AI Creates Materials Faster Than Labs Can Test Them

AI systems are now designing millions of new materials, including semiconductors and batteries, according to research published in August 2026 ^[1][3]. MIT researchers found that generative AI can predict candidate materials faster than laboratories can physically build and test them ^[2][3]. This creates a "bottleneck of abundance" where computational predictions outpace experimental capacity ^[1].

Scientists are proposing a "materials AI evidence passport" to standardize how these predictions are verified ^[1]. The CrysVCD model from MIT specifically helps bridge this gap by predicting materials that are more likely to work in real-world conditions ^[2][3]. This new approach aims to reduce the time spent on failed laboratory experiments ^[2].