Based on a Quality of Service (QoS) profile, administrators can define thresholds for the vGPU per workload. NVIDIA A100 has become the gold standard for training AI models in the cloud.