{
"title": "Windows 11's August 2025 Update KB5063878 Causing NVMe SSDs to Vanish Under Heavy Writes",
"content": "A mid-August cumulative security update for Windows 11 24H2 is causing a subset of NVMe solid-state drives to vanish from the system during sustained write operations, throwing users into a spiral of blue screens, missing drives, and potential data corruption. The patch, identified as KB5063878 (OS Build 26100.4946) and released on August 12, 2025, has triggered scattered but reproducible reports that some SSDs become unresponsive or temporarily disappear when transferring large files—often around 50 GB or more. While Microsoft has yet to officially list the storage failures as a known issue, independent testers and community forums have lit up with detailed symptom patterns and urgent mitigation advice.
A Troubling Pair of Storage Regressions
The current uproar follows an earlier, separate incident tied to the initial Windows 11 24H2 feature rollout in late 2024. That first wave affected Western Digital SN770 and SN580 DRAM-less NVMe drives, causing repeated Blue Screen of Death (BSOD) crashes during normal use. The root cause, as documented in a Microsoft Q&A thread, stemmed from how the new operating system version allocated Host Memory Buffer (HMB)—a portion of system RAM used as a cache for DRAM-less SSDs. Windows 11 24H2 increased the maximum HMB allocation to 256 MB, but the WD SN770 firmware only supported 64 MB of host cache; the SN580 was designed for 200 MB. The mismatch pushed the controllers beyond their limits, leading to stalls and read/write failures. A community-discovered registry tweak that capped HMB allocation provided a temporary fix for many users, and Western Digital later issued firmware updates to address the compatibility issue.
Now, KB5063878 introduces a new, more insidious storage regression. Rather than just triggering BSODs on boot or during light workloads, the August 2025 cumulative update causes certain NVMe SSDs to abruptly drop off the Windows storage topology under heavy sequential write stress. The failure mode is distinct: the drive disappears from Device Manager and Disk Management mid-operation, and in some cases, files written during the outage become corrupted or unrecoverable. Reboots may temporarily restore visibility, but the underlying instability persists until the offending update is rolled back or a firmware fix is applied.
Symptoms: How the Failure Presents
Community reports and independent testing paint a consistent picture of the problem. Affected users typically experience:
- Sudden disappearance of an NVMe SSD from the operating system while performing large file transfers, game installations, or disk cloning operations.
- Event Viewer logs showing storage controller errors, NVMe command timeouts, or unexpected device removals.
- SMART attributes and controller telemetry becoming unreadable to vendor diagnostic tools immediately after the incident.
- Following a reboot, the drive may reappear, but files that were being written at the time of failure are often corrupted or missing.
- In some instances, the system may stall completely and require a hard reset, though Blue Screens are less common in this regression compared to the earlier HMB-related crashes.
Which Drives Are Affected?
Pinpointing an exact list of vulnerable models is challenging because the issue is not tied to a single vendor or controller. However, community-collated data points to clusters involving specific Phison controller families—commonly found in many DRAM-less NVMe SSDs from vendors like Corsair, PNY, TeamGroup, and even some Samsung models. Western Digital drives, particularly the SN770 and SN580 models that suffered under the initial 24H2 HMB flaw, are once again showing up in user reports, but this time alongside devices from other manufacturers that use similar Phison controllers. Isolated reports also mention drives with non-Phison controllers, indicating that the Windows storage stack change may be exposing firmware edge cases across multiple vendors rather than a single point of failure.
It is crucial to interpret these community lists as investigative starting points, not definitive recalls. Many variables—firmware version, drive capacity, system configuration, and even PCIe signaling—can influence whether a given SSD exhibits the regression. Until Microsoft or individual vendors publish official compatibility tables, caution is warranted for any DRAM-less NVMe drive paired with Windows 11 24H2 and the August 2025 cumulative update.
The Technical Landscape: HMB and Controller Edge Cases
While the exact mechanism behind the KB5063878 regression is still under investigation, several plausible technical threads emerge from the evidence. At the highest level, the update appears to alter how the Windows storage stack handles buffered I/O, command ordering, or timing in a way that stresses NVMe controller firmware during sustained writes. Three interrelated theories dominate the discussion:
Host Memory Buffer (HMB) allocation changes – Even though the earlier HMB issue primarily caused BSODs, it demonstrated that changes in how Windows allocates system memory for SSD caching can destabilize DRAM-less controllers. The August update may have introduced additional HMB policy tweaks—either intentional or as a side effect of other optimizations—that once again push certain controllers past their safe operational limits.
Buffered I/O or NVMe command timing regressions – Under heavy write loads, an SSD controller must juggle host commands, internal metadata updates, wear-leveling, and garbage collection. A small shift in kernel buffering strategies or command queue depth could create timing collisions that the controller firmware cannot handle, resulting in a firmware hang and the drive effectively dropping off the PCIe bus.
Controller firmware edge cases under high utilization – Phison-based controllers, while widely used and generally reliable, have distinct recovery and error-handling routines. Aggressive sequential writing can saturate internal buffers and mapping tables, and if the host OS modifies its I/O pattern even slightly, it could expose latent firmware bugs that cause the controller to become unresponsive. The fact that similar failures have appeared on non-Phison drives suggests that the root cause is likely a host-side change that uncovers multiple firmware weaknesses.
Conclusive attribution will require joint telemetry from Microsoft and SSD vendors. Community reproductions and diagnostic artifacts are valuable for building hypotheses, but they lack the deep integration needed to isolate whether the trigger is a specific NVMe command sequence, a timing regression, or a misconfigured HMB policy.
Vendor and Microsoft Responses
As of this writing, Microsoft’s official KB page for KB5063878 does not list storage failures as a known issue. The update was delivered through the standard Windows Update channel and includes the usual monthly security and quality fixes. However, the company has a history of using its Known Issue Rollback (KIR) mechanism to silently disable problematic code paths when regressions are confirmed, and it has previously stationed upgrade blocks on devices with incompatible hardware. The Windows Release Health dashboard is the primary place where such mitigations will appear if Microsoft enacts them.
On the vendor side, Western Digital—whose drives were front and center in the earlier 24H2 HMB incident—has engaged with affected users and previously released firmware updates that resolved many of the BSOD reports. Other SSD manufacturers, including those whose drives may be implicated in the new regression, have been more circumspect. Several third-party diagnostic and dashboard tools have been updated to assist in capturing the failure state, but official firmware patches targeting the KB5063878 regression have not yet materialized. The community-driven nature of the investigation means that users are often the first detectors, with vendors and Microsoft playing catch-up.
Practical Guidance for Users and IT Admins
The immediate priority for anyone running Windows 11 24H2 with an NVMe SSD is to avoid data loss. The following tiered checklists synthesize vendor guidance, community mitigations, and Microsoft update mechanics.
Immediate Consumer Checklist
- **