Rare-Earth Price Shocks: Where Magnet Cost Pass-Through Stops and Supply Risk Starts

Key Takeaways

• Price shocks dilute fast. The IEA puts rare earths at about 40% of permanent-magnet cost but under 1% of average vehicle value; tripling rare-earth prices would add roughly 0.1% to a car.
• The binding risk is availability. China held about 94% of sintered magnet production in 2024, and Reuters reported on September 4, 2026 that some Chinese suppliers had stopped shipping to U.S. buyers months before the November 10 control deadline.
• Oxide prices rose; magnet prices did not. In SMM data, Pr-Nd oxide gained 7.0% from January 15 to September 3, 2026, while four China-domestic NdFeB grades fell 1.2% to 1.8% (NextMSC calculation across different grades).
• Minor rare earths moved most. Erbium rose 80% and holmium 23% over the same window, consistent with precautionary buying ahead of the suspended second-wave controls.
• Wind carries the largest dollar exposure. Modeled oxide input cost is about $127 per EV against about $16,500 per MW of wind capacity; AI servers and humanoids cannot yet be costed from public bills of materials.

Why the Question Is Live in 2026

Permanent magnets account for about 95% of rare-earth consumption by value (IEA), and NdFeB magnets sit inside EV traction motors, wind generators, data-center equipment and robot actuators. The chain runs from mining through separation, refining, metal and alloy production and magnet manufacturing to the finished motor. A price shock enters at the oxide stage and meets a different cost base at each step after it.

Source basis: Reuters, U.S. BIS and White House items as compiled in the NextMSC research file; dates as reported.
Source: https://www.nextmsc.com/industry-deep-dive/rare-earth-price-shocks-magnet-cost

Reading the calendar. The sequence shows supplier behavior changing before any formal embargo. Lead times, licence-approval pace and willingness to ship are earlier signals than the headline commodity price.

The 2026 Price Spectrum and Sensitivity Cases

SMM's August 20, 2026 weekly review put Pr-Nd oxide near CNY 717,000–720,000 per tonne, dysprosium oxide near CNY 1.40–1.41 million and terbium oxide near CNY 6.63–6.65 million. At a USD/CNY reference of about 6.7204, the midpoints convert to roughly $106.9, $209.1 and $988 per kilogram. These are China-referenced oxide proxies, not finished-magnet prices or global procurement prices; terbium oxide priced at about 9.2 times Pr-Nd oxide.

Source: SMM Aug 20, 2026 weekly review; NextMSC conversion and sensitivity cases, which are scenarios, not forecasts.
Source: https://www.nextmsc.com/industry-deep-dive/rare-earth-price-shocks-magnet-cost

A 100% rise in oxide prices does not imply a 100% rise in a vehicle, turbine or server. What matters is the magnet content per unit and the magnet's share of total cost.

What the Workbook Shows: Oxide Up, Magnets Flat

NextMSC's source-traceable SMM dataset allows a direct test of pass-through at the first two stages. Between January 15 and September 3, 2026, China-domestic oxide quotes moved unevenly, and finished-magnet quotes moved in the opposite direction.

Source: SMM weekly reviews of Jan 15 and Sep 3, 2026; midpoint = (low + high) / 2; NextMSC calculation.
Source: https://www.nextmsc.com/industry-deep-dive/rare-earth-price-shocks-magnet-cost

Source: SMM NdFeB price reviews; NextMSC calculation. Ce-containing grades differ in composition from the oxides above.
Source: https://www.nextmsc.com/industry-deep-dive/rare-earth-price-shocks-magnet-cost

Interpretation. This is the pass-through gap in data form: a 7% oxide rise coincided with 1–2% lower magnet quotes. The dataset cannot say why. Contract pricing, inventory, scrap supply and weak demand may all play a part, and the grades are not like-for-like. It does show that oxide moves are not automatically visible in magnet prices over this window.

The ex-China basis tells a different story. On September 2, SMM listed neodymium oxide at an average of $147/kg FOB China and $232/kg CIF Rotterdam, a 58% gap. Freight and insurance explain part of it; provenance and qualification explain more, which is why Fastmarkets launched ex-China assessments for Dy and Tb in March 2026.

