Discovered Materials raises $9 million to develop AI-powered semiconductor materials discovery

A newly-launched startup called Discovered Materials emerged from Y Combinator and closed a $9 million seed round led by Lightspeed India Partners to develop AI agents that identify new materials for more efficient semiconductor manufacturing. The company uses Anthropic's language models within a custom software pipeline that generates material candidates and verifies their feasibility using physics simulations. The startup released hundreds of newly discovered materials alongside its Material Discovery Bench, a benchmark platform for tracking how frontier AI models approach material discovery challenges. Founders Advaith Sridhar and Akash Ramdas are leveraging Ramdas' background in materials science from Stanford to solve a critical problem in AI infrastructure: excessive heat generation in chips running advanced AI workloads. While competitors like MatNex, SandboxAQ, and CuspAI pursue similar approaches, Discovered Materials is focusing specifically on thermal efficiency for semiconductors. The team claims to have already discovered several materials matching properties of those used by major chipmakers, though specific details remain confidential pending patent applications.