Nov 28, Paris: AI Health Hack #2 with Alan, Qovery and AWS. Applications close Nov 21

Which tools show the real cost of moving to Google Cloud before you commit? Migration Center, the Pricing Calculator, RaMP and what they all miss

Google Cloud Migration Center, the Pricing Calculator, the Cost Estimation API and RaMP funding estimate your run-rate on Google Cloud, and none of them prices the move itself. Here is what each tool actually does in 2026, how CloudZero, Vantage, Cloudability and Infracost fit, and the egress, dual-running and people costs you have to add yourself.

Morgan Perry
Co-founder
NOV 19, 2026 · 19 MIN
Which tools show the real cost of moving to Google Cloud before you commit? Migration Center, the Pricing Calculator, RaMP and what they all miss

Key points:

  • Google's own tools estimate your run-rate on Google Cloud. Migration Center (free) discovers AWS, Azure, VMware and on-premises assets and produces cost estimates, TCO reports and detailed pricing reports. The Pricing Calculator and the Cost Estimation API (preview) price a target architecture, with your contract prices if you link a billing account. None of them prices egress out of your current cloud, the months of dual running or your engineers' time.
  • StratoZone no longer exists. Google acquired it in 2020, stopped selling it on 30 July 2024 and shut it down on 30 September 2024. Its discovery and assessment work now lives in Migration Center.
  • RaMP changes the math more than any calculator. Google's Rapid Migration and Modernization Program funds partner and consulting work and pays service credits of 25% of incremental eligible spend, capped at the lesser of 30% of the projected annual run rate and $3M per workload, for workloads with a projected annual run rate of at least $60,000. Incentives for a workload are void if it is not fully migrated before the agreement expires.
  • Third-party FinOps tools need billing data. CloudZero, Vantage and Cloudability read billing exports, so before the move they baseline your current cloud and during the move they track both bills. Infracost is the exception: it prices Terraform changes for 24 Google Cloud services before you apply them.
  • The most accurate estimate is a measured pilot. Run one representative environment on GKE in your own Google Cloud account for two to four weeks and read the billing export. With Qovery, the same app deploys to GKE alongside AWS or Azure without a rewrite of your delivery setup.

Qovery · Infrastructure control plane
Kubernetes, operated through one governed API
Learn more

Every Google Cloud migration I have watched starts with a spreadsheet that says the new cloud will be cheaper. Most of the time the run-rate number is roughly right. The surprise comes from the lines nobody put in the spreadsheet: the replication traffic out of the old cloud, two full bills for a quarter, and the platform engineers who spent half a year on the move instead of the roadmap.

I am Morgan, co-founder of Qovery. We deploy applications on Kubernetes in our customers' own cloud accounts, on AWS, Google Cloud, Azure and Scaleway, so I see a lot of teams in the middle of moving between them. This article answers one question: which tools give you the real cost of moving to Google Cloud before you sign, and what you still have to model yourself.

Getting this wrong is common. Flexera's 2026 State of the Cloud survey of more than 750 cloud decision makers puts wasted cloud spend at 29%, the first increase in five years, and 85% of respondents name managing cloud spend as a top challenge (Flexera).

This piece is Google-specific on purpose. For a provider-neutral framework, read our cloud migration cost analysis with the formula and hidden costs and the 7-step framework for leaving a managed platform. For the per-service price comparison, see is Google Cloud cheaper than AWS.

Why is a Google Cloud migration estimate usually wrong?

A Google Cloud migration estimate is usually wrong because it prices the steady state on Google Cloud and leaves out the transition. The run-rate is the part tools are good at. The transition is where the money goes.

A complete estimate has three parts:

  • Target run-rate. What you will pay per month on Google Cloud once everything has moved: Compute Engine or GKE, Cloud SQL, storage, network, licenses, support. Google's tools model this well.
  • One-time move costs. Data transfer out of the current provider, migration tooling, partner or consulting fees, re-platforming work, new observability and security tooling.
  • Transition costs. Running both clouds in parallel for months, cross-cloud traffic between half-migrated services, the engineering time diverted from product work, and the commitments you keep paying on the old cloud.

