Windows 11 Build 28120.3032 Brings Full-Wipe Cloud Rebuild

Microsoft detailed the September 18 release in its Windows Insider documentation, while Windows Report highlighted the recovery feature alongside fixes for Hyper-V folder sharing and an Easy Anti-Cheat startup error. The company has previously introduced Cloud rebuild in the broader Experimental channel, but build 28120.3032 is the point at which the option arrives in the Experimental 26H1 branch.

For an Insider running the affected build, the appeal is straightforward. A damaged Windows installation can no longer be trusted to supply its own recovery files; conventional local reset options may depend on components stored on that same damaged disk. Cloud rebuild instead obtains the selected Windows build plus drivers through Windows Update, then sends the machine through the standard out-of-box setup experience.

Cloud rebuild is a full wipe, not a repair install​

The new option appears in Windows Recovery Environment, or WinRE, under Troubleshoot > Recovery and uninstall > Cloud rebuild in build 28120.3032. A user can reach WinRE automatically after repeated startup failures or manually from Advanced startup while Windows still runs. From there, Cloud rebuild establishes a network connection, displays the intended Windows build, edition, and language, and asks the user to acknowledge data loss before it begins.

That last screen deserves more attention than it will get. Microsoft’s Cloud rebuild documentation says the process reformats the system disk and removes locally stored files, accounts, applications, and settings. It is not equivalent to an in-place repair installation, nor is it a “keep my files” reset option. Files stored only on the PC are gone unless they were backed up before the failure; cloud-synced OneDrive data survives because it resides elsewhere, but it must still be available and synchronized to be useful.

The feature therefore fills a gap between a standard reset and hands-on reimaging. For a home user, it can replace the familiar scramble to locate another computer, download an ISO, create installation media, enter firmware boot settings, and reinstall Windows. For an IT team with properly configured Autopilot and Intune enrollment, it can provide a cleaner route back to a managed state after the device completes setup.

Microsoft’s documentation says an Autopilot-registered, Entra-joined and Intune-managed device can reconnect during setup, receive its assigned policies and apps, restore supported settings through Backup for Organizations, and make OneDrive files available once the user signs in. That is an operational advantage over a technician-led manual rebuild, but only for organizations that already have that deployment plumbing in place. Cloud rebuild cannot manufacture an endpoint-management strategy after a failure.

Internet access and driver coverage are the practical limits​

The name “Cloud rebuild” is accurate: the recovery workflow is dependent on the cloud. WinRE needs a usable internet connection, and the device must have networking support available before Windows itself has been reinstalled. Wired Ethernet connects automatically, according to Microsoft. Wireless is more restrictive: Microsoft’s current preview documentation specifies WPA-Personal Wi-Fi support in WinRE, rather than presenting it as a universal Wi-Fi solution.

That restriction could matter in offices, schools, hospitals, and other managed environments where enterprise Wi-Fi authentication is common. A laptop that can only reach the network through an enterprise SSID may still require Ethernet, a compatible temporary network, or traditional installation media. The recovery screen itself cannot solve an authentication or driver problem that prevents it from getting online.

Microsoft also requires the manufacturer to have included a compatible WinRE networking driver and to have the necessary device drivers published through Windows Update. A missing storage or networking driver can stop the rebuild even though the PC is otherwise perfectly capable of running Windows 11. Microsoft documents error 0xc1900200 for cases involving either Windows 11 hardware eligibility or a required driver absent from Windows Update; it specifically calls out TPM configuration as a common cause and identifies missing drivers as another.

That creates an important distinction between reinstalling Windows and recovering a particular PC. The downloaded operating-system image may be available, but a recovery process still depends on a working network adapter in WinRE and suitable drivers for the hardware after installation. Administrators supporting unusual storage controllers, older hardware, or tightly controlled driver catalogs should keep tested USB recovery media available. Cloud rebuild reduces reliance on that media; it does not eliminate it.

Microsoft’s guidance also warns users not to interrupt the download or installation. A power loss or manual restart during the rebuild can leave the device unable to boot, at which point Microsoft still directs users toward a recovery drive, Windows installation media, or vendor-supplied recovery media. Keep the system on AC power throughout the process.

The 26H1 release is still an Experimental flight​

Build 28120.3032 is not a general Windows 11 update. Microsoft labels the branch Experimental, says features can arrive with limited documentation, and uses Controlled Feature Rollout to expose changes gradually to only a subset of eligible Insiders before broader availability within that channel. Installing the build does not guarantee that every Insider will see every new capability immediately.

There is another limitation in the version name. Microsoft described Windows 11 26H1 in June as a targeted release for particular new silicon, including Qualcomm Snapdragon X2 Series devices, and generally recommended that other Insiders remain on the default Windows core version. The 26H1 Experimental branch is therefore a poor place to treat a recovery tool as production-ready infrastructure.

