How do we price hardware we already own?

Either give KubeOn unit rates per node pool, or enter the purchase price, depreciation period and monthly power, space and support. KubeOn spreads that cost over the nodes it covers.

Can on-prem and cloud clusters share one report?

Yes. Allocation, budgets, reports and chargeback treat every cluster the same way, whatever priced it.

What network access does a data center cluster need?

Outbound HTTPS on port 443 to KubeOn. KubeOn is SaaS only, so clusters with no internet egress at all cannot be covered.

On-premises and data centers

Kubernetes cost allocation for clusters that never had a bill

On-prem clusters run on hardware that was paid for years ago. KubeOn prices that capacity with your rate cards or your hardware cost, then allocates it to teams the same way it does in the cloud.

Book a demo See the setup

Pricing

Data readyUpdated 12 min ago

Last 7 days▾

Pricing

Clusters with no cloud bill are priced here: by a rate card for cores, memory and GPUs, or by hardware cost spread over its life, so they appear in allocation, budgets and reports like any cloud cluster.

Priced cluster	Rate card	Nodes priced	Per hour	Per month
dc1-prod	dc1-frankfurt	26 of 26	$38.28	$27,942.95

Rate cards

Rate card

dc1-frankfurt

Colocation contract, Frankfurt · from 2026-07-01 · 2 versions

vCPU-hour: $0.0240
GiB-hour: $0.0032
Storage GiB-month: $0.0500

NVIDIA A100 $1.18/hany other GPU $0.90/h

Hardware you own

Hardware profile

A100 GPU servers$3,612.40 /node/mo

dc1-prod · nodes a100-* · 8 × NVIDIA A100

Depreciation $2,391.25Power and cooling $504.58Colocation $552.00Support $164.57

$114,780 over 48 months · 3.2 kW at $0.16/kWh, PUE 1.35

values.yaml · name the rate card at install Copy

pricing:
  rateCard: dc1-frankfurt

What KubeOn reads

What KubeOn needs in your data center

Your data center

Your rate cardOr amortized hardware cost
Outbound HTTPSNo inbound port
OpenShift, Rancher, Tanzu or k3sThe agent sends a snapshot every hour

KubeOn hub

Fully managed by UnitVane

1Load your rates
2Join them with each snapshot
3Allocate every node's cost
4Show idle capacity

What you get

Cost per team and labelPods, namespaces and teams
Priced with your ratesThe same report as any cloud
Savings to act onAs pull requests, or applied by KubeOn

Pricing

Rate cards per node pool or label (vCPU-hour, GiB-hour, GPU-hour, storage GiB-month), or hardware purchase price over a depreciation period plus power, space and support.

Network

Outbound HTTPS from each cluster to KubeOn, which runs everything else. No inbound port.

Agent identity

A cluster token from KubeOn, in a Kubernetes Secret. Snapshots go to your organization in KubeOn and nowhere else.

Runs onRed Hat OpenShiftRancher RKE2VMware Tanzuk3skubeadm on bare metalMirantis Kubernetes Engine

Setup

Three steps to the first allocation

Cluster inventory appears within the first hour. Allocated costs follow as soon as billing data is in.

01

Define your rates

Write a rate card per node pool, or enter hardware cost and let KubeOn amortize it.

rate-cards/dc1-standard.yaml

# rate-cards/dc1-standard.yamlapiVersion: kubeon.io/v1kind: RateCardmetadata:  name: dc1-standardspec:  currency: USD  effectiveFrom: "2026-01-01"  # Option 1: unit rates, per node pool or node label  rates:    - match: { nodeLabel: "node.kubernetes.io/pool=general" }      vcpuHour: 0.0312      memoryGiBHour: 0.0041    - match: { nodeLabel: "node.kubernetes.io/pool=gpu" }      vcpuHour: 0.0312      memoryGiBHour: 0.0041      gpuHour: 1.18  storage:    - storageClass: ceph-rbd      gibMonth: 0.06  # Option 2: amortized hardware, spread across the nodes it covers  hardware:    - match: { nodeLabel: "kubeon.io/rack=r12" }      purchasePrice: 184000      depreciationMonths: 48      monthly:        power: 1450        space: 900        support: 610

02

Install the agent

The same chart, with the rate card named in the values.

kubeon-agent.yaml

# kubeon-agent.yaml (on-premises)clusterName: dc1-prodkubeon:  existingSecret: kubeon-agent-tokenpricing:  rateCard: dc1-standard   # the rate card that prices this cluster's nodes

03

Allow outbound HTTPS

The agent sends snapshots to KubeOn over port 443. Nothing else in the data center needs to change.

Helm

helm upgrade --install kubeon-agent oci://registry.kubeon.io/kubeon/charts/kubeon-agent \  --version 0.6.0 \  --namespace kubeon --create-namespace \  -f kubeon-agent.yaml

Highlights

Built for how data centers are priced

Rate cards

Unit rates per node pool or label, effective-dated so a new rate applies from its start date and earlier days keep theirs.

Hardware amortization

Purchase price over a depreciation period, plus power, space and support.

GPU pricing

GPU-hour rates for accelerator nodes, with idle GPU time reported.

Hybrid views

On-prem and cloud clusters in the same allocation, budget and report.

Outbound only

One HTTPS destination, no inbound port, no Service or Ingress.

Self-hosted models

GPUs priced by rate card, with idle time shown beside each one.

FAQ

On-premises questions

Other platforms

Amazon EKS and AWS

CUR 2.0 with split cost allocation data, read in place with Amazon Athena. Agents send to KubeOn with a cluster token, with no IAM role.

Learn more

Azure AKS

Cost Management exports read with a managed identity. Agents send to KubeOn with a cluster token.

Learn more

Google GKE

Cloud Billing export in BigQuery with GKE cost allocation labels. Agents send to KubeOn with a cluster token.

Learn more

Let's get started

A 30-minute walkthrough on your own billing data, for your clouds, your clusters and your model providers.

Book a demo Plan an evaluation