What removing a gearbox costs in rare earths

Offshore wind moved toward direct-drive generators to lose a failure-prone gearbox, and the published intensity figures show what the swap puts back in its place.

Reviewed by Peter Uppal

A very large open steel generator ring standing on a low cradle, its inner face lined with a continuous run of flat rectangular pole pieces. — Illustrative artwork: a direct-drive generator rotor of the kind used in wind turbine manufacturing. Not a facility, equipment or material connected to this project. · Illustration · Osmond Hub

The short version

A geared wind turbine turns a slow rotor into a fast generator; a direct-drive turbine deletes the gearbox and turns the generator slowly instead, which takes a very large permanent magnet. IRENA states that a megawatt of direct drive wind turbine capacity 'may require around 500 kilogrammes (kg) of permanent magnets, a third of which is REEs (notably, for direct drive offshore turbines)'(opens in a new tab) — a doubly hedged figure, and the clearest published statement of what the trade actually is.

Two ways to make electricity out of slow rotation

The problem is a mismatch. The US Department of Energy describes drivetrains where generators are connected to gearboxes that speed up rotation from 'the relatively slow speed of the turbine's blades (typically 5–15 rotations per minute for a modern machine), to the high speeds (1,000–1,800 rotations per minute) needed to generate electricity'(opens in a new tab). A separate DOE page gives the low-speed shaft as spinning between 8–20 rotations per minute(opens in a new tab); both figures are ranges, both are DOE's, and they are quoted here as published rather than reconciled.

Gearboxes are the part that breaks. The DOE drivetrain page states plainly that having all of those moving parts makes the gearbox 'one of the highest-maintenance parts of a wind turbine'(opens in a new tab), and that direct drive systems 'do not require a gearbox and therefore have fewer moving parts'(opens in a new tab). Removing it removes the maintenance.

What replaces it is mass and magnet. DOE describes direct-drive generators as producing power using 'a giant ring of permanent magnets that spin with the rotor'(opens in a new tab), and states part of the cost of that directly: direct drives usually use permanent magnets(opens in a new tab) and typically 'require heavier generators than geared machines for a given turbine capacity'(opens in a new tab). A further clause is attributed to the same page — that those magnets 'require expensive, heavy, rare earth materials such as neodymium and dysprosium' — and it is in neither retrieved DOE extract, so it is given here unquoted and unlinked, even though it is the clause that carries the cost point. The hedges — usually, typically — are the source's own.

Why offshore, specifically

The published explanations do not all say the same thing, and the difference is instructive. IRENA attributes the offshore preference to size and weight: eliminating the gearbox 'makes possible smaller and lighter turbines… making it more competitive in offshore applications'(opens in a new tab). The IEA's framing folds in operating cost, describing permanent-magnet synchronous generators as machines that dominate the offshore market 'due to their lighter and more efficient attributes as well as lower maintenance costs'(opens in a new tab). The 2025 JRC study puts reliability first, describing direct drive as allowing enhanced reliability, reduced maintenance needs, and simplified design(opens in a new tab).

The intuitive argument — that maintenance is hard when the machine is far out at sea — is not what any of these sources actually says, and it is not asserted here.

A gearbox is a maintenance liability you can reach. A magnet is a supply liability you cannot substitute.

What the intensity figures show

Beyond IRENA's 500 kg per megawatt, the JRC's 2020 report on raw materials for wind and solar gives magnet composition on a basis it attributes to a third party: on average a permanent magnet contains '28.5% neodymium, 4.4% dysprosium, 1% boron and 66% iron' and 'weighs up to 4 t'(opens in a new tab). The dysprosium fraction is small and matters out of proportion — IRENA notes that dysprosium is 'less than 1% of all REEs' and that its supply 'is even more constrained than the neodymium supply, and this may pose a problem for the energy transition'(opens in a new tab). May pose; the source does not say it will.

The architecture is a lever on all of this. The same 2020 report, again citing a third party, states that a hybrid drive can reduce neodymium use 'from 186 kg/MW installed capacity to just 62 kg/MW, compared with turbines that employ direct-drive permanent magnet systems'(opens in a new tab) — a gearbox and a smaller magnet, rather than one or the other. Magnet-free machines exist as a category: the JRC describes electrically excited synchronous generators, which 'use an external electrical source to magnetise the rotor', noting that such configurations 'involve complex control systems and higher maintenance needs compared to permanent magnet systems'(opens in a new tab), and superconducting machines as a cutting-edge development 'still in the experimental phase' and 'not yet widely deployed'(opens in a new tab).

Even the IEA's demand outlook is written as a conditional rather than a forecast of technology: it describes a case in which manufacturers 'are assumed to gradually switch to non-magnet technologies' and developers adopt hybrid configurations(opens in a new tab) — the assumption is the scenario, not a prediction.

How much of the market this is

Less than it sounds, and the published shares measure different things. IRENA reports that permanent magnet turbines drove three-quarters of the world's offshore installations in 2018(opens in a new tab) — one year's installations, not the fleet. The JRC reports that in 2022 turbines with permanent magnets represented 42% of the onshore wind market in China, 25% in the world, 21% in the EU, and only 2% in the USA(opens in a new tab) — the onshore market, on a different basis again. No source read here gives a share of total installed capacity, and none is constructed.

On the mineral end, Osmond's own reporting defines the relevant fraction: its preliminary monazite concentrate is described as ~19.4% TREO excluding yttrium, including 25% MREO(opens in a new tab), the release's own approximation mark included — MREO being the magnetic rare-earth oxides, defined in that release as neodymium, praseodymium, dysprosium and terbium. Those four elements are the ones a magnet chain buys.

Exploration results and mineralogical estimates only. Orión has no JORC-compliant Mineral Resource or Reserve; maiden MRE and Scoping Study pending, targeted Q3 CY26.

Related