Discovered Materials raises $9M to find cooler chip materials with AI agents
Original: Discovered Materials is playing AI whack-a-mole to hunt cooler chips
Why This Matters
AI-driven material discovery could reduce data center energy costs by enabling cooler, more efficient chips at scale.
Discovered Materials, a Y Combinator alumnus, closed a $9M seed round led by Lightspeed India Partners to use AI agent swarms and physics simulations to discover new semiconductor materials that reduce heat in AI chips.
Discovered Materials has raised $9 million in seed funding from Lightspeed India Partners, Peak XV Partners, and angels including Paul Graham to accelerate AI-driven semiconductor material discovery. The startup, founded by Advaith Sridhar and Akash Ramdas (Stanford materials science PhD), deploys Anthropic-powered AI agents in a custom software pipeline that generates material candidates and validates them using trained foundational physics simulation models. The system runs 24/7 on cloud infrastructure, enabling thousands of material guesses per day — compared to roughly 20 per day during Ramdas' PhD research. The company today released hundreds of new material examples and its 'Material Discovery Bench,' a benchmark tracking how frontier models approach this challenge. Competitors include MatNex, SandboxAQ, and CuspAI, but Discovered Materials is focusing specifically on thermal properties of semiconductor materials. The firm says it has already found materials matching properties used by major chipmakers, though details remain undisclosed. Lightspeed partner Hemant Mohapatra described the challenge as 'playing whack-a-mole with atomic structures,' noting that thermal, electrical, and manufacturing properties must all converge for a material to be viable. Monetization plans include patenting novel materials' use in GPUs or the manufacturing processes derived from them.