Discovered Materials, a months‑old San Francisco startup that applies AI models to discover new materials for advanced chips, has announced a prize competition totaling $1 million. The company assembled an academic advisory group and defined five open problems in materials science, offering pooled awards for solutions.
Why the challenge was created
While frontier AI models have recently solved notable mathematical problems and advanced some areas of the life sciences, they have not yet fully addressed the complex physics of advanced materials. Advaith Sridhar, a co‑founder of Discovered Materials, said materials problems are harder to specify precisely than math problems because they rarely admit a simple yes/no answer.
To create measurable targets for progress, the startup recruited university researchers from leading institutions to identify the field’s ‘‘grand challenges.’’ The total prize pool for solving these problems is $1 million.
The five problems
The five challenges Discovered Materials issued are:
- A theory explaining cuprate superconductivity — why certain materials exhibit zero electrical resistance at ultra‑low temperatures.
- A numerical method to predict the structures of interfaces between silicon and silicon dioxide, a material pairing present in nearly all semiconductors.
- A numerical method to map grain structure to thermal properties, enabling better thermal‑management materials.
- A theory predicting why illuminating materials can change their behavior, particularly around superconductivity.
- A numerical method to improve Density Functional Theory (DFT), a core computational technique used to model atomic and molecular structures in materials science.
Felipe Jornada, a Stanford professor who helped formulate the challenges, said that explaining these observed but currently unexplained phenomena would deepen our understanding of physics and directly influence how electronics are built, how energy is stored, and how future materials are manufactured.
Early AI tests and the competition’s purpose
Before announcing the challenge publicly, Discovered Materials ran frontier lab models on the problems. The company gave AI agents internet access and other resources to work autonomously for several days on each problem; according to the company, the models spent days on the tasks but did not make much headway. Sridhar noted that it is not crucial to the company whether the ultimate solver is Discovered Materials, a university researcher, or a frontier AI lab — the goal is to set clear goalposts that indicate real progress in the domain.
The company plans to track attempts and submissions to monitor ‘‘whether the models are climbing in this domain.’’
Scientific approach and business context
The challenge reflects shifts in scientific inquiry as AI begins to uncover insights and techniques human researchers had not anticipated. Sridhar described the approach as ‘‘a new way of thinking about science,’’ where verification criteria and challenges are defined first, and then others generate ideas that explain the problems. He emphasized that asking the right questions is nontrivial and that university professors were instrumental in shaping the correct questions.
Discovered Materials’ core business focus is developing new materials for use in semiconductors. The startup partners with major chipmakers on research projects while also funding its own scientific efforts. When the company announced its seed round in August, Sridhar told TechCrunch he hoped to produce a useful material within a year; he says timelines have accelerated somewhat and that he is now more confident the company can find industrially useful materials in under a year.
Next steps
By publicly defining these five challenges and offering financial incentives, Discovered Materials aims to stimulate research and measure whether AI‑driven methods can solve some of the hardest open problems in materials science. The company will observe submissions and broader research activity to see if AI models progress on these targets.
