Get a Demo

AWS Spot Instances

AWS Spot Instances are spare EC2 compute capacity available at a discount versus on-demand pricing, but can be reclaimed by AWS with two minutes of notice when that capacity is needed elsewhere.

How It Works

AWS maintains pools of unused EC2 capacity across instance types and Availability Zones. Spot Instances let you bid for access to that spare capacity. When AWS needs the capacity back, it sends a two-minute interruption notice and terminates or hibernates your instance. The discount reflects the trade-off: you pay less, but you accept the possibility of interruption at any time. Prices fluctuate based on supply and demand within each capacity pool, and AWS publishes current Spot prices by instance type and region.

Why It Matters for Cloud Cost

Spot Instances can reduce EC2 compute costs meaningfully for workloads that are designed to tolerate interruptions. Batch processing jobs, CI/CD pipelines, data analytics tasks, and stateless application tiers are common fits. Workloads that require continuous uptime, such as production databases or latency-sensitive APIs, are poor candidates. Without a deliberate strategy for handling interruptions, teams that run Spot Instances without fault-tolerant architecture can experience data loss or job failures that cost more to recover from than the discount was worth.

Key Characteristics

  • Spot Instances draw from spare EC2 capacity pools and are subject to availability fluctuations at any time.
  • A two-minute interruption notice is provided before AWS reclaims the instance, allowing workloads to checkpoint or drain.
  • Pricing varies by instance type, Availability Zone, and current demand, and changes independently of on-demand rates.
  • Spot capacity is an AWS-specific mechanism. Azure calls its equivalent Azure Spot Virtual Machines, and GCP calls its equivalent Spot VMs (formerly Preemptible VMs).

How Usage AI Handles This

Usage AI focuses on commitment-based savings through Savings Plans and Reserved Instances, which target baseline on-demand compute rather than interruptible workloads. For teams looking to reduce the on-demand cost of their always-on EC2, Fargate, and Lambda usage, the Usage flex savings plan delivers 40 to 60% savings with no interruption risk and no upfront cost.

See how Usage AI saves 30 to 50% on AWS, GCP, and Azure.

Common Questions

Are Spot Instances right for production workloads?

Spot Instances are best suited for fault-tolerant, interruption-aware workloads such as batch jobs, data pipelines, and stateless services. Production workloads that require consistent uptime should use on-demand instances, Reserved Instances, or Savings Plans to guarantee availability.

How does Spot pricing compare to other AWS discount options?

Spot pricing is variable and reflects real-time capacity availability. Commitment-based options like AWS Compute Savings Plans (up to 66% off on-demand) and AWS EC2 Instance Savings Plans (up to 72% off on-demand) offer predictable discounts in exchange for a usage commitment rather than interruption risk.

What happens if my Spot Instance gets interrupted mid-job?

AWS sends a two-minute warning before reclaiming a Spot Instance. Applications built to handle this signal can checkpoint progress, drain connections, or gracefully shut down. Without interruption handling built into the workload, any in-progress work will likely be lost.

Related Terms

AWS Reserved Instances

AWS Reserved Instances (RIs) are a commitment-based pricing model that lets you reserve capacity for AWS services in exchange for discounts of up to 72% compared to on-demand rates.

Azure Cost Management

Azure Cost Management is Microsoft's native toolset for tracking, analyzing, and reducing spend across Azure subscriptions, resource groups, and services.

AWS Compute Savings Plan

An AWS Compute Savings Plan is a flexible pricing model that saves up to 66% vs on-demand rates in exchange for a commitment to a consistent hourly compute spend over one or three years.