Discovered Materials, a San Francisco-based startup from the Y Combinator spring 2026 batch, has launched a $1 million prize pool. The challenge invites AI models to solve five specific, previously unresolved problems in materials science.
The company collaborated with an advisory group of academics from global universities to define these challenges. The goal is to establish benchmarks for evaluating the progress of AI models in understanding complex physical phenomena that currently lack clear explanations.
Prior to the public announcement, Discovered Materials tested frontier AI models on these problems. The agents were given internet access and autonomous resources but failed to make significant progress over several days of testing.
The startup focuses on developing new materials for semiconductors and partners with chip manufacturers. Founder Advaith Sridhar noted that the company's timeline for discovering industrially useful materials has accelerated, with expectations now set for less than a year.
