A catalog of AI models in Microsoft Foundry that you can discover, compare, and deploy using Azure’s built‑in tools for evaluation, fine‑tuning, and inference
Hello Sondre,
Greetings!
Thanks for raising this question in Q&A forum.
You've done a great job researching this — and you're absolutely right in your findings. The core reason for this gap is that Claude models on Microsoft Foundry currently run on Anthropic-hosted infrastructure, not on Azure's own regional infrastructure. This is different from how Claude works on AWS Bedrock and GCP Vertex AI, where the inference happens within the cloud provider's own EU-region infrastructure. As a result, the usual data residency and compliance guarantees that EU customers expect are not yet available for Microsoft Foundry.
Here's a clear picture of where things stand and what you can do right now:
Step 1: Understand the Current Limitation
When you deploy Claude via Microsoft Foundry today, the actual inference request is routed to Anthropic's own servers — regardless of which Azure region you select. This means data residency, GDPR compliance guarantees, and "no data sent to Anthropic" assurances do not apply to the Foundry path yet. This is officially acknowledged in Anthropic's own documentation as you correctly pointed out.
Step 2: Check Anthropic's Regional Compliance Page Regularly
Anthropic has listed Microsoft Foundry EU support as "Coming 2026" on their regional compliance page. Unfortunately, no more specific date has been publicly announced beyond that. You can keep an eye on updates here: claude.com/regional-compliance
Step 3: Use AWS Bedrock or GCP Vertex AI in the Interim
If EU data residency is a hard compliance requirement for your organization right now, the recommended workaround is to use Claude via AWS Bedrock (e.g., EU Frankfurt region) or GCP Vertex AI (e.g., EU regions), where Anthropic models run fully on the provider's own infrastructure with proper data residency guarantees already in place.
Step 4: Set inferenceProvider Correctly in Your Code
If you switch to Bedrock or Vertex, make sure your Cowork or API integration explicitly sets:
"inferenceProvider": "bedrock"
or
"inferenceProvider": "vertex"
This ensures your requests go through the provider's compliant EU infrastructure and the data residency guarantees apply.
Step 5: Raise a Microsoft Support Ticket for a Roadmap Update
Since this is a roadmap and timeline question specific to Microsoft Foundry's infrastructure plans, the most direct path to getting a more concrete timeline is to open a Microsoft Azure Support Request and ask specifically about the roadmap for native Azure-hosted Claude inference in EU Data Zone regions like Sweden Central or Germany West Central. Microsoft product teams sometimes share roadmap details through official support channels that aren't publicly posted.
Step 6: Watch for Azure Foundry Release Notes
Subscribe to the Azure updates page at azure.microsoft.com/en-us/updates and filter for "Foundry" or "Claude" to get notified as soon as native EU infrastructure support is announced.
If this answer helps you kindly accept the answer which will help others who have similar questions.
Best Regards,
Jerald Felix.