Anthropic commits $30 billion to Azure, partners with NVIDIA

Anthropic has forged a major partnership with Microsoft and NVIDIA that reshapes the AI infrastructure landscape. The company has committed to purchase $30 billion in Azure computing capacity and can contract up to one additional gigawatt of capacity—a measure of power that reflects the scale of the data centers required to train and serve advanced models.

The deal also brings Claude, Anthropic’s model family, to Microsoft Foundry, the platform Microsoft uses to offer AI models to businesses and developers. NVIDIA and Microsoft, meanwhile, will invest up to $10 billion and $5 billion, respectively, in Anthropic. The transaction values the company at roughly $350 billion.

Claude comes to Azure without leaving AWS

Anthropic is keeping Amazon as its primary cloud provider and training partner. But the agreement with Microsoft reduces its operational dependence on a single infrastructure provider and gives Azure a much larger role in Claude’s growth.

For enterprise customers, the change is easy to understand: they will be able to select Anthropic models from Microsoft Foundry alongside other options in Azure’s catalog. The initial lineup will include Claude Sonnet 4.5, Claude Opus 4.1, and Claude Haiku 4.5—three variants designed for different needs: a balance of quality and cost, complex tasks, and speed, respectively.

Microsoft has also confirmed that it will continue providing access to Claude across its Copilot family, including GitHub Copilot, Microsoft 365 Copilot, and Copilot Studio. That is particularly important for companies already using Microsoft tools that want to deploy Claude without building a separate integration or moving their data and workflows to another provider.

With this move, Claude becomes the only frontier model available across the three major cloud platforms. That does not require companies to use it, but it removes a common barrier: until now, choosing a model often also meant choosing a particular cloud.

NVIDIA gains a strategic customer for its next generation

The announcement’s second major component is a direct technology partnership between Anthropic and NVIDIA. The two companies will work on design and engineering to optimize Anthropic’s models for NVIDIA’s architecture while adapting future NVIDIA architectures to Claude workloads.

The stated goal includes improving performance, efficiency, and total cost of ownership. The latter goes beyond the price of the chips: it includes electricity, cooling, networking, servers, and data center operations. In generative AI, reducing those costs is just as important as increasing model capacity, because they determine how much it costs to serve millions of users or deploy an assistant across a large company.

The initial commitment covers up to one gigawatt of capacity based on NVIDIA Grace Blackwell and Vera Rubin systems. Grace Blackwell combines processors and graphics accelerators to run AI models at scale; Vera Rubin is NVIDIA’s next announced platform for this type of infrastructure.

Microsoft diversifies its bet on models

For Microsoft, the alliance expands a strategy that has so far been associated primarily with OpenAI. The company will continue offering OpenAI technologies, but it is adding an infrastructure, product, and investment relationship with another leading language-model developer.

That does not mean the models will be interchangeable. Each family offers different capabilities, costs, usage limits, and tools. What changes is the range of choice for Azure customers: they will be able to compare alternatives within the same enterprise environment, with its security controls, billing, and data management.

The combined investment of up to $15 billion also shows that the AI competition is no longer fought solely on model quality. It is being fought over access to chips, electricity, data centers, and cloud platforms capable of supporting them. Anthropic gains the capacity to grow; NVIDIA secures a deep relationship with one of the largest buyers of compute; and Microsoft strengthens Azure as a showcase for more than one advanced AI provider.

The next challenge will be turning that technical scale into useful, affordable services for businesses. Claude’s availability on Azure opens that door, but its ultimate value will depend on how Microsoft and Anthropic integrate the models into real-world programming, productivity, and automation tools.