Science · 2026-11-05
A Y Combinator startup offers $1M for AI solutions to five unsolved materials science problems
Discovered Materials, a months-old San Francisco startup, is offering a $1 million prize for explaining five scientific mysteries about materials needed for chips, batteries and electronics.
What happened
Discovered Materials, a months-old San Francisco startup, has announced a challenge for AI models. The company uses AI to search for new substances for advanced chips. It has pledged a total prize pool of $1 million for solving five important materials science problems.
The company is part of Y Combinator's spring 2026 batch, the well-known startup accelerator.
The problems in simple terms
Academics from leading universities worldwide helped define the tasks. They include: explaining why some materials have zero electrical resistance at extremely low temperatures (a property called superconductivity); predicting how atoms arrange at the boundary between silicon and silicon dioxide, a combination found in almost all semiconductors; predicting heat conduction from a material's internal grain structure to design better heat dissipation; explaining why light can change how materials behave, especially around superconductivity; and improving density functional theory, a key method scientists use to model the structure of atoms and molecules.
Why it matters
According to Felipe Jornada, a Stanford professor who helped develop the challenges, there are phenomena in materials science that scientists observe but cannot explain. Explaining them would improve understanding of physics and shape how future electronics, energy storage and materials are made.
The format is also significant. In math, answers are often clearly right or wrong, but in materials science it is harder to define a correct solution. So the company wants the challenges to serve as benchmarks: tracking how teams tackle them will show whether AI is actually getting better in this field.
No success yet
Before announcing the challenges, Discovered Materials gave frontier AI models internet access and other resources and let them work autonomously for several days. The models spent several days on each problem without making much progress.
The article does not give detailed participation rules. The full list of problems and details can be found on the company's page.
Useful from the source
- Discovered Materials grand challenges — The company's page listing five open science problems and the prize pool.
Sources: 🔗 TechCrunch — AI