Microsoft today announced a strategic partnership with Anthropic, marking the first time the AI startup’s models are being integrated into Microsoft Foundry. Under the agreement, Anthropic has committed to purchasing $30 billion worth of Azure computing capacity and plans to scale its additional computing infrastructure to 1 gigawatt.

Microsoft Foundry customers now gain access to Anthropic’s cutting-edge Claude models, including Claude Sonnet 4.5, Claude Opus 4.1, and Claude Haiku 4.5. Despite this integration into Microsoft’s AI infrastructure, Amazon will remain Anthropic’s primary cloud provider and training partner.

As part of this collaboration, NVIDIA and Anthropic also announced a joint effort to optimize Anthropic’s models for future NVIDIA architectures. Anthropic has pledged to utilize NVIDIA’s Blackwell and Vera Rubin systems, delivering up to 1 gigawatt of computational power.

In support of these initiatives, NVIDIA will invest up to $10 billion in Anthropic, while Microsoft is contributing $5 billion.

This new alliance between Microsoft and Anthropic follows closely on the heels of OpenAI’s recent restructuring into a for-profit entity and the signing of an updated agreement with Microsoft. The revised OpenAI deal extends Microsoft’s intellectual property rights to OpenAI models through 2032 and includes provisions covering models developed after the achievement of Artificial General Intelligence (AGI). While defining and verifying AGI has historically been a contentious issue in Microsoft’s OpenAI partnership, both parties have now agreed to have an independent panel of experts validate this milestone.

Additionally, Microsoft and OpenAI have relaxed certain exclusivity clauses in their collaboration, enabling OpenAI to partner with third parties and release open-weight models. Microsoft has also relinquished its right of first refusal as OpenAI’s preferred compute provider.

These shifts in Microsoft’s relationship with OpenAI have paved the way for deeper collaboration with Anthropic. In recent months, Microsoft has increasingly incorporated Anthropic’s models into its Copilot services. Notably, Microsoft’s new AI model selector in Visual Studio Code shows a preference for Anthropic’s Claude 4 over GPT-5. Moreover, Microsoft has recently rolled out Anthropic’s Claude Sonnet 4 and Claude Opus 4.1 to Microsoft 365 Copilot users.