AKS Pricing Guide: Costs Explained & Examples Azure Kubernetes Service (AKS) takes control-plane management off your plate. But "free AKS" doesn't mean a free cluster. Worker nodes, disks, load balancers, logs and other Azure services are billed separately, and they usually make up most of the invoice.

The final bill depends on control-plane tier, node count, VM size, region, workload behavior, storage, networking, monitoring and discount model. Many teams budget for the cluster they designed, then pay for the one that actually runs.

This guide covers the current tiers, each cost component, two worked examples, a budgeting process and the charges people miss. All prices are USD, East US, public pay-as-you-go retail rates. Rates change, so confirm them on Microsoft's pricing pages and in the Azure Pricing Calculator for your region before committing.

Key Takeaways

  • The AKS control plane is $0 on Free, or a per-cluster hourly fee on Standard and Premium.
  • Worker-node VMs and attached services drive most of the monthly bill.
  • Low-cost setups suit dev, test and interruptible work; production justifies higher spend for uptime.
  • Budget from real utilization, not the initial cluster design.

How Much Does AKS Cost? (Pricing Overview)

AKS has no single monthly price. Calculate the control-plane charge separately from worker-node compute, storage, networking, monitoring, security and other services.

Typical AKS Cost Range

Instead of one universal range, use three planning profiles:

  • Entry-level dev/test: a small cluster, low-cost Linux nodes, little persistent storage, minimal add-ons.
  • Standard production: multiple nodes, a paid control-plane tier, persistent volumes, load balancing, monitoring and availability needs.
  • Advanced or enterprise: several node pools, high-memory or GPU VMs, heavy traffic, multi-zone design, extensive observability.

Each profile should be priced with current list rates, a stated region and written assumptions. Estimates typically cover compute, disks, and networking; they exclude taxes, support plans and usage you can't predict, such as bursts of egress or log volume.

AKS Control-Plane Tiers

Tier East US, per cluster-hour Intended use API-server SLA
Free $0 Dev/test, learning, evaluation Best effort, no financially backed SLA
Standard $0.10 Production requiring an SLA 99.95% with availability zones; 99.9% without
Premium $0.60 Production plus long-term support 99.95% with availability zones; 99.9% without

According to Microsoft's tier documentation, the SLA covers the API server, not your workloads. Premium adds Long Term Support: one community-support year plus one LTS year, roughly 24 months from GA, and you must select the LTS plan explicitly.

The Free tier is separate from Azure's free-account offer ($200 for 30 days plus 12 months of selected services). That offer has its own eligibility and expiry rules.

Standard and Premium fees apply per cluster, not per node. One cluster at $0.10 per hour is about $73 a month at 730 hours. Fifteen dev, staging and preview clusters on Standard add up quickly.

Cost Breakdown of AKS

Control plane (recurring). The tier fee above. AKS manages the API server and cluster management; higher tiers add SLA and support benefits.

Worker-node compute (recurring). Node pools run on Azure VMs billed by family, size, OS, region, count and runtime.

VM Linux per hour Windows per hour
D2s v5 (general purpose) $0.096 $0.188
E4s v5 (memory-optimized) $0.252 $0.436
B2s (burstable) $0.0416 $0.0496

B-series VMs can't serve as AKS system pools.

Persistent storage (recurring or usage-based).

  • Premium SSD P10: $19.71 per disk-month
  • Standard SSD E10: $9.60 per disk-month, plus transactions
  • Premium SSD v2: $0.081/GiB-month, with extra charges for IOPS above 3,000 and throughput above 125 MB/s
  • Snapshots bill separately for used data; Azure Files is a separate share service

Persistent disks bill whether or not data fills them. Ephemeral storage disappears with the pod or node.

Networking and data transfer (usage-based).

  • Standard Load Balancer: $0.025/hour for the first five rules, $0.01 per extra rule, $0.005/GB processed
  • Standard public IPv4: $0.005/hour
  • NAT Gateway: $0.045/hour plus $0.045/GB
  • Internet egress: first 100 GB/month free, then $0.087/GB for the next 10 TB
  • Cross-region within the US: $0.02/GB outbound
  • Inter-AZ traffic in the same region: no data-transfer charge

Monitoring, security, registry and add-ons.

