Anthropic has announced the general availability of its Claude family of language models in Microsoft Foundry, enabling organisations to access and deploy these models directly in Microsoft Azure environments.
The release, effective immediately, brings Claude Opus 4.8 and Claude Haiku 4.5 to Azure customers via the Messages API, with integration into existing Azure identity, billing, network, and governance systems.
The models are hosted in Azure, giving enterprises the ability to process inference in either a global or US data zone. This addresses data residency requirements for teams operating under specific regulatory or compliance constraints.
Anthropic operates inference and serves as the data processor for customers running Claude in Microsoft Foundry.
With this expansion, teams can manage access through Microsoft Entra ID, apply Azure role-based access controls, and enforce governance policies already in use within their organisation.
Usage and billing are consolidated, with Claude Consumption Units (CCUs) appearing as a single line item on the Azure invoice.
Customers with a Microsoft Enterprise Agreement can use their Microsoft Azure Consumption Commitment (MACC) towards the cost. Per-model usage detail remains unchanged within the Foundry platform.
Zero data retention is offered for high-sensitivity workloads. When enabled, Anthropic does not retain API prompts or outputs once calls have been completed.
According to Microsoft, this is aimed at meeting the privacy and regulatory requirements of sectors needing strict data handling and confidentiality.
There are two configuration options for deployment. Customers may choose to run Claude models hosted on Azure to leverage Azure-native authentication, governance, and billing, including use of a US data processing zone for additional residency control.
Alternatively, the “hosted on Anthropic” option, previously called Foundry Preview, remains available for customers seeking features or model variants not yet enabled in Azure-hosted environments. Microsoft and Anthropic have stated their intention to align capabilities between both deployment modes in the future.
Microsoft foundry models Azure product lead Steve Sweetman wrote in an official Microsoft blog post that: “Enterprises can build with Claude through their existing Azure account, using the authentication, billing, networking, governance, and data controls their teams already trust.
“Instead of solving for infrastructure, teams can focus on building agentic applications that run their work with Claude, in the environment where they already operate.”
Developers can access core model features such as prompt caching, extended thinking, and tool streaming through the Messages API. These capabilities support workflows including code development, agentic task management, and complex reasoning.
Foundry Agent Service enables Claude to act as a reasoning engine, coordinating multi-step planning, tool use, and automation across enterprise applications.
Anthropic’s Claude models in Microsoft Foundry are powered by Nvidia’s latest GB300 Blackwell Ultra GPUs, using Nvidia Quantum-X800 InfiniBand networking to deliver inference performance. This deployment builds on the strategic partnership announced by Microsoft, Nvidia, and Anthropic in November 2025.
Nvidia stated: “With Claude in Foundry running on Nvidia GB300 NVL72 systems with Nvidia Quantum-X800 InfiniBand networking, enterprises can now build and run more powerful agentic systems, including autonomous and specialised sub-agents that can work across business domains to perform advanced tasks.”
Nvidia is collaborating with Anthropic to integrate developer tools directly into the Anthropic platform. This allows enterprises to extend Claude agents with domain-specific skills and control their operation using the Secure Agent Workspace Reference Design, which specifies identity, credential, network access, and runtime policy management at the infrastructure level.
The move aims to address a key concern for enterprises, which is the ability to deploy, secure, and govern generative AI models in familiar cloud environments without additional procurement hurdles.
Foundry’s Azure-native setup means that teams can incorporate Claude models into their existing workflows, governance frameworks, and compliance measures without separate infrastructure. This includes organisations building autonomous agents for highly regulated industries or cross-domain business processes.
Recently, Anthropic introduced Claude Tag, an AI agent currently in beta for Claude Enterprise and Claude Team users within Salesforce’s Slack platform.
