Microsoft Reduces Reliance on OpenAI Models to Cut AI Costs

A Strategic Shift at the Heart of Microsoft's Ecosystem

AI costs have become an increasingly heavy burden for major technology companies, prompting many to reassess their strategies and rein in spending. Against this backdrop, recent reports reveal that Microsoft has begun actively reducing its reliance on AI models sourced from OpenAI and Anthropic, gradually replacing them with internally developed models of its own.

MAI Models Enter Word and Excel

According to Bloomberg, Microsoft has started deploying its proprietary models — known as MAI — to handle a portion of user requests within Word and Excel, two of its most widely used applications worldwide. The company had previously promoted Office 365 as being largely powered by its two primary external providers.

This does not, however, mean Microsoft is abandoning those partnerships entirely. The company continues to rely on external models in specific contexts, but is simultaneously working to expand its internal ecosystem and strengthen its independent capabilities.

Seven New Models Announced at Build Conference

In line with this direction, Microsoft announced the launch of seven new MAI models at its annual Build conference last month, including:

• A model specialized in agentic coding (Agentic Coder)
• A text-to-image generation model
• Additional models targeting a range of enterprise workplace applications

Microsoft declined to comment further when contacted by specialist media outlets inquiring about these developments.

A Broader Wave of Rationalization Sweeping the Tech Sector

Microsoft is not alone in this trend — it is part of a wider phenomenon reshaping the AI industry. Following a period of rapid spending earlier this year, numerous major companies have begun making more cautious and measured decisions. Reports indicate that Amazon, Meta, Uber, and Accenture are also reconsidering the scale of their AI service expenditure.

Some Silicon Valley companies have even begun looking to Chinese AI models as a lower-cost alternative, though this comes with serious questions around security and privacy.

What Does This Shift Mean?

This trend reflects a growing maturity in how large technology institutions approach AI. After a period of fascination and unconditional adoption, these companies are moving into a more measured phase defined by careful cost-benefit analysis. If Microsoft — with its substantial resources and expertise — succeeds in developing its own models at scale, it may well encourage more companies to follow the same path, potentially redrawing the map of AI provider dependency in the years ahead.

✦ بقلم فريق دروب أيديا