Skip to content

Instant result, no sign-upUpdated Oct 2026

Cloud cost savings estimator.

Enter your monthly cloud bill and how it splits. Get a conservative savings range from rightsizing, scheduling, spot, commitments and storage tiering.

Your cloud bill

USD

AWS, Azure or Google Cloud, before credits.

%

VMs, containers, functions.

%

Object, block, file, backups.

%

Dev, test, staging.

%

Batch, CI, stateless workers.

Levers to include

Estimated savings a month

Estimated savings $4,600–8,100 a month.

That is 23% to 40% of your bill, or $55.2K–97K a year.

Bill today

$20K

Bill after changes

$11.9K–15.4K

  • Rightsizing$1,000–2,200
  • Scheduling and autoscaling$1,200–1,800
  • Spot or preemptible$600–700
  • Savings plans and reservations$1,300–2,100
  • Storage tiering$600–1,400

Optional

Email me the full report.

  • Savings per lever, low and high, monthly and yearly
  • Which part of your bill each lever touches
  • The order we would tackle them in
  • A printable summary for your engineering and finance leads

No spam. Your inputs go to one consultant.

How it works.

  1. 01

    Enter your bill

    Your total monthly spend across AWS, Azure or Google Cloud, before credits, and roughly how much of it is compute and storage.

  2. 02

    Describe the compute

    How much runs in non-production environments and how much is interruptible work such as batch jobs, CI runners or stateless workers.

  3. 03

    Choose the levers

    Switch levers on or off to see what each is worth. They are applied in order, so no dollar is saved twice.

  4. 04

    Read the range

    The low end uses cautious rates, the high end still sits well under the providers' published maximums. Most teams land in between.

Assumptions and sources.

Every default you can change, and where each figure comes from. Figures are as of 2026.

AssumptionValue usedSource
Rightsizing8-18% of computeFlexera 2026 State of the Cloud Report: respondents estimate 29% of IaaS and PaaS spend is wasted. We assume only part of that is recoverable by resizing.
Scheduling and autoscaling35-60% of non-production computeRunning dev and test 12 hours a day on weekdays uses 60 of 168 weekly hours, 64% fewer. The range allows for environments that must stay on.
Spot or preemptible50-70% of interruptible computeAWS: EC2 Spot Instances offer up to 90% off On-Demand prices. Azure Spot VMs and Google Cloud Spot VMs work the same way. Spot capacity can be reclaimed at short notice.
Savings plans and reservations20-35% of steady production computeAWS: Savings Plans save up to 72% (EC2 Instance Savings Plans) or up to 66% (Compute Savings Plans) versus On-Demand. Our range assumes one-year terms and partial coverage.
Storage tiering15-35% of storageAWS: S3 Intelligent-Tiering saves up to 40% in its Infrequent Access tier and up to 68% in Archive Instant Access. Only data that is rarely read moves down.
Order of leversRightsize, then split computeRightsizing shrinks compute first; scheduling, spot and commitments then apply to separate slices of what is left, and tiering to storage only.
Not includedNetwork, databases, licences, our feesData transfer, managed database tuning and licence optimization can add savings but vary too much to estimate this way. Your cloud bill is separate from any engineering fees.

Where cloud money usually leaks

Most waste is not exotic. It is servers sized for a launch-day peak that never came, development environments running all weekend, snapshots and old volumes nobody deleted, and steady workloads paying On-Demand prices for years. In Flexera's 2026 survey, respondents estimated that 29% of their IaaS and PaaS spend was wasted, up from the year before as AI workloads made spending harder to predict.

The order of work matters. Commit to savings plans or reservations after rightsizing and scheduling, not before, or you lock in a discount on capacity you were about to remove. Our cloud cost optimization engagements follow that sequence, and our guide to ways to cut your cloud bill explains each step.

Keeping the savings

One-off clean-ups drift back within months unless someone owns the bill. Tag resources by team and environment, show each team its own spend, set budgets and anomaly alerts, and review commitments quarterly. This is the core of FinOps: engineering, finance and product sharing responsibility for cloud cost.

Architecture changes can save more than any discount: moving spiky workloads to serverless, consolidating underused clusters on Kubernetes, or caching to cut data transfer. Those need a closer look at your workloads than a calculator can give.

Cloud savings questions, answered

Something else on your mind? Ask a consultant and get a reply within one business day.

How much can I realistically save on my cloud bill?

It depends on how much optimization has already been done. Teams that have never rightsized or bought commitments often find the largest savings; mature FinOps teams find less. This estimator uses conservative rates well under the providers' maximums, so treat the high end as achievable rather than guaranteed.

Savings plans or reserved instances: which is better?

On AWS, Compute Savings Plans are the most flexible, applying across instance families, regions and services such as Fargate and Lambda, with a slightly smaller discount than EC2 Instance Savings Plans or Reserved Instances. Azure and Google Cloud offer similar commitment options. Commit only to the steady baseline that remains after rightsizing.

Is it safe to use spot instances?

Yes, for work that can be interrupted and retried: batch processing, CI builds, rendering, data pipelines and stateless services behind a load balancer with enough capacity elsewhere. Spot capacity can be reclaimed at short notice, so it is not suitable for single databases or stateful services without careful design.

What is rightsizing?

Matching instance types and sizes to what workloads actually use. It relies on a few weeks of CPU, memory and network metrics, and often includes moving to newer instance generations that give more performance per dollar.

How does scheduling non-production environments save money?

Development and test environments are often only used in working hours. Stopping them at night and at weekends cuts their running hours by more than half. Infrastructure as code makes it easy to recreate whole environments on demand instead of leaving them running.

Does storage tiering affect performance?

It can. Colder tiers cost less to store but may add retrieval fees or latency. Intelligent tiering services move objects automatically based on access, which suits data with unknown or changing access patterns. Lifecycle rules suit predictable data such as logs and backups.

Does this work for Azure and Google Cloud?

Yes. The levers are the same on every major provider: rightsizing, scheduling, spot or preemptible capacity, commitment discounts and storage tiers. The source figures in the assumptions table are AWS's because it publishes clear maximums.

Why are network and database costs not included?

They vary too much by architecture to estimate from a total. Data transfer, managed databases and licences can hold real savings, but finding them needs a look at your actual bill and workloads.

How long does a cloud cost optimization take?

Quick wins such as deleting idle resources and scheduling environments often land in the first weeks. Rightsizing needs a few weeks of metrics, and commitments are best bought once usage has settled at its new, lower level.

Can Nexzem review our cloud bill?

Yes. We review AWS, Azure and Google Cloud accounts, deliver the savings and set up the tagging, budgets and reporting that keep them. Start on the contact page or read about our managed cloud services.

Want the real number for your bill?

Share read-only billing access or a recent invoice. A consultant replies within one business day with the first savings we would go after.