An AWS Savings Plan is a pricing model that gives you a lower price than On-Demand in exchange for committing to a consistent amount of usage, measured in dollars per hour, for a one or three year term. Usage above your commitment is billed at On-Demand rates.

The main types

PlanWhat it coversHeadline saving (per AWS)
Compute Savings PlansEC2 regardless of instance family, size, Region, OS or tenancy, plus Fargate and LambdaUp to 66%
EC2 Instance Savings PlansOne instance family in one Region (for example M5 in N. Virginia)Up to 72%
Database Savings PlansEligible database services such as Aurora and RDS, across RegionsUp to 35% (serverless); AWS's launch post gives up to 20% for provisioned instances

Database Savings Plans, launched in December 2025, are available for a one-year term with no upfront payment. Compute and EC2 Instance plans offer one or three years, and you can pay All Upfront, Partial Upfront or No Upfront.

Limits worth knowing

  • They are a commitment. A plan can't be cancelled during its term. AWS does allow a return of a plan with an hourly commitment of $100 or less, bought in the past seven days and in the same calendar month, per its FAQ.
  • They don't reserve capacity. If you need guaranteed capacity, for example in a recovery Region, use On-Demand Capacity Reservations; your Savings Plan discount applies to them.
  • They don't overlap with Reserved Instances. A Savings Plan doesn't apply to usage already covered by Reserved Instances.
  • They follow the baseline, not the peak. Committing to your highest spend means paying for hours you don't use.

Choosing a commitment

AWS recommends an hourly commitment based on your historical On-Demand usage, and AWS Cost Explorer reports coverage and utilisation once a plan is running. A sensible first step for a UK finance firm is to cover only the steady, always-on baseline, and revisit after a quarter of data. Our post on Savings Plans vs Reserved Instances compares the two in detail, and our AWS cost optimisation service sizes commitments against your actual usage.