Anthropic announced that its Claude series of large models has been officially opened to enterprise customers on Microsoft Azure, with the underlying computing power being supported by NVIDIA's new generation Blackwell Ultra GB300 GPU platform. This cooperation aims to create an enterprise-level AI agent system with stronger autonomy and higher performance for the rapidly growing "intelligent agent" (Agentic) enterprise application scenarios.
As agentic workloads continue to grow in importance in the field of AI, NVIDIA is working with leading companies to provide end users with faster and stronger model capabilities. In this cooperation, NVIDIA and Anthropic joined hands to deploy the Claude model to run on Microsoft Azure and provide it with a high-performance inference and training environment through the Blackwell Ultra GB300 platform.
Currently, the Claude models available on Microsoft Azure include Claude Opus 4.8 and Claude Haiku 4. Both models use Azure to complete a series of platform-level support such as billing, authentication, and governance. Anthropic said it will continue to expand its deployment scale and product array on Azure based on the existing basis to provide services to more enterprise users.
According to the official introduction, Claude in Foundry runs on infrastructure equipped with NVIDIA GB300 NVL72 systems and NVIDIA Quantum‑X800 InfiniBand networks, enabling enterprises to build and run more powerful intelligent agent systems. These systems can be composed of autonomous agents and sub-agents oriented to specific business areas, and can collaboratively complete complex task execution across business domains.
In this architecture, NVIDIA is working with Anthropic to deeply integrate NVIDIA's developer tools into the Anthropic technology stack to expand developer capabilities. Through NVIDIA's proven "agent skills" and relying on NVIDIA's accelerated computing platform, enterprises can give Claude agents more industry-specific capabilities, deeply embed them in business processes, and serve as the organization's "operating system-style" intelligent hub.
Enterprise customers can run Claude agents on Azure through the NVIDIA Secure Agent Workspace reference design, which provides a complete blueprint for running autonomous agents in governed environments. Under this framework, identities, network access, credentials, and runtime policies are strictly controlled at the infrastructure level, helping enterprises deploy intelligent agents while ensuring security and compliance.
Claude's deployment on Microsoft Foundry, relying on the NVIDIA GB300 GPU acceleration capabilities on Azure, further extends the strategic cooperation between Microsoft, NVIDIA and Anthropic in the field of enterprise AI. This cooperation stems from the strategic partnership announced by the three parties in November last year. The goal is to bring the Anthropic Claude model to the enterprise market more widely through the NVIDIA accelerated computing platform and improve the convenience of enterprises to obtain advanced large models and intelligent agent capabilities.