AWS vs Google Cloud pricing: what differs and how to compare

AWS and Google Cloud both bill most VMs per second and both sell one- and three-year commitments, but they reach a discounted rate differently. Google Cloud discounts some always-on VMs automatically; AWS does not. Data transfer, Kubernetes cluster fees and the way each data warehouse bills account for most of the rest.

This page explains those differences without quoting prices, because prices change and differ by region. To see real numbers for your own workload, price it on both clouds on the canvas: each quote comes from that cloud's own pricing engine.

How AWS and Google Cloud charge, side by side

How it is charged	AWS	Google Cloud
Without a commitment	On-Demand: billed for the time a resource runs (per second for many EC2 instances), with no commitment.	On-demand: VMs are billed per second after a short minimum, with no commitment.
Commitment discounts	Savings Plans (Compute or EC2 Instance, 1 or 3 years, with no, partial or all upfront payment) and Reserved Instances for EC2 and for managed services such as RDS, ElastiCache and Redshift.	Committed use discounts for 1 or 3 years: resource-based (a fixed amount of vCPU, memory and other resources in one region) or spend-based (an hourly spend across eligible services). Commitments are billed monthly.
Automatic discounts	None for EC2. Some services, such as S3 storage and data transfer, get cheaper per unit as monthly volume grows.	Sustained use discounts: applied automatically when eligible Compute Engine VMs run for a large part of the month. Not every machine series qualifies.
Spare capacity	EC2 Spot Instances: spare capacity at a discount, interrupted with a short notice when AWS needs it back.	Spot VMs: spare capacity at a discount, preempted when Google Cloud needs it back.
Windows Server and SQL Server	Licence-included Windows Server and SQL Server. Bringing your own licences generally needs Dedicated Hosts for Windows Server, or License Mobility through Software Assurance for products that allow it.	Premium images add the Windows Server or SQL Server licence to the VM price. Bringing your own Windows Server licences generally needs sole-tenant nodes; SQL Server can use License Mobility.
Data transfer	Data in is free. Data out to the internet is billed per GB in tiers, after a monthly free amount; traffic between Availability Zones and between regions is billed too.	Data in is free. Internet egress is billed per GB and depends on the network tier (Premium or Standard) and the destination; traffic between zones and regions is billed too.
Support plans	Basic support is included. Paid plans are mostly priced as a percentage of monthly AWS spend, with a monthly minimum.	Basic support is included. Paid Customer Care tiers (Standard, Enhanced, Premium) are priced from a monthly minimum and a share of spend.
Free offers	The AWS Free Tier. What a new account gets depends on when it was created, so check the current offer.	A Free Trial credit for new customers, and a Free Tier with monthly usage limits on some products.

Where an AWS and a Google Cloud quote usually differ

Discounts: Savings Plans and Reserved Instances, or committed and sustained use

EC2 has no automatic discount: the On-Demand rate stays until you buy Savings Plans or Reserved Instances. Savings Plans come as Compute Savings Plans, which follow usage across instance families and regions, and EC2 Instance Savings Plans, which are tied to one family in one region and go deeper. You can pay nothing, part or all of a commitment upfront.

Google Cloud applies sustained use discounts automatically to eligible VMs that run for a large part of the month, and sells committed use discounts on top: resource-based for a fixed amount of vCPU and memory in one region, or spend-based across eligible services. Commitments are billed monthly, with no upfront option. Compare the same term on both clouds, and check which machine series each engine chose, since not every Google Cloud series gets sustained use discounts.

Data transfer

Both clouds let data in for free and bill data out per gigabyte. Both also bill traffic between zones in the same region and between regions, so a design spread across zones adds transfer lines. On AWS a NAT gateway adds a per-gigabyte processing charge on traffic that passes through it. On Google Cloud the network tier, Premium or Standard, sets the internet egress rate. Put the expected outbound gigabytes per month in the request.

Kubernetes

Amazon EKS charges an hourly fee per cluster on top of the nodes. Google Kubernetes Engine charges a cluster management fee too, with a free tier that covers the fee for one zonal or Autopilot cluster per billing account, and its Autopilot mode bills for the CPU and memory your pods request rather than for nodes. Say whether you want nodes you manage or Autopilot-style billing, and how many clusters.

Data warehouses

Amazon Redshift bills for provisioned nodes by the hour, or for capacity used in its serverless option. BigQuery bills either on demand, by the data each query scans, or by reserved capacity, plus storage. The two models answer different questions, so describe the usage: how much data you store, how much you scan per month, or how much capacity you need.

Spare capacity

EC2 Spot Instances and Google Cloud Spot VMs both sell unused capacity at a discount and can take it back at short notice. They suit batch work and stateless workers, not a primary database.

Support and free offers

Paid AWS support is mostly priced as a percentage of the monthly bill with a minimum; Google Cloud's paid Customer Care tiers are priced from a monthly minimum and a share of spend. Neither is part of a workload quote. Both clouds have free offers for new customers and free usage within monthly limits on some products; the details change, so check the official pages.

How to compare AWS and Google Cloud on the canvas

Describe the workload once. In the box on the canvas, write what you want to run: the services, sizes, region, hours per month, operating system and licences, outbound data, and any commitment such as "1-year reserved". The more of it you say, the closer the two quotes match.
Price it on both clouds. Pick "All three" to send the same request to every engine at once, or price it on AWS first and use Mirror on Google Cloud on the quote card. Mirroring turns the priced quote into a provider-neutral description of the workload and lets the Google Cloud engine choose its own services and sizes.
Read the Compare lane. It lines the quotes up by role (compute, database, storage, network), with each cloud's monthly total and the difference against the cheapest. Lines that could not be priced live are flagged.
Change one assumption at a time. Add a commitment, switch the licences or the region, and price again, so you can see which choice moves the total.
Keep the result. Export to Excel or CSV, or share a read-only link with whoever signs off the budget.

An AI assistant can do the first two steps for you by writing an AI link. This one opens three Linux VMs in US East on a 1-year commitment on AWS and on Google Cloud, side by side:

https://cloudpricingcalculators.com/q/aws/ec2/size=m5large_qty=3_os=linux_region=useast1_commit=1yr/gcp/gce/size=n2standard2_qty=3_os=linux_region=useast1_commit=1yr?n=Linux%20VMs%20on%20AWS%20and%20Google%20Cloud%2C%201-year

An AI link is priced when you open it, from your own free daily allowance. Nothing is priced in advance, so the page itself quotes no prices.

Official pricing pages

Each provider publishes its own prices and terms; these pages are where the details above come from.

Amazon Web Services

Google Cloud

Pricing on one cloud only

If the choice is already made, the single-cloud calculators go deeper on that cloud:

AWS Price Calculator: itemized Amazon Web Services quotes from AWS published pricing, which you can save, share, export and compare with the other clouds.
GCP Price Calculator: itemized Google Cloud quotes from the Cloud Billing Catalog API, which you can save, share, export and compare with the other clouds.

All three calculators are listed on the calculators page, and common questions are answered in the FAQ.

Other comparisons

Estimates are informational. cloudpricingcalculators.com is not affiliated with Microsoft, Amazon or Google. Validate prices with the provider or an authorised partner before buying.