Key Takeaways
- Microsoft has unveiled a new strategy for Windows 11, focusing on hybrid intelligence.
- The strategy includes the general availability of Microsoft Execution Containers (MXC) on Windows 11.
- Larger AI models are being brought directly to Windows devices to enhance local processing.
Microsoft has announced a new strategy for Windows 11, centered around hybrid intelligence, which combines local and cloud AI to run tasks where they are most suitable while maintaining user and IT control.
The strategy encompasses various components, including Copilot+ PCs, developer tools, and systems designed for local AI workloads. Microsoft Execution Containers (MXC) have been made generally available on Windows 11, allowing organizations to control which files and networks an AI agent can access, with these rules enforced while the agent is running.
MXC works with Agent 365 and Intune, enabling IT teams to separate an agent’s identity and actions from those of the person using the device. Codex, GitHub Copilot, and OpenClaw already support MXC, with other AI tools expected to follow suit.
Microsoft is also bringing larger AI models directly to Windows devices. MAI Code 1.1 Flash, a 3-bit precision model, has reduced its size by nearly 80% while retaining a 256K local context window. Windows ML is adding llama.cpp support across GPU, NPU, and CPU hardware, and Microsoft is testing GitHub HydraFusion to route workloads between cloud and on-device models.
Copilot on Copilot+ PCs will use the hybrid AI layer to access local files and recent activity with user permission, performing actions in Windows and calling models running directly on the device. These capabilities are expected to roll out over the coming months across Copilot Home, Code, and Autopilot.
Windows Search is also being expanded to allow users to perform system actions directly from the taskbar, such as toggling settings, managing windows, and sending messages without manually opening relevant apps or menus.
Microsoft’s hardware strategy spans from always-on mini PCs to RTX Spark builder systems and DGX Station for Windows machines, with DGX Station for Windows expected later this year and designed to support AI models with up to 1 trillion parameters.
The broader strategy connects Windows, Copilot, GitHub, and Nvidia hardware into one platform designed for local AI and secure agents across personal PCs and shared enterprise systems, positioning Windows as a platform for running AI agents and large models on the machine itself.
