CloudBolt Launches GPU Optimization and Bill-Accurate GPU Allocation for Kubernetes at KubeCon + CloudNativeCon North America
The latest release of StormForge by CloudBolt makes GPU optimization generally available and adds bill-accurate GPU
Press Release Disclaimer: This is a press release distributed through the XPR Media network. It has not been independently verified by our newsroom.
ROCKVILLE, Md., Nov. 10, 2026 (GLOBE NEWSWIRE) — CloudBolt Software, a recognized leader in cloud and AI infrastructure, today announced general availability of GPU optimization in StormForge, its Kubernetes optimization product, along with bill-accurate GPU cost allocation, node pool cost reporting, and new reliability safeguards. Platform and FinOps teams can now see what each GPU workload and each node pool costs on the cloud bill, and get recommendations for reducing waste on those GPU nodes. StormForge has also added reliability–focused recommendations that account for CPU contention, namespace policies and Java memory settings. CloudBolt will demo the release at KubeCon + CloudNativeCon North America at booth #207.
GPU optimization, in early access since September, is now generally available, giving teams a per-workload view of GPU use and waste. This release also puts the bill behind that view. Seeing that a GPU workload is wasteful starts the conversation, but knowing what it costs, which team owns it, and which node pool it runs on is what gives teams the confidence to drive costs down. StormForge analyzes usage patterns on the node and recommends better-fit node types to reduce waste.
“GPU spend is growing faster than most teams’ ability to explain it,” said Yasmin Rajabi, chief operating officer at CloudBolt. “Platform teams need a number that matches the invoice before they can have a real conversation to drive change and they need automation they trust enough to stand behind. This release gives them both, whether that’s at the workload level or down to the individual node level, without overriding the guardrails they’ve already set.”
GPU Cost Allocation From the Bill
StormForge now allocates billed GPU spend to the namespaces, workloads and labels that used it, including on shared and time-sliced GPUs. Costs come from the cloud provider’s bill, not list-price estimates, so the numbers hold up in a chargeback conversation with finance. Paired with GPU optimization, teams see in one place where GPU capacity is wasted, who is causing it, and how to reduce it.
Node Pool Cost Reporting
StormForge already shows container-level, bill-accurate costs across the cluster. The new experience brings node pool cost and idle reporting together with Reserved Instance and Savings Plans benefits attributed in billing data. Node pool reporting gives insight to the cost of the machines underneath. Users can drill from every cluster down to a single node pool and instance type, and see what each pool costs, how much of it sits idle, how much runs on Spot, and how much Reserved Instances and Savings Plans cover. one view connects infrastructure cost, idle capacity, and purchasing coverage, so engineering and finance can work from the same data when deciding whether to consolidate, shift to Spot or buy commitments from the same data.
Rightsizing That Improves Reliability
StormForge recommendations now account for three conditions that commonly hinder autonomous rightsizing:
- CPU contention: When nodes run short on CPU, StormForge detects the pressure and adjusts recommendations, so latency-sensitive workloads get the CPU they need.
- LimitRange policies: StormForge reads the LimitRange constraints platform teams set on each namespace and only recommends values that comply with them.
- JVM heap: Improved memory recommendations for Java workloads, reducing out-of-memory errors.
Availability
To get started, visit https://www.cloudbolt.io/stormforge/gpu-optimization/. To see it live, visit CloudBolt at booth #207 at KubeCon + CloudNativeCon North America in Salt Lake City, Nov. 9-12.
About CloudBolt
CloudBolt Software helps enterprises and service providers turn cloud complexity into operational advantage by bringing intelligent automation, governance, and optimization to public, private, multi-cloud and AI infrastructure. CloudBolt empowers platform engineering, IT operations, and finance teams with the insight and automation they need to run cloud environments at scale, efficiently, continuously, and with confidence. Learn more at www.cloudbolt.io.
Media Contact:
Caroline Statile
Scratch Marketing + Media for CloudBolt
CloudBolt@scratchmm.com
Media gallery
