Spot VMs are Google Compute Engine instances that run on excess cloud capacity at discounted rates and can be preempted at short notice, designed to run fault-tolerant and interruption-tolerant workloads while reducing compute spend.
Published Oct 12, 2026
| Aspect | Spot VMs (GCP) | AWS EC2 Spot Instances | Oracle OCI Preemptible Instances |
|---|---|---|---|
| Starting Price | Variable by region and machine type; published on the Spot pricing page | Variable by region and instance type; spot prices published and updated frequently | Flat discount model: consistent published discount off on-demand in all regions (per OCI documentation) |
| Appropriate workloads | Fault-tolerant batch/HPC, autoscaled stateless services, interruption-aware ML workflows | Similar workloads; often used with Auto Scaling groups, diversified instance pools, and AWS Spot tooling | Workloads that can tolerate interruptions and prefer predictable discount math |
| Key Strengths | Vendor-reported substantial headline discounts; integration with GCP services such as GKE, Batch, Dataproc, and Dataflow; no fixed maximum lifetime while capacity is available (vendor statement) | Large ecosystem of spot tooling and published pricing mechanics; commonly cited short interruption notice; extensive documentation and advisories | Consistent published discount, and preemption events surfaced through OCI events system |
| Key Weaknesses | Very short preemption notice; excluded from SLA; capacity scarcity for some instance shapes (notably GPUs) has been reported; premium image licensing costs are not discounted | Availability and interruption risk; may require sophisticated diversification strategies and tooling | Smaller ecosystem compared with AWS and GCP; discount magnitude may be lower than some vendor headline claims; interruptions remain possible |