Microsoft made Anthropic's Claude models generally available on Azure via Microsoft Foundry, running on Nvidia GB300 Blackwell Ultra GPUs. The deployment targets enterprise agentic AI workloads, including autonomous agents and domain-specific assistants.
Key facts
- Claude models now GA on Azure via Microsoft Foundry.
- Runs on Nvidia GB300 Blackwell Ultra GPUs.
- Targets enterprise agentic AI workloads.
- First Blackwell deployment for Anthropic in a major cloud.
- Microsoft invested $2.7B in Anthropic in 2025.
Microsoft and Anthropic have expanded their partnership to bring Claude models to Azure, now generally available through Microsoft Foundry. The inference runs on Nvidia's GB300 Blackwell Ultra GPUs, marking the first time Anthropic's frontier models deploy on Blackwell architecture in a major public cloud [According to the source](https://news.google.com/rss/articles/CBMiuAFBVV95cUxNamkzakVyMWwxSjViaFpIY09NMktvcG9KUzRoZ3pkNlFzbU5SSTRsMHhtYUlmQUhXLUZ6VE5qSjk1SnJQczFURVhRcG1kR1cwSm5sdEwxVWlKZ0taanNFbVBYdDRxUWdobzNDN2xVX3BNRUYzeEszZl8xYUsyVWZXMmxSZ214RUhZWnNmeUpkSDBGRzJBS3lMWG5PR0xoM2NKak1ucjNadktCckplT1Z0SEFvaWR1ek9v?oc=5].
The move gives Azure-native enterprises direct access to Claude without leaving Microsoft's ecosystem, competing directly with Azure OpenAI Service which hosts GPT-4o and o3 models. Microsoft has historically invested in both OpenAI and Anthropic — the latter received a $2.7B investment from Microsoft in 2025 [per public filings].
Key Takeaways
- Microsoft launched Claude models on Azure with Nvidia GB300 Blackwell GPUs, targeting enterprise agentic AI.
- First Blackwell deployment for Anthropic in a major cloud.
Why GB300 Blackwell Matters
Nvidia's GB300 Blackwell Ultra GPUs, announced in 2025, deliver up to 30x inference performance over H100 for large language models, per Nvidia's own benchmarks. The Azure deployment uses these chips for Claude inference, not training. Anthropic trains its models on its own clusters, including those built with Google TPUs and Nvidia hardware [according to Anthropic's infrastructure disclosures].
Enterprise Agent Use Case
The announcement explicitly targets "autonomous and domain-specific AI agents" — a category that includes Claude Code, Anthropic's terminal-native coding agent which scored 88.6% on SWE-bench Verified with Opus 4.8 [per Anthropic's published benchmarks]. Microsoft Foundry offers toolchains for building and deploying such agents, including integration with Azure AI Studio and Copilot Studio.
Competitive Dynamics
This is the closest Microsoft has come to hosting a non-OpenAI frontier model at native Azure scale. Google Cloud already offers Claude via Vertex AI, and Amazon Bedrock hosts Claude as well. The Azure deal gives Microsoft a hedge against OpenAI's increasingly independent go-to-market strategy, while giving Anthropic access to Microsoft's enterprise sales channel — a channel that closed over $60B in commercial cloud revenue in fiscal 2025 [per Microsoft's Q4 2025 earnings].
Neither Microsoft nor Anthropic disclosed pricing, committed compute terms, or whether Claude on Azure will see preferential API pricing compared to direct Anthropic API access.
What to watch
Watch for Azure Marketplace adoption metrics for Claude in Q3 2026, and whether Microsoft bundles Claude tokens into existing Azure OpenAI commitments. Also track Anthropic's IPO timeline — the $1T+ valuation target announced June 27, 2026 may accelerate if Azure enterprise seat counts cross 100K.
Source: news.google.com
