Anthropic pledges $30B in Azure compute as Claude moves into Copilot

Microsoft logo (public domain) via Wikimedia Commons

Anthropic pledges $30B in Azure compute as Claude moves into Copilot

Microsoft is adding Anthropic's Claude models to Copilot alongside OpenAI, backed by billions in investment and a new agentic tool

Microsoft spent years as the most famous OpenAI backer in tech. Now its flagship AI assistant has a second brain on call, and that brain belongs to Anthropic.

Microsoft has been weaving Anthropic’s Claude models into its Copilot ecosystem. Users can now pick Claude for coding help, research and workflow management instead of defaulting to OpenAI.

How Claude ended up inside Copilot

The rollout started on September 24, 2025. Copilot users gained the option to select Claude variants, including Sonnet 4 and Opus 4.1, next to the existing OpenAI models.

Claude models became the default for many commercial customers. Its models remain available inside Copilot, so this is an expansion of the menu, not a replacement of the chef.

Advertisement

For most commercial users outside the EU, EFTA and the UK, Claude is switched on by default. Inside those regions, an administrator has to turn it on manually.

The reason is regulatory. Claude’s availability in Europe is restricted because of compliance with local data boundary rules.

The money behind the partnership

The relationship got considerably more expensive on November 18, 2025. That day, Anthropic pledged to secure $30 billion in Azure compute capacity.

Microsoft committed up to $5 billion to Anthropic, while NVIDIA pledged $10 billion.

On March 9, 2026, Microsoft went further and launched Copilot Cowork. It is an agentic tool built from Anthropic’s technology.

Not putting every egg in one basket

Microsoft is also building its own proprietary models, known as MAI. It continues to use OpenAI technology. And it has reportedly trimmed its projected internal spending on Claude by more than one-third.

What this means for Microsoft, Anthropic and enterprise buyers

For Anthropic, the deal is a distribution win as much as a financial one. Copilot sits inside the Microsoft tools that many businesses already pay for. Being the default model for many commercial customers gives Claude a reach that would be hard to build through direct sales alone.

The reported cut of more than one-third in projected internal Claude spending is worth watching closely. Combined with the MAI effort, it hints that Microsoft wants to own more of its AI stack over time.

There are trade-offs for enterprise customers. More options mean more decisions for IT departments, more governance questions and more data-handling reviews. The European admin requirement is a preview of how compliance rules can fragment the experience across regions.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.
Anthropic pledges $30B in Azure compute as Claude moves into Copilot
Anthropic pledges $30B in Azure compute as Claude moves into Copilot

Microsoft is adding Anthropic's Claude models to Copilot alongside OpenAI, backed by billions in investment and a new agentic tool

Microsoft logo (public domain) via Wikimedia Commons

Microsoft spent years as the most famous OpenAI backer in tech. Now its flagship AI assistant has a second brain on call, and that brain belongs to Anthropic.

Microsoft has been weaving Anthropic’s Claude models into its Copilot ecosystem. Users can now pick Claude for coding help, research and workflow management instead of defaulting to OpenAI.

How Claude ended up inside Copilot

The rollout started on September 24, 2025. Copilot users gained the option to select Claude variants, including Sonnet 4 and Opus 4.1, next to the existing OpenAI models.

Claude models became the default for many commercial customers. Its models remain available inside Copilot, so this is an expansion of the menu, not a replacement of the chef.

Advertisement

For most commercial users outside the EU, EFTA and the UK, Claude is switched on by default. Inside those regions, an administrator has to turn it on manually.

The reason is regulatory. Claude’s availability in Europe is restricted because of compliance with local data boundary rules.

The money behind the partnership

The relationship got considerably more expensive on November 18, 2025. That day, Anthropic pledged to secure $30 billion in Azure compute capacity.

Microsoft committed up to $5 billion to Anthropic, while NVIDIA pledged $10 billion.

On March 9, 2026, Microsoft went further and launched Copilot Cowork. It is an agentic tool built from Anthropic’s technology.

Not putting every egg in one basket

Microsoft is also building its own proprietary models, known as MAI. It continues to use OpenAI technology. And it has reportedly trimmed its projected internal spending on Claude by more than one-third.

What this means for Microsoft, Anthropic and enterprise buyers

For Anthropic, the deal is a distribution win as much as a financial one. Copilot sits inside the Microsoft tools that many businesses already pay for. Being the default model for many commercial customers gives Claude a reach that would be hard to build through direct sales alone.

The reported cut of more than one-third in projected internal Claude spending is worth watching closely. Combined with the MAI effort, it hints that Microsoft wants to own more of its AI stack over time.

There are trade-offs for enterprise customers. More options mean more decisions for IT departments, more governance questions and more data-handling reviews. The European admin requirement is a preview of how compliance rules can fragment the experience across regions.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.