# Discovered Materials raises $9M to build AI agents that find better chip materials in months, not decades > Source: > Published: 2026-08-10 12:15:13+00:00 Via altair.com # Discovered Materials raises $9M to build AI agents that find better chip materials in months, not decades The Y Combinator startup wants to use swarms of AI scientists to solve the semiconductor industry's looming power and heat problems. Building a better chip isn’t just about shrinking transistors anymore. It’s increasingly about finding better stuff to make them out of. Discovered Materials, a San Francisco startup fresh out of Y Combinator’s Spring 2026 batch, just raised $9 million to tackle that exact problem with what it calls “AI scientists,” autonomous agents designed to discover novel materials for semiconductors. The pitch is straightforward: traditional materials discovery takes over a decade from lab curiosity to production-ready component. Discovered Materials wants to compress that timeline to months by deploying swarms of AI to handle candidate proposals, simulations, and physical testing in rapid succession. ## The team and the tech The company was co-founded by Akash Ramdas and Advaith Sridhar, a pairing that reads like a deliberate fusion of the two disciplines they’re trying to merge. Ramdas holds a PhD and completed a postdoc in materials science at Stanford, where his research led to discoveries of materials for nanoscale interconnects that were subsequently adopted by Intel and TSMC. Sridhar brings the AI chops, with a master’s degree from Carnegie Mellon and prior experience at Persona AI. The startup, which also operates under the name Matforge in certain contexts, currently has between 2 and 10 employees. Job postings for roles like Founding Process Engineer suggest it’s actively scaling up its team to match the ambition of its $9 million war chest. ## Why materials matter more than ever Modern AI chips are power-hungry beasts. Training large language models and running inference at scale generates enormous amounts of heat, and the materials used in packaging and interconnects directly determine how efficiently that heat can be dissipated. Better thermal interface materials, more conductive interconnects, and improved dielectric layers could meaningfully extend the performance ceiling without requiring yet another leap in transistor density. This is where Discovered Materials is positioning itself. Rather than competing with TSMC or Intel on fabrication processes, the company is targeting the upstream problem: identifying materials that make existing and next-generation chip designs more thermally efficient and energy-conscious. ## A growing field with high stakes Discovered Materials isn’t alone in betting that AI can accelerate the notoriously slow process of materials discovery. CuspAI is another startup that has attracted attention in the same space, and larger players like Google DeepMind have published research on using machine learning to predict crystal structures and material properties. What distinguishes Discovered Materials within this field is the specificity of its focus. Rather than pursuing general-purpose materials discovery, the company is laser-focused on semiconductor applications, particularly datacenter and fabrication facility challenges. Ramdas’s track record of discoveries that actually made it into production at Intel and TSMC suggests the team understands the gap between a promising lab result and something a chipmaker will actually use. The $9 million raise is substantial for a company at this stage. Early-stage materials science ventures typically burn through capital more slowly than software startups, since much of the initial work is computational rather than physical. The funding should give the team enough runway to demonstrate that their AI agents can identify viable material candidates and validate them through simulation before moving to costly physical testing. **Disclosure:** This article was edited by Editorial Team. For more information on how we create and review content, see our [Editorial Policy](https://cryptobriefing.com/editorial-policy/).