Every electric vehicle on the road hides a dirty secret inside its motor. The magnets that make it spin depend on rare earth elements — and China controls nearly 90% of the global supply. One research team just used AI to crack that problem wide open.

The Supply Chain Everyone Ignores

Here’s the uncomfortable math: over 86% of EV motors sold in 2024 used rare earth permanent magnets. Each vehicle packs roughly 1.5 kilograms of neodymium iron boron (NdFeB) — the strongest permanent magnets money can buy. They’re in your phone, your MRI machine, your wind turbine, and increasingly, your car.

China doesn’t just mine rare earths. It controls nearly 90% of all processing and refining capacity globally. Western nations ceded this ground decades ago because the work is environmentally hazardous and chemically nasty. Beijing took on those risks and built an empire.

An empire with teeth. China has imposed targeted export bans on heavy rare earth elements and modified dual-use export controls to restrict magnet exports. According to S&P Global, supply bottlenecks are expected to persist throughout 2026. The rare earth magnet market for EV motors in Europe and the US alone could hit $5 billion by 2035.

That’s a lot of money flowing through a single chokepoint.

Teaching AI to Read 67,000 Papers

A team at the University of New Hampshire, led by doctoral student Suman Itani, took a different approach to finding alternatives. Instead of the traditional method — synthesize a compound, test it, repeat for decades — they built an AI system that could read thousands of scientific papers and automatically extract experimental data about magnetic materials.

That extracted data trained predictive models to do two critical things: determine whether a material is magnetic, and calculate its Curie temperature — the point where a material loses its magnetism. For EV motors running at high temperatures, that second property is everything.

The result: the Northeast Materials Database, containing 67,573 magnetic compounds. And buried in that data, 25 compounds nobody had previously flagged as high-temperature magnets.

Their paper, published this week in Nature Communications, represents a paradigm shift in how we discover materials.

Why 25 Compounds Changes Everything

Twenty-five sounds modest until you learn the context. No entirely new permanent magnet has been identified from known compounds in recent memory. NdFeB magnets were developed in 1984 — independently by General Motors and Sumitomo Special Metals — and nothing has fundamentally displaced them since.

Four decades of incremental improvements on the same technology. Finding 25 new high-temperature candidates in a single study, pulled from data already sitting in published papers, isn’t just a result. It’s proof that the methodology works.

The real breakthrough isn’t the compounds. It’s the process. If AI can mine existing literature to surface overlooked materials this effectively, the pace of materials discovery just shifted gears entirely.

The Geopolitical Race Underneath

This research lands amid a full-blown scramble over critical minerals. A Council on Foreign Relations report warned that “the rare-earth magnet supply chain remains structurally exposed to upstream leverage, pricing power, and strategic coercion from China.”

Jack Hidary, CEO of Alphabet spin-off SandboxAQ, recently argued that AI and quantum computing could create synthetic substitutes that bypass the traditional 10-to-20-year timeline for bringing a new mine online. But moving from lab breakthroughs to large-scale manufacturing — the kind Beijing perfected over decades — remains an enormous gap.

The UNH research, funded by the U.S. Department of Energy, fits squarely into this strategic picture. It’s science with a national security mission.

What This Means for Your Wallet

If even a few of these 25 candidates prove viable:

Cheaper EVs. Neodymium prices are volatile and climbing. Heavy rare earths like dysprosium and terbium — used to make NdFeB work at high temps — are even pricier. Alternative magnets could meaningfully cut motor costs.

Stable supply chains. Automakers like Tesla, BMW, and BYD wouldn’t live and die by one country’s export policies. Several manufacturers have explored rare-earth-free motor designs, but performance trade-offs kept most of the market dependent on NdFeB.

Faster clean energy. Wind turbines use massive rare earth magnets. Offshore wind, with its larger direct-drive generators, is especially dependent. Alternatives could unblock deployment in regions bottlenecked by supply concerns.

Broader applications. The same AI methodology could discover materials for batteries, semiconductors, superconductors — any domain where the design space is too vast for traditional experimentation.

The Long Road Ahead

Let’s be clear: having a database and having a commercial magnet are very different things. Each candidate needs synthesis, real-world testing, manufacturing optimization, and scale-up. That takes years.

But the old way — manually searching an essentially infinite design space — was taking decades and producing nothing new. AI just compressed the discovery phase from impossible to manageable. The engineering challenges remain, but at least now we know where to look.

The Bottom Line

This isn’t the flashy AI story. No chatbot, no deepfake, no robot. It’s AI doing what it arguably does best: finding needles in haystacks too big for humans to search alone. Those needles could reshape trillion-dollar industries, redraw geopolitical battle lines, and make the clean energy transition faster and cheaper.

The race to replace rare earth magnets is one of the most consequential and least discussed technology challenges of our time. It sits at the intersection of AI, materials science, climate policy, and great-power competition.

As of this week, we have 25 new leads.