  • Log Analytics: $2.30/GB ingested at the standard tier
  • Managed Prometheus: $0.16 per 10M samples ingested
  • Defender for Containers: $0.00941/vCore-hour
  • Container Registry: $0.1666/day Basic, $0.6666 Standard, $1.6666 Premium
  • Application Gateway Standard v2: $0.20/gateway-hour plus $0.008/capacity-unit-hour

AKS cost components and Azure service charges breakdown

Key Factors That Affect AKS Cost

Rate optimization and usage optimization are different jobs. A cheaper VM rate won't help if the cluster runs idle capacity or racks up avoidable spend on logs, disks, and traffic.

Cluster Size, Workload Demand, and Node Pools

Node count, pod requests, replica counts, scaling limits, and separate workload pools all drive compute spend. The system pool needs at least 2 nodes, and Microsoft recommends 3 for production, each with at least 4 vCPUs and 4 GB of memory.

A cluster sized for peak demand pays for idle capacity the rest of the time, especially when a node pool's minimum count is set too high.

VM Family, Size, and Operating System

Match the family to the workload, not the hourly price alone:

  • General purpose for balanced web and API workloads
  • Memory-optimized for caches and in-memory data
  • Burstable for light, spiky dev workloads (not system pools)
  • GPU for ML and rendering

Windows nodes cost more: a D2s v5 runs $0.188 per hour versus $0.096 on Linux. Eligible Azure Hybrid Benefit can reduce Windows licensing costs. Also check that your SKU is available in your chosen region.

Scaling and Utilization Strategy

  • Horizontal Pod Autoscaler changes replica counts
  • Vertical Pod Autoscaler adjusts pod CPU and memory requests
  • Cluster Autoscaler adds or removes nodes based on pending pods
  • Node auto-provisioning (Karpenter-based) picks VM SKUs and quantities for pending workloads

Base the settings on observed demand and your service levels. Aggressive scale-down can cause thrashing; conservative settings leave nodes idle.

Region, Availability, and Network Architecture

Disk and VM prices vary across US regions, and zone-redundant storage costs more than LRS. Latency, residency, and compliance can rule out the cheapest region. Availability zones raise the API-server SLA to 99.95%, and same-region inter-AZ transfer is free. Cross-region replication, public exposure, and high egress are where network costs climb.

Discounts, Commitments, and Optional Services

Option Discount Trade-off
Reservations Up to 72%, 1 or 3 years Locked to VM family and region
Savings Plan Up to 65%, 1 or 3 years Hourly commitment billed regardless of usage
Spot VMs Up to 90% Secondary pools only; can be evicted

AKS reservations savings plans and Spot VM comparison

Rightsize before committing. Add-ons such as monitoring, registry, ingress, and security can shift the total even when the control-plane fee stays the same.

Low-Cost vs High-Cost AKS: What's the Difference?

  • Performance: budget setups use smaller or burstable nodes and tight scaling; premium ones add larger, specialized or redundant capacity.
  • Reliability: low-cost fits non-critical work; higher tiers and zone-redundant designs give stronger availability and extended support.
  • Long-term value: the cheapest hourly rate gets expensive when it causes throttling, downtime, manual intervention, excess egress or repeated rebuilds.

How to Estimate the Right AKS Budget

Start with workload requirements, not a target monthly price. Write down every assumption before opening the Azure Pricing Calculator.

Inputs to Collect Before Estimating

  • Target US region and number of clusters
  • Environment type and expected uptime
  • Node-pool roles, VM family and OS
  • Average and peak pod demand
  • Storage capacity, traffic and egress
  • Observability, security, backup and availability needs
  • Which workloads are steady, bursty, interruptible or non-production

Building the Estimate

Add AKS management, configure each node pool separately, then add disks, load balancers, IPs, bandwidth, monitoring, registry and security. Use this formula:

Monthly budget = control-plane fees + (node hours × VM rate) + storage + networking and egress + monitoring and add-ons + taxes or support

Worked Example 1: Dev/Test Cluster

Assumptions: East US, Free tier, 2 D2s v5 Linux nodes, 10 hours a day on 22 workdays (220 hours), two 128 GiB Standard SSD disks, one public IP.

