Claude in Microsoft Foundry is now generally available, hosted on Azure and operated by Anthropic. Azure customers can build production applications and agents with Claude Opus 4.8 and Claude Haiku 4.5, all within their Azure environment. It works with the authentication, billing, and governance you already have in place, and for eligible customers, usage draws down your Microsoft Azure Consumption Commitment. Prompt caching and extended thinking are supported today, with more capabilities on the way. Read more: https://lnkd.in/e8XqwUd7
This is great news! One question: Today we use Azure OpenAI because guaranteed EU-hosted inference is a key compliance requirement for us. The direct Anthropic API doesn’t currently offer an EU-only inference option. Will Microsoft Foundry enable Claude to run with guaranteed EU-hosted inference? That would make it a very compelling option for many European companies.
While it is exciting to see Anthropic now available within Microsoft Foundry, it is very, very disappointing to see it's not available in the European or UK region. And considering what's going on between Anthropic and the US government, you would have thought they might have started prioritising Europe a little bit more, but they still seem to be lagging behind on that side of things.
Great milestone — having Claude available natively within Microsoft Foundry removes a real adoption barrier for enterprises already standardized on Azure governance and billing. Being able to draw down existing Azure Consumption Commitment makes the business case even easier for teams evaluating Claude for production workloads. Curious to see what capabilities land next beyond prompt caching and extended thinking.
Still a market place model with Anthropic being the operator, which will ruffle some feathers in German enterprise. OpenAI models being Azure native are much easier to get approved...
Anthropic For ‘hosted on Microsoft Azure’ with a selected data residency region (e.g., EU data zone), does ‘hosted on Azure’ guarantee that prompts and outputs are processed exclusively within that region ? https://learn.microsoft.com/en-us/azure/foundry/responsible-ai/claude-models/data-privacy#hosted-on-azure
The availability of Claude in Microsoft Foundry is a strategic win for enterprise architects. It finally bridges the gap between high-performance reasoning models and the governance, security, and consumption frameworks that enterprises rely on. Choosing between models is no longer just a performance trade-off; it's about integrating the best intelligence into a unified, secure stack. A significant step forward for the agentic ecosystem.
The combination of enterprise governance, prompt caching, and extended thinking is compelling. As AI moves into production, operational considerations are becoming just as important as benchmark performance.
For the .NET and Azure shops I've worked with, the part that actually unblocks things isn't the model, it's the boundary it runs inside. The blocker for Claude in those orgs was almost never capability. It was procurement and data governance — security wouldn't sign off on sending data to another vendor's endpoint, and there was no clean budget line for it. Running it in-tenant with the auth, billing and governance already in place, drawing down the Azure commit, quietly removes both. That's a bigger deal than another benchmark point. Before I move a production agent over, the thing I'd want pinned down: prompt caching and extended thinking are listed, but are tool use and MCP at full parity with the first-party API today, or part of the "more capabilities on the way"?
Enterprise AI succeeds when innovation meets governance. Bringing Claude into Azure lowers the barrier to building production ready AI applications at scale. Exciting progress.
Head of Transformation office @ CETIN
1moHosted on Azure with all data processing possible within EU?