Startup Uses AI to Discover Materials for Chip Cooling

Processor overheating, especially during AI workloads, forces data centers to allocate massive resources to cooling systems. Discovered Materials offers a solution: leveraging AI to identify new materials that reduce chip heat dissipation. The startup recently secured $9 million in a seed funding round led by Lightspeed India Partners, Peak XV Partners, and angel investors, including Paul Graham.

Founders Adwait Sridhar and Akash Ramdas combine expertise in materials science and AI agent development. Their platform uses Anthropic models to generate material candidates, which are then validated through physical simulations. According to Sridhar, the system can process thousands of hypotheses daily, whereas traditional methods allowed for only about 20 variations to be tested.

Discovered Materials has already identified hundreds of new materials and introduced the Material Discovery Bench tool to evaluate model efficacy in this field. However, transitioning from lab to mass production remains difficult: even promising materials may fail commercial viability due to technological constraints. Investors highlight the startup’s ability to rapidly validate hypotheses in lab settings as a key advantage.

In the future, the company plans to patent discovered materials and license them to chip manufacturers. Founders anticipate the first patentable results within a year, though real-world adoption may take significantly longer. To date, no AI-discovered material has reached mass-market adoption, but the technology is advancing rapidly.