AI Database Identifies 25 New High-Temperature Magnets, Offering Path Beyond Rare Earths

Researchers at the University of New Hampshire have built an AI-powered database of 67,573 magnetic compounds — and used it to identify 25 previously unrecognised materials that function as high-temperature magnets. The work, published in Nature Communications, targets one of the clean energy transition's most uncomfortable dependencies: rare earth elements.

Rare earth magnets are essential components in electric vehicle motors, wind turbine generators, and countless electronic devices. They are what make modern EVs efficient, what allow wind turbines to generate power without gearboxes, and what enable everything from smartphones to MRI machines to function. But the supply chain is heavily concentrated — China controls roughly 60% of mining and 90% of processing — and extraction is environmentally damaging, often involving toxic chemicals and significant landscape destruction.

Finding alternatives has been a strategic priority for governments and manufacturers alike. The US Department of Energy has funded multiple programmes through its Critical Materials Institute, and the EU's Critical Raw Materials Act explicitly targets reduced dependency on Chinese rare earth supply chains.

The AI system works by reading and extracting data from thousands of published scientific papers, building a searchable repository of magnetic properties that no human team could compile manually. The database doesn't just catalogue known materials — it identifies patterns and relationships that reveal candidates researchers had overlooked or never thought to test.

Among the 25 new candidates, several show magnetic characteristics comparable to existing rare earth materials at the temperatures required for industrial applications. This is crucial: many alternative magnets lose their magnetic properties at the operating temperatures found inside EV motors and turbine generators, which is why rare earth magnets have remained dominant despite their problematic supply chain.

The approach itself is a powerful demonstration of how AI can accelerate materials discovery. Traditional materials science relies on synthesis-and-test cycles that can take years per candidate. By computationally screening tens of thousands of compounds simultaneously, the AI compresses what might have been decades of trial-and-error into months of targeted investigation.

Key Facts

• Database covers 67,573 magnetic compounds (Nature Communications)
• 25 previously unrecognised high-temperature magnets identified
• China controls approximately 60% of rare earth mining and 90% of processing
• Rare earth magnets are used in roughly 90% of EV motors
• Global rare earth market valued at approximately $10 billion annually, projected to triple by 2030

Why This Matters

The clean energy transition depends on magnets. Every EV motor, every direct-drive wind turbine, every electric aircraft concept requires powerful permanent magnets — and right now, that means rare earth elements controlled predominantly by one country. Any disruption to that supply chain — whether from geopolitical tension, trade restrictions, or environmental regulation — could slow the very technologies the world needs to decarbonise.

This research doesn't solve the problem today, but it opens a credible pathway toward solving it. If even a few of the 25 candidates can be synthesised and manufactured at scale, the clean energy transition becomes more resilient, more geographically distributed, and less dependent on environmentally destructive mining.

What We Don't Know Yet

The 25 candidates are computationally identified — they have not yet been synthesised or tested in real-world applications. The gap between a database prediction and a working magnet inside an EV motor is substantial.

Performance at actual operating temperatures in real motors and turbines may differ from database predictions. Computational models, however sophisticated, cannot capture every variable that affects material behaviour under stress.

The timeline from discovery to commercial application for new magnetic materials is typically 10 to 15 years. Existing rare earth supply chains have significant incumbent advantages in cost and manufacturing integration that new materials must overcome.

The database relies on published scientific literature, which may contain errors or gaps. Not all magnetic materials have been equally studied, meaning the database may have blind spots.

Sources: ScienceDaily · University of New Hampshire
Published 21 February 2026 · Category: Science & Technology