What each one actually does
Reserved capacity means Savings Plans or Reserved Instances. They discount roughly 20 to 72 percent against on demand in exchange for a one or three year commitment, and they bill whether or not the capacity is used. Compute Savings Plans flex across instance family, size, region, and across EC2, Fargate, and Lambda; Reserved Instances and EC2 Instance Savings Plans lock to a narrower scope for a slightly deeper rate.
Autoscaling changes the number of running instances in response to demand. It eliminates the waste of paying for instances you do not need at 3am, but it does nothing to the per hour rate. An autoscaled fleet running entirely on demand is elastic and expensive at the same time.
How to blend them
Picture your demand as a curve over a month. The bottom band, the floor your usage rarely goes below, should be covered by flexible Compute Savings Plans: it is predictable, so commitment risk is low and the discount is pure win. The volatile band above the floor should ride autoscaling on demand, or spot for fault tolerant work. You commit to the trough, not the average, because committing to the average means paying for capacity that is idle whenever demand dips.
A worked example
Indicative figures, verified against the client's billing data, anonymized. A European SaaS company autoscales between 40 and 120 EC2 instances, with a floor of 40.
| Approach | Covered by commitment | Indicative monthly cost |
|---|---|---|
| All on demand | 0 instances | 168,000 USD |
| Commit the average (80) | 80 instances, often idle | 149,000 USD plus stranded commitment |
| Commit the floor (40) | 40 instances, always used | 121,000 USD |
Committing the floor cuts roughly 28 percent with no stranded commitment, because every committed instance is one autoscaling never turns off. Committing the average looks cheaper on paper but pays for idle capacity whenever demand dips below 80.
Your next step
Find your true floor from the last several months of usage, commit to it with flexible Savings Plans, and let autoscaling handle the rest. For the full method read the AWS cost optimization guide, and for neighbouring detail see on demand versus commitment on AWS and the biggest AWS waste categories. To apply it, our AWS cost optimization service turns the plan into verified savings.
Put a defensible number on your cloud spend.
No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.
The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.