What is AWS cost optimization? AWS cost optimization is the ongoing practice of reducing your Amazon Web Services bill without hurting performance or reliability. It combines four levers: rightsizing resources to real usage, choosing cheaper pricing models such as Savings Plans and Spot Instances, removing idle resources, and running an AWS FinOps process that gives every dollar a clear owner.
The fastest way to reduce AWS costs is to rightsize oversized resources, move steady workloads to Savings Plans, run flexible jobs on Spot, clean up storage and logs, and give every resource an owner. Quick wins such as gp3 volumes, log retention, and scheduling need no application code changes.
The need is real. In Flexera's 2026 State of the Cloud report, respondents estimated wasted cloud spend at 29%, the first increase in five years, driven by surging AI workloads.1 That is nearly 30 cents of every cloud dollar.
At Redefine, we build and run client workloads through our AWS development services, and the same waste shows up in most accounts we review. Below are 12 AWS cost optimization strategies, a worked example, and a 90-day plan.
How Do You Reduce AWS Costs? 12 Ways at a Glance
Savings figures come from AWS pricing pages, cited in each section below.
| # | Way | Typical saving | Effort |
|---|---|---|---|
| 1 | Rightsize EC2 and RDS | About 20% of compute in our worked example | Medium |
| 2 | Savings Plans and Database Savings Plans | Up to 66% (Compute), 72% (EC2 Instance), 35% (databases) | Low |
| 3 | Spot Instances | Up to 90% | Medium |
| 4 | AWS Graviton | Up to 20% | Medium |
| 5 | S3 Intelligent-Tiering | 40% to 95% on cold data, by tier | Low |
| 6 | EBS gp2 to gp3 | 20% per GB | Low |
| 7 | Log retention and orphaned resources | $0.03 per GB-month of logs no longer stored | Low |
| 8 | NAT Gateway and data transfer | $0.045 per GB avoided | Medium |
| 9 | Schedule non-production | About 70% on scheduled resources | Low |
| 10 | Autoscaling and serverless | Pay for demand, not peak | High |
| 11 | Tags, budgets, and anomaly alerts | Prevents new waste | Low |
| 12 | AWS FinOps practice | Keeps savings in place | Medium |
Not sure which of the 12 matter most for your accounts? A technology stack assessment delivers a prioritized roadmap in 14 days.
What Is AWS Cost Optimization?
AWS cost optimization means paying only for the capacity you need, at the lowest price AWS offers for it. Cost optimization is one of the six pillars of the AWS Well-Architected Framework, and good AWS optimization is an ongoing practice, not a one-time cleanup.
The first step is knowing who owns each line on the bill. An IT consulting review can map cost ownership across accounts and teams.
Why Does Your AWS Bill Keep Growing?
AWS bills grow because resources are easy to launch and easy to forget:
- Lift-and-shift sizing: servers keep their on-premises peak size.
- No shutdown rules: dev and test run all night.
- On-Demand by default: steady workloads pay full price.
- Logs kept forever: CloudWatch Logs stores log data indefinitely by default.2
Oversized environments often start on migration day. Redefine's cloud migration services include FinOps work that typically targets 25% to 40% savings within 90 days.
How Do You Reduce AWS Compute Costs? (Ways 1 to 4)
1. Rightsize EC2 and RDS Instances
Rightsizing matches each instance to its real CPU, memory, and network use. AWS Compute Optimizer flags oversized instances and recommends smaller types.
How to do it
Redefine's working rule: start with instances under 40% average CPU over 14 days. Drop one size, watch a week, repeat.
