When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works.

Microsoft brings Claude to its own Azure infrastructure, powered by Nvidia GB300 Blackwell

Claude is now generally available on Azure-hosted infrastructure powered by Nvidia GB300 Blackwell Ultra systems.
Microsoft Claude

Anthropic's Claude models are now generally available in Microsoft Foundry on Azure and are running on Nvidia's GB300 Blackwell Ultra systems.

Nvidia wrote in its announcement that the models are hosted on Microsoft Azure and accelerated by GB300 Blackwell Ultra GPUs, with Quantum-X800 InfiniBand networking used to support larger agentic systems and specialized sub-agents that can operate across business domains.

This is great for customers and enterprises that want to build autonomous and domain-specific AI agents using Claude without moving outside Microsoft’s cloud platform.

Microsoft currently offers Claude models in Foundry in two forms: “Hosted on Azure,” which runs end-to-end on Azure infrastructure and is generally available, and “Hosted on Anthropic infrastructure,” which remains in preview. This separation is quite important for organizations that have procurement, compliance, data processing, or internal governance requirements tied to Azure.

Anthropic currently has 11 Claude models listed in Microsoft Foundry, including Opus 4.8, Sonnet 4.6, and even the unavailable Mythos and Fable models. Billing is handled through Claude Consumption Units (CCUs).

Microsoft says CCU is an invoicing unit for Claude models in Foundry, with token usage converted using Anthropic’s published per-model token rates.

The usage is billed through Azure Marketplace just like models from other distributors and appears on the customer's Azure invoice, while eligible spend can count against a Microsoft Azure Consumption Commitment.

For starters, GB300 NVL72 is a rack-scale, fully liquid-cooled system that combines 72 Blackwell Ultra GPUs and 36 Grace CPUs. Nvidia has listed 37TB of fast memory, 130TB/s of NVLink bandwidth, and FP4 Tensor Core performance of up to 1,440 petaflops with sparsity.

The deal is also part of a three-way partnership between Microsoft, Nvidia, and Anthropic. Under the deal, Anthropic has committed to buying $30 billion in Azure compute capacity and contracting additional capacity up to one gigawatt. Nvidia and Microsoft also said they would invest up to $10 billion and $5 billion in Anthropic, respectively.

Samsung Interactive Displays
Next Article

Samsung introduces new AI classroom tools and interactive displays at ISTELive 2026

WhatsApp Username
Previous Article

WhatsApp is getting usernames, and you can reserve your preferred one now

0 Comments

Load the comments and join the conversation!

Read the comments, ask the editors questions, show respect and join the conversation.

Click here