Nvidia does not publicly report a consolidated "Compute Utilization" percentage in its annual or quarterly financial filings. This metric is not a standard financial disclosure.
The figures provided are industry estimates for Nvidia GPUs in data centers running AI and high-performance computing (HPC) workloads.
GPU utilization is a real-time operational metric that varies significantly based on the specific workload (e.g., AI training, inference), application, and end-user.
Estimates from industry sources and analysts for AI workloads in major cloud data centers typically place average utilization rates in the 60% to 70% range. A 2021 Nvidia presentation showcased a specific, highly-optimized research cluster reaching 90% utilization, highlighting it as a key goal.
