Spot Instances
What are Spot Instances?
Spot instances are spare cloud compute capacity offered at large discounts compared with on-demand prices, which the provider can reclaim with short notice when it needs the capacity back.
Updated November 2026 · 5 min read
Spot instances let customers use the spare capacity in a cloud provider's data centers at a fraction of the normal price. Providers always keep extra capacity to handle demand, and rather than leave it idle, they sell it at steep discounts. AWS describes savings of up to 90% compared with on-demand rates. The catch is that the provider can reclaim the capacity when it needs it, giving the customer only a short warning, such as two minutes on AWS. Microsoft Azure and Google Cloud offer similar options called spot virtual machines.
Because instances can be interrupted, spot capacity suits workloads that can tolerate interruptions or be restarted easily, such as batch processing, data analysis, testing, rendering and many containerized and stateless applications. For IT financial management, spot instances offer some of the largest discounts available in the cloud, but only for workloads designed to use them, so the savings depend on engineering choices as much as purchasing decisions.
Definition of Spot Instances
Spot instances are compute instances that use unused cloud capacity and are priced below on-demand rates, with no commitment, on the condition that the provider may interrupt them when the capacity is needed elsewhere.
Unlike reserved instances and savings plans, spot instances require no term or upfront payment. Their discount comes from accepting interruption risk rather than from committing to future usage.
How Spot Pricing Works
Spot prices vary by instance type, region and availability zone, and they change over time based on long-term supply and demand. Customers pay the current spot price for the time their instances run.
When the provider needs capacity back, it issues an interruption notice, two minutes on AWS and typically shorter on some other platforms, after which the instance is stopped, hibernated or terminated. Applications must handle this gracefully.
Availability varies. Some instance types in some locations are rarely interrupted, while others are reclaimed often, so diversifying across instance types and zones improves reliability.
Suitable Workloads
Good candidates include batch jobs, big data processing, continuous integration and testing pipelines, image and video rendering, machine learning training that can checkpoint progress, and stateless web or container workloads that run on many instances behind a load balancer.
Poor candidates include single-instance databases, stateful applications that cannot be restarted easily and workloads with strict deadlines that cannot tolerate delay.
Container orchestration and managed services increasingly handle spot interruptions automatically, making spot capacity practical for a wider range of workloads.
Importance of Spot Instances in IT Financial Management
Spot instances offer deeper discounts than commitment options and require no long-term risk. For suitable workloads, they can reduce compute costs dramatically.
They complement other rate optimization tools. Spot capacity is not a replacement for commitments but an addition to them, used for workloads where interruption is acceptable.
An organization's best mix typically combines commitments for steady usage, spot capacity for interruption-tolerant work and on-demand capacity for the rest.
Large Savings Without Commitment
Because spot instances need no commitment, savings begin immediately and stop if usage stops. There is no risk of unused commitments. This makes spot capacity a useful option for workloads whose future is uncertain, where a long commitment would be too risky.
For organizations with large batch or analytics workloads, moving that work to spot capacity can be one of the most valuable single optimization steps.
Spot usage also reduces the amount of steady usage that needs to be covered by commitments, which can change how commitment coverage targets are set.
Engineering Requirements
Spot savings depend on applications being designed for interruption. This may require checkpointing, retry logic, distributing work across many instances and using orchestration tools that replace interrupted capacity automatically.
These changes take engineering effort, so the business case should weigh the savings against the work required and the operational risk.
Because engineering teams make these choices, spot adoption is a good example of usage and architecture decisions that FinOps encourages engineers to own, supporting cloud cost optimization.
How to Use Spot Instances
Using spot instances effectively involves identifying suitable workloads, designing for interruption and monitoring both savings and reliability.
Starting with low-risk workloads, such as test environments, builds confidence before moving to production use. It also allows teams to learn how often interruptions occur for the instance types they use.
Identifying and Preparing Workloads
Review workloads for interruption tolerance, flexibility in timing and ability to run on several instance types. Batch, testing and stateless container workloads are usually the first candidates.
Make applications resilient to interruption by saving progress regularly, handling interruption notices and using managed services or orchestration that can rebalance capacity automatically.
Allow several instance types and availability zones, which reduces the chance of all capacity being interrupted at once. Many providers offer tools that select from a pool of instance types automatically, choosing those least likely to be interrupted.
Monitoring and Mixing
Track spot savings, interruption rates and any effect on job completion times or service levels. Adjust instance choices if interruptions are frequent. Some instance types are far more stable than others. Share these measures with the teams that own the workloads, so they can judge whether the savings justify the trade-offs.
Use a mix of spot and on-demand capacity for workloads that need a guaranteed minimum, such as running a baseline on demand or under commitments and scaling with spot.
Report spot savings separately so their contribution to overall savings is visible. Showing teams the savings from their spot usage also encourages them to look for other suitable workloads.
Examples of Spot Instances in IT Financial Management
The examples below show spot instances in practice.
Both use illustrative figures. The first moves a batch workload, and the second mixes spot with other capacity.
In each, workload design determined how much could be saved. Both started small and expanded.
Batch Data Processing
A company runs nightly data processing jobs costing $60,000 a month on demand. The jobs are split into many small tasks that can be retried if interrupted.
Moving the jobs to spot instances across several instance types reduces costs to about $18,000 a month, a 70% saving. Occasional interruptions add a few minutes to some runs but do not affect morning reporting deadlines. The data team adds monitoring of interruption rates and keeps an on-demand fallback for nights when spot capacity is scarce.
Mixed Capacity for a Web Service
Another company runs a stateless web service on a container platform. It keeps a baseline of capacity covered by savings plans and adds spot capacity for traffic peaks, with on-demand capacity as a fallback.
Peak capacity costs fall by about 60%, and automatic replacement of interrupted instances keeps the service within its availability target. The company extends the same approach to two other stateless services over the following quarter.
Conclusion
Spot instances use spare cloud capacity at large discounts but can be interrupted at short notice. They require no commitment and suit workloads that are fault-tolerant, flexible in timing or able to run across many instances.
In IT financial management, spot instances offer some of the largest available savings, but they depend on applications designed for interruption. Identifying suitable workloads, building in resilience, diversifying instance types and monitoring savings and reliability help organizations use spot capacity safely.
Related terms
Reserved Instances
Reserved instances are a cloud pricing option in which customers commit to using a specific type of compute capacity for one or three years in exchange for a significant discount compared with on-demand rates.
Savings Plans
Savings plans are a cloud pricing option in which customers commit to a fixed hourly spend on compute for one or three years, receiving discounted rates on eligible usage up to that amount.
Commitment Coverage
Commitment coverage is the percentage of eligible cloud usage or spending that is covered by commitment discounts, such as reserved instances, savings plans or committed use discounts, rather than billed at on-demand rates.
Cloud Cost Optimization
Cloud cost optimization is the ongoing practice of reducing cloud spending and improving its value, by removing waste, matching resources to needs, using the right pricing options and designing efficient architectures.