Here is the formula I use with teams. Every variable has to come from a source you can name:

TEXT
Real 3-year cost on Google Cloud =
    36 x target monthly run-rate (after CUDs you will actually buy)
  + egress out of the current cloud (one-time bulk + ongoing replication)
  + months of dual running x current monthly bill
  + engineering hours x loaded hourly cost
  + partner, tooling and training fees
  - RaMP credits and funds you can realistically claim
  - remaining commitments you can stop paying on the old cloud

Google's tools fill the first line and help with the second-to-last one. The rest of this article covers which tool fills which line.

What does Google Cloud Migration Center actually estimate?

Google Cloud Migration Center is Google's free discovery, assessment and cost estimation service for moving to Google Cloud. "You can use Migration Center to run a cost estimate and to create an assessment at no cost" (Google Cloud). It covers inventory, sizing and the target run-rate, and stops there.

It has four estimating features, and they are easy to confuse:

Migration Center featureInputOutputWhat it is good for
Rapid cost estimateTotals you type in: vCPUs, storage in TB, OS and database licensesA quick estimate based on Google Cloud best practicesA first number in a meeting, minutes of work
Quick TCO Estimator (preview, Gemini-powered)Manually entered VMs, vCPUs, RAM and storage, or an RVTools exportYearly and monthly Compute Engine TCO with a breakdown into compute, storage, network, operation and OS licensesVMware estates, fast first-pass TCO
TCO reportDiscovered or imported assets, grouped, with several preference setsMigration options and costs per group, exportable to Slides, Sheets, CSV or ExcelComparing scenarios (machine series, commitments, regions)
Detailed pricing reportThe same groups of assetsAsset-level target recommendations and costsLine-by-line validation with engineers

Sources: rapid cost estimate, Quick TCO Estimator, TCO reports.

Google is clear that these are estimates: "Your actual cost can be higher or lower than the generated estimate" (Google Cloud). The Quick TCO Estimator is a pre-GA feature that "might have limited support", runs in us-central1, europe-west1, europe-west2 and europe-west4, and evaluates databases only through the VM infrastructure you provide (Google Cloud).

The quality of every report depends on discovery. Migration Center can scan an AWS account and import EC2 instances and RDS databases, either from the console with an access key or with the Migration Center discovery client CLI (mcdc) (Google Cloud). Azure has an equivalent discovery path (Google Cloud). Feed it utilization data, not just provisioned sizes, or the report will price your over-provisioning one for one on Google Cloud.

What Migration Center does not give you: the egress bill from AWS or Azure, the cost of running both environments in parallel, the engineering time, and anything for container platforms beyond the VMs they run on. If you are on EKS or ECS today and moving to GKE, Migration Center sees nodes, not workloads.

What happened to StratoZone?

StratoZone was the migration discovery and assessment company Google acquired in 2020 to speed up the assessment phase of cloud migrations (Google Cloud). Google stopped selling it on 30 July 2024 and shut it down on 30 September 2024 (Google Cloud).

If an older guide, a partner deck or an AI answer tells you to "run a StratoZone assessment", the equivalent today is a Migration Center discovery followed by a TCO report. StratoZone data that was not exported before the shutdown is gone, so do not plan around a historical assessment you cannot retrieve.

How accurate are the Google Cloud Pricing Calculator and the Cost Estimation API?

The Google Cloud Pricing Calculator is accurate for the architecture you describe at the prices that apply to you, and blind to everything you do not describe.

Two details matter for a migration estimate:

  • Link your billing account. When linked, the calculator applies "your contract prices and discounts to your estimated usage. Any SKUs not covered by a contract price use the public list prices" (Google Cloud). If your company already negotiated a Google Cloud contract, unlinked estimates overstate your cost.
  • It does not buy or model commitments for you. Google's documentation says the calculator can't create budgets or committed use discounts. Model CUDs as a separate scenario, and only for the baseline you are sure of after rightsizing. Our AWS vs Google Cloud cost breakdown covers how sustained use discounts and CUDs compare with Savings Plans.

