A materials science startup founded by two IIT Madras alumni has announced a $1 million prize pool for solutions to five difficult research problems. The challenge is intended to attract researchers and teams working on questions that could affect semiconductors, electronics and energy technologies.
The company, Discovered Materials, was co-founded by Advaith Sridhar and Akash Ramdas. It is developing AI-based methods to discover materials for semiconductor chips.
The prize is divided across five problems, with the largest award set at $400,000. The challenge has no fixed deadline; awards will be made only when a proposed solution meets the scientific criteria established for the relevant problem.
Five questions in materials science
The problems include understanding cuprate superconductivity, predicting electrical properties at semiconductor interfaces and improving the speed and accuracy of quantum chemistry calculations.
The other challenges focus on how light changes the behaviour of materials and how heat moves across interfaces between solid materials.
These questions are relevant because materials underpin technologies used in computing, energy and manufacturing. Better understanding of how materials conduct electricity, respond to light or transfer heat could help researchers develop more efficient components.
But the problems are difficult for a reason. They involve complex interactions between atoms, energy and structure, and many observed effects are not yet fully explained.
How the prize is structured
The largest prize, $400,000, is assigned to the cuprate superconductivity challenge. Two other problems carry awards of $200,000 each, while the remaining two offer $100,000 apiece.
The challenge is open to eligible individuals, teams and organizations. Rather than setting a single deadline, the company says it will award prizes when solutions meet the scientific requirements.
That structure puts emphasis on verification. A promising result would need to be supported by evidence that other researchers can examine, and the criteria for each challenge will shape what counts as a valid solution.
The prize also aims to make some research questions more visible outside specialist laboratories. It can attract attention, but money alone does not guarantee a breakthrough.
A test for AI-assisted research
Before launching the challenge, the startup tested advanced AI models on the five problems. The models were allowed to work as autonomous agents with access to online research resources.
The company reported that the systems made little progress after several days. That result illustrates the difference between AI producing plausible text and AI solving a difficult scientific question with a result that can be independently verified.
AI tools can help researchers search literature, generate hypotheses or analyze data. But scientific discovery also requires experiments, careful measurement and reproducibility. In materials science, a prediction must ultimately correspond to a real material and a result that holds up under testing.
Why the challenge matters
The contest is also an attempt to define what progress in AI-driven science should look like. Instead of judging a model by how convincing its response sounds, researchers can evaluate whether it contributes to a solution that survives scrutiny.
The problems are connected to potential improvements in electronics, semiconductor manufacturing and energy technology. Still, any practical application would depend on follow-up research and development after a solution is found.
The startup raised $9 million in seed funding in August to support its work on semiconductor materials. The prize challenge adds a public-facing component to that effort by inviting researchers to tackle questions the company cannot solve alone.
Whether the challenge produces a breakthrough remains uncertain. Its immediate contribution is to put five hard scientific questions into a structured competition and invite the wider research community to take them on.