CloudBolt Adds Ability to Optimize GPU Consumption by Kubernetes Workload
Cloud Native Now, Thursday, September 24th, 2026
CloudBolt's new StormForge capability tracks GPU consumption per Kubernetes workload to cut cost and overprovisioning.
CloudBolt has launched a StormForge capability giving granular visibility into GPU utilization across Kubernetes clusters by tracking consumption at the individual workload level.
Written by Mike Vizard, the article explains the solution addresses limits in NVIDIA's Data Center GPU Manager by mapping GPU processes directly to Kubernetes pods, letting IT teams track resource consumption and costs by cluster, namespace, and workload.
This is especially valuable for time-sliced GPUs, where standard monitoring tools fall short. By offering detailed cost breakdowns and optimization recommendations, the platform helps organizations reduce overprovisioning as AI workloads increasingly run on Kubernetes infrastructure.