Microsoft Turns to Anthropic, Scales Back OpenAI Reliance

According to The Information, Microsoft plans to reduce its reliance on OpenAI and increase its investment in Anthropic. The company, best known for its Claude large language model, is widely regarded as the strongest in the programming domain.

Two people familiar with the matter revealed that Microsoft will integrate Anthropic’s technology into certain AI functions within Office 365 and pay for its use.

However, Microsoft is not abandoning OpenAI entirely — instead, it will blend technologies from both AI unicorns. One source noted that for automation tasks, such as Excel financial functions or generating PowerPoint presentations from client instructions, Anthropic’s latest model outperformed OpenAI. In particular, presentations created by Claude Sonnet 4 were considered more aesthetically appealing.

The change is expected to be announced in the coming weeks. Pricing for Office AI features will remain unchanged at $30 per user per month. OpenAI will continue to support some Office 365 Copilot functions, but more advanced tasks will increasingly be handled by Anthropic’s large models.

Microsoft has not disclosed sales data for Office 365 Copilot. However, its latest earnings showed that more than 100 million users have tried at least one Copilot product, including the consumer-facing Copilot and the developer-focused GitHub Copilot.

With over 430 million paying Office 365 subscribers, if just 1% were to spend $30 per month on AI features, the potential annual revenue could exceed $1 billion.

It is worth noting that cracks have begun to emerge in the Microsoft-OpenAI relationship. OpenAI has sought to restructure itself into a for-profit entity, but faced resistance from Microsoft, which wants the restructuring to safeguard its investment interests. Negotiations have been difficult, and OpenAI executives even considered accusing Microsoft of monopolistic behavior during the partnership.