Can we move AI Service resource between subscriptions from same tenant

Rohit Shinde1 20 Reputation points
2025-10-08T08:27:35.4+00:00

We are facing the issue while migrating the Azure AI foundry resources from PayG subscription to CSP subscription, let me know is there any alternate solution for it.

Azure OpenAI in Foundry Models

Answer accepted by question author
Anonymous
2025-10-09T06:28:44.9+00:00

Hello Rohit Shinde1,

A direct move of Azure AI resources from a Pay-As-You-Go to a CSP subscription is not supported. Thanking Zafer KAYA for prompt response.

I would like to expand on why this limitation exists and provide a more detailed strategy for the re-creation process.

Actually, a Direct Move Isn't Supported as it's a fundamental constraint related to Azure's commercial and billing architecture.

  • Pay-As-You-Go (PayG): This is a direct billing relationship between you and Microsoft.
  • Cloud Solution Provider (CSP): This is a billing and support relationship managed through a Microsoft Partner.

Because the provider, billing agreement, and support model are fundamentally different, a resource cannot be seamlessly "moved" between them. The resource's identity and billing metadata are tied to its original subscription type. Therefore, the only supported path is to re-create the resources under the new CSP agreement.

Recommended: The Re-creation strategy approach

While it requires planning, this "migrate by re-creating" approach is a standard cloud practice that ensures a clean setup in the new environment.

  1. Audit and Document: Before you begin, thoroughly document the configuration of all your AI services in the source (PayG) subscription. This includes deployed models (e.g., gpt-4o-mini), versions, endpoint names, content filter settings, and any connections to other services like Azure AI Search or Storage Accounts.
  2. Deploy New Resources: In the target CSP subscription, create the new, empty AI Service resources (e.g., a new Azure OpenAI resource, a new AI Search, etc.).
  3. Migrate Configurations and Data: This is the most detailed step:
    • For Azure OpenAI, you will re-create your model deployments. For any fine-tuned models, you will need to re-run the fine-tuning job using your original dataset pointed at the new resource.
    • For Azure AI Search, you can migrate your indexes by exporting the index schema as JSON and using the import wizard or an SDK to re-create it. You will then need to re-run the indexers to populate the data.
    • For other services like Azure AI Vision or Speech, which are often stateless, you mainly need to manage the new keys and endpoints.
  4. Update and Test: Update your client applications and services with the new endpoints and API keys from the resources in the CSP subscription. Thoroughly test the entire workflow to ensure it functions correctly before decommissioning the old resources.

Please accept the answer and upvote, for remediation of other community members.

Was this answer helpful?

0 comments No comments

1 additional answer

Sort by: Most helpful
  1. Zafer KAYA 335 Reputation points MVP
    2025-10-08T08:44:23.6933333+00:00

    Moving Azure AI resources (like AI Foundry or AI Services) between subscriptions is only supported within the same tenant and only for certain resource types.

    For Pay-As-You-Go → CSP migrations, most AI resources (including Azure OpenAI and AI Foundry) cannot be directly moved due to provider and billing restrictions.

    Workaround:

    Recreate the resource in the target (CSP) subscription.

    Export and re-deploy your configuration, models, and data.

    Update client connections to the new resource.

    Reference:

    Move resources between subscriptions

    Was this answer helpful?

    0 comments No comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.