EKS, AKS, GKE, and OKE all charge for the worker nodes you run, and the underlying compute dwarfs the control plane fee, so the cheapest managed Kubernetes is usually the one whose node discount programme and network pricing best fit your workload. Control plane fees are roughly $0.10 per cluster per hour on EKS and GKE standard, free on the basic tiers of AKS and OKE, all indicative; the larger differences come from Savings Plans and Graviton on AWS, Reservations and Hybrid Benefit on Azure, Committed Use Discounts and sustained use on GCP, and flexible shapes with cheaper egress on OCI.
Here is how each layer compares: the control plane fee, the node compute and its discounts, the networking and add on charges, and a worked example of how the same fleet prices out across the four.
Do the control plane fees actually matter?
Each provider charges for the managed control plane differently, but the amounts are small next to the nodes. EKS charges per cluster per hour, around $0.10 indicative, with a higher rate once a version enters extended support. GKE charges a comparable per cluster management fee on Standard, and bills Autopilot by the resources your pods request rather than by node. AKS offers a free control plane tier and a paid Standard tier with a financially backed uptime guarantee, again around $0.10 per hour indicative. OKE runs basic clusters with no control plane fee and charges only for enhanced clusters. For a cluster running dozens of nodes, the control plane is a rounding error; choosing a provider on that fee alone is optimising the wrong line.
Where does the real cost difference come from?
The worker nodes are virtual machines, and the price you pay for them is set by each provider's commitment and architecture programmes. That is where clusters of equal size diverge by double digit percentages.
| Provider | Node discount levers | Standout cost factor |
|---|---|---|
| AWS (EKS) | Savings Plans and Reserved Instances, plus Graviton Arm instances | Graviton plus a Compute Savings Plan over the steady node floor |
| Azure (AKS) | Reservations, the Azure Savings Plan, Hybrid Benefit, Spot | Hybrid Benefit reuses Windows and SQL licences against node VMs |
| GCP (GKE) | Committed Use Discounts and automatic sustained use discounts | Sustained use applies without a contract; Autopilot bills per pod request |
| OCI (OKE) | Universal Credits, flexible compute shapes, Ampere Arm | Per core and per gigabyte flex sizing plus materially cheaper egress |
The pattern repeats across providers: cover the steady floor of nodes with a commitment for the deep discount, run the variable layer on Spot or preemptible capacity where the workload tolerates interruption, and move to Arm shapes where the application supports them. The provider that wins is the one whose programme matches how predictable your node floor is and whether your stack is Arm ready.
What about networking and add on charges?
The charges that surprise buyers sit outside the node price. Load balancers, NAT gateways, and inter zone traffic add up quickly on the hyperscalers, and data egress out of the cluster is billed per gigabyte. OCI prices egress materially lower than AWS, Azure, and GCP, which matters for clusters that serve large volumes of data or replicate across regions. Control plane logging, managed add ons, and observability agents each carry their own cost on every provider. Two clusters with identical nodes can differ by a wide margin once a chatty service mesh, cross zone traffic, and heavy egress are counted, so model the data path, not just the compute.
A worked example
A media company modelled a forty node production cluster across all four providers. The control plane fees differed by under a hundred dollars a month and did not change the ranking. What moved the total was the node layer: moving the steady thirty node floor to Arm shapes with a commitment cut compute by roughly a third on the providers that supported the workload, and the egress heavy delivery path made OCI noticeably cheaper on the data line. The decision came down to Arm readiness and egress volume, not the headline control plane price. Figures are verified against billing data and anonymised; rates are indicative and depend on region, term, and architecture.
Frequently asked questions
Which managed Kubernetes service is cheapest?
Is the EKS or GKE control plane fee worth avoiding?
Does running Arm nodes save money on every cloud?
Choose the cluster on total cost, not the brochure
We model managed Kubernetes cost across AWS, Azure, GCP, and OCI on every layer, control plane, nodes, commitments, and egress, as an independent advisory that takes zero provider commissions and answers only to you. Our guarantee: we reduce your cloud spend or we reimburse our service fee, on a Fixed Fee or a no risk Gainshare basis. Download the cloud cost optimization playbook, read the deeper Kubernetes cost guide and the cross cloud cost optimization guide, and on add on charges see control plane and add on costs. For monthly buyer side analysis, subscribe to The Cloud Spend Navigator.
Put a defensible number on your cloud spend.
No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.
The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.