Startup Uses AI to Accelerate Discovery of Semiconductor Materials

Indian startup Discovered Materials is aiming to address a critical bottleneck in the semiconductor industry by accelerating the discovery of new materials using artificial intelligence. The company recently secured $9 million in its seed funding round, with participation from investors including Lightspeed India, Y Combinator, and Peak XV Partners.

The startup, founded by materials scientist Akash Ramdas and AI researcher Advaith Sridhar, employs AI agents to speed up the identification of novel materials for semiconductor applications. Traditionally, material discovery can take years, but Discovered Materials believes AI can enable them to screen a significantly larger pool of potential alternatives.

The semiconductor industry is facing increasing performance limits as the approach of shrinking transistors becomes more challenging. This shift necessitates new materials that can enable next-generation chips, particularly in areas like power electronics and graphics processing units (GPUs), where increased power consumption generates more heat, making thermal management a key concern.

Discovered Materials' AI system proposes new material structures, assesses their properties, and refines candidates based on its learning. The system can screen over 2,000 potential material candidates daily, a substantial increase compared to the approximately 20 candidates a human researcher can evaluate.

A significant challenge remains in verifying the practical manufacturability and commercial viability of AI-proposed materials. The startup is developing a verification and experimental layer to distinguish theoretically promising candidates from those that can realistically be produced and deployed in fabrication facilities.