Dysprosium oxide closed at $2,147.18/kg on 2 October 2026, up 373.05% year-to-date, while the same element traded at $453.90/kg only one year earlier [S1]. That 4.7x move in a single input reverberates straight through to the price of finished NdFeB magnets, because dysprosium is not a trace additive: it is the element that keeps a neodymium magnet from losing its field above roughly 80 to 150 degC.
The reason neodymium and dysprosium oxide prices act as the dominant cost driver for magnetic material is mechanical, not financial. The NdFeB tetragonal Nd2Fe14B phase provides the high saturation magnetization (Js around 1.6 T) that defines the energy product, but the magnetocrystalline anisotropy field that resists thermal demagnetization collapses rapidly with temperature. Dysprosium substitutes onto the neodymium sites and raises the anisotropy field, which is why a 6 to 10 wt% Dy addition is standard in motor-grade grades rated above 150 degC. Every kilogram of dysprosium oxide that goes into the alloy carries the oxide price plus conversion yield losses, so a 4x spike in oxide price shows up as a roughly proportional spike in magnet price within one to two procurement quarters.
What actually moves when oxide prices move
Neodymium oxide averaged $56/kg in 2024 and an estimated $73/kg in 2025 per USGS data cited in 2026 industry reporting [S4]. On top of that, dysprosium oxide climbed from $711.40/kg five years ago to $2,147.18/kg in October 2026, a +201.82% five-year move and +745.28% over ten years [S1]. The result is that dysprosium now carries a larger dollar weight per finished magnet than neodymium in most traction-motor and wind-generator grades, even though the dysprosium content by mass is roughly one-tenth that of neodymium.
Cost composition inside a typical Dy-containing NdFeB blank breaks down roughly as: neodymium oxide (60 to 70% of raw-material cost), dysprosium oxide (20 to 30%), iron and boron precursors (5 to 10%), with praseodymium substituting for part of the neodymium in commercial "didymium" or NdPr feed. Magnet-sintering, grain-boundary diffusion, and machining contribute an additional fixed cost per kilogram that does not move with oxide markets, so the variable oxide share is the lever the spot price actually pulls.
Why supply concentration makes the link tight
China mined about 69% of the world's rare-earth oxide in 2025 (270,000 of 390,000 tonnes), and the four largest producers supplied about 95% of that total, according to USGS figures cited in March 2026 [S4]. The 2010 to 2011 dysprosium and neodymium spike, triggered by China-Japan export quota tensions, pushed rare-earth oxide prices up by an order of magnitude within months and permanently reset the post-crisis floor higher [S3][S5].
That same source concentration is the structural reason magnetic material pricing cannot decouple from oxide spot prices. Heavy rare-earth processing capacity sits almost entirely in Chinese separation facilities, and terbium, dysprosium, and yttrium are recovered as byproducts of the same ion-exchange circuits that produce neodymium. A bottleneck or export-licence change at one of those facilities flows directly into oxide price, then into alloy price, then into magnet blank price, with each step adding roughly one to two months of inventory lag. Wind turbine generator magnets already drove one of these cycles: industry reporting from 2018 onward warned that neodymium oxide was on track to reach roughly $100/kg by 2025 if supply could not diversify, a level that has since been exceeded [S3].
Cost drivers a buyer can actually influence
Four inputs move NdFeB magnet price in order of magnitude: rare-earth oxide spot price (largest swing), grade selection (N35 to N52, SH to UH temperature class), order volume and dimensional tolerance (tooling amortization), and grain-boundary diffusion versus alloy-addition dysprosium route.
On the substitution side, Sumitomo Metal Mining markets samarium-iron-nitrogen (SmFeN) magnet materials, branded Wellmax, that contain neither neodymium nor dysprosium and use samarium oxide, a rare-earth whose price returned to pre-2011 levels by around 2015 and has held flat since, according to vendor-published case material [S5]. SmFeN carries lower Br than NdFeB (typically 0.8 to 1.1 T versus 1.1 to 1.4 T for N-series NdFeB), so the substitution is not drop-in; it fits motors and sensors where the existing ferrite design fell short on torque density. For buyers who cannot tolerate another oxide cycle, ferrite remains the cheapest baseline at the cost of roughly one-third the energy product, and ferrite supply is decoupled from rare-earth oxide markets entirely.
Comparing the main options on price exposure
Three magnet families are realistically available to a buyer in October 2026. Ferrite (strontium or barium hexaferrite) has the lowest unit cost (typically $3 to $10/kg finished magnet), zero rare-earth content, and a Br of 0.2 to 0.45 T, which forces larger magnetic circuits. NdFeB with dysprosium additions delivers Br 1.1 to 1.4 T, costs $50 to $150/kg finished depending on grade and Dy content, and inherits the full oxide-price exposure documented above. SmFeN sits in the middle, with finished-magnet cost typically $40 to $90/kg, Br 0.8 to 1.1 T, and oxide-cost exposure limited to the samarium line, which has not spiked since 2011 [S5].
The decision grid is straightforward. If the application already accepts the size and weight of a ferrite magnetic circuit, ferrite wins on both purchase price and total cost of ownership, because there is no rare-earth feedstock risk. If the application needs the energy density of NdFeB but the duty cycle stays below 100 to 120 degC, a low-dysprosium or dysprosium-free NdFeB grade (such as N35SH with grain-boundary diffusion) cuts oxide exposure sharply without redesigning the magnetic circuit. If the operating temperature, miniaturization, or torque density rule out both ferrite and low-Dy NdFeB, SmFeN is the only widely commercialized rare-earth option that does not inherit the current dysprosium price curve. Procurement teams weighing these trade-offs are also weighing total cost of ownership, since installation, drive electronics, and downtime on motor rewinds often exceed the magnet blank cost over a 15 to 20 year equipment life.
Limits, failure modes, and what the data does not cover
One reason NdFeB pricing cannot be hedged with long-dated oxide contracts is that the heavy-rare-earth byproduct supply is governed by the throughput of the neodymium-praseodymium separation train. Dysprosium is not mined on its own; it is recovered as a small fraction of the same ore body, so production cuts aimed at NdPr can leave dysprosium physically short even when neodymium output is steady [S1][S3]. A second limit is that grain-boundary diffusion only works on thin, simple shapes: complex rotor magnet arcs lose the benefit of selective Dy placement and revert to alloy-addition economics.
A third point buyers routinely miss: price quotes for "sintered NdFeB" vary by a factor of 3 to 4 across suppliers at any given oxide price, because tooling amortization, scrap rate, and Dy-alloying route are all negotiable. Comparing two magnet quotations without normalizing to the same Dy content, the same temperature class (N, M, H, SH, EH, UH), and the same dimensional tolerance will produce apparent "savings" that disappear the first time the oxide market moves 20%. For a related read on how finished-part tolerances and process route drive delivered cost in a different industry, see the discussion of Dac cost per tonne 2026: $400-$600 today, $100 the long-term target, which traces the same input-cost-versus-finished-price pattern in carbon-removal procurement.
Trackable signals to watch over the next two quarters: any change in Chinese rare-earth export licensing reported via Ministry of Commerce notices, USGS Mineral Commodity Summaries revisions to 2025 mine production, and spot dysprosium oxide prints relative to the $2,147.18/kg level cited on 2 October 2026 [S1]. A pullback toward the 2024 dysprosium baseline of $403.20/kg would compress NdFeB magnet quotes within one procurement cycle; a sustained move above $2,500/kg would push more motor designs into SmFeN or heavy-ferrite territory [S5].
Spec-level background on the components involved: checking and testing apparatus.