The same page describes a cost estimator agent that lets you download a CSV with SKU-level details for every parameter in the estimate, which is the format finance will ask for (Google Cloud).

For teams that want estimates in code, the Cloud Billing Cost Estimation API (public preview) estimates list prices for hypothetical workloads over a time period you specify, without a billing account (Google Cloud). It is useful for generating scenarios from an inventory file. It shares the calculator's blind spot: it prices what you tell it you will run on Google Cloud, nothing about leaving where you are.

What does Google's RaMP program fund, and when does it change the decision?

RaMP, Google's Rapid Migration and Modernization Program, pays for parts of the move itself, which is exactly where the calculators are silent. For a workload above the threshold, RaMP can change whether a migration pays back in year one or year three.

What the program states in its user guide (Google Cloud):

  • Eligibility. "The estimated Projected Annual Run Rate must be at least $60,000 USD per workload to qualify for RaMP", based on a data-driven RaMP assessment.
  • Funds. Partner Services Funds, Google Cloud Consulting Services Funds, and Training and Implementation Services Funds pay for assessment, business case and implementation work.
  • Service credits. "Each workload is eligible for 25% of the incremental eligible spend, capped at the lesser of 30% of Projected Annual Run Rate and $3M USD", disbursed "within 45 days of each calendar quarter-end beginning with the quarter-end falling on June 30, 2026".
  • The catch. "If a workload is not fully migrated before the RaMP Agreement expiration, all future incentives for that workload growth are void", and workloads where migration work has already started cannot be rolled into a new agreement.

Two practical consequences. First, talk to Google or a partner before you start migrating, because work already underway may no longer qualify. Second, model RaMP credits as a range in your estimate, with the low end assuming a delay. A migration that slips past the agreement date loses the credits exactly when you are paying for both clouds.

Which third-party tools help, and what do they need before they can help?

Third-party cost tools fall into two groups: FinOps platforms that read billing data, and infrastructure-as-code tools that price a change before it is applied. Only the second group works on Google Cloud before you have a Google Cloud bill.

CloudZero reads Google Cloud costs from the Cloud Billing detailed usage export to BigQuery, through a read-only service account; new exports can take 24 to 48 hours to deliver first data (CloudZero docs). Its strength is unit cost (cost per customer, per feature), and AnyCost brings in costs CloudZero does not connect to natively.

Vantage connects to Google Cloud with a read-only service account and ingests costs across projects, labels and regions into cost reports (Vantage docs), with forecasting on top (Vantage docs).

IBM Apptio Cloudability supports Google Cloud (Apptio) and offers forecasting, budgeting, rightsizing, container cost allocation and commitment management (Apptio). In June 2026 Apptio announced updates including IBM Apptio Data Center Total Cost of Ownership and Cloudability Intelligent Forecasting (Apptio via PR Newswire).

Infracost prices Terraform changes in pull requests, before anything is applied. It supports 24 Google Cloud services with charged resources, including google_compute_instance, google_container_cluster, google_sql_database_instance and google_storage_bucket, uses on-demand list prices by default, and has gaps such as E2 custom machine types and sole-tenant VMs (Infracost). For usage-based costs like storage and egress it applies predefined values that "attempt to set each usage-based cost as $5/month for common configurations" unless you override them in infracost-usage.yml (Infracost). If you migrate by writing Terraform for Google Cloud, Infracost is the only tool in this list that prices the target from your actual code.

How I would use each one in a Google Cloud migration:

  • Before the move: point CloudZero, Vantage or Cloudability at your current AWS or Azure bill to get a clean, per-service baseline. That baseline is the "before" column of the business case, and the source of your dual-running cost.
  • During the move: keep both clouds in the same FinOps tool, so you see the combined bill every week instead of two invoices a month apart.
  • While writing the target: run Infracost on every Terraform pull request with a real infracost-usage.yml for storage and egress, so the $5 defaults do not hide your largest usage lines.
Measure GKE costs on your own account before you commit.
Qovery deploys your apps to GKE in your own Google Cloud account, next to what you run on AWS or Azure today, so a pilot gives you real billing data instead of an estimate. Start free with one cluster of your own.

