American giants Microsoft and NVIDIA have agreed to invest up to $15 billion in AI model developer Anthropic. In exchange, the company will receive giant cloud resources on Azure and joint work on the hardware and Claude models, which should become available to even more business customers.
American giants Microsoft and NVIDIA have agreed to invest up to $15 billion in AI model developer Anthropic. In exchange, the company will receive giant cloud resources on Azure and joint work on the hardware and Claude models, which should become available to even more business customers.
According to Anthropic, the startup has committed to buying $30 billion worth of Azure computing resources from Microsoft and may expand the order to one gigawatt of data center capacity. These resources will run on NVIDIA’s next-generation Grace Blackwell GPUs and upcoming Vera Rubin line, which will be tuned to Claude’s workloads.
For the first time, Anthropic and NVIDIA are forming a deep technology partnership, with engineering teams working together to optimize both Claude models and future NVIDIA architectures. The goal is to reduce the cost of running models, accelerate inference and training, and make better use of data center energy.
Microsoft is also expanding Claude’s integration into its ecosystem. Claude’s Frontier models Sonnet 4.5, Opus 4.1, and Haiku 4.5 will be available to Azure customers through Foundry and other services, and will remain as one of the «engines» for GitHub Copilot, Microsoft 365 Copilot, and Copilot Studio. For enterprise users, this means more model choices within a single cloud.
The financial part of the alliance provides that NVIDIA will invest up to $10 billion in Anthropic, and Microsoft up to $5 billion. At the same time, the company emphasizes that Amazon remains its main cloud partner for model training, and the new agreement with Microsoft and NVIDIA complements, rather than replaces, the existing cooperation.
Market analysts are already calling Anthropic’s alliance with Microsoft and NVIDIA another round of the «arms race» in AI: cloud platforms are trying to secure the most powerful models, and chip manufacturers are trying to secure long-term customers for gigantic amounts of computing.
Previously, dev.ua reported on how Chinese cyberspies used the Claude Code AI tool to hack into approximately 30 major companies and government agencies. Anthropic called it the first cyberespionage campaign driven by artificial intelligence.
