ABOUT THIS EPISODE
Two best friends from college. One watched Claude go from helping with his code to writing nearly all of it, almost overnight. The other, Akash, was finishing a Stanford PhD in materials science. When they asked the model to work on some unpublished research Akash had spent a year on, it reached similar conclusions in weeks. That was the moment they decided to start a company.In this episode of Deeptech Decoded, Nihal Kurth talks with Advaith Sridhar, co-founder of Discovered Materials (YC), about how two people in a garage lab in Oakland made thermal interface materials that, by their own tests (in their lab and at a third-party facility in Livermore), match products Dow, Honeywell and Henkel have sold for over 20 years. And why Advaith believes the bottleneck in semiconductors is no longer just computation, but the materials that have to survive exploding power levels.Advaith says NVIDIA’s next-generation chips are heading toward 4,000 watts, a single chip putting out the heat of roughly four US houses. The materials in chips today were never designed for that. He and Akash are betting that AI agents, combined with real lab iteration, can compress the time it takes to find and qualify new thermal materials. He also explains why he thinks the “self-driving lab” consensus is wrong.Chapters:00:00 – Intro03:42 – Akash, the PhD defense, and a happy accident04:27 – The month Claude took over his codebase05:46 – Three weeks against a year of unpublished Stanford work07:40 – GNoME: 50,000 materials, +30% overnight10:16 – 118 elements, 60 usable, 5.5 million combinations12:05 – Diamond and graphite: why structure beats composition14:56 – Materials against drug discovery: where the money actually sits17:50 – An R&D lab, not a factory: making hits, not one-hit wonders20:25 – Inverse design: benchmarking Opus, Fable and GPT-5.623:03 – Why the same machine on the same day gives different results25:11 – The heat wall: 1,000W Blackwell to 4,000W Rubin Ultra29:24 – Thermal interface materials: 40% of the resistance30:00 – Three months, $500K, and a garage lab in Oakland30:53 – Why the self-driving lab wasn't the win34:17 – Computation widens the funnel, experiment is the throughput37:31 – Why they paused the TIM track41:30 – Still thermal, but material-agnostic44:15 – Feedback loops: talking to chip manufacturers and the YC partners46:33 – The investor who only wanted to see one slide52:32 – Staying sane: a co-founder, the Bay Area, a partner54:26 – The tech stack: no VS Code, Claude Code, Supabase, Fortran to Python57:32 – What YC changed his mind about: distribution may beat product1:01:10 – Elias from Zettascale asks about the next 10 years1:02:43 – World models, digital twins, and why it's still early1:06:10 – From PM back to engineering: the reset he's glad he madeWhere to find Advaith:LinkedIn → https://www.linkedin.com/in/advaith-sridhar/Discovered Materials → https://discoveredmaterials.com
