# Discovered Materials

> Discovered Materials builds AI scientists to discover new materials for the semiconductor industry, compressing the lab-to-fab timeline from years to weeks. Backed by a $9M seed round led by Lightspeed, with Y Combinator, Peak XV, Paul Graham, Gokul Rajaram, and Thariq Shihipar.

## About
- Discovered Materials (operated by Matforge Inc.) uses swarms of AI agents to find novel semiconductor materials
- Finding novel materials today takes 10+ years of lab work; Discovered Materials aims to find 10x better alternatives in months
- Thesis: heat is the bottleneck for AI chips — new materials (e.g. better thermal interfaces and dielectrics) can bring back Moore's-law-style scaling
- Discovered phase-change thermal interface materials among 500+ previously unknown materials surfaced by AI models
- Focus areas: datacenters and fabs
- Located in Mountain View, CA

## Founders
- Akash Ramdas, Co-Founder & CEO — PhD in Material Science from Stanford, postdoc in novel materials for nanoscale electronics. His work on novel interconnect materials has received widespread publicity. https://www.linkedin.com/in/akashramdas/
- Advaith Sridhar, Co-Founder & CTO — Former research engineer at Luma Labs and Persona AI, where he worked on post-training, evaluations and agentic harnesses for foundation models. MS in AI from Carnegie Mellon University. Best Outgoing Student at IIT-Madras. https://www.linkedin.com/in/advaith-sridhar/

## Research
- Material Discovery Bench: https://discoveredmaterials.com/research/ — measures frontier AI models' ability to discover new materials for the semiconductor industry. Models have surfaced 500+ previously unknown materials, verified with DFT and attempted in a real lab; generating plausible synthesis routes remains the hard part. Includes an interactive explorer of candidates, synthesis grades, and expert commentary.
- Code and data: https://github.com/advaith-dm/materials-autoresearch

## Models & tools
- HBM Thermal Resistance (interactive): https://discoveredmaterials.com/models/hbm_thermal_resistance/ — an interactive model of thermal resistance in HBM-on-GPU stacks that shows how improving the DRAM dielectric's thermal conductivity raises the thermally-limited DRAM stack height. Based on 2025 IEEE IEDM research (doi:10.1109/IEDM50572.2025.11353711).

## Contact
- Email: founders@discoveredmaterials.com
- Website: https://discoveredmaterials.com
- X: https://x.com/discoveredmat
- LinkedIn: https://www.linkedin.com/company/discoveredmaterials/
- Careers: https://www.ycombinator.com/companies/discovered-materials/jobs
- Location: Mountain View, CA