Cloud cost optimization is the practice of cutting cloud spend that creates no value, without slowing down the teams that build on the cloud. Done well, it lowers your AWS, Azure or Google Cloud bill and makes every dollar you keep spending easier to justify.
This guide covers the 12 strategies that consistently save the most money, in the order a FinOps practitioner would apply them, plus a checklist you can use this week.
What is cloud cost optimization?
Cloud cost optimization means matching what you pay for to what you actually need. It has two halves. The first is using fewer or smaller resources: deleting what is idle, resizing what is oversized and turning off what nobody uses at night. The second is paying a lower price for the resources you keep, through commitment discounts, spot capacity and better licensing.
The FinOps Foundation groups this work under its "Optimize Usage & Cost" domain, which includes the capabilities Usage Optimization, Rate Optimization, Architecting & Workload Placement, and Licensing & SaaS. If you are new to the discipline, start with What Is FinOps? A Plain-Language Guide for Cloud Teams.
Why cloud waste is going up again in 2026
Flexera's 2026 State of the Cloud report estimates that 29% of cloud spend is wasted. That is the first increase in five years, and Flexera links it to the growth of cloud-based AI workloads. Managing cloud spend is the top cloud challenge, cited by 85% of the more than 750 cloud decision-makers surveyed.
The FinOps Foundation's State of FinOps 2026 survey, which covers organizations responsible for more than $83 billion in cloud spend, tells the same story: workload optimization and waste reduction remain top priorities, and 98% of respondents now manage AI spend, up from 63% a year earlier.
For a company spending $500,000 a year on cloud, 29% waste is about $145,000 a year that buys nothing.
Usage first, then rate
The order matters. If you buy a three-year commitment for a server that is twice the size it needs to be, you lock in the waste at a discount. So FinOps teams clean up usage first, then commit to the steady baseline that is left. Strategies 1 to 6 below reduce usage. Strategies 7 to 9 lower the rate. Strategies 10 to 12 make the savings stick.
Usage optimization: stop paying for what you don't use
1. Delete idle and orphaned resources
Every cloud account collects leftovers: unattached disks, old snapshots, idle load balancers, unused public IP addresses and test databases nobody remembers creating. They are the easiest savings because removing them changes nothing for users. Run a monthly report of resources with no traffic or no attachments, confirm the owner, then delete or archive.
2. Rightsize oversized instances
Engineers often pick a large instance "to be safe" and never revisit it. Look at CPU and memory use over at least two weeks. If a server peaks at 20% CPU, a smaller size or a different instance family will usually do the same work for less. AWS Compute Optimizer, Azure Advisor and Google Cloud's recommender all suggest rightsizing targets for free.
3. Schedule non-production environments
Development and test environments rarely need to run at night or on weekends. An environment that runs 12 hours a day, five days a week, is on for 60 of the 168 hours in a week. Turning it off the rest of the time cuts its compute cost by roughly 64%. Most teams automate this with a scheduler and a tag such as "schedule = office-hours".
4. Move storage to the right tier
Storage grows quietly and rarely shrinks. Use lifecycle rules to move old data to cheaper tiers. On AWS, S3 Intelligent-Tiering automatically moves objects not accessed for 30 days to a tier that saves up to 40%, and after 90 days to one that saves up to 68%. For block storage, AWS gp3 volumes offer up to 20% lower price per GB than gp2, often with no performance trade-off.
5. Watch data transfer
Data transfer charges are among the hardest costs to read on a cloud bill. Traffic between regions, between availability zones and out to the internet is billed separately, and chatty architectures multiply it. Keep services that talk to each other in the same zone where resilience allows, cache content with a CDN, and review your NAT gateway costs.
6. Autoscale, and go serverless where it fits
A fixed fleet sized for peak traffic is idle most of the day. Autoscaling groups, Kubernetes autoscalers and serverless services such as AWS Lambda, Azure Functions and Cloud Run let capacity follow demand, so you pay for the work done rather than for capacity on standby.
Rate optimization: pay less for what you keep
| Lever | AWS | Azure | Google Cloud | Max discount (published) |
|---|---|---|---|---|
| Commitment discounts | Savings Plans and Reserved Instances | Reservations | Committed use discounts | AWS: up to 72%. Azure: up to 72% |
| Spare capacity | Spot Instances | Spot Virtual Machines | Spot VMs | AWS: up to 90% |
| Storage tiering | S3 Intelligent-Tiering | Blob Storage access tiers | Cloud Storage classes | AWS: 40% to 95%, by tier |
| Licensing | License Included or bring your own | Azure Hybrid Benefit | Bring your own license | Depends on your agreements |
Maximum discounts are the providers' published "up to" figures. Your actual savings depend on region, term, payment option and workload.
