Major adjustment of Microsoft's AI strategy: Reduce burdens and costs, Excel and Outlook take the lead in "de-dependence" In order to optimize operating costs and build a more competitive self-developed AI technology matrix, Microsoft has recently launched an important strategic adjustment in its core software products

Major adjustment of Microsoft's AI strategy: Reduce burdens and costs, Excel and Outlook take the lead in "de-dependence" In order to optimize operating costs and build a more competitive self-developed AI technology matrix, Microsoft has recently launched an important strategic adjustment in its core software products. According to people familiar with the matter, Microsoft is gradually introducing its self-developed MAI artificial intelligence model into Excel and Outlook to replace the OpenAI and Anthropic models that it relied heavily on before. Currently, these two heavyweight office software have tens of thousands of AI tasks independently completed by MAI models every week. Although this action accounts for a small proportion of Microsoft's overall AI calls, it sends a clear signal: Microsoft has achieved a phased breakthrough in the research and development process of independently building high-level AI models. Mustafa Suleyman, head of Microsoft AI models, publicly stated in June this year that the company is increasing the frequency of MAI use to reduce high expenditures on third-party models. For a long time, the "AI computing power bill" behind office auxiliary products such as Copilot has been a major expense for Microsoft. Although Microsoft can enjoy certain technology purchase discounts through its close partnership with OpenAI, this is not a long-term solution. Through self-developed model technology, Microsoft aims to take the initiative in future AI layout and avoid being constrained by the pricing strategies formulated by leading AI laboratories. The current strategic layout does not stop there. In addition to Excel and Outlook, Microsoft's MAI model has been put into use in GitHub Copilot. According to Suleyman, in the next few months, the company will further expand the application scope of its self-developed transcription model and gradually integrate the technology into Teams video conferencing and other product lines, thereby building a more autonomous and controllable AI ecosystem. via AI News (author: AI Base)