AI scientists to discover semiconductor materials
Accelerating scientific discovery
Tl:dr We’re building AI scientists to discover new materials for the semiconductor industry. If you know Packaging/thermal engineers at chip companies like NVIDIA, AMD, Groq, Cerebras, Google, Intel, Samsung, TSMC etc, please put us in touch!
—----------
Discovering a material with commercial viability is usually a story of persistence.
The incandescent lightbulb filament came not from a single guess, but from Edison trying 6000+ candidate materials over 20 years. The first transistor ever made in Bell labs was famously not with silicon, but germanium.
(left) The first transistor, made from Germanium. (right) some materials Edison tried as lightbulb filaments
The same holds true in the semiconductor industry. We really need better materials for our chips, but finding them takes 10+ years of lab work.
At Matforge, we build AI agents to accelerate this process of finding new materials for the semiconductor industry. Our agents work across the entire materials discovery process - coming up with new candidates, synthesizing them and testing them in physical labs.
The founders - Advaith has a Masters in AI from Carnegie Mellon, and was most recently a research engineer at Luma Labs, building agent harnesses and working on frontier models. Akash has a Masters, PhD and PostDoc from Stanford University in materials discovery for the semiconductor industry. We’ve known each for the last 11 years, and have even written a book together!
Breakthrough materials will still take many, many experiments. But for the first time, a small team might just have the tools to speed up this search and win.
