AWS Fargate Spot
AWS Fargate Spot is a pricing option for AWS Fargate that runs containerized tasks on spare compute capacity at up to a 70 percent discount compared to standard Fargate pricing, in exchange for the possibility that AWS may interrupt tasks…
Definition
AWS Fargate Spot is a pricing option for AWS Fargate that runs containerized tasks on spare compute capacity at up to a 70 percent discount compared to standard Fargate pricing, in exchange for the possibility that AWS may interrupt tasks with a two-minute warning when that capacity is needed elsewhere.
Overview
AWS Fargate is a serverless compute engine for containers that removes the need to provision or manage EC2 instances for running ECS or EKS workloads. Fargate Spot extends this model by letting customers run interruptible tasks on spare AWS capacity at a substantial discount, following the same underlying economic model as EC2 Spot Instances but abstracted away from any instance management, since Fargate itself has no visible servers to interrupt — only the tasks running on it. When AWS needs to reclaim Spot capacity, it sends a two-minute interruption notice to the affected task via Amazon EventBridge before stopping it, giving applications a short window to gracefully drain connections, checkpoint state, or hand off work. Because interruption is possible at any time, Fargate Spot is designed for fault-tolerant, stateless, or checkpointable workloads rather than long-running stateful services that cannot tolerate sudden termination. Fargate Spot is configured per capacity provider within an ECS cluster, and tasks can use a capacity provider strategy that mixes standard Fargate and Fargate Spot — for example, running a baseline of on-demand Fargate tasks for guaranteed availability while scaling additional capacity on Fargate Spot to absorb traffic spikes cost-effectively. This blended approach is a common pattern for balancing cost savings against reliability requirements. Typical suitable workloads include batch processing jobs, CI/CD build runners, image and video processing pipelines, and other tasks that can be retried or resumed if interrupted, where the discount meaningfully reduces total compute spend without threatening availability of user-facing services.
Key Features
- Runs Fargate tasks on spare AWS capacity at up to ~70% discount
- Provides a two-minute interruption warning via Amazon EventBridge
- No EC2 instances to manage — Fargate abstracts the underlying servers
- Configured as a capacity provider within an ECS cluster
- Can be blended with standard Fargate using capacity provider strategies
- Best suited for fault-tolerant, stateless, or checkpointable workloads
- Available for Amazon ECS; not identical to EC2 Spot pricing mechanics
- Requires no bidding — pricing is a fixed discount off standard Fargate rates
Use Cases
Alternatives
Frequently Asked Questions
From the Blog
Commitment discounts and spot capacity explained
Decide what to commit and what to run on spare capacity: how commitment discounts work, which workloads tolerate interruption, and how to size a commitment.
Read More AI & TechnologyHow to Spot Overfitting Early in a Fine-Tuning Run
Overfitting shows up as validation loss rising while training loss keeps falling, and as outputs that reproduce training examples verbatim. This covers what to watch during a run, how to build a validation set that can detect the problem, and when to stop, roll back or fix the data instead.
Read More Data ScienceHow to read a Spark physical plan and spot the expensive step
Read a Spark physical plan by answering three questions rather than parsing the whole tree: where are the exchanges, did the filter reach the scan, and which join did the optimiser pick. Those three answers account for most of the cost difference between a fast query and a slow one.
Read More Data ScienceOverfitting and Regularization Explained
Overfitting is when a model memorizes training data instead of learning general patterns. Regularization fights it. Learn to spot, measure, and prevent both.
Read More