Microsoft’s August 12, 2025 cumulative update for Windows 11 24H2, KB5063878 (OS Build 26100.4946), is causing certain NVMe SSDs to abruptly disappear from the system during sustained file transfers, and in some cases, the drives never come back. Independent testers and multiple specialist outlets have reproduced the issue consistently: copy around 50 GB of data or more in one go, and the target SSD may drop out of File Explorer, Disk Management, and Device Manager, often with corrupted files and unreadable controller telemetry. The update, which was meant to deliver routine fixes and security patches, has instead surfaced a narrow but devastating storage regression that Microsoft had not acknowledged at the time the first reports emerged.
A Routine Patch with a Dangerous Surprise
KB5063878 is a standard cumulative update for Windows 11 version 24H2, released through Windows Update, WSUS, and the Microsoft Update Catalog. Its support page lists only installation guidance, with the blunt note that Microsoft was “not currently aware of any issues with this update.” That phrasing now looks painfully optimistic. Within days of rollout, users on forums and expert channels began reporting identical failures: large file copies or game installations would abruptly stop, the target NVMe drive would vanish from the OS, and subsequent reboots would not always bring it back.
Specialist outlets like Tom’s Hardware, Club386, and IT Pro, along with industry commentators such as BornCity, quickly picked up the story and ran controlled tests. They confirmed the symptom fingerprint—controller hang, disappearance from the OS, corrupted transfers—and narrowed the trigger to sustained sequential write workloads of roughly 50 GB or more. In several reproductions, the drives became completely unresponsive until power-cycled, and files written during the failure window were truncated, garbled, or simply gone.
How the Failure Unfolds
Users report the same chilling sequence. You start transferring a large video project, a disk image, or a game installation to an NVMe SSD. Everything looks normal for tens of gigabytes. Then, the progress bar freezes, Windows throws an error, and the drive icon vanishes. Opening Disk Management shows only remaining drives. Device Manager shows either no trace of the NVMe device or a yellow warning icon. Vendor diagnostic tools like WD Dashboard or Crucial Storage Executive cannot communicate with the drive; SMART data returns as unreadable or all zeroes.
A reboot sometimes restores the drive’s visibility temporarily, but not always. In a minority of reports, the drive remains inaccessible even after a cold boot, requiring vendor-specific recovery tools or professional data recovery. The forum aggregations note that “files written during the incident may be partially written, corrupted, or lost,” underscoring the real risk to data integrity.
Why Sustained Writes Expose Firmware Bugs
Modern NVMe SSDs hide enormous complexity behind a tiny M.2 form factor. They combine NAND flash, a multi-core controller, DRAM or Host Memory Buffer (HMB) caching, and sophisticated firmware that manages wear leveling, garbage collection, and error correction. Short, bursty desktop workloads rarely stretch these systems; sustained sequential writes, however, stress all of them at once.
Under continuous write pressure, the controller must manage an ever-growing mapping table, flush data from volatile caches, juggle garbage collection cycles, and throttle thermally—all while the host OS pours more data through DMA queues. If the firmware encounters an unhandled state—perhaps a race condition between internal tasks or an exhaustion of HMB resources—it may lock up entirely. The host then sees no response, times out, and removes the device from the storage stack.
Previous Windows 11 24H2 patches triggered BSODs on some DRAM-less SSDs because of changes to HMB allocation. KB5063878 appears to touch a similar host–firmware interface but in a different way. Community testers observed that the failure typically occurs when controller utilization hits around 60% and after roughly 50 GB of written data. That pattern suggests a caching or resource management edge case rather than a simple bug in the NVMe driver. When the controller goes silent, SMART values become unreadable because the host cannot even query the device.
Which SSDs Are Most at Risk?
Early evidence points strongly toward DRAM-less NVMe SSDs using Phison controllers. Several affected models from Western Digital, SanDisk, and other brands share Phison E-series controller architectures, many without on-board DRAM. These drives rely heavily on HMB to borrow a slice of system RAM for mapping tables. Any host-side change that alters HMB timing or size can rattle their firmware. Earlier in the 24H2 lifecycle, certain WD and SanDisk models needed firmware updates to fix HMB-related blue screens; this new regression feels like a sibling, triggered by sustained stress rather than boot-time initialization.
However, reports are not confined to a single controller family. Some community tests have flagged other NVMe drives, and a few HDDs have been mentioned, though those likely represent separate issues. No consolidated, vendor-verified list of vulnerable models exists. The forums stress that “claims that a single controller family is the exclusive culprit remain premature.” Until OEMs publish advisories, any DRAM-less NVMe SSD on an updated 24H2 machine should be treated as potentially vulnerable.
What You Should Do Right Now
If you’ve installed KB5063878, stop heavy sequential writes immediately. Don’t copy that 4K video project to an internal NVMe drive; don’t unzip a massive game repack; don’t clone a disk. Break large operations into chunks smaller than 50 GB where absolutely necessary, but ideally, postpone them entirely.
Back up critical data now. Target an external USB drive or a cloud service, not another internal NVMe that might share the same vulnerability. If your important files are only on a potentially affected SSD, make a verified copy before you do anything else. Backups are your only reliable defense against write-time corruption.
