Discovered Materials puts $1M behind five open materials science problems

An AI startup is paying $1 million for answers to five open problems in materials science. The bounty is an alternative to adding more people to its own payroll. TechCrunch reports that Discovered Materials announced the prize on 5 November. The company offers $200,000 for each of the five problems. Discovered Materials is a Y Combinator Spring 2026 startup. It closed a seed round in August 2026. Advaith Sridhar and Akash Ramdas are the founders. The company is based in San Francisco.

The five problems

TechCrunch lists the five problems in full. Two are deep theories. One asks for a theory of cuprate superconductivity. Cuprate superconductors carry current with no resistance at low temperatures. Another asks for a theory of how light changes a material's behaviour, superconductivity in particular. Three are numerical methods. One asks for a way to predict the structure where silicon meets silicon dioxide. That junction sits inside every chip. One asks for a method that maps a metal's grain structure to how well it carries heat. The last asks for an improvement to Density Functional Theory. DFT is the main workhorse for modelling atoms and molecules.

Each problem is deep. None has a standard answer in the field. Sridhar tells TechCrunch that "it's more difficult to specify problems clearly in materials science because it's not like math". Open problems in math are stated in a paragraph. The company has written each materials problem in enough detail that an outside team can check an answer.

What Discovered Materials sells, and why this bounty fits

Discovered Materials builds AI models to find new chip materials. It partners with chip manufacturers on specific research projects. TechCrunch reports that the company also funds its own internal science work. No customers are named in the TechCrunch piece.

The bounty fits a pattern that an AI-first company can choose over hiring. A specialist spending a career on cuprate superconductivity costs the salary of a senior researcher every year. The returns are uncertain. A $200,000 prize paid only on a verified answer buys a lottery ticket across the field. Sridhar tells TechCrunch that his timeline has come forward. He says "I'm more confident now" about seeing some of these problems solved. The bounty is the company's bet on that confidence.

What a research-led AI company can take from this

Two things are worth copying before the five problems are solved.

First, write down the specific scientific or engineering question your company depends on. Put it in enough detail that an outsider can check an answer. Most companies carry a loose list in heads. The act of writing it tightens the question itself. Second, decide which items on that list are worth paying to answer and which are worth hiring to answer. The two routes attract different people and run on different clocks.

Source

• A YC startup is putting $1M behind five unsolved materials science problems for AI, TechCrunch.