Anthropic is currently in discussions with Microsoft regarding the potential use of the software giant's custom AI chips. The negotiations aim to help the AI startup secure additional computing power to accommodate rising demand for its Claude chatbot and Claude Code programming tools, according to people familiar with the matter. No formal agreement has been finalized yet.

The talks center on Microsoft’s custom Maia AI processors. Microsoft unveiled its second-generation Maia 200 chip in January and reports indicate the chips are currently operating within the company's data centers in Arizona and Iowa. During an April earnings call, CEO Satya Nadella stated that the Maia 200 platform delivers over 30% improved tokens per dollar compared to existing silicon in the company's current infrastructure. While Microsoft is actively developing its own hardware to reduce dependency on third-party semiconductor providers like Nvidia, the Maia platform is still expanding in availability and market maturity compared to established alternatives from competitors such as Amazon and Google.

Anthropic has faced recent challenges regarding infrastructure capacity. CEO Dario Amodei noted earlier this month that the startup has struggled with compute availability. To address these needs, the company has entered several major agreements, including a $1.25 billion monthly commitment to SpaceX for compute capacity through May 2029. Anthropic also maintains a 10-year, $100 billion deal with Amazon to utilize its custom Trainium chips and utilizes Google’s tensor processing units. The potential integration of Microsoft’s hardware would add a significant new supplier to the startup's existing network of cloud and hardware partners. Microsoft, which previously invested $5 billion in Anthropic and committed to further infrastructure spending, would gain a major commercial partner to test the scale of its custom silicon platform.

Anthropic is currently in discussions with Microsoft regarding the potential use of the software giant's custom AI chips. The negotiations aim to help the AI startup secure additional computing power to accommodate rising demand for its Claude chatbot and Claude Code programming tools, according to people familiar with the matter. No formal agreement has been finalized yet.

The talks center on Microsoft’s custom Maia AI processors. Microsoft unveiled its second-generation Maia 200 chip in January and reports indicate the chips are currently operating within the company's data centers in Arizona and Iowa. During an April earnings call, CEO Satya Nadella stated that the Maia 200 platform delivers over 30% improved tokens per dollar compared to existing silicon in the company's current infrastructure. While Microsoft is actively developing its own hardware to reduce dependency on third-party semiconductor providers like Nvidia, the Maia platform is still expanding in availability and market maturity compared to established alternatives from competitors such as Amazon and Google.

Anthropic has faced recent challenges regarding infrastructure capacity. CEO Dario Amodei noted earlier this month that the startup has struggled with compute availability. To address these needs, the company has entered several major agreements, including a $1.25 billion monthly commitment to SpaceX for compute capacity through May 2029. Anthropic also maintains a 10-year, $100 billion deal with Amazon to utilize its custom Trainium chips and utilizes Google’s tensor processing units. The potential integration of Microsoft’s hardware would add a significant new supplier to the startup's existing network of cloud and hardware partners. Microsoft, which previously invested $5 billion in Anthropic and committed to further infrastructure spending, would gain a major commercial partner to test the scale of its custom silicon platform.