Google DeepMind and the Wellcome Sanger Institute launched an AI genomics consortium to build large genomic datasets intended to train predictive biological machine learning models. Google.org and DeepMind will commit $5 million per year over five years, with the consortium designed to create shared training resources for broader life-science research communities. The initiative builds on prior partnerships, including an AI genomics fellowship and efforts to strengthen AI capability in lower- and middle-income countries. Key takeaway: the consortium formalizes a dataset-building strategy aimed at accelerating AI-driven discovery rather than restricting efforts to model benchmarking alone.