Use your SSD manufacturer’s toolbox—WD Dashboard, Crucial Storage Executive, Corsair SSD Toolbox, etc.—to identify the exact model and firmware version. Check the vendor’s website for firmware updates and release notes. If a new firmware explicitly addresses “drive dropouts under heavy write load” or similar language, apply it, but only after you’ve made a backup. Firmware flashes carry their own small risk, so follow the vendor’s instructions precisely.
For IT Administrators: Hold Your Rollouts
If you manage updates through WSUS, SCCM/MECM, or Intune, put KB5063878 on hold for broad deployment. Set up a test ring with a representative sample of your hardware—especially machines with DRAM-less NVMe drives—and exercise them with controlled, sustained sequential writes. Monitor for drive dropouts, Event Viewer warnings, and vendor diagnostic logs. Expand the update only after you’ve verified that your fleet is not affected or after applying any vendor firmware fixes.
Increase testing rigor. Add a script that copies a 100 GB file across the local SSD and checks for completeness and SMART health afterward. Capture the output of Get-PhysicalDisk and Get-StorageReliabilityCounter before and after. These steps can surface failures in the staging environment instead of a production director’s workstation.
Remember that backup targets matter. If your backup job writes to the same model of SSD currently under suspicion, you risk corrupting the backup during the same failure window. Use network-attached storage, a different SSD family, or the cloud for critical backups until the situation stabilizes.
Microsoft has already applied a Known Issue Rollback for an unrelated WSUS deployment issue with this KB, demonstrating that they can and will push targeted mitigations. It’s plausible they’ll issue a compatibility hold for vulnerable SSD hardware IDs if OEM telemetry warrants it.
Diagnosis and Recovery if Your Drive Disappears
Despite all caution, you might still encounter the bug. Don’t panic-reboot. The first priority is preserving evidence and data.
- Capture logs immediately. Open Event Viewer and navigate to Windows Logs > System. Look for disk, controller, or NVMe warnings with timestamps that match the failure. Export the log. Run your vendor’s diagnostic tool in read-only mode if possible and save any output, even if it reports comm failures.
- Do not repeatedly reboot. While a reboot sometimes makes the drive visible again, rapid power-cycling can overwrite volatile controller state and complicate later forensic recovery. Instead, try a single clean cold boot (shut down, wait, power on) after collecting logs.
- Clone the drive if it reappears. If the disk becomes accessible, even briefly, use a tool like
ddor your vendor’s cloning utility to create a sector-level image to a healthy drive. This preserves everything—good sectors and corrupted areas—before any repair attempts that might modify data. - Contact vendor support. Provide the logs, drive serial number, firmware revision, and a description of what you were doing. Vendors often have low-level diagnostic commands that can revive a locked-up controller or pull internal logs for engineering analysis.
- Consider professional recovery. If the data is absolutely critical and the drive remains dead, stop all DIY efforts. Professional recovery firms can sometimes repair controller electronics or read NAND directly, but every amateur attempt reduces the chances of success.
What’s Next: Fixes on the Horizon
The industry has seen this movie before. When a Windows update exposes a firmware edge case, the resolution usually follows one of three paths, often all at once.
Vendor firmware updates. If the root cause lies in how the controller handles prolonged cache pressure or HMB management under stress, drive makers will release firmware patches. The earlier 24H2 HMB saga was resolved exactly that way, with WD, SanDisk, and others issuing updates that adjusted their firmware’s interaction with the host. Users should monitor their vendor’s support pages and apply updates as soon as they are available—after verifying backups.
Microsoft mitigation. If the issue stems from a host-side change—perhaps to the NVMe driver, storport, or HMB allocation policy—Microsoft can either release an out-of-band fix or use Known Issue Rollback to disable the offending change on affected hardware. They have also placed compatibility holds in the past, preventing the update from installing on machines with specific SSD firmware. Expect a formal entry on the Windows release health dashboard if telemetry confirms widespread impact.
Combined approach. Most complex storage bugs require both sides to adapt. The operating system works around the most egregious firmware quirks, and drive vendors tighten their code to handle the new OS behavior. Historically, this double fix arrives within weeks, not days, so patience and interim workarounds are necessary.
In the meantime, independent testers and community trackers will continue to expand the list of affected models and firmware revisions. Keep an eye on the forums and specialist press for updates, but remember that only vendor-approved advisories should guide your firmware update decisions.
The Bigger Picture
KB5063878 is a stark reminder that cumulative updates are not just bundles of nice-to-have fixes. They touch the deepest layers of the OS, including the storage subsystem, and can interact with hardware in unexpected ways. SSD firmware has become an intricate layer of software between your data and the physical NAND, and the host’s cooperation—through HMB, trim commands, and queue management—is essential to that firmware’s stability.
The episode also reinforces timeless IT hygiene. Robust backups remain the single most effective defense against any sort of data corruption or loss. Staged rollouts, whether in a home lab or a global enterprise, catch such regressions before they can eat through the most critical systems. And when something does go wrong, methodical diagnosis—stop heavy writes, collect logs, image the drive, contact the vendor—outweighs frantic button-mashing every time.
Windows 11 24H2 users with NVMe SSDs should assume nothing until either Microsoft or their drive manufacturer says otherwise. Keep your backups current, avoid marathon write sessions, and watch for firmware updates. The fix will come, but until it does, treat your SSD with a little extra caution.