NVIDIA Data Science Academic Grant Program

About this program

NVIDIA Corporation invites proposals for innovative data science research that leverages CUDA‑X libraries (e.g., cuPy, cuDF, cuML, cuGraph, cuOpt) across themes such as data processing and analytics, operational research and routing, statistical methods, graph analytics and GNNs, and visual analytics. Selected principal investigators may receive GPU compute resources (up to 30,000 H100 80 GB hours or equivalent, maximum 8 concurrent GPUs) or up to 2 DGX Spark units, with GPU hours expiring six months after award and final allocations set by an NVIDIA awards panel. Eligible applicants are full‑time faculty at accredited institutions that award Ph.D. research degrees; postdocs and graduate students must be sponsored by a qualifying faculty member, individuals may submit one proposal per quarter (max four per year) and may receive no more than one award per calendar year. Proposals must follow NVIDIA’s template and be no longer than four pages (appendices excluded), and awardees must acknowledge NVIDIA support and report publications, presentations, open‑source releases, and related activities via the NVIDIA grant portal.