Instance Core running
The vCPU half of a virtual machine. Google bills CPU and memory as separate SKUs, which is why a single VM appears as several lines on your bill and why per-instance costs are not directly visible without joining them back together.
Billed: Per vCPU-hour, priced per machine family and region. RAM is billed as a separate SKU.
What triggers this charge
- Every hour a VM is in the RUNNING state, whether or not it is doing anything.
- Machine family and region, both of which change the rate materially.
- GKE nodes, Dataproc workers, Cloud Composer environments, and managed instance groups — all of which are Compute Engine underneath and bill through this SKU.
How to reduce it
Buy committed use discounts for the steady baseline
One- and three-year commitments are substantially cheaper than on-demand and apply automatically to matching usage. Size them to the floor of your usage, not the average.
Use Spot VMs for interruptible work
Batch jobs, CI runners, and stateless workers survive preemption fine and attract a much lower rate. The Spot SKU is a separate line, so the saving is visible in the bill.
Stop non-production instances out of hours
A dev VM running nights and weekends is idle about 70% of the time. Scheduled stop/start removes most of that with no capacity risk.
Right-size before you commit
Committing to oversized instances locks in the waste for years. Check utilisation first.
Rates
Related charges
Now find out how much of this you are actually paying.
Finitizer breaks your Google Cloud bill down to the SKU, shows how each one has moved since last week, and separates rate changes from usage changes — so you know whether to talk to engineering or to procurement.