Line item Calculation Monthly
Control plane Free tier $0.00
Nodes 2 × 220 h × $0.096 $42.24
Disks 2 × $9.60 $19.20
Public IP 730 h × $0.005 $3.65
Total about $65

This excludes transactions, egress, logs and taxes. Run the nodes 24/7 and compute climbs to about $140 (2 × 730 × $0.096), so scheduled shutdowns cut it by more than half. Stop or scale-to-zero policies help further, though system pools can't be stopped individually.

Worked Example 2: Production Cluster

Assumptions: East US, Standard tier, 3-node system pool and 3-node app pool (all D2s v5, 730 hours), three P10 disks, one load balancer with 5 rules, one public IP, 200 GB egress, 20 GB of logs.

Line item Calculation Monthly
Control plane 730 × $0.10 $73.00
Nodes 6 × 730 × $0.096 $420.48
Disks 3 × $19.71 $59.13
Load balancer 730 × $0.025 $18.25
Public IP 730 × $0.005 $3.65
Egress 100 GB × $0.087 $8.70
Logs 20 GB × $2.30 $46.00
Total about $629

Swap to E4s v5 and the node line becomes 6 × 730 × $0.252 = $1,103.76. Double the log volume and logging reaches $92. A reservation or savings plan lowers the node line, but only after you've rightsized.

AKS production cost scenario comparison by VM and logging choice

Validate the Estimate After Deployment

Compare the calculator estimate with Azure Cost Management data by subscription, resource group, cluster, node pool and service. Standard and Premium clusters can also use AKS cost analysis for namespace-level allocation.

Actual spend shifts with autoscaling, pod scheduling, disk creation, log volume, traffic, failed jobs, snapshots and temporary resources. Review the budget regularly.

What Most People Miss When It Comes to AKS Cost

Focusing only on the control-plane fee gives a misleading budget. Compute, storage, network and observability usually dominate.

Easy-to-overlook charges:

  • System-node overhead (the 2-node floor)
  • Unattached disks, which keep billing after VM deletion
  • Persistent volumes retained after workload deletion (reclaimPolicy: Retain keeps the Azure disk after the claim is removed)
  • Snapshots
  • Public IPs and extra load balancer rules
  • Egress and high-volume container logs

An estimate reflects provisioned capacity; the bill reflects what actually ran. A CNCF microsurvey of just over 100 respondents found 70% cite overprovisioning as an overspend factor, though it isn't AKS-specific. Reviewing idle and oversized block storage belongs in your FinOps governance.

AKS hidden charges and overprovisioning risk overview

For enterprise teams, Lucidity offers an autonomous, multi-cloud block-storage management platform. Its Lumen product gives visibility into every volume and flags four idle-disk types: unattached, reserved, unmounted and zero-I/O.

Lucidity's internal analysis found these can account for up to 70% of unused block-storage spend. AutoScaler expands and shrinks eligible disks with zero downtime. The platform complements, not replaces, compute rightsizing, Azure billing controls and workload autoscaling.

Conclusion

AKS pricing is a control-plane tier plus the Azure infrastructure and services the cluster consumes. A trustworthy budget accounts for node pools, utilization, storage, networking, monitoring, region, availability, and commitments.

The right cost balances reliability, performance, operational effort, and long-term utilization, not the lowest headline price.

Frequently Asked Questions

What is the pricing for Azure Kubernetes Service (AKS)?

The control plane is Free ($0), Standard ($0.10 per cluster-hour) or Premium ($0.60 per cluster-hour) in East US. Nodes and associated services bill separately, so check current rates for your region.

How much does Kubernetes cost?

Kubernetes itself is open-source software. Self-managed clusters still incur compute, storage, networking, and operations costs; AKS adds a tier-based management option, with infrastructure billed separately.

Is AKS free to use?

The Free tier removes the control-plane charge. Worker nodes, storage, networking, monitoring and other Azure resources are still billed.

What are the main costs in an AKS cluster?

Control-plane fees, worker-node VMs, persistent storage, networking and egress, monitoring, security, container registry and add-ons. Nodes are typically the largest line.

How can I reduce my AKS costs?

Rightsize nodes, use autoscaling, schedule shutdowns for non-production, and run interruptible work on Spot. Commit to reservations after rightsizing, clean up unused storage and cap log volume.

Does AKS pricing include worker nodes?

No. Worker nodes are Azure VMs billed separately from the control plane, by VM type, size, OS, region, runtime and purchasing model.