Microsoft Tests Project Zenith, a Windows 11 Build Ready for AI Coding Out of the Box

Microsoft announced Project Zenith on September 4, 2026, a new Windows 11 variant designed as a ready-to-work environment for AI developers from the very first boot. Instead of manual system configuration, a Zenith-equipped PC is meant to come with development tools already set up and the ability to run large language models locally, without a cloud connection.

Microsoft describes Zenith as an experience tied to specific devices rather than a separate Windows edition. In practice, this means a specially prepared system image that hardware makers will be able to install on machines meeting certain memory and bandwidth thresholds. The company has not yet announced a specific market release date or pricing for the software itself, focusing instead on presenting the concept and its first hardware partner.

A PC Free of Clutter

Zenith removes from the desktop and Start menu the elements developers usually disable manually, such as prompts about recently used files or sync notifications. File Explorer shows file name extensions, hidden files and full folder paths by default, and the system supports long file paths without the extra registry configuration many developers previously had to enable themselves.

Windows Terminal and Visual Studio Code are pinned to the taskbar by default, and the set of programming languages, runtimes and version control tools can be fully customized by swapping out the default components for others. Microsoft's aim is to cut down the time developers lose configuring a new machine before getting to actual work.

AI Models Without the Cloud

The key element of Project Zenith is the ability to run AI models with more than 30 billion parameters locally. This requires hardware with at least 64 GB of unified memory and bandwidth exceeding 250 GB per second, which in practice means AMD Ryzen AI Max-class processors, equipped with integrated memory shared between the CPU and GPU instead of a traditional graphics card with its own VRAM.

Project Zenith will let developers run models above 30 billion parameters locally and without limits - Logan Ayer, Vice President of the Windows Platform at Microsoft

Such a setup is meant to relieve developers of token costs paid for queries to cloud-based models and enable offline work or work in environments where sending code and data to external servers is undesirable for security reasons. Actual performance will, however, depend on the quantization of the specific model and the required context window, so not every AI task will be offloaded from the cloud to the same degree.

First Hardware and Agent Security

The first device with Project Zenith is a mini-desktop built on the AMD Ryzen AI Halo platform, equipped with a Ryzen AI Max+ 395 processor, 128 GB of LPDDR5x-8000 memory, a 2 TB SSD and integrated Radeon 8060S graphics. At Micro Center, this configuration is priced at $3,999.99, placing it in the premium hardware segment, far from a mass-market office laptop.

The system also includes a mechanism called Microsoft Execution Containers, which isolates AI agent activity from the rest of the operating system. Microsoft cautions, however, that the tool is still in an early testing phase, and some automatically generated access rules are too permissive, so the current profiles should not be treated as a full-fledged security boundary.

What It Means for Polish Companies

For Polish development teams and companies investing in workstations for AI departments, Project Zenith signals a direction in which PC makers are starting to design hardware for running large models locally, not just for accessing cloud providers' APIs. The entry cost remains high, but for companies processing sensitive customer data, the ability to keep the model and data on a single machine may prove more important than the hardware price itself.

For now, the tools underlying Zenith are available independently through the Windows Developer Configurations repository, and WSL containers have been in public preview since June 29, 2026. Microsoft has not ruled out extending the program to other partner devices, potentially including mini PCs from manufacturers other than AMD.