Microsoft is preparing to bring Anthropic’s AI into Office 365 apps, a move that changes how the company powers AI features inside its productivity suite. According to The Information, Microsoft will pay to use Anthropic’s technology in Word, Excel, Outlook, and PowerPoint alongside OpenAI’s models.
The decision matters because Microsoft had previously relied solely on OpenAI for AI in that part of its software business. Now, the company is making room for another major model provider inside some of its most widely used workplace tools.
A broader model strategy for Office 365
The reported Anthropic deal means Microsoft is no longer treating OpenAI as the only AI supplier for Office 365 apps. Anthropic’s systems will help support new features in Word, Excel, Outlook, and PowerPoint, according to the source article.
That does not mean OpenAI disappears from the picture. The source states that Anthropic’s technology will work alongside OpenAI’s, rather than replacing it outright. Microsoft spokesperson Michael Collins also told TechCrunch:
“As we’ve said, OpenAI will continue to be our partner on frontier models and we remain committed to our long-term partnership,”
For users, the important point is not the name of the model behind every feature. It is that Microsoft appears to be choosing models based on where it believes they perform best. The Information says Microsoft leaders believe Anthropic’s latest models, specifically Claude Sonnet 4, are stronger than OpenAI’s in certain functions, including creating aesthetically pleasing PowerPoint presentations.
That detail shows why model choice can become a product decision. A spreadsheet, a document, an inbox, and a presentation each ask different things from AI. If one model is better suited to a specific workflow, Microsoft has a reason to mix providers rather than depend on a single one.
The OpenAI relationship is changing
The Anthropic move comes as Microsoft and OpenAI are dealing with a more complicated relationship. The source article describes a growing rift between the two companies, connected to OpenAI’s own infrastructure projects and a potential LinkedIn competitor.
Microsoft is also negotiating a new deal with OpenAI to secure access to its AI models after a pending for-profit restructuring. The Information says Microsoft’s Anthropic deal is not a negotiating tactic, according to the TechCrunch article.
That distinction is important. The reported reason is not simply leverage in talks with OpenAI. It is also product performance: Microsoft leaders are said to see Claude Sonnet 4 as better in some Office-related functions.
Still, the timing makes the shift more significant. Microsoft and OpenAI remain partners, but the relationship is no longer defined only by dependence. Microsoft is keeping access to OpenAI models in view while also making space for Anthropic in core productivity software.
Microsoft has already started diversifying
The Anthropic deal is not Microsoft’s first step beyond OpenAI. The source article notes that OpenAI is the default model, but Microsoft already offers other models through GitHub Copilot.
Those models include xAI’s Grok and Anthropic’s Claude. That means Microsoft has already been giving developers access to model choice in one major product area. Office 365 would bring that same broader approach into workplace apps used for writing, analysis, email, and presentations.
Microsoft is also trying to build more self-reliance. The company recently introduced its first two in-house models: MAI-Voice-1 and MAI-1-preview.
Taken together, those moves point to a layered AI strategy:
- OpenAI remains central as a partner on frontier models.
- Anthropic becomes more important for selected Office 365 features.
- Other models are already available through GitHub Copilot.
- Microsoft’s own models add another path toward self-reliance.
This is not a retreat from AI partnerships. It is a shift toward using several of them at once, while also developing internal options.
OpenAI is moving in its own direction
OpenAI is also seeking to step out from under Microsoft’s influence, according to the source article. Last week, OpenAI launched a jobs platform to take on Microsoft’s LinkedIn.
The Financial Times also reported that OpenAI is set to begin mass production on its first AI chips in partnership with Broadcom in 2026. The source article explains the implication: OpenAI could potentially run training and inference on hardware it controls, rather than depending on Microsoft’s Azure setup.
Those moves matter because they touch areas where Microsoft has strategic interests. LinkedIn is a Microsoft property, and Azure has been central to Microsoft’s AI infrastructure role. If OpenAI builds more of its own infrastructure and enters jobs, the partnership becomes more complicated.
Microsoft’s Anthropic deal fits into that context. It gives Microsoft another source of AI technology for Office 365 while OpenAI expands into areas that reduce its own dependence on Microsoft.
What this signals for AI in productivity tools
The immediate story is about Anthropic AI entering Office 365 apps. The larger story is about how major software platforms may choose AI systems going forward.
Rather than treating one model provider as the answer for every feature, Microsoft appears to be selecting among providers and internal models. That approach can let the company match different AI systems to different tasks, such as presentation design, coding assistance, or voice-related capabilities.
The source article does not say when Anthropic-powered Office 365 features will appear, how they will be branded, or which users will see them first. It also notes that TechCrunch reached out to Anthropic for comment.
What is clear from the available facts is that Microsoft’s AI stack is becoming less dependent on one partner. OpenAI remains part of the plan, but Anthropic, xAI’s Grok, Claude in GitHub Copilot, and Microsoft’s own MAI-Voice-1 and MAI-1-preview now sit inside a broader picture.
For Office 365, that could make the software less about a single AI relationship and more about a portfolio of models. For Microsoft, it is a way to keep OpenAI close while making sure its most important productivity apps are not tied to only one source of AI.