Accelerating the
lab-to-fab timeline

Intelligence has a heat problem

AI chips have a major heat problem. GPUs today handle heat fluxes of ~140 W/cm², higher than a space shuttle nose cone re-entering the earth’s atmosphere, with each generation generating more heat than the last. This heat is why datacenters consume so much power and water; they need it to stay cool during operation.

The amount of heat produced by a chip and the speed at which it dissipates are both influenced by the materials used to make it. Today’s materials are at their limit, and new materials can improve chips by orders of magnitude. However, discovering a new material and getting it into a fab takes years and hundreds of millions of dollars. Most ideas die in this valley of death, between a science experiment and a fab.

Autoresearch for the lab

AI agents compress months of inter-disciplinary scientific work to days. Over the 3 months of our Y Combinator batch, we simulated, synthesized and tested thermal interface materials that match the performance of products the world’s largest chemical companies have sold and guarded as trade secrets for over 20 years.

We’re also releasing Material Discovery Bench, an open-source benchmark for AI-driven materials discovery. We’ve built the benchmark in collaboration with experts from IBM, IMEC, Stanford and Cambridge, and it tests frontier model ability on a relevant, real-world materials problem. The benchmark has many verifiers for grading model ability, and we will continue building more in both simulation and experiment.

Bring back Moore’s law

Advances in compute have driven most of technological progress in the last 50 years. That engine stalled in the mid-2010s, and we intend to restart it.

The headroom is enormous. Chips today are at least 10,000x less power efficient than the human brain, and our goal is to close this gap by accelerating material discovery.

Meet the team

We’re Advaith and Akash. Akash has a PhD in Material Science from Stanford University, and has spent the last 11 years researching new materials for semiconductor chips. His work on new nanoscale interconnects was Stanford Engineering’s most popular story of 2025. Advaith studied AI at Carnegie Mellon and was a research engineer building video models and agents at Persona AI (acquired) and Luma Labs. We both met over 10 years ago at IIT-Madras.