Illumina Launches Billion-Cell CRISPR Dataset for AI-Driven Drug Discovery, Built with AstraZeneca, Merck, and Lilly
Illumina has unveiled the Billion Cell Atlas, a large-scale CRISPR perturbation dataset aimed at accelerating drug discovery by enabling AI training on real biological responses of more than 1 billion individual cells across 200 disease-relevant cell lines. Developed under its new BioInsight division, the Atlas is being built in partnership with AstraZeneca, Merck, and Eli Lilly as founding participants. Its launch coincides with CEO Jacob Thaysen’s presentation at the J.P. Morgan Healthcare Conference 2026.
Data is collected from over a billion single cells drawn from a wide range of tissue origins—including lung, liver, brain, pancreas, blood, and muscle—representing disease models in oncology, neurology, immunology, cardiometabolism, and rare genetic disorders.
The initiative is the first phase of a planned 5-billion-cell resource over three years, announced in February 2025, intended to become the most extensive reference of human disease biology to date. Founding partners Merck, AstraZeneca, and Eli Lilly will use the Billion Cell Atlas to train AI models on biologically grounded data, investigate how genetic perturbations affect disease-relevant cell types, and identify actionable targets and biomarkers to inform drug development across complex and underserved indications.
The Billion Cell Atlas is the first dataset released by Illumina’s BioInsight unit, a newly formed division focused on data infrastructure for AI-driven drug discovery. Its scale is enabled by Illumina’s Single Cell 3' RNA prep platform, which allows millions of cells to be profiled in a single run. The project is set to produce 20 petabytes of single-cell transcriptomic data within a year, processed via the DRAGEN hardware-accelerated pipeline and hosted on Illumina Connected Analytics for cloud-based analysis.
Over the past decade, large shared datasets have become a central layer of infrastructure for biology and drug discovery. Population-scale genomic and phenotypic resources such as UK Biobank and multi-million-cell single-cell efforts like Tahoe Therapeutics and the Arc Institute’s Arc Virtual Cell Atlas, and the Human Cell Atlas, among others, are being used as reference maps of human tissues and cell types for disease research and target discovery. For a broader view of how such data platforms are converging into a new infrastructure layer for drug development, see our deep dive The Infrastructure Layer: 30 Platforms Powering Human-Relevant Drug Development.
Topic: Industry Movers
