The Arc Institute inaugurated the Virtual Cell Challenge, a global AI competition encouraging development of machine learning models that predict how cells respond to gene perturbations. Utilizing a unique dataset of single-cell transcriptomics involving 300,000 human embryonic stem cells with 300 genetic modifications, the challenge tests models on their ability to generalize beyond training data with limited samples, reflecting authentic biological complexity. This initiative aims to accelerate AI adoption in biology by fostering model standardization and providing high-quality benchmarking, ultimately catalyzing drug discovery and functional genomics research.