Anthropic has agreed to purchase $30 billion worth of Azure compute capacity and may scale further with an option for up to one gigawatt of additional compute. On Tuesday, Microsoft, NVIDIA and Anthropic unveiled a set of new partnerships designed to transform how Anthropic’s Claude models are scaled, deployed and delivered across major cloud platforms.
At the heart of the announcement is Anthropic’s move to scale its Claude AI systems on Microsoft Azure, leveraging NVIDIA’s advanced infrastructure to power the next phase of its model expansion.
Anthropic has committed to buying $30 billion in Azure compute capacity, with an additional option to secure up to one gigawatt of extra compute power. This investment will broaden Claude’s reach for Azure enterprise customers and further expand the model choices available through Microsoft Foundry.
In addition, the deal includes investment commitments of up to $10 billion from NVIDIA and up to $5 billion from Microsoft to support Anthropic’s continued growth and development.
This investment builds on Anthropic’s $13 billion Series F round in September, led by ICONIQ, which placed the company’s post-money valuation at $183 billion. Fidelity Management & Research Company and Lightspeed Venture Partners also served as co-leaders in that round.
We’re focused on making Claude more accessible to organisations developing with AI,” Anthropic CEO Dario Amodei said during a joint conversation with Microsoft’s Satya Nadella and NVIDIA’s Jensen Huang.
For the first time, Anthropic and NVIDIA have established a deep technology partnership centred on co-design and joint engineering. The two companies will optimise Claude models for greater performance and efficiency on NVIDIA’s architectures, while NVIDIA will refine its future chips to better support Anthropic’s workloads.
Anthropic’s compute agreement with NVIDIA will span Grace Blackwell and Vera Rubin systems, with capacity scaling to as much as one gigawatt.
Microsoft and Anthropic are also broadening their existing partnership, extending Claude’s frontier models—Sonnet 4.5, Opus 4.1 and Haiku 4.5—to Microsoft Foundry customers. Microsoft will further integrate Claude across its Copilot ecosystem, including GitHub Copilot, Microsoft 365 Copilot and Copilot Studio. “We want developers and enterprises to have choice in the models they use,” Nadella said.
However, the companies clarified that Amazon remains Anthropic’s primary cloud provider and core partner for model training.
Huang noted that the expanded partnership is designed to fuel the next stage of AI advancement as global demand for compute power accelerates.
Anthropic also recently unveiled a $50 billion investment in U.S. computing infrastructure, partnering with Fluidstack to build new data centres in Texas and New York, with more locations on the way. These facilities are purpose-built for Anthropic’s workloads and will underpin the company’s ongoing AI research and development.
The project is expected to generate roughly 800 permanent positions and 2,400 construction jobs, with the new sites coming online through 2026. It also supports the Trump administration’s AI Action Plan, which focuses on bolstering domestic AI leadership and technology infrastructure.