Four Pillars of Cost Pass-Through

• Magnet intensity. Wind generators can need hundreds of kilograms of NdFeB per MW; humanoids spread small loads across many compact actuators. The metric that matters is kilograms of magnet per unit of equipment.
• Composition. DOE describes NdFeB as about 69% iron, 1% boron and 30% rare-earth metals by weight. Nd and Pr dominate; Dy and Tb are added for heat resistance, so mass sets the size of exposure while composition decides which price matters.
• Heavy rare earths. Small Dy and Tb additions carry outsized value because Tb oxide trades near nine times Pr-Nd per kilogram.
• The pass-through gap. Between oxide and finished hardware, the shock meets other costs, margins, contracts, inventory and redesign. The share that survives, not the commodity move, is the number to track.

Four-Application Cost Model

Source: NextMSC model; AIChE Journal 2026; MDPI Minerals review; William Blair 2026; SMM Aug 20 reference prices. Upstream oxide cost only; excludes processing, assembly, logistics and margins.
Source: https://www.nextmsc.com/industry-deep-dive/rare-earth-price-shocks-magnet-cost

Reading the model. The same price move produces a modest per-vehicle exposure, a much larger absolute cost in a wind turbine and an unresolved but strategically important exposure in AI servers. Cross-checks agree at order-of-magnitude level only: DOE-VTO cites 0.7–0.8 kg NdFeB per 100 kW and up to 150 kg NdPr per MW (2019-era), and the 1.7 kg NdPr EV figure is a single 2017 teardown.

Humanoids and Tesla. Reuters reported that Tesla's Optimus production was affected by China's 2025 export restrictions, a case where the risk appeared first as a licensing constraint rather than a price line.

Supply Concentration and the Cost of Diversifying

China's dominance grows downstream, so the chokepoints are separation, metallization, alloying and magnet manufacturing rather than ore. After China's April 2025 controls, IEA reported that automakers outside China cut utilization or paused lines, and importing-country prices for controlled elements stayed elevated, up to six times Chinese domestic levels in Europe.

Source: IEA Rare Earth Elements and Global Critical Minerals Outlook 2026; OECD Inventory of Export Restrictions 2026.
Source: https://www.nextmsc.com/industry-deep-dive/rare-earth-price-shocks-magnet-cost

By 2035 announced non-China mining capacity could reach nearly 50 kt of rare-earth content, but planned refining stays under 40 kt and downstream metal, alloy and magnet capacity totals about 18 kt. Higher prices also pull forward efficiency, substitution and recycling, a long-run release valve most relevant to EVs and wind, where drivetrain design can change magnet needs.

Outlook: What to Monitor

• November 10, 2026: whether the suspended second-wave controls are extended, modified or resume. Erbium, holmium and thulium buyers carry the most direct exposure.
• Licence approvals and supplier willingness to ship for Dy, Tb and Y buyers, ahead of any formal announcement.
• Oxide-to-magnet pricing: whether the January–September gap closes as contracts reset.
• Ex-China benchmarks: reconciliation between SMM and Fastmarkets assessments for Dy and Tb.
• Teardown and BOM disclosure for AI servers and humanoids, the only route to unit-cost models.

As of this snapshot, no public evidence establishes a confirmed percentage rise in AI-server, EV, wind-turbine or humanoid finished-product prices attributable to 2026 rare-earth volatility. That measurement gap, more than the commodity move, is the central analytical challenge.

Dataset Explorer and Export

Filter the company, policy, procurement, fleet and supply data compiled for this article, then export the full dataset as CSV. Every record is labeled as reported, calculated, forecast or target.

0 records shown

The CSV includes all 13 fields, including source document, source date and notes. NextMSC calculations are marked "Calculated".
Source: https://www.nextmsc.com/industry-deep-dive/rare-earth-price-shocks-magnet-cost

Methodology and Data Note

Data cut-off: September 8, 2026 (workbook audit date); published October 3, 2026; planned refresh quarterly. Market prices come from SMM and are kept in source-reported units and bases (China domestic, FOB China, CIF Rotterdam); NextMSC does not treat them as interchangeable. CNY/t converts to USD/kg by dividing by the stated USD/CNY reference and by 1,000; midpoint = (low + high) / 2. Engineering benchmarks (IEA, OECD, DOE-VTO, AIChE Journal, MDPI, William Blair) stay separate from prices, and rare-earth input cost = sum of (content x reference price), upstream only. Sensitivity cases are scenarios, not forecasts. Known limits: SMM states that non-public data are model-processed and for reference only; the Aug 20 reference prices are not rows in the verified workbook, which holds Jan 15, Sep 3 and Sep 7 domestic prices; SMM's Sep 7 live USD quotes (for example Pr-Nd oxide near $96.7/kg) use a different basis from NextMSC's CNY conversion and are not reconciled here; and the FX reference (Aug 25) post-dates the price date (Aug 20).