AWS Compute Optimizer is free for its default recommendations and aggregates across an entire AWS Organization when you opt in from the management or delegated administrator account. It analyzes EC2 instances, EBS volumes, Lambda functions, ECS services on Fargate, and Auto Scaling groups, then flags each as over- or under-provisioned with an estimated monthly saving. Enable it org-wide, turn on the paid enhanced infrastructure metrics only where the longer history pays, review by finding and savings, validate against the performance-risk score, then right-size and commit on the clean baseline. Recommend first, resize second, commit last.
Last updated: June 2026 · Written by Morten Andersen and reviewed by Fredrik Filipsson · See, Cut, Lock, Run method
AWS Compute Optimizer is a free AWS service that uses machine learning over Amazon CloudWatch metrics to recommend the optimal resource configuration for your workloads. Run on a single account it is useful; run across an AWS Organization from the management account it becomes a right-sizing dashboard for the whole estate, surfacing the over-provisioned resources that waste the most money in one view. This guide is part of our complete guide to AWS cost optimization, the cluster pillar it links up to.
What is AWS Compute Optimizer and what does it cost?
Compute Optimizer is AWS's right-sizing recommendation engine, and its default recommendations are free. It analyzes utilization across Amazon EC2 instances, EC2 Auto Scaling groups, Amazon EBS volumes, AWS Lambda functions, and Amazon ECS services on Fargate, and returns a recommended configuration with a projected utilization and an estimated monthly saving for each. The free tier bases recommendations on the trailing 14 days of CloudWatch metrics. The one paid feature is enhanced infrastructure metrics, which extends the analysis lookback to up to three months for more stable recommendations on workloads with longer cycles, billed per resource-hour. Confirm the current enhanced-metrics rate and supported resource types against the AWS Compute Optimizer pricing page, since features and pricing change over time.
How do I enable Compute Optimizer across an AWS Organization?
Enable it once from the top of the organization so recommendations from every member account aggregate into a single view. Work through these steps:
- Enable org-wide from the management account. In the AWS Organizations management account, or a delegated administrator account, opt in to Compute Optimizer for all member accounts. This gives you one aggregated set of recommendations instead of logging into each account separately.
- Turn on enhanced infrastructure metrics where it pays. Enable the paid three-month lookback selectively, on production accounts with seasonal or weekly cycles where 14 days misses the real peak, and leave the free 14-day analysis on everything else.
- Review recommendations by finding. Filter resources by Over-provisioned and Under-provisioned findings across EC2, EBS, Lambda, Fargate, and Auto Scaling groups, and sort by estimated monthly savings so the biggest wins surface first.
- Validate against the risk and performance data. Read each recommendation's performance-risk score and projected utilization before acting, and set aside workloads whose spiky or seasonal pattern the lookback window did not capture.
- Act, then commit on the clean baseline. Right-size the validated resources, let utilization settle for a week or two, then buy Savings Plans or Reserved Instances against the lower baseline rather than the old oversized one.
That last step is the heart of our method and our AWS cost optimization service: right-size before you commit. Compute Optimizer now recommends Graviton instance types too, so pair it with the sibling guide how to migrate EC2 workloads to Graviton without code changes to capture both the sizing and the architecture saving.
Compute Optimizer tells you the right size; it does not buy your commitments. Resize first, let the new utilization stabilize, then commit. Buying Savings Plans against the pre-right-sizing baseline locks a discount to capacity you are about to remove.
How do I act on Compute Optimizer recommendations at scale?
Act at scale by triaging recommendations into low-risk batch changes and individually reviewed exceptions, rather than treating every line equally. The low-risk majority, idle and clearly over-provisioned instances, oversized EBS volumes, and Lambda functions with too much memory, can be resized in batches per account, ideally during a change window with a rollback path. The exceptions are workloads with a high performance-risk score, a spiky pattern, or a known seasonal peak, which deserve a human read and sometimes the enhanced three-month lookback before any change. Export the recommendations to Amazon S3 for analysis, tag the actioned resources so you can measure realized savings against estimates, and feed the result into your reporting. For the broader governance frame, this is the kind of continuous oversight our complete guide to AWS cost optimization places in the operate-and-govern steps of the method.
Sitting on a pile of right-sizing recommendations?
Our AWS cost audit turns Compute Optimizer output across every account into validated, actioned right-sizing, then commits on the clean baseline so the saving is realized, not just estimated. On the performance model, you pay only from realized savings. No savings, no fee.
Book a cloud cost audit →What are the limits of Compute Optimizer?
Compute Optimizer is a strong starting point, not a complete FinOps program, and knowing its limits keeps you from over-trusting it. It recommends sizing from utilization but does not see application-level constraints, licensing tied to core counts, or business cycles outside its lookback window, so a recommendation can be technically correct and operationally wrong. It does not cover every resource type, it does not buy or manage commitments, and its savings estimates are projections that depend on you actually resizing and the workload behaving as the history suggests. Use it to find candidates and quantify the opportunity, then apply judgment, validation, and the commit-last discipline on top. The right-sizing it surfaces is one lever among several; clearing zombie spend, covered in how to find and delete unused Elastic IPs and snapshots on AWS, runs alongside it.
The AWS Cost Optimization Field Guide includes the right-sizing workflow we run on Compute Optimizer output across an organization, so estimated savings become realized savings.
Frequently asked questions
Is AWS Compute Optimizer free?
Yes, the default recommendations are free, based on the trailing 14 days of CloudWatch metrics. The only paid feature is enhanced infrastructure metrics, which extends the analysis lookback to up to three months and is billed per resource-hour. Enable the paid feature selectively where the longer history changes the recommendation.
How do I enable Compute Optimizer for all accounts?
Opt in from the AWS Organizations management account or a delegated administrator account, which enables it for all member accounts and aggregates their recommendations into one view. This avoids logging into each account separately and is the right way to run it across an estate.
What resources does Compute Optimizer cover?
It covers Amazon EC2 instances, EC2 Auto Scaling groups, Amazon EBS volumes, AWS Lambda functions, and Amazon ECS services on Fargate, and it now includes Graviton instance type recommendations. It does not cover every AWS resource, so treat it as a strong starting point rather than full coverage.
Should I act on every Compute Optimizer recommendation?
No. Review each recommendation's performance-risk score and projected utilization first, and set aside workloads with spiky or seasonal patterns the lookback window missed. Batch the low-risk, clearly over-provisioned changes, and give the high-risk exceptions individual review before resizing.
The short version
Compute Optimizer is free for default recommendations and aggregates across an AWS Organization when you opt in from the management account. Enable it org-wide, add enhanced metrics only where the longer history pays, review by finding and savings, validate against the risk score, then right-size and commit on the clean baseline. It finds the candidates; judgment and the commit-last discipline realize the saving. When you want the recommendations turned into a lower bill across every account, that is the work 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.