Parse Biosciences and Tahoe Therapeutics announced a collaboration to generate a 300 million‑cell single‑cell perturbation atlas to train artificial‑intelligence models. Parse will produce the sequencing using its GigaLab services while Tahoe contributes genetically and chemically perturbed samples designed to map cellular responses across tissues and disease states. The atlas aims to power virtual cell models that predict therapeutic outcomes and to accelerate discovery of drug targets and biomarkers. The project builds on prior 100M‑cell efforts and reflects strong investor and industry momentum to scale biological datasets for foundation‑model training. Leaders said the resource will improve predictive AI for drug discovery and translational biology, while highlighting the technical and cost challenges of producing such large single‑cell datasets.