Microsoft appears to be moving GitHub Copilot toward a more self-reliant future by developing its own AI coding model. The shift signals a broader effort to depend less on OpenAI and build more of the company’s AI stack in-house.
The plan is expected to be discussed at the Build developer conference in San Francisco next week. At the same time, Microsoft is facing a more crowded market for AI coding assistants, with Anthropic’s Claude Code becoming a more serious rival.
According to The Information, Microsoft is building several internal AI models for different tasks. The lineup reportedly includes systems for coding, image generation, voice, reasoning, and transcription.
Coding appears to be the central focus. The technology is expected to become an important part of GitHub Copilot, Microsoft’s flagship tool for developers.
A broader shift in Microsoft’s AI strategy
The move reflects a larger change in Microsoft’s AI direction. In recent months, Microsoft and OpenAI have reportedly revisited parts of their partnership structure as both sides look to reduce dependence on each other, according to The Verge.
That development gives Microsoft a stronger incentive to expand its own model capabilities. With internal models, the company would have more control over product direction and technical development.
GitHub Copilot currently relies on technology from multiple AI companies, including OpenAI, Anthropic, and Google. But the rise of more coding assistants has made the market far more competitive than it was in the early wave of generative AI.
Anthropic is putting pressure on the market
The strongest pressure appears to be coming from Anthropic. Claude Code has gained momentum and is pushing Microsoft to move more aggressively on an in-house alternative.
Microsoft itself recently gave thousands of employees access to Claude Code, according to reports. Workers across different divisions reportedly used it to experiment with AI-assisted software development.
Its internal popularity grew quickly because it made coding tasks easier to try, even for non-technical employees. That broad accessibility helped make it a useful testing tool inside the company.
Still, Microsoft’s direction is now changing. The company is reportedly planning to reduce internal use of Claude Code by the end of June and steer teams toward a Copilot-based command-line tool instead.
Cost and control are also part of the equation
Competition is not the only factor shaping the decision. The Verge reported that reducing reliance on third-party AI software could help Microsoft lower operating costs as it enters the next fiscal year.
A move toward in-house models also brings strategic advantages. With fewer external dependencies, Microsoft would have more flexibility over integrations, feature development, and Copilot’s business direction.
Reuters also reported that Microsoft has explored acquiring AI startups to strengthen its internal expertise. The goal is to improve its chances of building advanced AI systems that can compete with the best offerings in the market.
That push comes as Anthropic continues to grow quickly. Its expansion has been supported by the popularity of Claude Code and its other AI products.
Anthropic recently announced a new $65 billion funding round. Its valuation was said to have risen to nearly $965 billion, placing it ahead of several major rivals in the AI industry.
The company also disclosed an annualized revenue run rate of around $47 billion. That figure highlights how quickly demand for AI coding tools is expanding across the market.
For Microsoft, Build could become an important stage to show that it is no longer just a distributor of partner technology. If the new coding model becomes a centerpiece of the event, GitHub Copilot’s competition with OpenAI and Anthropic will enter a new phase.