Microsoft’s own Cloud rebuild documentation is blunter: the feature is in preview and should be evaluated only on non-production devices. That warning is sensible. Recovery tooling has an unusually high bar because it is used when a system is already in a degraded state, often when the user cannot test alternatives or retrieve missed data.

The feature also should not be confused with Microsoft’s Quick Machine Recovery work, which is intended to help devices recover from certain widespread Windows startup problems. Cloud rebuild is a full local operating-system reinstall initiated through recovery; it does not diagnose the cause of a boot failure, preserve the existing application state, or selectively reverse a bad update. It is the more drastic option.

Hyper-V and Easy Anti-Cheat fixes remove two immediate irritants​

The same build fixes a Hyper-V issue affecting applications that use HCS-managed virtual machines. Microsoft says Windows host folders shared with Linux guest VMs through Plan9 sometimes failed to appear in the guest or appeared but could not be accessed. For developers relying on Linux VMs and host-mounted source trees, that failure could look like missing project content or a permissions problem even when the share configuration was correct.

Plan9 folder sharing is a technical detail, but it matters in workflows where a Windows host exposes source code, build artifacts, or configuration files to Linux guest environments. The fix is welcome, although Microsoft has not published a broader root-cause analysis or identified the specific HCS-managed applications most affected. Organizations using that setup should validate mounts after installing the preview rather than assuming every file-sharing scenario is covered.

Microsoft also marks the Easy Anti-Cheat issue as resolved. The prior build, 28120.3002, listed a known problem in which Easy Anti-Cheat could fail to launch with error 0xc000007b, a Windows application-startup status often associated with invalid executable or library loading. The release notes do not name the affected games or explain the underlying correction, so the practical test remains simple: users who hit the error should install 28120.3032 and retry the game before rebuilding runtimes, reinstalling the title, or changing anti-cheat files manually.

Treat it as a recovery option to test before disaster​

Cloud rebuild’s real contribution is removing the installed Windows image from the recovery chain. When the operating system is corrupt enough that local recovery is questionable, obtaining the image and drivers externally is a defensible design. The hidden dependency is that the machine must still enter WinRE, connect to an acceptable network, meet Windows 11 requirements, and find its needed drivers through Windows Update.

For now, build 28120.3032 gives 26H1 Experimental Insiders a useful tool to trial on spare hardware, particularly on a device with Ethernet and a verified OneDrive or organizational backup. It does not replace tested backups, recovery media, or documented re-provisioning procedures. When the machine will not boot is the worst possible time to discover that its Wi-Fi is unsupported in WinRE or that its storage driver was never published to Windows Update.

Update: Report says Cloud rebuild may gain drive sanitization and remote IT controls (September 19, 2026)​

According to TheWinCentral, Microsoft has added optional drive sanitization to Cloud rebuild in newer Experimental-channel testing. Unlike the existing reformat process, the reported option would overwrite storage sectors to make previously deleted local data substantially harder to recover—an important distinction for device retirement, reassignment, or incident response.

The outlet also reports that Microsoft has expanded the Recovery Configuration Service Provider so administrators can configure or initiate Cloud rebuild remotely through MDM platforms such as Intune. If confirmed in Microsoft’s documentation, that would shift Cloud rebuild beyond a user-operated WinRE recovery feature toward a practical remote remediation option for managed endpoints.

These capabilities would not make Cloud rebuild less destructive: sanitization would make data recovery even less feasible, while remote triggering would require IT teams to set clear authorization, backup, and enrollment safeguards before using it on employee devices.

Update: Remote Cloud rebuild may support customized target images (September 21, 2026)​

Techzine Global reports that Microsoft’s newer Experimental build 26340.9502 adds more detailed Recovery CSP controls for remotely initiated Cloud rebuilds. Beyond triggering a reinstall, IT administrators could reportedly choose the device’s current Windows build or specify a different target build.

The report says a customized rebuild may also define the Windows edition, system language, and cumulative-update version, with support for an unattended installation file. That would make the process more suitable for standardized endpoint recovery without requiring an administrator on site.

For Windows admins, the practical implication is a potential shift from emergency remote wiping to controlled remote reprovisioning. These controls remain Insider testing features, and organizations would still need to validate driver availability, WinRE networking, enrollment behavior, and data-protection approvals before relying on them.

Update: Cloud rebuild adds per-drive sanitization choices (September 21, 2026)​

According to Windows Report, build 26340.9502 lets users choose between standard file removal and “Remove & sanitize files” before a Cloud rebuild. The latter asks the storage hardware to securely erase data before Windows is reinstalled, while the standard option is faster but may leave recoverable remnants.

The report says systems with multiple disks can apply sanitization to only the Windows drive or to every installed drive. That makes the feature more useful for device retirement and reassignment, but also raises the risk of unintentionally erasing secondary data volumes.

Windows Report also notes that Windows relies on the storage device’s reported completion status and cannot independently verify that a hardware-level sanitization operation succeeded. IT teams handling regulated or highly sensitive data should therefore validate their device firmware and disposal procedures rather than treating the new option as standalone proof of secure erasure.