Claude models go live on Microsoft Foundry via Azure
Wed, 1st Jul 2026 (Today)
Anthropic's Claude models are now generally available in Microsoft Foundry running on NVIDIA GB300 Blackwell Ultra GPUs hosted on Microsoft Azure, expanding infrastructure options for organisations building AI agents in Azure environments.
The availability brings together Anthropic's language models, Microsoft's cloud platform and NVIDIA's latest accelerated computing hardware to support enterprise deployments of autonomous and domain-specific AI agents.
Enterprise access
The deployment uses NVIDIA GB300 NVL72 systems connected through NVIDIA Quantum-X800 InfiniBand networking. The configuration is designed to support AI inference workloads for enterprise applications that require multiple specialised AI agents operating across different business functions.
The companies said the platform is intended for organisations developing autonomous agents that can perform complex tasks across enterprise environments. The infrastructure is also aimed at supporting domain-specific AI systems tailored to particular industries or operational requirements.
The release extends enterprise access to Claude through Microsoft Foundry, Microsoft's platform for developing and deploying AI models and applications within Azure.
Agent tools
NVIDIA and Anthropic are also expanding their collaboration through the integration of NVIDIA tools into Anthropic's software stack.
The integration is intended to allow developers to equip Claude-based agents with domain-specific functions using NVIDIA verified agent skills. These tools are designed to enable enterprises to create AI agents that perform specialised business tasks while taking advantage of NVIDIA's accelerated computing platform.
According to NVIDIA, the combination gives organisations a way to develop AI systems that are more closely aligned with internal business processes and operational requirements.
The companies said enterprises can build specialised sub-agents that operate across different business domains while remaining part of a broader AI workflow.
Secure deployment
Organisations deploying Claude on Azure can also use the NVIDIA Secure Agent Workspace Reference Design.
The reference design provides a framework for running autonomous AI agents within controlled enterprise environments. It includes infrastructure-level controls covering identity management, network access, credentials and runtime policy.
The approach is intended to support organisations that require governance and security controls when deploying AI systems across enterprise operations.
By applying these controls at the infrastructure layer, enterprises can manage how AI agents access internal systems and business data while maintaining operational policies.
The availability of the reference design complements Microsoft's Azure platform by providing a structured deployment model for organisations introducing autonomous AI into production environments.
Partnership expands
The general availability follows an existing collaboration between Microsoft, NVIDIA and Anthropic to expand enterprise access to Claude models on NVIDIA accelerated computing within Azure.
That collaboration focuses on making Anthropic's AI models available through Microsoft's cloud ecosystem while using NVIDIA hardware to support enterprise-scale inference workloads.
The latest deployment combines Microsoft's AI development environment, Anthropic's Claude models and NVIDIA's GB300 Blackwell Ultra architecture into a single offering for Azure customers.
The companies said the platform targets enterprises seeking to develop autonomous AI systems that operate across business functions while remaining within Azure-native environments. The release also expands the range of AI models available through Microsoft Foundry for organisations building production AI applications.