Microsoft has started replacing AI models from OpenAI and Anthropic with its own AI models in apps like Excel and Outlook to reduce AI costs. The company is now using its own MAI (Microsoft AI) models to answer tens of thousands of AI prompts every week in Excel and Outlook, according to Bloomberg, citing a person familiar with the matter.
Earlier, these AI tasks in Excel and Outlook mostly depended on AI models from OpenAI and Anthropic. The person who shared this information spoke on the condition of anonymity because the matter is internal. This is the first time the scale of Microsoft’s MAI model usage in workplace apps has been revealed.
Microsoft AI plans
Microsoft did not officially comment on the report, as a company spokesperson declined to comment. Right now, Microsoft’s own AI models still handle only a small part of the company’s overall AI work, according to Bloomberg. Even though the share is small, the move shows Microsoft is becoming more capable of building strong AI models at a lower cost.
Microsoft wants to reduce the amount of money it spends on outside AI companies by relying more on its own technology. Microsoft’s AI chief Mustafa Suleyman said in June that the company was already trying to reduce spending on Anthropic by increasing the use of MAI models. Microsoft uses a huge number of AI “tokens” every day to power products like Copilot.
Also read: Xbox cuts 3,200 jobs, divests five studios in major Microsoft overhaul
Why costs matter
AI tokens are the basic units used to measure how much AI computing is being used. Microsoft currently gets a large amount of AI technology from OpenAI at discounted rates because of their long-term partnership. However, Microsoft knows this discounted arrangement may not last forever. Because of that, Suleyman’s team is preparing Microsoft to become less dependent on OpenAI and other outside AI companies.
The goal is to avoid paying whatever prices leading AI companies may charge in the future. At Microsoft’s annual Build developer conference in June, the company introduced seven new AI models. One of those new models can match the coding performance of an older Anthropic AI model called Opus 4.6, but at a lower cost. Suleyman openly said, “We pay a lot of money to Anthropic — so our goal is to reduce and ultimately eliminate that cost,” according to Bloomberg.
New AI tools
Microsoft’s MAI models are already available inside GitHub Copilot, which helps software developers write code with AI. Suleyman also said Microsoft will soon start using its own AI transcription model in Teams video meetings and other Microsoft products in the coming months.
According to CryptoBriefing, Microsoft’s move shows it wants to keep more of the AI business and profits inside the company instead of depending on outside AI providers. This strategy could weaken Anthropic’s position in the AI market because Microsoft may buy fewer AI services from it. Investors may now have less confidence that Anthropic will reach its expected high valuation by December 31, 2026.
What happens next
Investors should now closely watch how OpenAI and Anthropic respond to Microsoft’s strategy. Any new partnerships, funding announcements, or business updates from Anthropic could affect how investors value the company, according to CryptoBriefing. Microsoft’s success in rolling out its own AI models across more products will be important in shaping the future AI market.
Overall, Microsoft’s latest move signals a bigger push to become more independent in AI, lower long-term costs, and reduce its reliance on OpenAI and Anthropic.
