Skip to content
IFAI

AI models

Anthropic brings Claude to Microsoft Foundry

Published Reach 48

Anthropic just made its Claude models generally available inside Microsoft Foundry on Azure. Enterprise teams can now deploy Claude Opus 4.8 and Haiku 4.5 directly through their existing Azure identity and governance setups, running on NVIDIA GB300 GPUs. Even better, usage draws down directly from Microsoft Enterprise Agreement commitments.

Anthropic brings Claude to Microsoft Foundry
Image: claude.com

For months, enterprise developers have faced the same headache: engineering wants to build on Claude, but finance and compliance want everything kept inside their existing Microsoft Azure commitments.

Anthropic and Microsoft just removed that friction. Claude is now generally available in Microsoft Foundry, hosted directly on Azure.

Enterprise governance meets Claude

Starting today, teams can run Claude inside their Azure environments using existing identity, networking, and governance policies. Anthropic operates the actual inference and serves as the data processor, but the workflow lives within Azure.

Crucially for enterprise procurement, usage appears on a single consolidated invoice and burns down existing Microsoft Enterprise Agreement commitments. For organizations with strict data residency mandates, Anthropic is offering a dedicated US data zone.

Foundry currently offers two operational tracks. The hosted on Azure option gives you full Azure compliance and billing integration, while the hosted on Anthropic track (formerly the Foundry Preview) provides earlier access to the full API feature suite and newer models before parity is reached.

Opus 4.8, Haiku 4.5, and GB300 silicon

At launch, two models are available via the Messages API: the flagship Claude Opus 4.8 and the lightweight Claude Haiku 4.5. Both support core enterprise tools including prompt caching and extended thinking.

Under the hood, the deployment is powered by NVIDIA's GB300 GPUs. NVIDIA noted that its own internal teams rely on autonomous AI agents running Claude daily for technical workflows.

Other early enterprise feedback highlights serious throughput. Testing platform Momentic says it serves millions of tokens per minute running Opus on Azure Foundry, while a nuclear industry case study reported compressing a 200 human-day safety evaluation into a single day.

Why it matters

This is a massive distribution win for Anthropic and a clear signal of where enterprise AI is heading.

For enterprise teams, this removes the single biggest barrier to adopting Claude: legal and billing friction. You do not need to sign a new vendor contract or justify external cloud spend when you can simply draw down pre-committed Azure dollars.

For the broader industry, it shows Microsoft doubling down on being an open infrastructure broker rather than an exclusive shop for OpenAI. Even with Anthropic's deep ties to Amazon and Google, Microsoft is giving Claude premium real estate on its newest GB300 chips.

We think the Azure-hosted path is an obvious choice for Fortune 500 engineering leaders already locked into Microsoft ecosystems. The only real trade-off right now is feature lag behind Anthropic's direct API, but for production systems where compliance beats bleeding-edge features, this solves almost every operational headache.

Sources (1)

  1. [1]
    claude.com(opens in a new tab)Originalclaude.com

Best of AI, daily