What Microsoft announced
The announcement is dated November 17, 2026. It says GCP support for the connector is now generally available. It also says the GCP connector is GA alongside other Arc updates in Microsoft's Ignite 2026 roundup, as WindowsForum's own coverage noted.
This isn't a new product. The GCP connector entered public preview about a year ago. At that point it offered agentless inventory that projects GCP resources into Azure, plus Arc onboarding for GCP VMs. The GA release builds on that workflow and adds Azure Cloud Shell integration. You can now authenticate to Google Cloud and run setup commands from the Azure portal, without preparing a separate execution environment.
The connector doesn't migrate anything. It is a management and governance layer, and your workloads stay in Google Cloud.
GA versus preview
A "GCP is GA" headline can overstate things. Microsoft's solution table marks several items as preview:
| Solution | What it does | Status |
|---|---|---|
| Inventory | Shows supported GCP resources in Azure and lets you query them with Azure Resource Graph | Part of GA announcement |
| Arc VM onboarding | Installs the Azure Connected Machine agent on eligible GCP VMs | Part of GA announcement |
| Arc GKE onboarding | Discovers GKE clusters and connects them to Azure Arc-enabled Kubernetes | Public Preview |
| Security Baselines | Audits machines against CIS Benchmarks and the Azure security baseline using Machine Configuration | Preview |
| SQL Server licensing and ESU | Manages SQL Server licensing and Extended Security Updates subscriptions | Requires Arc VM onboarding |
| Windows Server ESU | Attaches an existing ESU license to eligible Arc-enabled Windows Server machines | Requires Arc VM onboarding |
Security Baselines, SQL Server licensing and ESU, and Windows Server ESU all depend on Arc VM onboarding. Each also carries its own configuration, eligibility and licensing requirements.
The documentation hasn't caught up
Several Microsoft Learn pages still carry the preview label. The overview lists "Google Cloud Platform (GCP) (preview)" in the VM onboarding documentation, and other pages I reviewed say the same. The Learn pages in the research packet were last updated between April and November 2026. They may simply predate the announcement.
I'd treat the November 17 Microsoft Community Hub post as the authority on GA status. Check the support terms and SLA for your own subscription before relying on it for production.
What each solution does
Inventory
Inventory scans your GCP project periodically and creates Azure representations of the resources it finds. You can query them through Azure Resource Graph next to Azure and AWS resources. One example is finding every Azure, AWS and GCP resource with a given tag. You can also apply Azure tags and Azure Policy to the representations.
Coverage is limited to supported services. The packet lists BigQuery datasets, Cloud Functions, Compute instances, GKE clusters, Storage buckets and Cloud SQL instances. Don't assume it sees everything in your GCP estate.
Arc VM onboarding
The connector discovers eligible VMs and installs the Azure Connected Machine agent. Once the VMs are Arc-enabled, you can use services such as Azure Monitor and Microsoft Defender for Cloud. Those services still need to be configured, and they carry their own pricing.
With periodic sync on, a newly created VM that meets the prerequisites and your filters is onboarded automatically. If onboarding fails for an eligible machine, Microsoft's documentation says it is retried at a later sync. A machine that already has the agent installed but shows as disconnected is not retried that way. You have to restore connectivity, restart the agent or reinstall it.
Windows Server and SQL Server ESU
The Windows Server ESU solution attaches a license to machines as the connector onboards them. You don't link licenses machine by machine.
The documentation sets strict license requirements:
- The license must already exist in Azure.
- It must be for Windows Server 2016.
- It must be Standard SKU with virtual cores, and it must be activated.
- Virtual-core licensing requires a minimum of eight virtual cores per VM.
- The Standard SKU applies even when the guest operating system is Datacenter edition.
The solution attaches an existing license and doesn't create one. Billing follows the cores provisioned on an activated license, so you need to manage core counts as machines are added, upgraded or retired. A single license covers up to 10,000 cores.
Attaching a license only makes a machine eligible for patches. It still needs a patching tool such as Azure Update Manager, WSUS, Microsoft Update or Configuration Manager. To confirm enrollment, run azcmagent show on the machine. The Extended Security Updates section should report Active.
Setup steps
- In the Azure portal, go to Azure Arc, then Management, then Multicloud connectors.
- Select Create, then Create GCP connector.
- Enter the subscription, resource group, a unique connector name and a supported Azure region. Provide the GCP project ID and project number. For an organization account, provide the organization ID instead.
- Choose your solutions. For Inventory, pick the services, a sync schedule and regions. For Arc onboarding, pick regions and an optional GCP label filter.
