Scientists couldn’t solve these problems for years. Now an IIT Madras startup is offering $1 million to anyone who can

Two IIT Madras alumni are putting $1 million on the table for anyone who can solve five longstanding problems in materials science, in a challenge that aims to test how far artificial intelligence and researchers can push scientific discovery, Tech crunch reported.

Advaith Sridhar and Akash Ramdas, who met more than a decade ago at IIT Madras, co-founded Discovered Materials, a San Francisco-based startup working on AI-powered methods to discover new materials for semiconductor chips.

The startup, which participated in Y Combinator's spring 2026 batch, has now launched the prize challenge after working with academics from leading universities to identify five scientific problems that have remained unsolved despite years of research.

What are the five problems?

The five challenges cover some of the hardest questions in materials science.

They include understanding cuprate superconductivity, predicting the electrical properties of semiconductor interfaces and finding faster and more accurate ways to carry out quantum chemistry calculations.

The other two challenges focus on understanding how light changes the behaviour of materials and predicting how heat travels across interfaces between solid materials.

The problems are not just academic. Finding answers could help researchers develop more efficient electronic devices, improve semiconductor manufacturing and advance energy-related technologies.

Don't Miss: The ₹80 lakh gamble: How a Bengaluru couple turned samosas into a ₹200 crore business

The biggest prize is $400,000 for solving the cuprate superconductivity problem. Two other challenges carry prizes of $200,000 each, while the remaining two offer $100,000 each.

The competition is open to eligible individuals, teams and organisations. There is no fixed deadline, but prizes will be given only when a proposed solution meets the scientific criteria laid down for the challenge.

The AI test did not produce easy answers

Before opening the challenge to researchers, Discovered Materials tested advanced AI models on the five problems. The models were allowed to work as autonomous AI agents, with access to the internet and other research resources.

Yet the results showed how difficult these questions remain.

The AI models made little progress even after working independently for several days, TechCrunch reported.

“How do we know AI is making progress?” Sridhar asked, highlighting one of the biggest questions around using AI for scientific research.

Stanford professor Felipe Jornada, who helped develop the challenges, said, “Across materials science, there are phenomena we observe but cannot explain.”

For Sridhar, the competition is also about creating a way to measure whether AI and researchers are actually making progress on difficult scientific questions, rather than simply producing answers.

The startup has already raised $9 million in seed funding in August to advance its research into semiconductor materials.