The Tiny Magnet Quietly Strangling the Clean Energy Revolution

Everyone has been looking at the wrong component. The debate around electric vehicles has fixated on lithium, cobalt, battery chemistry, and charging infrastructure. Those are real issues. But the component that could quietly derail the entire clean energy transition sits much smaller, weighs almost nothing, and almost nobody talks about it in mainstream coverage.

It is a magnet. A rare earth permanent magnet, to be specific. And the supply chain behind it is fragile in ways that should keep energy analysts up at night.

The good news: a team at the University of New Hampshire has built something that could change the equation entirely, using AI trained on a century’s worth of materials science literature.

Why Rare Earth Magnets Are the Hidden Chokepoint in EV Motors

When you press the accelerator in an electric vehicle, the response comes from a traction motor. That motor almost certainly contains a rare earth permanent magnet, typically made from neodymium, praseodymium, or dysprosium. These elements produce the strongest magnetic fields of any known materials, which is exactly why engineers love them.

The U.S. Department of Energy notes that nearly all hybrid and plug-in electric vehicles use rare earth permanent magnets in their traction motors. The International Energy Agency confirms that these same elements are essential for the generators inside wind turbines. Without them, both technologies either perform worse or cost significantly more to engineer around.

The geopolitical problem compounds the engineering one. China controls the vast majority of global rare earth processing capacity. That concentration creates price volatility, export restriction risks, and supply vulnerabilities that no amount of battery chemistry improvement can fix.

So while headlines debate solid-state batteries and lithium prices, the magnet problem hums along quietly, baked into the cost of every new EV rolling off an assembly line.

The NEMAD Database: 67,573 Materials and One Very Specific Mission

Suman Itani and colleagues at the University of New Hampshire did not set out to write another paper about magnets. They set out to build infrastructure. The result is NEMAD, a public database designed specifically to accelerate the discovery of magnetic materials that do not require rare earth elements.

The numbers behind it are impressive. NEMAD contains 67,573 entries. It covers 84 different elements across the periodic table. It tracks 15 distinct material features per entry. To build it, the team compiled roughly 100,000 article identifiers from journals published by Elsevier and the American Physical Society, then used AI to extract, clean, and structure the data.

The AI models trained on this database achieved 90% accuracy when classifying whether a material is ferromagnetic, antiferromagnetic, or non-magnetic. For engineers, that distinction is everything. Ferromagnetism is what makes a permanent magnet work. Getting that classification right is the first filter in any materials discovery pipeline.

The team also built a model to predict Curie temperature, the point at which a magnetic material loses its magnetism when heated. This matters enormously in real-world applications. An EV motor runs hot. A wind turbine generator in a desert climate runs hotter still. A magnet that loses its properties at 200°C is useless in those environments.

The best Curie temperature prediction model achieved an R² of 0.87 with a mean absolute error of just 56 kelvin. That level of precision means researchers can filter candidates computationally before ever synthesizing a single gram of material in a lab.

“The aim is to reduce dependence on rare-earth elements” by identifying alternative magnetic materials that can perform comparably without the geopolitical and environmental costs.

— Suman Itani, lead author, University of New Hampshire

GaFe2Co4Si and the 25 Candidates That Could Reshape Motor Design

The most striking output from this research is a list. The AI system identified 25 previously unreported high-temperature ferromagnetic candidates, compounds that no published study had flagged before as promising magnet materials.

One stands out immediately. The compound GaFe2Co4Si was identified as a predicted ferromagnetic material with a Curie temperature of approximately 1,005 kelvin. That converts to 732 degrees Celsius, or 1,350 degrees Fahrenheit. For context, most electric motor operating temperatures stay well below 200°C. A magnet that remains stable at over 700°C has enormous thermal headroom.

The significance here is not just one compound. It is the method. Traditional materials discovery relied on chemists manually scanning literature, drawing on intuition, and synthesizing candidates one by one. That process takes years per material. A database of 67,573 entries processed by machine learning can surface candidates in hours.

This is the accelerant that materials science has needed. The periodic table has not changed. What changed is humanity’s ability to find useful combinations within it, at speed, with quantified confidence.

The compounds in question, including GaFe2Co4Si, are built from iron, cobalt, silicon, and gallium. None of these are rare earth elements. Iron and cobalt are globally abundant. Silicon is one of the most common elements on Earth. Gallium is less common but not geopolitically concentrated in the way neodymium is. A motor built around such a magnet would have a fundamentally different supply chain story.

What Non-Rare-Earth Magnets Would Mean for Wind Power and EV Costs

The economic logic is straightforward once you follow the supply chain. Wind turbines using direct-drive generators, the most efficient modern design, rely heavily on rare earth permanent magnets. A large offshore turbine can require several hundred kilograms of neodymium-iron-boron magnetic material. Multiply that across thousands of turbines in a single offshore wind farm and the supply constraint becomes visible fast.