🚨 HyProMag USA has published a PFS to quadruple its planned US magnet capacity to ~4,200 t/yr. The economics look strong on paper. The document itself tells you where the hard part is: feedstock. 🧵 Mkango (AIM/TSX-V: MKA) released the HyProMag USA Pre-Feasibility Study today, prepared by PegasusTSI and BBA USA. I read all eight pages. What it says, what it does not, and the scale-up question. 🏭 The plan🇺🇸 🏭🏭🏭Three new plants (South Carolina, Nevada, a second Texas site) plus a Central Texas Hub, on top of the Northlake, Texas plant already in detailed engineering 🏭🧲Each plant: feedstock processing + HPMS (hydrogen processing of magnet scrap) + unfinished sintered blocks → shipped to the Hub for VIM/strip casting, grain boundary diffusion, machining, plating, coating, QA and customer qualification 🎯Target by 2034: ~6,000 t/yr salable material = 4,192 t magnets + 1,305 t NdFeB alloy powder + 503 t other. Northlake alone: ~1,048 t magnets (1,526 t total) 🏗️Construction of the three new plants "commencing in 2030, 2032 and 2034", so read the 2034 figure as a run-rate target, not 2034 output 💵 The numbers (three new plants only, Northlake excluded) 📍Post-tax NPV (7% real): US$2,062m for Case 1 (recycled feedstock only) to US$1,854m for Case 2 (recycled + primary alloy) 📍IRR 36% / 31%; payback 2.3 / 3.0 years; Profitability Index 3.3x / 2.8x 📍Capex (AACE Class 3): US$623m / US$656m 📍AISC in 2030: US$37.4/kg (Case 1) to US$41.6/kg (Case 2) against a weighted realised price of US$100.5/kg 📍Price deck = Adamas Intelligence Q2 2026 forecasts. NPV excludes any government incentives or carbon pricing 📍Northlake re-cut: NPV US$723m (from US$797m in June), capex US$167m, Phase 1 commissioning mid-2028. Combined platform NPV US$2.6–2.8bn ♻️ Feedstock: the real scale-up question Case 1 (recycled only) wins on every metric: cheaper to build, cheaper to run, faster payback, higher NPV. Scrap is cheaper than primary alloy. The catch is volume. • The release does not state how much scrap the platform needs. Output cannot exceed input: ~6,000 t/yr of salable NdFeB means well over 6,000 t/yr of contained magnet in feedstock once losses are counted (my arithmetic, not in the document) • Scale check: a hard disk drive carries 10–20 g of NdFeB. At that rate, 6,000 t/yr is 300–600 million drives a year. HDDs can start Northlake; they cannot feed four plants. Rotors, MRI machines, wind generators and appliance scrap have to carry the expansion, and that is exactly where HyProMag USA says its feedstock discussions have moved • US e-waste is estimated to hold ~18,000 t/yr of NdFeB on paper (Chatham House), most of it landfilled. End-of-life magnet collection is below 15% globally (IEA). Globally ~90,000 t of NdFeB enters waste streams in 2030, 35,000 t of it in consumer electronics (Adamas) • Status today: "feedstock inventories currently under contract and in storage" plus active procurement, expected to support Phase 1 commissioning in 2028. No tonnage disclosed. Three Inserma HDD separation systems at ILS sites, one now mobile • HyProMag USA's own January 2026 concept-study release said the expansions "would be conditional on securing adequate feedstock and offtake" • That is why Case 2 exists: primary NdFeB alloy as a swing feedstock, with VIM and strip casting at the Hub 🧪 Where does the primary alloy come from? • The PFS says "primary NdFeB alloy as a feedstock". Alloy, not oxide. Alloy means someone has already done oxide → metal → alloy • The release points to "potential future access to primary sourced feedstock from Mkango pursuant to certain Mkango supply rights". Songwe Hill (Malawi) and the proposed Pulawy separation plant (Poland) would produce carbonate and oxide, not alloy. The metallisation step between them is not described • Grain boundary diffusion, the route that lifts high-grade output from ~10% to ~20%, needs dysprosium or terbium. The release does not say where the heavy rare earths come from 🧲 Magnets • 4,192 t/yr by 2034 is roughly 9% of 2025 US NdFeB demand (~48,000 t, Arthur D. Little via Reuters) • First US finished magnets planned H1 2027 from up to 20 t of blocks sent from HyProMag's UK and German operations; HPMS section at Northlake Q2 2028 per the July 2026 release • "Product performance data and customer validation are underway". Sample magnets have gone to US customers. No offtake agreement has been announced • DFARS angle: the only recycling carve-out in the 2027 rule is an NdFeB magnet "manufactured from recycled material if the milling of the recycled material and sintering of the final magnet takes place in the United States". HPMS powder plus US sintering is that configuration 🏢 Who owns what HyProMag USA = 50% CoTec, 50% HyProMag Ltd. HyProMag Ltd is 100% Maginito, which is 79.4% Mkango / 20.6% CoTec. Mkango's look-through interest is ~40%. 👀 Bottom Line A PFS is a study, not a plant. The returns are strong on paper because recycled scrap is cheap, but the scrap volume behind 6,000 t/yr is a large multiple of anything disclosed as contracted, the primary-alloy fallback needs an oxide-to-alloy step the document does not describe, and the GBD upgrade needs heavy rare earths. The company says it in its own risk factors: "availability of suitable feedstock" and "the ability to secure customer offtake and feedstock supply arrangements". Feedstock is the gating item. Everything else in the study follows from it. @MkangoResources #RareEarths #Magnets #NdFeB #Recycling #HyProMag #Mkango #CriticalMinerals #DFARS #SupplyChain #CircularEconomy