Discovered Materials is playing AI whack-a-mole to hunt cooler chips

Data centers consume large amounts of electricity due to AI workloads, necessitating cooling systems. This issue has spurred entrepreneurs to use AI in solving the problem it created. Discovered Materials, a startup, aims to use AI agents to find new materials for more efficient integrated circuits. The company recently secured a $9 million seed round, led by Lightspeed India Partners, following an emergence from Y Combinator.

Founders Advaith Sridhar and Akash Ramdas have created a software pipeline that uses Anthropic models to generate material leads and then employs physics models to simulate and verify the potential of these candidates. Ramdas' doctoral work in materials science from Stanford and Sridhar's experience in AI agents have enabled their software to run thousands of simulations daily.

Discovered Materials unveiled hundreds of new materials and a "Material Discovery Bench" to monitor how frontier models tackle this challenge. The startup believes that focusing on the thermal issues of semiconductor materials is the way to success. However, the engineering trade-space poses a challenge as promising materials might be difficult to manufacture or compromise electrical properties.

Discovered Materials' founders hope to patent valuable candidates and license them to chipmakers. Despite the excitement, there is no evidence of AI-discovered drugs or materials making a commercial impact, with Insilico Medicine's Renterosib being the closest example. The bottleneck for AI materials science lies in filtering and synthesizing candidates, not just discovering them.