2. Commit With Savings Plans or Database Savings Plans
Compute Savings Plans cut up to 66% and flex across instance families, Regions, Fargate, and Lambda. EC2 Instance Savings Plans reach 72% but lock you to one instance family.3 Database Savings Plans, launched December 2025, save up to 35% across Aurora, RDS, DynamoDB, and more.4
AWS pricing models compared
| Pricing model | Max discount vs On-Demand | Commitment | Best for |
|---|---|---|---|
| On-Demand | None | None | Spiky or new workloads |
| Compute Savings Plans | Up to 66% | 1 or 3 years | Flexible compute |
| EC2 Instance Savings Plans | Up to 72% | 1 or 3 years | One fixed instance family |
| Database Savings Plans | Up to 35% | 1 year | Changing database fleets |
| Reserved Instances | Varies by service | 1 or 3 years | Stable, unchanging databases |
| Spot Instances | Up to 90% | None | Interruptible workloads |
How to do it
Commit to your steady baseline, not your peak. Redefine usually covers 60% to 80% of stable usage, which locks in AWS cost savings and leaves room to rightsize.
Watch out for
Buying Savings Plans before rightsizing locks in oversized capacity for the whole term.
3. Run Flexible Workloads on Spot Instances
Spot Instances cost up to 90% less than On-Demand,5 but AWS can reclaim them with a two-minute interruption notice.6 Use Spot for batch jobs, CI/CD runners, and stateless containers.
4. Move to AWS Graviton
AWS Graviton instances cost up to 20% less than comparable x86 EC2 instances and use up to 60% less energy.7 Most Linux, container, and managed database workloads move with little code change.
Smaller services are easier to rightsize, so teams often pair cost work with a monolith to microservices migration.
How Do You Cut AWS Storage and Logging Costs? (Ways 5 to 7)
5. Use S3 Intelligent-Tiering and Lifecycle Rules
S3 Intelligent-Tiering moves objects to cheaper tiers as access drops, saving up to 40% in Infrequent Access, 68% in Archive Instant Access, and 95% in Deep Archive Access.8 Use S3 Storage Lens to find your largest buckets first.
6. Switch EBS Volumes From gp2 to gp3
EBS gp3 volumes cost 20% less per GB than gp2, with a 3,000 IOPS baseline at any size.9 You can change the type in place, with no downtime.
7. Set Log Retention and Delete Orphaned Resources
CloudWatch Logs storage costs $0.03 per GB per month in US East,10 and logs never expire by default. Set a retention period on every log group, then delete unattached EBS volumes, old snapshots, idle Elastic IPs, and unused AMIs.
Pro tip
Set log retention in your infrastructure-as-code templates so every new log group gets a limit.
Data lakes are often the largest S3 buckets in an account. Our data and analytics team keeps hot data small.
How Do You Cut Network and Idle Resource Costs? (Ways 8 to 10)
8. Cut NAT Gateway and Data Transfer Fees
In US East (N. Virginia), a NAT Gateway costs $0.045 per hour plus $0.045 per GB processed, while gateway VPC endpoints for S3 and DynamoDB cost nothing.11 Route that traffic through endpoints, keep chatty services in one Availability Zone, and serve public content through Amazon CloudFront.
9. Schedule Non-Production Environments
A dev environment used 10 hours a day, 5 days a week, needs only 50 of 168 weekly hours. Scheduling it off the rest cuts its compute cost by about 70%, using AWS Instance Scheduler.
10. Autoscale and Go Serverless Where It Fits
Auto Scaling, AWS Fargate, and AWS Lambda match capacity to demand, not peak. Redefine's DevOps and CI/CD services build test environments that exist only during each run.
"If it runs when nobody uses it, it should not be running."Rule of thumb
How Does AWS FinOps Keep Costs Down? (Ways 11 and 12)
11. Tag Everything and Set Budgets With Alerts
Tag every AWS resource with an owner, team, and environment, and activate those tags as cost allocation tags. Add AWS Budgets alerts and Cost Anomaly Detection to catch spikes in days, not at month end.
12. Build an AWS FinOps Practice
AWS FinOps puts engineering, finance, and product in one monthly spend review. Flexera found 63% of organizations rely on a FinOps team,1 and 98% of FinOps teams in the State of FinOps 2026 survey now manage AI spend.12
The FinOps Foundation's three phases
Inform
Show each team what it spends.