How do the tools compare side by side?

The table below compares what each tool covers at the moment you need a number, which is before you sign anything with Google. Prices and features are from each vendor's own pages as of November 2026.

ToolStageInput it needsPrices target run-rate on Google CloudPrices egress out of your current cloudPrices dual running and people timeCost to use
Migration Center (estimate, TCO, detailed pricing)Pre-commitDiscovery of AWS, Azure, VMware or on-premises assets, or typed totalsYes, compute, storage and licenses; network in the Quick TCO EstimatorNoNoFree
Google Cloud Pricing CalculatorPre-commitAn architecture you describeYes, with contract prices if billing account linkedOnly Google-side egress you enterNoFree
Cloud Billing Cost Estimation API (preview)Pre-commitWorkloads defined in codeYes, list pricesNoNoFree API
RaMPPre-commit to post-migrationRaMP assessment, $60k+ projected annual run rate per workloadVia the assessmentNo, but funds offset move costsPartly, through partner and consulting fundsProgram, not a tool
InfracostWriting the targetTerraform code plus usage fileYes, for 24 Google Cloud servicesOnly if you set usage valuesNoOpen source CLI, paid cloud tier
CloudZeroBaseline, then actualsBilling exportsNo, until Google Cloud bills existShows it after it happensShows both bills once both are connectedPaid
VantageBaseline, then actualsBilling data via read-only accessNo, until Google Cloud bills existShows it after it happensShows both bills once both are connectedFree up to $2,500 of tracked cloud spend, then paid plans (Vantage)
IBM Apptio CloudabilityBaseline, then actuals and forecastBilling dataNo, until Google Cloud bills existShows it after it happensForecasts from historyPaid

No single tool gives you the real cost of moving to Google Cloud before you commit. Migration Center plus the calculator gives you the destination, a FinOps tool gives you the origin, and you build the bridge between them yourself.

Which costs do all of these tools miss?

All of these tools miss the costs of the transition itself: egress out of the old cloud, dual running, and people. These three lines regularly add up to more than the first year of savings, so they deserve as much rigor as the run-rate.

Egress during the migration

Moving data into Google Cloud is free on the Google side, because Google does not charge for ingress (Google Cloud). The bill comes from the provider you are leaving.

AWS offers free data transfer out for customers moving all of their data off AWS, granted as credits after a request to AWS Support: "Once approved, we will provide credits for the data being migrated" (AWS News Blog). Read the condition carefully. It covers a full exit, approved in advance. It does not obviously cover a hybrid end state where some workloads stay on AWS, or months of continuous replication while services move one at a time. We covered the leave-and-return symmetry in leaving AWS for Google Cloud.

Model three egress lines separately:

  1. Bulk copy: total GB of object storage, database dumps and archives, priced at your current provider's egress rate, minus any approved exit credit.
  2. Ongoing replication: change data capture and incremental sync for the weeks between first copy and cut-over.
  3. Cross-cloud chatter: API calls and queries between services already on Google Cloud and services still on the old cloud. This line grows the longer the migration takes.

Dual running

During a safe migration you pay for both clouds. The old environment stays live until the new one has carried production traffic long enough to trust it, and any reserved instances or Savings Plans on the old side keep billing until they expire. Take your current monthly bill from your FinOps baseline, multiply it by a realistic overlap in months, and add the Google Cloud ramp on top. Then check the expiry date of every commitment on the old cloud, because a three-year commitment signed last year is a cost the migration cannot remove.

People time

People time is the largest line most estimates leave out. Count the hours to re-create infrastructure as code, rebuild CI/CD for the new target, migrate secrets and IAM, re-instrument monitoring, run load tests, rehearse the database cut-over and train the on-call team. Then multiply by a loaded hourly cost, and add the opportunity cost of the product work that did not ship.

