One email each morning — yesterday's AI, sortedGet it in your inbox

METAL LAB

New Materials to Tame Chip Heat: AI Agents Search Thousands of Times a Day

Discovered Materials raises $9 million seed round to find new materials that solve semiconductor thermal problems

실험실 가운을 입은 두 남성이 나란히 웃고 있다

이미지: TechCrunch Startups

Summary

  • Discovered Materials raised a $9 million seed round led by Lightspeed India Partners
  • The company combines Anthropic's models with its own physics simulation model to explore thousands of new material candidates per day
  • The company unveiled hundreds of example materials and the "Material Discovery Bench," which measures frontier model performance on this task
투자 규모
900만 달러 시드 라운드
주요 투자자
Lightspeed India Partners 주도, Peak XV Partners·Paul Graham·Gokul Rajaram·Thariq Shihipar 참여
창업자
Advaith Sridhar, Akash Ramdas (Y Combinator 출신)
기술 방식
앤트로픽 모델 기반 에이전트로 후보 물질 도출 후 자체 물리 시뮬레이션 모델로 검증
경쟁사
MatNex, SandboxAQ, CuspAI 등이 유사 사업을 진행 중

AI startup Discovered Materials announced it has raised a $9 million seed round to find new materials that could reduce heat buildup in semiconductors. The round was led by Lightspeed India Partners, with participation from Peak XV Partners and angel investors Paul Graham, Gokul Rajaram, and Thariq Shihipar. The company received the investment after going through the Y Combinator program.

The founders are Akash Ramdas, who holds a PhD in materials science from Stanford, and Advaith Sridhar, who built agent development experience at Persona AI and Luma Labs. The two built a system that uses a custom agent pipeline powered by Anthropic's models to generate candidate materials, then runs simulations with their own trained physics-based model to verify how effective those candidates actually are.

"During his PhD, [Ramdas] might have made around 20 guesses a day. Now the agents run around the clock in the cloud, trying thousands a day," Sridhar explained.

Alongside the funding announcement, the company also unveiled the "Material Discovery Bench," which includes hundreds of examples of candidate materials and tracks how well frontier models perform on this problem. The company said it has already discovered several materials whose properties align with those used by major chipmakers, though it did not disclose further details.

While companies like MatNex, SandboxAQ, and CuspAI are pursuing similar materials discovery businesses, Discovered Materials has chosen to focus specifically on semiconductor thermal issues. Hemant Mohapatra of Lightspeed, who led the round, described the challenge by saying, "It's a bit of playing whack-a-mole with atomic structures," adding that a material may reduce heat or improve heat dissipation but still prove difficult to manufacture or end up compromising electrical properties — a persistent difficulty in this field.