Finding a material with the right properties, such as a better battery component or a lighter alloy, has traditionally meant slow trial and error. AI models trained on known materials predict the properties of combinations that have never been made, letting researchers shortlist the most promising candidates.
This narrows a vast search space so laboratory effort focuses where success is likeliest, speeding up fields from energy storage to electronics. The predictions guide rather than replace experiments, because a promising calculation must still be synthesised and tested, and unusual chemistries remain hard to model, but the approach is changing how discovery begins.