Discovered Materials uses AI agents to hunt for cooler chip materials

Discovered Materials, a startup emerging from Y Combinator, has raised $9 million in seed funding to use AI agents to discover new semiconductor materials that can help reduce heat generation in chips, addressing a key challenge in data center energy consumption.

Why chip cooling matters

As AI workloads grow, chips are generating more heat, forcing data centers to consume significant electricity for cooling. This thermal problem is a bottleneck for performance and sustainability. Discovered Materials aims to tackle this by using AI to rapidly generate and test new materials that could lead to more efficient integrated circuits.

The startup’s software pipeline, built on Anthropic models, generates material leads and then uses physics-based simulations to verify their potential. The founders claim this approach can test thousands of candidates per day, compared to the 20 or so a human researcher might manually evaluate.

Founders and funding

Founded by Advaith Sridhar and Akash Ramdas, the company combines expertise in AI agents and materials science. Ramdas holds a doctorate from Stanford, while Sridhar previously worked on AI agents at Persona AI and Luma Labs. The seed round was led by Lightspeed India Partners, with participation from Peak XV Partners and notable angel investors including Paul Graham.

Lightspeed partner Hemant Mohapatra, who led the investment, described the challenge as “playing whack-a-mole with atomic structures,” where materials must meet multiple criteria—thermal, electrical, and manufacturability—simultaneously to be useful.

Industry context

Discovered Materials is entering a growing field of AI-driven materials discovery, competing with companies like MatNex, SandboxAQ, and CuspAI. However, the startup is focusing specifically on semiconductor thermal issues, a niche it believes offers a clear path to commercial impact.

While the company has already identified several promising materials, it has not yet disclosed details. The founders say they plan to patent the use of these materials in GPUs or the manufacturing processes, licensing them to chipmakers.

Challenges ahead

Despite the promise, AI-discovered materials have yet to achieve widespread commercial deployment. The closest example is in pharmaceuticals, where Insilico Medicine’s AI-discovered drug reached Phase II trials. In materials, candidates like rare-earth-free magnets from MatNex have been identified but not scaled.

Mohapatra notes that the bottleneck is not finding candidates but “filtering them correctly and synthesizing them.” This requires wet-lab experimentation, which cannot be accelerated. Sridhar acknowledges that “a lot of this will involve actually going into wet labs and making things as well,” a process that inherently takes time.

Conclusion

Discovered Materials’ approach to using AI for chip materials discovery is a promising step toward addressing the thermal challenges of modern computing. While the technology is still in its early stages, the startup’s focused strategy and strong backing position it well to contribute to the future of more efficient semiconductors.

FAQs

Q1: What is Discovered Materials?
Discovered Materials is a startup that uses AI agents to discover new semiconductor materials, specifically to reduce heat generation in chips. It recently raised $9 million in seed funding.

Q2: How does the AI technology work?
The company uses AI models to generate material candidates, then runs physics-based simulations to verify their properties. This allows them to test thousands of materials per day, far more than traditional methods.

Q3: Why is this important for data centers?
Chips running AI workloads generate significant heat, requiring energy-intensive cooling systems. Finding materials that dissipate heat more efficiently can reduce energy consumption and improve performance.