Only Compute Savings Plans cover Lambda and Fargate; EC2 Instance Savings Plans and Reserved Instances do not. A Compute Savings Plan commits a fixed dollar-per-hour of compute spend and applies it automatically across EC2, Fargate, and Lambda. The rates differ: up to 17 percent on Lambda, up to 50 percent on Fargate, up to 66 percent on EC2. The Lambda discount covers duration and provisioned concurrency, not requests; the Fargate discount covers vCPU and memory. Size the commitment to the combined steady baseline, never the peak.
Last updated: June 2026 · Written by Fredrik Filipsson and reviewed by Morten Andersen · See, Cut, Lock, Run method
An AWS Savings Plan is a commitment to spend a fixed dollar amount of compute per hour for one or three years in exchange for a discount below on-demand rates. There are two kinds, and the distinction is the whole point of this page: an EC2 Instance Savings Plan discounts a specific instance family in a Region and reaches nothing else, while a Compute Savings Plan is flexible and is the only Savings Plan that applies to AWS Lambda and AWS Fargate as well as EC2. If your serverless spend is large enough to have a steady floor, a Compute Savings Plan can take a real bite out of it. This article is part of our complete guide to AWS cost optimization, the cluster pillar it links up to.
Which Savings Plan covers Lambda and Fargate?
Only the Compute Savings Plan covers Lambda and Fargate. The EC2 Instance Savings Plan, despite the broader-sounding name, is locked to a chosen EC2 instance family in one Region and never applies to serverless. A Compute Savings Plan, by contrast, commits you to a dollar-per-hour figure and then AWS automatically applies that commitment to your highest-discount eligible usage across EC2 of any family or Region, Fargate, and Lambda. That flexibility is why it carries a slightly smaller maximum EC2 discount than the instance-specific plan, and it is the trade worth making when serverless is part of your footprint. AWS documents exactly how the commitment is applied in the Savings Plans user guide. For the EC2-side decision, see our Savings Plans vs Reserved Instances comparison.
How much do Savings Plans save on Lambda and Fargate?
The savings rate depends on the service, and serverless sits below EC2. AWS Compute Savings Plans discount Lambda by up to 17 percent, Fargate by up to 50 percent, and EC2 by up to 66 percent, with the exact figure depending on term length and payment option. The reason serverless saves less is that AWS already prices Lambda and Fargate with much of the elasticity baked in, so the commitment buys a thinner margin. Verify the current rates on the Compute Savings Plans pricing page before you model a commitment, since the published maximums can change.
| Service | Max discount (Compute SP) | What is discounted | What is not |
|---|---|---|---|
| AWS Lambda | Up to 17% | Duration (GB-seconds), provisioned concurrency | Per-request charges |
| AWS Fargate | Up to 50% | vCPU and memory per hour | Data transfer, ephemeral storage above the free amount |
| Amazon EC2 | Up to 66% | Instance hours, any family or Region | EBS, data transfer |
The verdict: a Compute Savings Plan is worth buying when you run a steady serverless baseline alongside EC2, but model it on the discountable portion only. Lambda request charges and Fargate data transfer keep billing at full rate, so they should never be inside the commitment math.
What exactly does the Lambda discount apply to?
The Lambda discount applies to duration and provisioned concurrency, not to requests. Lambda bills in two parts: a per-request charge and a duration charge measured in GB-seconds, plus a separate charge if you reserve provisioned concurrency. A Compute Savings Plan discounts the duration and provisioned concurrency components, which for most steady workloads are the larger share, but it leaves the per-request charge at the standard rate. This is why a function that is invoked millions of times but runs for a few milliseconds each may see little benefit, while a function that runs longer or holds provisioned concurrency benefits more. The closely related sibling guide on right-sizing compute, Graviton4 C8g, M8g, and R8g compared, covers the architectural saving you can stack underneath a commitment on the EC2 and Fargate side.
Carrying serverless spend with no commitment on it?
Our AWS cost audit models the discountable floor of your combined EC2, Fargate, and Lambda spend, recommends the right Compute Savings Plan term, and proves the saving against a clean baseline. On the performance model, you pay only from realized savings. No savings, no fee.
Book a cloud cost audit →How do I size a Compute Savings Plan for serverless?
Size to the steady baseline of your combined discountable compute, never the peak. Because a Compute Savings Plan pools EC2, Fargate, and Lambda under one dollar-per-hour commitment, the right number is the floor that all three together rarely drop below, so the commitment is always fully used. Start with the Savings Plans recommendations in AWS Cost Explorer as a reference point, then commit conservatively, typically a one-year no-upfront plan, so you retain flexibility as serverless usage shifts. Leave variable and spiky usage on demand. Buying to the peak guarantees the commitment sits partly unused during quiet hours, which quietly erases the discount. The disciplined approach to building this number is covered in our guide on how to build an AWS Savings Plan purchase strategy.
The AWS Cost Optimization Field Guide includes the commitment-sizing worksheet we use to separate the discountable serverless floor from the variable spend that should stay on demand.
Frequently asked questions
Do Savings Plans cover Lambda and Fargate?
Only Compute Savings Plans cover Lambda and Fargate. EC2 Instance Savings Plans and Reserved Instances do not. A Compute Savings Plan commits you to a dollar-per-hour of compute spend and automatically applies that commitment across EC2, Fargate, and Lambda.
How much do Savings Plans save on Lambda and Fargate?
AWS Compute Savings Plans discount Lambda by up to 17 percent and Fargate by up to 50 percent, versus up to 66 percent on EC2. The Lambda discount applies to duration and provisioned concurrency; the Fargate discount applies to vCPU and memory.
Does a Lambda Savings Plan cover requests?
No. The Lambda discount applies to duration charges (GB-seconds) and provisioned concurrency, not to the per-request charge. Request charges are billed at the standard rate regardless of the Savings Plan.
How should I size a Compute Savings Plan for serverless?
Commit only to the steady baseline of your combined EC2, Fargate, and Lambda spend, not the peak. Use Cost Explorer's Savings Plans recommendations as a starting point, start conservative with a one-year no-upfront plan, and leave variable spend on demand so the commitment never sits unused.
The short version
Savings Plans reach serverless only through the Compute Savings Plan, which discounts Lambda by up to 17 percent and Fargate by up to 50 percent while also covering EC2. The Lambda discount touches duration and provisioned concurrency but not requests; the Fargate discount touches vCPU and memory but not data transfer. Pool the three services, commit to the steady floor, and leave the rest on demand. When you want the discountable baseline of your serverless and EC2 spend modeled and the commitment sized correctly, that is part of what our AWS cost optimization service delivers.
Cloud pricing and service behavior change frequently. Verify the specifics in this guide against the providers’ own current documentation and the FinOps Foundation: AWS pricing ↗, AWS documentation ↗ and FinOps Foundation Framework ↗. This article also reflects Cloud Cost Room’s hands-on, vendor-neutral engagement experience.