Skip to main content

AI Agent 'Elements Claw' Discovers New Superconductors, Speeding Up Materials Science

AI Steps into the Lab: A New Era for Superconductor Discovery

For decades, finding new superconducting materials has been a slow, painstaking process—a lot of trial and error, years of work, and often, a bit of luck. But that might be changing. On July 3, a team from Alibaba DAMO Academy, Renmin University, and the University of Chinese Academy of Sciences introduced Elements Claw, an AI agent built specifically to hunt for superconductors. And it's already delivering results.

Think of it as a tireless, super-smart research assistant that never sleeps. Unlike traditional methods that rely on combing through limited databases (the international SuperCon database, for instance, has only about 2,000 entries after decades of work), Elements Claw taps into a massive library of 125 million molecules and crystal structures. It's powered by a 10-billion-parameter atomic foundation model called Elements, which can predict a material's superconducting potential with astonishing accuracy—an AUC score of 0.996 and critical temperature predictions within 1 Kelvin.

Image

From Screen to Lab: AI That Designs and Tests

What makes Elements Claw stand out is that it doesn't just crunch numbers. It mimics the full workflow of a human scientist: reading literature, assessing whether a material can actually be synthesized, designing experiments, and even learning from its own findings. In a recent test, the AI sifted through 2.4 million crystal structures and flagged 68,000 promising candidates—all in just 28 GPU hours. That's a speed that would take human researchers years, if not decades.

And the proof is in the lab. The team has already synthesized and verified four new superconducting materials: HfZrRe4 (which AI designed from scratch), plus Hf21Re25, Zr4VRe7, and Zr3ScRe8 (discovered by correcting and reanalyzing existing data). Their critical temperatures reach up to 6.5K—a solid step forward.

What This Means for Materials Science

Rong Yu, head of Science Intelligence at DAMO Academy, says these results confirm the huge potential of AI agents in material discovery. To help the field move faster, the team has released the full dataset of 2.4 million stable crystals. Professor Huang Wenbing from Renmin University adds that this framework could be adapted to find other key materials, like solid-state battery electrolytes, catalysts, and thermoelectric materials.

In short, AI is no longer just a helper—it's becoming a discoverer. And that could open up a whole new chapter in science.

Key Points

  • Elements Claw is the first AI agent designed specifically for discovering superconducting materials.
  • It screened 68,000 candidates from 2.4 million structures in just 28 GPU hours.
  • Four new superconductors have been synthesized, including one AI-designed material (HfZrRe4).
  • The team has released the dataset of 2.4 million stable crystals to accelerate research.
  • This AI framework could be applied to other materials like battery electrolytes and catalysts.