Compute Engine rightsizing
Machine types matched to real CPU and memory utilization, including custom machine type opportunities.
Product · For GCP
GCP spend, rightsized automatically. Shipped as a pull request.
Jetscale AI connects read-only to your Google Cloud projects and continuously finds Compute Engine, storage, and GKE opportunities, each one delivered as a PR-ready Terraform diff against your own repo.
Compute Engine rightsizing Committed use discounts GKE tuning
What gets optimized
Every corner of your GCP bill, covered.
The Analyzer Agent scans across services, not just Compute Engine, so waste hiding in storage, clusters, and commitments gets found too.
Persistent disk optimization
Oversized or unattached persistent disks, and disk-type mismatches (SSD where standard would do) found automatically.
GKE tuning
Node pool sizing and pod resource requests re-tuned to actual usage, the same evidence-backed approach used across every cluster Jetscale connects to.
Committed use discounts
Commitment coverage checked against actual usage patterns, so steady-state workloads aren't paying on-demand rates.
Cloud Storage tiering
Objects moved to Nearline, Coldline, or Archive based on real access patterns, not a one-time policy guess.
Spot & Preemptible VMs
Eviction risk simulated before recommending a move off on-demand pricing, the same OptimizeSpot logic used on AWS and Azure.
Pre-verified savingsConnect read-only
A dedicated read-only identity. Nothing more.
Google Cloud connections follow the same pattern used across every provider Jetscale supports: a dedicated read-only identity, scoped to the cost, billing, and resource-inventory data Jetscale needs, with no permissions to modify or delete infrastructure. Every new connection goes through a validation step before it goes live, confirming the permissions are correctly scoped and nothing more.
See it in action
From finding to merged Terraform.
pull/307 · rightsize-analytics-instance.tf
resource "google_compute_instance" "analytics_worker" {- machine_type = "n2-standard-8"+ machine_type = "n2-standard-4" zone = "us-central1-a" project = "prod-analytics"}
OptimizeSpot · eviction-risk simulation
workloadbatch-etl / us-central1
simulation30d usage replayed against Spot & Preemptible pools
eviction risk
low
Safe to migrate: On-demand → Spot VM
Running more than GCP?
Jetscale AI covers your whole footprint.
For AWS
EC2/EBS rightsizing, S3 lifecycle tiering, Spot via OptimizeSpot, and VPC endpoint gaps.
Learn more →For Azure
VM and managed disk rightsizing, Blob tiering, and Reserved Instance alignment.
Learn more →Cloud waste
The full story on how waste is found, ranked, and why it keeps coming back without continuous monitoring.
Learn more →Get started
See your number before you spend a dollar.
Read-only connection. Evidence-backed findings. PR-ready fixes. Invoiced only when savings land.