7. Commit to your steady baseline
Once usage is clean, look at the compute you run around the clock. Commitment discounts trade flexibility for price: AWS Savings Plans save up to 72% compared with On-Demand, Azure reservations up to 72% compared with pay-as-you-go, and Google Cloud offers committed use discounts. Commit only to the baseline you are confident will still exist for the whole term, and review coverage and utilization every month.
8. Use spot capacity for interruptible work
Spot capacity is spare capacity the provider can reclaim at short notice, sold at a deep discount. AWS Spot Instances are available at up to 90% off On-Demand prices. Batch jobs, CI/CD pipelines, data processing and stateless containers are good candidates. Production databases are not.
9. Optimize licensing
Software licenses can cost more than the servers they run on. Check whether you can bring licenses you already own, for example with Azure Hybrid Benefit for Windows Server and SQL Server, switch to open-source alternatives, or consolidate SaaS subscriptions that overlap.
Make the savings stick
10. Tag and allocate every dollar
You cannot optimize what nobody owns. A consistent tagging policy (team, application, environment, cost center) lets you show each team its own spend. Teams that see their numbers act on them. Aim to allocate most of your bill to an owner, and report the untagged share as a metric to shrink.
11. Set budgets and anomaly alerts
A forgotten GPU cluster or a runaway loop can burn a month's budget in a weekend. Set budgets per account or team, and turn on the anomaly detection built into AWS, Azure and Google Cloud so the right person hears about an unexpected spike within hours, not at month end.
12. Track unit costs, not just the total
A rising cloud bill is not a problem if the business is growing faster. Measure cost per customer, per transaction or per order. If the total goes up while the unit cost goes down, your optimization is working. That is the shift from cutting costs to managing value, which is the core idea of FinOps.
Cloud cost optimization checklist
Use this list for a first 30-day pass on any cloud account.
Common mistakes to avoid
Buying commitments before rightsizing. You lock in waste for one to three years.
Treating optimization as a one-time project. New resources appear every day. FinOps is a loop: inform, optimize, operate.
Cutting without the engineers. Savings that break performance or slow releases get reversed. The people who build the systems should own the changes.
Chasing the total instead of value. The goal is spending wisely, not spending as little as possible.
Learn cloud cost optimization the FinOps way
Cost optimization is a core part of the FinOps Certified Practitioner (FOCP) exam. Our Mastering FinOps for Practitioners course covers it in short animated lessons, with a timed 50-question exam simulator. Engineers who want to go deeper into rightsizing, commitments and architecture can take Mastering FinOps for Engineers, built for the FinOps Certified Engineer exam.
Not ready to buy? Test yourself with our free 10-question FOCP practice test, or estimate your AI workload costs with the free AI cost simulator.
Cloud Institute Canada is an independent training provider and is not affiliated with or endorsed by the FinOps Foundation.
Cloud cost optimization FAQ
What is cloud cost optimization?
It is the ongoing practice of reducing cloud spend that creates no business value, by using fewer or smaller resources and by paying lower rates for the resources you keep.
How much can you save with cloud cost optimization?
It depends on how mature your practice is. Flexera estimates that 29% of cloud spend is wasted in 2026, so organizations starting from scratch often find savings in that range. Individual levers go further: AWS Spot Instances are up to 90% cheaper than On-Demand, and Savings Plans up to 72%.
Where should I start?
Start with visibility: tag resources and find out who owns what. Then remove idle resources, rightsize, and schedule non-production environments before buying any commitment discount.
Is cloud cost optimization the same as FinOps?
No. Optimization is one part of FinOps. FinOps also covers visibility, allocation, forecasting, budgeting and the collaboration between engineering, finance and business teams.
Sources
Flexera, 2026 State of the Cloud Report press release, March 18, 2026
USU, 6 takeaways from the State of FinOps Report 2026
FinOps Foundation, FinOps Framework capabilities
AWS Savings Plans
Amazon EC2 Spot Instances
Amazon S3 Intelligent-Tiering
Amazon EBS General Purpose volumes
Azure Reserved Virtual Machine Instances