Microsoft is reportedly bringing Anthropic’s Claude models into Office 365, signaling a shift from its exclusive reliance on OpenAI for AI-powered productivity tools.
Quick Summary – TLDR:
- Microsoft plans to integrate Anthropic’s Claude models into Office 365 apps like Word, Excel, Outlook, and PowerPoint.
- This marks a major diversification of Microsoft’s AI strategy, previously dominated by OpenAI’s models.
- Microsoft reportedly believes Claude Sonnet 4 outperforms GPT-5 in tasks like generating PowerPoint presentations.
- Pricing for Microsoft 365 Copilot will remain unchanged despite added costs for using Anthropic via AWS.
What Happened?
Microsoft is expanding its AI partnerships by licensing models from Anthropic to enhance Office 365’s Copilot features. The decision reflects both performance-driven choices and strategic diversification, reducing its dependence on OpenAI, which has powered most of Microsoft’s AI features to date.
Microsoft Brings Anthropic Into Office
After years of relying primarily on OpenAI for Copilot in Office apps, Microsoft is reportedly integrating Anthropic’s Claude models into Word, Excel, Outlook, and PowerPoint. This shift was reported by The Information and confirmed by sources involved in the rollout.
🚨 Claude can now create Excel, PDFs and Word docs and will now be used by Microsoft in O365.
— Deedy (@deedydas) September 10, 2025
Here, I asked it to generate Elon Musk’s resume in LaTeX, and it one-shotted it. It also:
— Built a financial model of Google from scratch
— Built a PDF and slide deck report for H-1B… pic.twitter.com/tacjDwTkJN
- Microsoft believes Claude Sonnet 4 performs better than GPT-5 in specific Office use cases
- In particular, Claude’s output in PowerPoint presentations is said to be more aesthetically pleasing
- Developers found Claude more effective in automating complex Excel functions
Though Microsoft remains OpenAI’s largest investor, with over $13 billion committed, the company appears eager to broaden its AI base. It has already introduced its own in-house models like MAI-Voice-1 and MAI-1-preview, and integrated external models such as xAI’s Grok into platforms like GitHub Copilot.
No Rift, But Some Tensions
The timing of Microsoft’s Anthropic partnership comes amid reports of growing strain with OpenAI. Disagreements around OpenAI’s restructuring into a public benefit corporation and its push into areas like job platforms and custom AI chips have signaled that the two tech giants are increasingly seeking independence.
Still, Microsoft insists there is no bad blood. “As we’ve said, OpenAI will continue to be our partner on frontier models and we remain committed to our long-term partnership,” said Microsoft spokesperson Michael Collins.
This isn’t Microsoft’s first AI detour from OpenAI. GitHub Copilot has already incorporated other providers, and Azure now hosts multiple third-party models, including DeepSeek.
AWS in the Middle
Anthropic’s Claude models will be accessed through Amazon Web Services (AWS), which is Anthropic’s primary cloud partner and a major investor. That means Microsoft will effectively pay AWS to run a competitor’s models in its own productivity suite.
Notably, Microsoft pays nothing extra to use OpenAI’s models as part of their partnership. Even so, the company will not increase pricing for its Office 365 Copilot features. The current rate of $30 per user per month will stay in place.
SQ Magazine’s Takeaway
Honestly, I think this is a pretty smart and bold move by Microsoft. While their partnership with OpenAI was a clear win early on, it makes sense not to put all their eggs in one basket. Bringing Anthropic into the mix means better AI performance where it matters most: real-world tools we use every day. Plus, it shows that Microsoft is willing to go the extra mile (even paying AWS, their competitor) to give users better features. That’s big. This could also be a wake-up call for OpenAI to stay ahead in quality and innovation, not just brand dominance.
