AI startup Discovered Materials raises $9M for cooler semiconductors

👁 116 reads

Discovered Materials has closed a $9 million seed round to use AI agent swarms in the search for semiconductor materials that run cooler than those currently used in chips, the startup announced on August 10. Lightspeed India Partners led the round, with Peak XV Partners and angel investors Paul Graham, Gokul Rajaram, and Thariq Shihipar also participating.

The company was founded by Advaith Sridhar, who previously worked on agents at Persona AI and Luma Labs, and Akash Ramdas, who holds a doctorate in materials science from Stanford. Together they built a software pipeline that uses Anthropic models in a custom harness to generate candidate materials, then runs those candidates through foundational physics models the team trained in-house to simulate and verify whether they are actually worth pursuing.

The productivity gap with traditional research is stark. During his PhD, Ramdas evaluated around 20 material candidates a day. The current system runs thousands of guesses a day, according to Sridhar, with agents operating continuously on cloud infrastructure across directions that Ramdas defines. Alongside its funding announcement, the startup released examples of hundreds of new materials and a “Material Discovery Bench” designed to track how frontier models perform on this challenge.

The search problem

The core difficulty is that a promising material rarely excels across all the dimensions chipmakers need at once. Thermal performance, manufacturability, and electrical properties must converge simultaneously. Hemant Mohapatra, the Lightspeed partner who led the round, described it as “a bit of playing whack-a-mole with atomic structures,” adding that “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.”

Discovered Materials says it has already identified several materials that match the properties of substances used by major chipmakers, though it declined to share specifics. The startup is not alone in the space - MatNex, SandboxAQ, and CuspAI are pursuing comparable efforts - but Discovered Materials is betting that a narrow focus on the thermal problems of semiconductors gives it an edge.

When viable candidates emerge, the business plan is to patent either the use of the materials in GPUs or the manufacturing processes that derive from them, and then license those patents to chipmakers. Sridhar said he expects to have materials worth patenting within the next year. Mohapatra acknowledged that the underlying capability of predicting novel substances will likely be commoditized as models improve, and argued that Ramdas’ domain expertise and the team’s ability to rapidly validate candidates in a lab setting are the more durable advantages.

The broader field has yet to produce a commercially deployed material discovered primarily through AI, though early-stage results - including rare-earth-free permanent magnets from MatNex - suggest the pipeline is generating scientifically credible candidates.

// read moreFrench startup Kog raises seed funding to speed LLM inference on GPUsAI ToolsAI code-testing startup Blacksmith raises $45M at $550M valuationMagazineLawsuit Accuses xAI of Training Grok on Child Sexual Abuse MaterialNews

Kinetyra