V100 compute grants are open for applications

Applications are open for the V100 compute track. This post covers how the process works so applicants know what to expect.

The V100 track is grant-led

V100 capacity is allocated through review rather than self-serve purchase. Applicants describe the workload, the research question, and the expected duration, and the review looks at fit rather than ability to pay.

Enterprise B200 capacity works differently — that track is quote-based and handled through direct intake.

What the review looks at

• The research question. What is being investigated, and why does it need GPU compute?
• Workload shape. Expected runtime, GPU class, and whether the job checkpoints.
• Timeline. When the capacity is needed and for how long.
• Open science commitment. What will be published, and under what terms.

The last point carries real weight.

What applicants commit to

Successful applicants agree to publish results, methodology, and where possible the code needed to re-execute the analysis. Negative results count — an experiment that did not work is a contribution when the method is documented.

Timeline

Applications are reviewed in batches. Expect a response within a few weeks. Incomplete applications are held for clarification rather than rejected.