Google Kubernetes Engine
GCP Google Kubernetes Engine (GKE) — commitment brief
Data sourced: March 2026. Verify current figures at the Google Cloud Pricing Calculator.
Coverage summary
GKE workloads are covered by compute flexible CUDs — the same spend-based commitment instrument used for Compute Engine. GKE Standard nodes run on Compute Engine VMs and are eligible for compute flexible CUDs (28% at 1-yr, 46% at 3-yr) across all eligible VM families. GKE Autopilot resources are also covered by the compute flexible CUD starting with the expanded program. Legacy GKE Autopilot-specific CUDs are no longer available for new purchases but existing active commitments remain honored.
What is covered: Node compute costs for GKE Standard clusters (via Compute Engine VM CUDs) and GKE Autopilot pod resource costs (vCPU and memory). The compute flexible CUD applies across GKE, Compute Engine, and Cloud Run on the same billing account.
What is not covered: GKE cluster management fee ($0.10/hr per cluster for Standard, charged separately), Persistent Disk storage, network egress, Cloud Load Balancing, node provisioner Spot/preemptible VMs, and GPU nodes not covered by compute flexible CUDs (A2, A3, G2 require resource-based CUDs instead).
CUD types
GKE is covered exclusively by the compute flexible CUD (spend-based commitment at the billing-account level). GKE Standard nodes consume Compute Engine SKUs and benefit from the same 28% (1-yr) and 46% (3-yr) flat-rate discounts as other eligible VM families. GKE Autopilot resources are also eligible under the expanded compute flexible CUD program. The legacy GKE Autopilot-specific spend-based CUD (20% at 1-yr, 45% at 3-yr) is no longer available for new purchases as of 2025.
Machine type / SKU coverage
GKE Standard — General Purpose nodes (N1, N2, N2D, N4, E2, T2D, T2A)
✅ Yes
~28%
~46%
Via Compute Engine VM pricing
GKE Standard — Compute Optimized nodes (C2, C2D, C3)
✅ Yes
~28%
~46%
Via Compute Engine VM pricing
GKE Standard — HPC Optimized nodes (H3, H4D)
✅ Yes
~17%
~38%
H3/H4D flex CUD: 17% (1-yr), 38% (3-yr)
GKE Standard — Memory Optimized nodes (M1, M2, M3)
✅ Yes
no flex discount
~63%
M-series flex CUD: no 1-yr discount; 3-yr only at 63%
GKE Standard — Accelerator Optimized (A2, A3, G2)
❌ No
—
—
Resource-based CUD only; not covered by flex CUD
GKE Autopilot (vCPU and memory)
✅ Yes
~28%
~46%
Covered under expanded compute flexible CUD; legacy Autopilot CUD no longer sold
GKE Standard — Spot / Preemptible nodes
❌ No
—
—
Spot pricing; no commitment instruments
GKE cluster management fee
❌ No
—
—
Flat per-cluster fee, not eligible for CUDs
Sustained Use Discounts (GKE)
GKE Standard nodes running on eligible VM families (N1, N2, N2D, C2, M1, M2) automatically earn Sustained Use Discounts (SUDs) on the portion of compute usage not covered by a CUD. SUDs accumulate throughout the billing month and max out at 20–30% for continuous monthly usage depending on family. The optimal strategy is to purchase compute flexible CUDs for your predictable baseline node consumption, and let uncommitted burst usage earn SUDs automatically. GKE Autopilot workloads and E2/T2D/N4 family nodes do not earn SUDs.
Regional availability
Available in all commercial GCP regions. The compute flexible CUD covering GKE is a billing-account-level spend commitment with no regional restrictions — it applies automatically across all eligible GKE node and Autopilot usage in any region alongside Compute Engine and Cloud Run. However, certain node types have restricted availability: A2 and A3 GPU nodes are limited to specific zones, and M-series memory-optimized nodes are only available in select regions, mirroring Compute Engine constraints.
⚠️ CUD availability varies by machine type and region. Always verify at GCP regions and zones before purchasing.
Archera
Google Kubernetes Engine is within Archera's commitment management scope. Archera manages GKE coverage through Compute Engine compute flexible CUDs — automatically sizing commitments across your combined GKE and Compute Engine footprint, monitoring utilization at the billing-account level, and wrapping commitments in a GRI/GSP to eliminate downside risk on over-commitment.
Sources
⚠️ Discount percentages are approximate and region/machine-type-dependent. Always verify with the Google Cloud Pricing Calculator.
Last updated
Was this helpful?