Harell Data will run training and inference workloads using Nvidia A100 and Hopper GPUs
CoreWeave announced on September 23, 2026, a multi-year agreement with Harell Data. Under the deal, Harell Data will deploy its platform on CoreWeave Cloud to run AI model training, fine-tuning, and inference workloads.
The partnership supports Harell Data as it scales its secure research platform for proprietary biotechnology datasets. The platform operates by bringing AI models directly to the data rather than moving datasets to models, ensuring raw scientific data never leaves the environment. Model training runs will execute on CoreWeave using Nvidia A100 and Nvidia Hopper GPUs.
Harell Data operates a marketplace model where compute is metered per run. Data owners receive a revenue share on every training job run against their datasets, while model developers can offer their trained models under a Models as a Service structure to collect fees per usage.
Jon Jones, chief revenue at CoreWeave, stated that biotechnology research teams require infrastructure that allows them to concentrate on scientific discovery rather than underlying hardware systems. CoreWeave, founded in 2017, completed its public listing on Nasdaq in March 2025.
Newsletter
Markets in your inbox, weekly
Latin America-focused analysis, investment themes and the week in finance.
Keep reading