YC start-up Discovered Materials puts $1 million in prizes behind five unsolved materials science problems

Discovered Materials, a San Francisco start-up from Y Combinator's spring 2026 batch, has launched a set of "Grand Challenges" in materials science and is offering $1 million in prize money. TechCrunch reported the launch on November 5. A solution to any one of the five problems is worth $200,000. The company presents the challenges as goals for "AI scientists", but its page does not say whether entrants must use AI.

The five problems

The challenges were drawn up with an advisory group of academics from universities around the world. Stanford professor Felipe Jornada helped develop them. On the company's site they are titled as follows: a predictive theory of cuprate superconductivity; interfacial electrical properties from processing history; breaking the speed–accuracy barrier in quantum chemistry; tuning materials with light; and predicting heat transport across solid interfaces. TechCrunch describes them in more detail. They include predicting the structure of interfaces between silicon and silicon dioxide, linking a material's grain structure to its thermal behaviour, and improving density functional theory, a widely used method for modelling atoms and molecules. Entries go in by email, and each challenge page lists what a submission has to include. Taking part means agreeing to official rules. The main page calls the challenges verifiable but does not explain how entries will be judged, and it gives no precise deadline.

Why a start-up is funding prizes

Co-founder Advaith Sridhar told TechCrunch that the problems are meant to serve as goalposts for whether AI models are actually advancing in the field, and that the company does not mind who solves them. Before the launch, the company gave frontier models internet access and other resources and let them work alone for several days on each problem. They made little progress, according to the report.

Its main business is finding new semiconductor materials, with a focus on heat in chips. When it raised a $9 million seed round in August, led by Lightspeed India Partners with Peak XV Partners also taking part, TechCrunch reported that its agents run on Anthropic models in a custom setup. Physics models trained by the company screen the candidates, and an in-house lab tests the most promising ones. Sridhar now says he is more confident of finding an industrially useful material within a year. That is a company goal, not a demonstrated result.