Researchers from Microsoft Research AI unveiled BioEmu, a generative deep learning model capable of rapidly predicting protein conformational ensembles with high accuracy. Published in Science, BioEmu simulates equilibrium protein behavior thousands of times faster than traditional molecular dynamics, capturing diverse functional motions relevant to drug development. Its property-prediction fine-tuning algorithm allows alignment with experimental data even when structural information is limited. BioEmu complements, but does not replace, molecular dynamics, marking a breakthrough for high-throughput protein research and therapeutic exploration.