According to Watchtower Beating monitoring, Microsoft plans to announce multiple in-house AI models at its annual Build developer conference in San Francisco next week. In order to regain lost ground in the AI coding market, Microsoft is focusing on unveiling a proprietary programming model to enhance the competitiveness of GitHub Copilot, countering the market erosion caused by Cursor and Claude Code.
In addition to the programming model, Microsoft is also planning to release in-house large models in various parameter specifications, covering vertical tasks such as transcription, reasoning, speech, and image processing. The research and development team under Microsoft's AI CEO Mustafa Suleyman has not yet had a model reach the top of industry mainstream rankings in the past two years, making this upcoming release a significant test of the team's in-house capabilities. Previously, Microsoft's in-house efforts were constrained due to an exclusive agreement with OpenAI that restricted Microsoft's authority to develop human-level AI until April 2026. It was not until April of this year that the two companies revised their collaboration terms, granting Microsoft research freedom once again.
In terms of software services, Microsoft has long relied on OpenAI's technology, licensed for free until 2032, and commercially purchased Anthropic models to power GitHub Copilot and 365 Copilot. The high cost of external procurement has forced Microsoft to raise prices for certain GitHub Copilot packages and impose usage limits on Anthropic models. By distributing lower-cost in-house models through the Azure cloud platform, Microsoft can offer developers an affordable alternative to Claude, thereby reducing the daily operational costs of AI features in Office productivity software. More importantly, about 6 years from now when the free licensing agreement with OpenAI expires, in-house large models will become a key asset for Microsoft to break free from expensive vendors and maintain core independence.
