Akash Ramdas, an alumnus of the Indian Institute of Technology Madras, has channelled years of materials science research into an artificial intelligence-driven startup aimed at one of the semiconductor industry’s most pressing constraints: the heat generated by increasingly powerful chips. Co-founding Discovered Materials with a fellow IIT Madras graduate, Ramdas has helped the company secure 9 million dollars in seed funding to develop AI agents that can discover and validate new materials far faster than traditional methods.
Ramdas graduated from IIT Madras in 2019 with a Bachelor of Technology degree in Metallurgical and Materials Engineering. He then moved to Stanford University, where he completed a master’s degree in Materials Science and Engineering followed by a PhD in Materials Engineering. His doctoral research focused on computational techniques and machine learning approaches to identify promising materials for nanoscale electronic devices. In one study, his work screened more than 15,000 metals using computational methods and identified around ten candidates predicted to outperform copper and other alternatives under consideration, including ruthenium. Elements of this research have been noted for their relevance to technology roadmaps at major chipmakers.
After completing his doctorate, Ramdas briefly continued as a postdoctoral researcher at Stanford before turning fully to entrepreneurship. In 2026 he co-founded Discovered Materials with Advaith Sridhar, an IIT Madras batchmate whose background includes advanced study in artificial intelligence and experience building autonomous agents. The pairing brings together deep materials expertise with AI systems knowledge, forming the foundation of the company’s approach.
The Challenge of Heat in Modern Chips
Contemporary artificial intelligence chips generate substantial heat, with fluxes reported to exceed 140 watts per square centimetre. Managing this thermal load has become a critical bottleneck for performance, energy efficiency and data-centre operations. Traditional materials discovery and qualification cycles can stretch over a decade, a timeline that is increasingly mismatched with the rapid advances in chip design and the escalating demands of AI workloads.
Discovered Materials seeks to compress this timeline dramatically. The company builds AI agents capable of exploring vast materials spaces, running simulations and generating candidates that can then be tested and refined. The initial focus is on thermal materials that could improve heat dissipation and enable more efficient packaging, including three-dimensional architectures that place logic and memory closer together. Broader ambitions extend across the semiconductor stack, covering interconnects, substrates and other components where new materials could unlock performance gains.
Funding and Early Progress
In August 2026 the startup announced a 9 million dollar seed round led by Lightspeed India Partners. Participants included Y Combinator, Peak XV Partners and angel investors such as Paul Graham, Gokul Rajaram and Thariq Shihipar. The capital is intended to expand the team, build laboratory infrastructure and scale the AI research agents.
Alongside the funding announcement, the company released hundreds of computationally discovered material candidates and introduced Material Discovery Bench, described as a benchmark for evaluating how well frontier AI models perform on real-world semiconductor materials-discovery problems. Early results cited by the founders include new thermal materials that, in their assessment, match the performance of products developed over years by large chemical companies. These materials have undergone third-party testing under relevant standards, and the company has reported interest from semiconductor manufacturers.
Bridging Research and Commercial Application
The path from computational discovery to materials that can be manufactured at scale remains long and complex. Discovered Materials positions its AI agents as tools that can generate high-quality candidates at a volume and speed far beyond what individual researchers could achieve manually. One co-founder has noted that while a researcher might evaluate a limited number of possibilities each day, the agents can explore thousands of candidates continuously in the cloud.
The business approach centres on intellectual property. The company plans to license both the materials it discovers and the processes required to manufacture them, aiming to partner with chipmakers rather than produce materials itself. This model reflects a recognition that success will depend on close collaboration with the semiconductor supply chain and rigorous validation of computational predictions in physical laboratories.
A Journey from Campus to Deep-Tech Entrepreneurship
Ramdas’s trajectory illustrates a growing pattern among researchers who combine domain depth with computational tools. His undergraduate foundation at IIT Madras, advanced training at Stanford, and exposure to both academic research and brief industry roles provided the technical grounding. The decision to found a startup rather than remain solely in research or corporate positions reflects a conviction that AI can materially change the pace of materials innovation in semiconductors.
Sridhar’s complementary expertise in artificial intelligence and autonomous systems completes the founding team’s capability set. Together they have framed the problem not merely as one of faster computation but as the creation of AI scientists that can propose, evaluate and prioritise materials candidates in ways that align with the practical constraints of chip manufacturing.
Looking Forward
The semiconductor industry continues to confront physical limits even as demand for computing power accelerates. Heat management sits among the most immediate of these constraints. By applying AI agents to materials discovery, Discovered Materials seeks to contribute solutions that could improve energy efficiency and enable denser, more powerful chip designs. The 9 million dollar seed funding provides resources to expand both computational capacity and laboratory validation.
Success will ultimately be measured by the number of materials that progress from computational prediction to qualified use in production environments. The company has taken early steps by releasing candidates and a public benchmark, inviting broader scrutiny and collaboration. For Ramdas, the work represents a continuation of research begun in academic settings, now directed toward industrial impact at a scale and speed that conventional methods alone cannot match.
The story of Discovered Materials underscores how specialised knowledge developed at institutions such as IIT Madras and Stanford can be translated into focused deep-tech ventures. With AI chips generating unprecedented thermal loads, the search for better materials has become both a scientific and commercial imperative. Through AI agents designed to accelerate that search, Ramdas and his co-founder aim to shorten the distance between laboratory insight and the fabrication facilities that produce the next generation of semiconductors.
Read more – Lpainhand.com