The smaller lines

  • Licenses: bring-your-own versus Google-provided Windows Server and SQL Server licenses change the run-rate a lot. Migration Center's estimate asks about them; make sure the answer matches your contracts.
  • Support plan: a Google Cloud support tier is a percentage or a fixed fee that the calculator does not add unless you include it.
  • Managed service equivalents: a DynamoDB table or an SQS queue has no one-to-one Google Cloud service. Budget the rewrite as engineering time.
  • New tooling: observability, security scanning and backup tools priced per cloud account or per node.

How do you build a pre-commit estimate you can defend?

You build a defensible estimate by combining Google's tools for the destination, a FinOps baseline for the origin, and a short measured pilot for the parts nobody can model. Here is the sequence I recommend.

  1. Baseline the current cloud. Connect CloudZero, Vantage or Cloudability to your AWS or Azure billing data, or use your provider's native cost explorer. Export 3 to 6 months of cost per service and per environment.
  2. Discover and size. Run Migration Center discovery on the source account, with utilization data. Generate a TCO report with at least two preference sets: on-demand, and with the CUDs you would actually buy after rightsizing.
  3. Price the target architecture. Rebuild the target in the Pricing Calculator with your billing account linked, or in Terraform with Infracost if you already write infrastructure as code. Include GKE or Cloud Run, Cloud SQL, storage, load balancing, network egress to your users and the support plan.
  4. Run a measured pilot. Deploy one representative environment (an API, a worker, a database, real traffic or a replay) on Google Cloud for two to four weeks. Turn on the Cloud Billing export to BigQuery from day one and compare the actual daily cost with the estimate for the same services.
  5. Add the transition. Egress (three lines), dual-running months, remaining commitments, engineering hours, partner fees.
  6. Subtract RaMP as a range. Ask Google or a partner for the RaMP assessment before the migration starts, and model credits at a conservative and an expected level.
  7. Decide with a payback date. Divide the one-time and transition costs by the monthly saving. If payback is longer than your planning horizon, write the business case around the strategic reasons for moving and drop the cost-saving argument.

The pilot in step 4 is the step teams skip most, and the one I trust most. A pilot gives you a Google Cloud bill with your own traffic, your own data sizes and your own inefficiencies, which no estimate can.

How does Qovery help you measure Google Cloud costs before you commit?

Qovery does not estimate migration costs, and I would not pretend otherwise. What Qovery does is make the pilot in step 4 and the dual-running period cheap to set up and short to run.

Qovery is an agentic DevOps platform that deploys and runs your applications on Kubernetes in your own cloud account: AWS, Google Cloud, Azure, Scaleway, or any existing Kubernetes cluster. Qovery does not host your apps, so the Google Cloud bill, your CUDs and any RaMP credits stay in your name.

For a Google Cloud migration that means:

  • One delivery setup on both clouds. The same service definitions, environment variables and git-push deployments target a GKE cluster in your Google Cloud account and your existing EKS or AKS cluster. You can clone an environment from AWS to Google Cloud to run the pilot, then compare its billing export with the original's.
  • A shorter dual-running period. When the delivery layer is identical on both sides, cut-over is a traffic switch and a database migration, and CI/CD stays as it is. Every week you remove from the overlap is a week of the old bill you stop paying.
  • Managed cloud services and infrastructure as code. Managed cloud services can be provisioned next to your apps with Blueprints or Terraform/OpenTofu services, and the Qovery Terraform provider keeps the whole setup in code that Infracost can price.
  • Cost visibility after the move. Qovery shows resource consumption per environment, and auto-stopping non-production environments removes a cost line that calculators assume runs 24/7.

The Qovery Free plan includes one cluster of your own and 1,000 deployment minutes a month, which is enough to run a GKE pilot of one environment. Cross-cloud networking between the old and new clouds is not built into Qovery; it is configurable with a VPN, Cilium Cluster Mesh or Tailscale, and its egress belongs in your estimate. Our guide on migrating from AWS to GCP without rewriting your app covers the cut-over sequence.

How do I set up a Google Cloud pilot with Qovery from Claude Code or Cursor?