- On the Configure Access page, download the generated Terraform template.
- Run the Terraform commands in GCP. Microsoft's setup page uses
terraform apply tf.plan. - Wait for the first scan. On average, it takes about one hour for resources to become available in Azure.
If you run Terraform before you finish the Azure portal flow, resources are scanned immediately. If you finish the portal flow first, the first scan takes a bit longer.
Permissions
- In Azure, you need the Contributor role. First-time use also requires registering the resource providers
Microsoft.HybridCompute,Microsoft.HybridConnectivity,Microsoft.AwsConnectorandMicrosoft.Kubernetes. - In GCP, Inventory offers a choice between Global Read and Least Privilege access.
- Global Read covers all resources and picks up new services without a new template.
- Least Privilege covers only the selected services. Adding services later means uploading a new Terraform template.
- Arc onboarding needs OSPolicyAssignment Admin permissions in GCP.
Costs and limits to check first
- The connector is free. Microsoft says its periodic read-only API calls don't incur GCP charges.
- The overall deployment isn't free. Azure services you use with it, such as Azure Monitor, are billed at their own rates.
- GCP VM Manager has usage costs. Arc onboarding requires it. It works through the OS Config agent, and the process uses VM Manager to enforce policies on your VMs through that agent.
- Onboarding can switch the OS Config agent on. The solution doesn't install the agent. If the agent is installed but isn't talking to the service, onboarding can enable that communication. That can change it from inactive to active and add GCP charges.
- Scan regions are limited. The documented GCP regions are us-east1, us-east4, us-central1, us-west1, us-west2, europe-west1, europe-west2, europe-west3, europe-north2, northamerica-northeast1, southamerica-east1 and australia-southeast1.
- Sovereign clouds are excluded. The connector isn't available in Azure Government or Microsoft Azure operated by 21Vianet.
- Deletions propagate. If a source GCP VM is removed, its Arc server representation in the
<PublicCloud>_<AccountId>resource group is deleted too. - An empty label filter onboards everything. Leaving the GCP label filter blank means every discovered VM that meets the prerequisites is eligible.
- Turning periodic sync off stops discovery. New VMs won't be found or onboarded.
Analysis
This is a sensible step for shops running Windows Server or SQL Server across Azure and GCP. Arc already offers a single control plane for monitoring, policy and security. This GA release brings GCP closer to the AWS side of the connector.
The ESU angle may matter most to Windows administrators. It is the part that benefits from the dependency on Arc VM onboarding, and the license rules are narrow. Still, Windows Server 2016 on a third-party cloud is the case where managing the license centrally pays off.
My cautions are about operations. Automatic onboarding with an open label filter is convenient, and it can also touch more VMs than you intended. The OS Config cost note is easy to miss. The Kubernetes and baseline pieces are still preview, so don't build compliance reporting on them yet. This is my own reading of the documentation, not hands-on testing.
A cautious rollout would be:
- Start with Inventory under Least Privilege access.
- Pilot Arc VM onboarding using a GCP label filter in one region.
- Check VM Manager and OS Config costs in GCP.
- Widen the scope only after that.
Microsoft's own Learn pages will show whether the preview wording is dropped over the coming weeks.
Update: Microsoft details six new Multicloud Connector solutions (November 18, 2026)
Microsoft’s Azure Arc team has now framed the GCP work as part of a broader Multicloud Connector expansion. In its November 17 announcement, Microsoft lists six added solutions: GKE onboarding, EKS onboarding, Windows Server ESU, SQL Server ESU, Security Baselines, and Power Cycle Management.
For GCP administrators, this confirms that GKE onboarding remains a preview feature, while its purpose is to discover eligible GKE clusters and connect them to Azure Arc-enabled Kubernetes. Microsoft says the resulting clusters can use supported Arc capabilities including GitOps, governance and monitoring.
The announcement also clarifies an AWS-side addition not covered in the original GCP-focused update: Power Cycle Management is in preview and lets authorized users start, stop or restart supported discovered AWS EC2 instances from Azure. It applies to connector-discovered instances, including inventory-only systems that have not been Arc-enabled.
That does not change the production-readiness caveat for Kubernetes onboarding or Security Baselines: those capabilities remain preview.
References
- Announcing General Availability of GCP support in the Azure Arc Multicloud Connector Azure Arc Blog · 2026-11-17T18:45:00+00:00
- Expanding Multicloud Management with Azure Arc Azure Arc Blog · 2026-11-17T18:45:00+00:00
- What is Multicloud connector enabled by Azure Arc? - Azure Arc | Microsoft Learn learn.microsoft.com