Optimize
Turn findings into tickets with owners.
Operate
Review savings and anomalies every month.
A Power BI cost dashboard, fed by AWS Data Exports (CUR 2.0), puts AWS spend by team next to the numbers finance already tracks.
Worked Example: Cutting a $40,000 Monthly AWS Bill
This illustrative example assumes $24,000 monthly compute (25% non-production), $2,000 of gp2 volumes, and 40 TB of monthly S3 traffic through a NAT Gateway.
| Step | Assumption | Monthly saving |
|---|---|---|
| Rightsize (Way 1) | Compute down 20%, to $19,200 | $4,800 |
| Schedule (Way 9) | 70% off the $4,800 non-production share | $3,360 |
| Compute Savings Plans (Way 2) | 30% off 70% of the remaining $15,840 | $3,326 |
| gp2 to gp3 (Way 6) | 20% off $2,000 of EBS | $400 |
| VPC endpoint (Way 8) | 40,000 GB not processed at $0.045 | $1,800 |
| Total (illustrative) | $40,000 down to about $26,300 | $13,686 (about 34%) |
If an application only scales by adding bigger servers, savings plateau. Legacy application modernization fixes that root cause.
What Are the Best AWS Cost Optimization Tools?
Cost Optimization Hub groups 18 types of AWS savings recommendations in one place.13
| Tool | Best for | Cost |
|---|---|---|
| AWS Cost Explorer | Top cost drivers | Free in console; $0.01 per API request14 |
| AWS Compute Optimizer | Rightsizing | No additional charge (enhanced metrics paid)15 |
| Cost Optimization Hub | One ranked savings list | Part of Billing and Cost Management |
| AWS Trusted Advisor | Idle resources | 56 checks free; all 482 need Business Support+16 |
| AWS Budgets | Spend alerts | Monitoring budgets free17 |
| Cost Anomaly Detection | Spotting spikes | No additional charge18 |
Tools only flag problems. Redefine's managed application support assigns a named engineer to act on them.
How Do AI Workloads Change Cloud Cost Optimization?
AI workloads on Amazon SageMaker, Amazon Bedrock, and GPU instances are now the fastest-growing line on many AWS bills, and Flexera links 2026's rise in cloud waste to surging AI workloads.1 The same 12 ways apply, with three twists:
- Scale inference endpoints down when traffic drops
- Train on Spot capacity with checkpointing
- Track cost per prediction, not GPU hours
Redefine's AI managed services monitor production AI for cost as well as accuracy.
A 90-Day AWS Cost Optimization Roadmap
Days 1 to 30: See the waste
- Turn on the tools. Cost Explorer, Cost Optimization Hub, Budgets, and Cost Anomaly Detection.
- Ship quick wins. Tag owners, set log retention, delete orphaned resources, move to gp3, and schedule dev.
Days 31 to 60: Right-size
- Rightsize and tier. Work through Compute Optimizer findings, then add S3 Intelligent-Tiering and VPC endpoints.
Days 61 to 90: Commit and govern
- Commit and review. Buy Savings Plans for the smaller baseline and start a monthly FinOps review.
Moving to AWS? Redefine's AWS cloud migration services set up tagging and budgets before the first workload moves.
Common AWS Cost Optimization Mistakes
- Buying Savings Plans before rightsizing: you lock in waste for years.
- Cutting costs without owners: unowned resources come back.
- Treating it as a one-off project: usage changes every sprint.
- Trading reliability for savings: a production database on Spot is not a saving.
"The cheapest AWS resource is the one you turn off. The most expensive is the one nobody owns."Cloud and FinOps Lead, Redefine
AWS cost savings should never cost you uptime. SLA-based support keeps response times in writing.
AWS Cost Optimization Checklist
Still in a data center? Our on-premise to cloud migration includes a three-year ROI view.
Frequently Asked Questions
What is AWS cost optimization?
