Microsoft's KB5094126 adds NPU monitoring for IT teams
Now IT admins can see NPU activity in Task Manager to verify AI hardware before rollout.
Microsoft’s June 2026 Windows 11 update, KB5094126, introduces native NPU monitoring in Task Manager for supported AI PCs. The update applies to Windows 11 versions 24H2 (build 26100.8655) and 25H2 (build 26200.8655). IT teams can now add optional NPU and NPU Engine columns on the Processes, Users, and Details pages, while the Details page also tracks dedicated and shared NPU memory. The Performance page can display neural engines built into GPUs. This shifts AI PC validation from vendor claims to verifiable endpoint checks, aligning with Microsoft’s Copilot+ PC definition of an NPU capable of over 40 trillion operations per second.
For enterprise deployment, IT administrators should start by confirming the correct build number, then enable NPU columns in Task Manager. Running a supported AI workload and observing NPU activity helps verify hardware, driver status, and policy-sensitive changes before broad rollout. Low NPU activity may indicate driver or routing issues but is not definitive proof of failure—Windows ML can route to CPU, GPU, or NPU depending on app configuration. The update also includes other changes: Windows Hello now defaults to face or fingerprint when available, Multi-App Camera allows multiple apps to share a camera stream, and two-character file search improvements. These features require pilot testing against existing policies. KB5094126 gives IT teams a concrete tool to validate AI PC readiness, reducing reliance on marketing claims.
- Task Manager now shows NPU and NPU Engine columns on Processes, Users, and Details pages for supported AI PCs.
- Update applies to Windows 11 24H2 (build 26100.8655) and 25H2 (build 26200.8655), plus includes changes from May 26 preview.
- IT teams can verify Copilot+ PC hardware (40+ TOPS), check driver status, and test AI workload offloading before enterprise rollout.
Why It Matters
Enables IT teams to independently verify AI PC hardware and software readiness before enterprise deployment.