Tempus AI announced September 13 an initiative to build a research platform containing 100,000 whole genomes linked to longitudinal clinical information, with a long-term goal of reaching one million genomes. The dataset will be the first de-identified multimodal whole-genome sequencing (WGS) resource built around disease populations and patient outcomes, designed specifically for AI model development.
The company said the initial dataset is already available through an Early Adopter Program, with general availability planned for mid-2027. Tempus will onboard additional members in waves as the dataset grows.
Whole genomes paired with outcomes
Existing population-scale genome programs draw largely from general populations and do not link genomic data with longitudinal disease and treatment outcomes. Tempus has built one of the largest multimodal real-world oncology databases and used it to support hundreds of drug development decisions. The company has also spent years structuring this data for AI-derived insights, building its own oncology foundation models and supporting model development for other AI researchers.
Moving beyond targeted gene panels into WGS, Tempus aims to create a deeper understanding of how the genome influences disease progression and treatment response. The new dataset will be integrated into the company's existing de-identified multimodal data environment, where researchers can access genomic information alongside clinical histories, imaging, pathology, and patient outcomes.
Through Tempus Lens, researchers and model developers can analyze the data and build and validate AI models without moving datasets between systems. The resource is built to pair whole-genome data with longitudinal information and a computational system for pre-training and post-training workflows - a model-ready environment for AI for Healthcare research.
What leadership says
"A large dataset is only valuable if you can turn it into insight," said Eric Lefkofsky, Founder and CEO of Tempus. "Tempus has spent the last decade building the infrastructure to connect diagnostics, multimodal clinical data and AI at scale. Adding whole genome data linked to longitudinal outcomes makes that platform even more powerful and gives researchers a richer foundation to build AI models, generate new insights and ultimately improve patient care."
The initiative builds on Tempus' existing position in real-world oncology data. The company has structured its data environment for AI for Science & Research, with the goal of enabling model developers to work with genomic and clinical data in a single system rather than moving files between platforms.
Why this matters for healthcare professionals
For clinicians, researchers, and healthcare data teams, this dataset represents a shift in how genomic data is organized for analysis. Instead of population surveys disconnected from clinical records, whole-genome sequences will be paired with treatment histories and outcomes. That structure could make it easier to ask questions about why certain patients respond to therapy and others do not - and to build predictive models that reflect real clinical trajectories rather than abstract genetic variation.
Your membership also unlocks:
