Savings Plans vs Reserved Instances in Practice

Reserved Instances commit to a specific instance family, size, and Region (Standard RIs) or allow limited attribute changes during the term (Convertible RIs), in exchange for a discount versus On-Demand pricing, for a 1 or 3 year term. This works well if you're highly confident your instance configuration won't change, but any meaningful architecture change (migrating to a different instance family, or from EC2 to Fargate/Lambda for part of a workload) can leave an RI commitment poorly matched to actual usage.

Savings Plans instead commit to a specific dollar-per-hour spend level for compute usage, and AWS automatically applies the discount to whatever matching usage actually occurs — Compute Savings Plans apply flexibly across instance families, sizes, Regions, and even to Fargate and Lambda usage, not just a fixed EC2 instance type. This flexibility is why Savings Plans have become the generally recommended default over traditional RIs for most customers: you get a comparable discount level to RIs, without needing to correctly predict your exact future instance configuration years in advance.

EC2 Instance Savings Plans offer a middle ground — a deeper discount than Compute Savings Plans, but committed to a specific instance family within a Region (still flexible across instance size and OS within that family), appropriate when you're confident about the instance family but want to preserve some flexibility on size.