A new startup from Y Combinator is offering a $1 million prize to solve five of the most stubborn problems in materials science, all to test the limits of AI. The company, Discovered Materials, aims to spur breakthroughs around topics such as superconductivity, semiconductor interfaces, and atomic-scale modeling.
What problems are on the line?
Working with top academics worldwide, Discovered Materials has defined five open challenges where progress has been elusive. The competition covers:
- Developing a theory that explains cuprate superconductivity—how certain materials conduct electricity without resistance at extremely low temperatures.
- Finding a numerical method that predicts how silicon and silicon dioxide interfaces arrange themselves—important since this material pair is foundational to nearly all semiconductors.
- Mapping how grain structure impacts thermal properties, enabling better heat dissipation in materials.
- Crafting a theory for how shining light on materials alters their behavior, particularly around superconductivity.
- Improving Density Functional Theory (DFT), a computationally intensive technique used to model atomic and molecular structures.
Why this matters now
Discovered Materials, founded earlier in 2026 and based in San Francisco, is applying frontier AI models toward materials discovery. While AI has made inroads in solving complex math and biology problems, many core physics challenges in solid-state and quantum materials remain uncharted territory. These issues lack precise definitions in the way math or logic puzzles do, complicating clear progress measurement.
To test what’s possible, the company has already unleashed AI agents—given internet access and compute—for days at a time to tackle each challenge. So far, these agents have made minimal headway. But Discovered Materials sees this effort as defining “goalposts” for the field, whether or not it’s this company or another research lab that ultimately breaks through.
The company focuses on materials important to semiconductors, partnering with chip manufacturers while conducting internal R&D. When Discovered Materials closed its seed funding in August, its founders gave themselves one year to find an industrially useful new material; today, they believe they’re ahead of that schedule.
These challenges aren’t just a showcase of ambition—they’re a bet on a new scientific method where AI models may uncover insights humans haven’t anticipated. By predefining the toughest unanswered questions, the hope is that creative solutions will emerge from unexpected places.
What this means: If any of the five challenges are solved, it could shift how electronics are built, energy is stored, and materials are manufactured. Success would push high-performance computing, quantum devices, and semiconductor design into new territory.