You can set up the pilot from your coding agent with Qovery's Agent Skills and MCP Server. Run these commands in order:

BASH|1. Install the Qovery Agent Skills globally (Claude Code, Cursor, Codex, OpenCode, Gemini CLI...)
curl -fsSL https://skill.qovery.com/install.sh | bash

# 2. Or install them for the current project only
curl -fsSL https://skill.qovery.com/install.sh | bash -s -- --project

# 3. Add the Qovery MCP Server to Claude Code over OAuth, read-only by default
claude mcp add --transport http qovery https://mcp.qovery.com/mcp --callback-port 4242

# 4. Only if you explicitly want write access (also needs write access enabled in the Qovery Console)
claude mcp add --transport http qovery "https://mcp.qovery.com/mcp?read_write=true" --callback-port 4242

Your agent asks you to approve each install command, and the MCP Server opens an OAuth login in your browser so it acts with your own Qovery permissions. Then ask in plain language, for example: "set up a GKE cluster in my Google Cloud account with Qovery and deploy the staging environment of this app there". The skills (qovery-onboard, qovery-deploy, qovery-optimize and others) do the setup and deployment; the MCP Server reads and operates what already exists. Start with the AI agent quickstart or the Claude Code integration.

Frequently asked questions
What tools show the real cost of migrating to Google Cloud before you commit?

Google Cloud Migration Center (free) discovers your current assets and produces cost estimates, TCO reports and detailed pricing reports, and the Google Cloud Pricing Calculator prices a target architecture with your contract prices when you link a billing account. Infracost prices Google Cloud Terraform changes before they are applied. None of these covers egress out of your current cloud, dual running or engineering time, so add those yourself and validate the estimate with a two to four week pilot on Google Cloud.

Is Google Cloud Migration Center free, and does it replace StratoZone?

Yes. Google states that you can run a cost estimate and create an assessment in Migration Center at no cost. StratoZone, which Google acquired in 2020, reached end of sale on 30 July 2024 and was shut down on 30 September 2024, and Migration Center is where its discovery and assessment functions live now.

How much funding can RaMP provide for a Google Cloud migration?

RaMP offers partner, consulting, and training and implementation funds, plus Google Cloud service credits of 25% of incremental eligible spend per workload, capped at the lesser of 30% of the projected annual run rate and $3M. A workload needs a projected annual run rate of at least $60,000 to qualify, and incentives are void for a workload that is not fully migrated before the RaMP agreement expires. Apply before migration work starts.

Do CloudZero, Vantage or Cloudability estimate Google Cloud migration costs?

Not before you have Google Cloud billing data. CloudZero, Vantage and IBM Apptio Cloudability read billing exports, so they are best used to baseline your current AWS or Azure costs before the move and to track both bills during it. For pre-commit estimates on Google Cloud, use Migration Center, the Pricing Calculator or Infracost.

What is the biggest hidden cost when migrating to Google Cloud?

Usually the transition: months of paying for both clouds, ongoing replication and cross-cloud traffic out of the old provider, and engineering time diverted from product work. AWS grants data transfer out credits for customers moving all of their data off AWS on request, but that does not cover a long hybrid phase. Shortening the dual-running period is often worth more than a better discount on the target.

Can I measure Google Cloud costs with a pilot instead of an estimate?

Yes, and it is the most reliable method. Deploy one representative environment to GKE or Cloud Run in your own Google Cloud account, enable the Cloud Billing export to BigQuery, and compare two to four weeks of actual cost with your estimate. Qovery deploys the same app to GKE alongside your current cloud, so the pilot runs your real workload without rebuilding your delivery pipeline.

Morgan Perry
About the author
Morgan Perry

Morgan co-founded Qovery and leads engineering. He writes about Kubernetes architecture, DevOps best practices, and building resilient infrastructure at scale.

Next step

Measure GKE costs on your own account before you commit.

Qovery deploys your apps to GKE in your own Google Cloud account, next to what you run on AWS or Azure today, so a pilot gives you real billing data instead of an estimate. Start free with one cluster of your own.