Discovered Materials Uses AI Swarms to Find Heat-Resistant Chip Materials

Overheating processors running AI workloads are a persistent headache for data centers—a problem that drives massive electricity consumption and complex cooling infrastructure. Now a startup called Discovered Materials is betting that the solution lies in using artificial intelligence itself to find better materials for building chips.

The company just closed a $9 million seed round from Lightspeed India Partners after graduating from Y Combinator, with backing from Peak XV Partners and notable angels including Paul Graham, Gokul Rajaram, and Thariq Shihipar.

How the AI Discovery Engine Works#

Discovered Materials has built a software system that orchestrates AI agents to explore candidate materials at scale. The process begins with Anthropic models generating potential material compositions, which then get fed into custom physics simulations trained by the founders to verify whether candidates might actually work.

The impact of this automation is striking. Co-founder Akash Ramdas—who holds a doctorate in materials science from Stanford—would manually test roughly 20 material combinations per day during his PhD research. The new system runs 24/7 on cloud infrastructure, allowing the AI agents to evaluate thousands of possibilities daily while exploring avenues that human researchers point them toward.

The Engineering Reality Check#

Finding materials that theoretically reduce heat or improve heat dissipation is only half the battle. The real challenge lies in manufacturing: A promising candidate might be practically impossible to fabricate into a working chip, or it might compromise critical electrical properties in the process.

Hemant Mohapatra, the Lightspeed partner who led this funding round, captured this tension: "A material is only useful in the real world if all of them converge at once, which is what makes this a really interesting search problem."

Why Discovered Materials Stands Out#

While competitors like MatNex, SandboxAQ, and CuspAI are pursuing similar AI-driven material discovery, Discovered Materials is narrowing its focus to the thermal challenges that plague semiconductor design. The company has already identified several materials that match the performance profiles of substances currently used by major chipmakers, though confidentiality prevents public disclosure of specifics.

The startup plans to pursue patent protection around either the use of new materials in GPUs or the manufacturing processes for chips built from them, then license these innovations to chipmakers. Sridhar expects to have patentable material candidates within a year.