AWS cost optimization is the ongoing practice of lowering your AWS bill without hurting performance.
How much can AWS cost optimization save?
AWS cost optimization savings depend on your starting point. Flexera respondents estimate 29% of cloud spend is wasted, and our illustrative worked example cuts about 34%.
What is the fastest way to reduce AWS costs?
The fastest AWS cost savings come from deleting orphaned resources, setting log retention, scheduling dev environments, and moving to gp3.
How do I find what costs the most on AWS?
To find your biggest AWS costs, group spend by service and tag in Cost Explorer, then check Cost Optimization Hub.
Should I use Savings Plans or Reserved Instances?
Compute Savings Plans suit most compute because they flex across instance families and Regions. For databases, compare Database Savings Plans, which flex, with Reserved Instances, which can suit a stable database.
Are Spot Instances safe for production?
Spot Instances are safe for stateless production workloads that tolerate a two-minute interruption, but not for databases.
What is AWS FinOps?
AWS FinOps is a practice where engineering, finance, and product share ownership of AWS spend.
Which AWS cost optimization tools are free?
The free AWS cost tools are Cost Explorer (console), Compute Optimizer, Cost Anomaly Detection, and monitoring budgets. Full Trusted Advisor checks need Business Support+.
What is the AWS Well-Architected cost optimization pillar?
The cost optimization pillar is one of six AWS Well-Architected pillars, covering financial management, usage awareness, cost-effective resources, demand, and ongoing optimization.
Should we hire help for AWS cost optimization?
Outside help for AWS cost optimization makes sense when your team lacks time or visibility. A short, fixed-scope review works best.
Redefine's consulting services run on fixed scope and fixed pricing.
Where Should You Start With AWS Cost Optimization?
Start with visibility, then quick wins, then commitments. AWS cost optimization is not about spending less. It is about spending on the right things, with an owner for every dollar.
Running more than one cloud? The same cloud cost optimization principles apply elsewhere, and our Azure consulting team uses this playbook for Microsoft workloads.
"A lean AWS bill is not a project you finish. It is a habit your team keeps."Cloud and FinOps Lead, Redefine
Want to know what your AWS bill should be?
Share read-only access to your billing data. We will map your accounts, flag waste, and give you a ranked list of savings with owners and timelines, even if the answer is that you are already in good shape.
Sources & Citations
Primary research and official AWS documentation used for the statistics and pricing figures in this guide:
- Flexera, "Flexera 2026 State of the Cloud Report: The Convergence of Cloud and Value" (March 2026)
- Amazon Web Services, "Working with log groups and log streams"
- Amazon Web Services, "Compute and EC2 Instance Savings Plans Pricing"
- Amazon Web Services, "Announcing Database Savings Plans with up to 35% savings" (December 2, 2025)
- Amazon Web Services, "Amazon EC2 Spot Instances"
- Amazon Web Services, "Spot Instance interruption notices"
- Amazon Web Services, "AWS Graviton Processor"
- Amazon Web Services, "Amazon S3 Intelligent-Tiering"
- Amazon Web Services, "Amazon EBS General Purpose Volumes"
- Amazon Web Services, "Amazon CloudWatch Pricing"
- Amazon Web Services, "Amazon VPC Pricing"
- FinOps Foundation, "State of FinOps 2026 Report"
- Amazon Web Services, "AWS Cloud Financial Management"
- Amazon Web Services, "AWS Cost Explorer Pricing"
- Amazon Web Services, "AWS Compute Optimizer Pricing"
- Amazon Web Services, "AWS Trusted Advisor"
- Amazon Web Services, "AWS Budgets Pricing"
- Amazon Web Services, "Enable AWS Cost Anomaly Detection (AWS Prescriptive Guidance)"
Umang Patel
Co Owner · Technology, Engineering, and Delivery
Redefine's cloud team builds, migrates, and runs AWS workloads for mid-market and enterprise companies, with FinOps reviews, rightsizing, and commitment planning built into every engagement.
