lake to fill the Dy line → 500 t Dy oxide but ~77 t Tb oxide: 13 t short of 90, 23 t short of 100 📍His caveat: not proof of a missing source, but "an exactly matched product slate requires an explanation". The Tb-richest basket, BRE at ~5.1:1, is the 2031 one ❓ His question to Carencotte Publish the REO accounting basis and element composition of the primary feed basket, the acceptance spec for each supplied intermediate, and the contracted feed schedule that supports both the Dy and the Tb targets, year by year through the ramp. Until then: "A late 2026 commissioning date does not promise immediate production of 500 tons of Dy oxide and 90–100 tons of Tb oxide." 👀 Bottom Line 🇫🇷🏭⚗️♻️🧲Caremag is a real plant with real money and a real customer. What it does not yet have, on the public record, is the heavy-enriched feed to run its heavy nameplate before 2029–2031. The 600 t is a mining number that gets read as a recycling number: the magnet circuit covers about a tenth of it. 🏭⚗️Separation capacity can be financed and built to a timetable. Feed with the right dysprosium and terbium in it cannot be conjured on the same timetable. That was the question in April. It is still the question. My April post: 👇 🔗xitter.gq/roblun1/status/2046127… #RareEarths #Dysprosium #Terbium #Caremag #Carester #CriticalMinerals #MagnetRecycling #HeavyRareEarths #France #Japan