GitHub and Microsoft are moving closer to the center of MCP, Anthropic’s standard for linking AI models with the places where data and software functions already live. The companies are joining the MCP steering committee and say they will support the protocol across major parts of their platforms.
The announcement was made at Microsoft’s Build 2025 conference on Monday. It arrives as MCP gains attention from major AI companies, with OpenAI and Google having said earlier this year that they would support the standard in their own AI products.
Why MCP matters for AI apps
MCP is designed to solve a practical problem in AI development: models often need access to outside systems to do useful work. Those systems can include business tools, software, content repositories, and app development environments.
Instead of treating each connection as a one-off integration, MCP provides a common way for developers to connect AI-powered applications to data sources. In plain terms, it gives developers a structure for letting a chatbot, workflow, or other AI app reach into approved systems when a task requires it.
The source article describes two key parts of that setup. Developers can expose data through MCP servers. They can also build MCP clients, such as apps and workflows, that connect to those servers on command.
That two-way model is important because it frames MCP as more than a simple data lookup mechanism. It is meant to help AI applications interact with systems where information and functionality already exist, while giving developers a defined pattern for building those connections.
Microsoft and GitHub move from support to governance
GitHub and Microsoft are not just adding compatibility. They are joining the steering committee for MCP, placing them inside the group that helps guide the standard’s direction.
Microsoft is GitHub’s corporate parent, and the two companies say they plan to deliver broad first-party support for MCP across their platforms and services. The source specifically names Microsoft Azure and Windows 11 as part of that effort.
For developers, first-party support matters because it can make MCP feel less like an outside add-on and more like a built-in path for connecting AI applications to existing tools. If the protocol is available across widely used platforms, developers can build around it with fewer assumptions about where it will run.
The broader industry context also matters. The source notes that OpenAI and Google said earlier this year that they would support MCP in their respective AI products. With GitHub and Microsoft now joining the steering committee, MCP is gaining support from companies that influence both AI product development and software development workflows.
Windows 11 will expose more app functionality to models
One of the clearest product implications is coming to Windows. Microsoft and GitHub say that, in the next few months, Windows will gain MCP integrations that let developers expose app functionality to MCP-enabled models.
The basic idea is that developers will be able to wrap selected features and capabilities in their apps as MCP servers and make them available for Windows. That could let an AI-powered application interact with app functions through MCP instead of requiring a separate custom bridge for every use case.
The source also says Microsoft is planning to expose certain Windows system functionalities as MCP servers. The examples named are File System, Windowing, and Windows Subsystem for Linux.
Those examples show how MCP could apply at several layers of the computing environment. It can connect AI models to application features, but it can also give models a structured way to interact with parts of the operating system environment when developers make those capabilities available.
Security and discovery are part of the push
Microsoft and GitHub are also contributing to the MCP standard itself. Their work is not limited to product integration.
Microsoft says its identity and security teams worked with Anthropic, the MCP steering committee, and the broader MCP community on an updated authorization specification. The goal is to let MCP-connected apps connect more securely to MCP servers.
According to the source, the updated authorization spec enables people to use trusted sign-in methods to grant AI-powered apps access to data and services. The examples given include personal storage drives and subscription plans.
That focus is central to MCP’s usefulness. If AI apps are going to connect with data sources and services, developers and users need a way to control access. Authorization is the layer that determines whether an AI-powered app can reach a particular resource and under what conditions.
GitHub is contributing on a different but related problem: discovery and management. GitHub says it worked with the MCP steering committee to design a registry service for MCP servers.
The registry service allows developers to implement public or private centralized repositories for MCP server entries. It is also intended to help with discovery and management of different MCP implementations and their associated configurations.
What this signals for developers
The announcement points to MCP becoming a more visible part of the AI development stack. A protocol for connecting models to data sources is most useful when developers can rely on it across tools, operating systems, cloud services, and app environments.
For now, the concrete details from the announcement are focused on committee participation, first-party platform support, Windows integrations, authorization work, and server registry design. Together, those pieces address several practical needs:
- Making MCP available across Microsoft and GitHub platforms.
- Letting developers expose app features and system functionality through MCP servers.
- Improving how MCP-connected apps receive permission to access data and services.
- Helping developers find and manage MCP server implementations.
The significance is not that every AI app will immediately work with every data source. The source does not make that claim. The more grounded takeaway is that major software and AI infrastructure companies are aligning around MCP as a common method for connecting models, applications, and systems.
As AI tools become more task-oriented, the quality of their connections to real software environments will matter. Microsoft and GitHub’s support gives MCP a larger role in that conversation, especially for developers building AI-powered applications on platforms such as Azure, Windows 11, and GitHub.