Discovered Materials Raises $9M to Use AI Agents in Search for Cooler Chip Materials
Discovered Materials says its AI-agent pipeline using Anthropic models can evaluate thousands of semiconductor candidates per day (vs ~20 during founder Akash Ramdas’s Stanford PhD). The 2026 $9M seed—led by Lightspeed India Partners with Peak XV and angel Paul Graham—targets thermally cooler chip materials, and it plans to patent viable GPU uses.
Discovered Materials, a startup focused on finding new semiconductor materials to reduce chip heat, announced in 2026 that it has closed a $9 million seed round led by Lightspeed India Partners, with additional investment from Peak XV Partners and angel investors Paul Graham, Gokul Rajaram, and Thariq Shihipar. The company emerged from Y Combinator before the raise.
The company was founded by Advaith Sridhar, who previously worked on AI agents at Persona AI and Luma Labs, and Akash Ramdas, who holds a doctorate in materials science from Stanford. Their software pipeline uses Anthropic models in a custom harness to generate candidate materials, then runs simulations using foundational physics models to verify whether those candidates are worth pursuing. The system allows the team to evaluate thousands of material candidates per day — compared to roughly 20 per day during Ramdas’ doctoral research.
AI chips running heavy workloads generate significant heat, driving up electricity consumption and cooling costs in data centers. Discovered Materials is betting that a narrow focus on the thermal properties of semiconductor materials gives it an edge over competitors such as MatNex, SandboxAQ, and CuspAI, which are pursuing broader material discovery efforts.
The startup says it has already identified several materials that match properties used by major chipmakers, though it has not disclosed details. When viable candidates are found, the company plans to patent their use in GPUs or the manufacturing processes involved, then license them to chipmakers. Sridhar said he hopes to have patentable materials within the next year.
Lightspeed partner Hemant Mohapatra, who led the round, cautioned that finding candidates is not the main obstacle. “Filtering them correctly and synthesizing them is the bottleneck,” he said. Sridhar echoed that point, noting that physical lab work — which cannot be accelerated — will remain a necessary part of the process. No AI-discovered material has yet reached commercial deployment at scale.