Discovered Materials, a newly launched startup, has secured $9 million in seed funding to develop innovative materials designed to address overheating issues in artificial intelligence chips. The company plans to leverage artificial intelligence agents operating continuously in cloud environments to identify and test potential new substances that could improve semiconductor efficiency and thermal performance.
The startup was founded by Advaith Sridhar and Akash Ramdas, whose backgrounds span artificial intelligence development and materials science research respectively. Their approach combines large language models from Anthropic with custom physics-based simulations to generate and validate material candidates at unprecedented speed. What previously required weeks of manual laboratory work can now be accomplished in days through automated exploration.
While Discovered Materials has already identified several promising materials with properties comparable to those used by major chipmakers, significant challenges remain before these discoveries reach real-world applications. Manufacturing constraints and competing material properties present complex engineering tradeoffs that must be resolved simultaneously. The company aims to patent its discoveries and license them to chipmakers, with hopes of achieving patentable materials within the next year.
Despite growing interest in AI-driven materials discovery, the field has yet to demonstrate widespread commercial success. Experts suggest that identifying viable candidates remains only part of the solution, with synthesis and practical implementation presenting the true bottlenecks in advancing the technology forward.
Originally reported by Techcrunch →
About the Author
