Microsoft and SSD controller designer Phison are jointly investigating a serious storage regression introduced by the August 2025 cumulative update for Windows 11 24H2 (KB5063878, OS Build 26100.4946). Multiple users and independent testers have reproduced a scenario where certain solid-state drives—especially those using Phison controllers—stop responding and disappear from the system entirely during sustained, large sequential write operations. In some cases, the drives return after a reboot with corrupted data, raising the specter of permanent data loss.

The issue first surfaced on Japanese social media and specialist forums shortly after Microsoft shipped the update on August 12. While the official KB article initially listed only security and quality changes, community-driven testing quickly pinpointed a consistent failure profile: when an SSD is subjected to a continuous write of around 50 GB or more—common when installing large games, extracting archives, or cloning disks—the drive may suddenly become unresponsive. File Explorer, Device Manager, and Disk Management all show the device as missing, and vendor diagnostic tools report unreadable controller telemetry.

Symptoms and Trigger Profile

Across numerous independent reproductions, the failure presents with a predictable set of symptoms. An SSD (and in a few isolated reports, some HDDs) stops acknowledging commands mid-write, effectively dropping off the PCIe bus. The system may continue to run, but the drive is no longer accessible. A reboot often restores visibility, but files that were being written at the moment of failure are frequently left incomplete or corrupted. In a minority of cases, the drive does not return without vendor intervention, and some users have reported that partitions became unreadable.

The common thread in successful reproductions is a heavy sequential write workload on a drive that is already moderately full—many reports cite a fill level above 50% to 60% as a factor that increases the likelihood of failure. Workloads that stress the drive’s SLC cache and metadata management appear to be the most reliable triggers. These include large game updates (which often download and decompress tens of gigabytes in a single operation), mass media copies, and disk cloning tasks.

Testers have also noted that DRAM-less NVMe SSDs using Host Memory Buffer (HMB) technology seem to be overrepresented among the affected models. This is a significant clue because HMB-equipped drives rely on a portion of system RAM for their mapping tables, and even a minor change in the host OS’s memory allocation or driver behavior could expose latent firmware bugs. However, drives with other controller chips (such as InnoGrit and Maxio) have also been mentioned in scattered reports, so the problem is not exclusively tied to one vendor.

Affected Hardware

While no official recall list has been published, community-collated data points strongly toward Phison-based NVMe SSDs. Phison, a major controller vendor, publicly acknowledged “industry-wide effects” from the August Windows updates and confirmed it is working with partners to investigate. Many affected models are consumer NVMe drives from various brands that use Phison controllers, particularly DRAM-less designs. Early reports from Japan initially highlighted specific gaming-oriented SSDs, but as global attention grew, the list expanded.

That said, the dataset remains heterogeneous. Some widely used brands—Samsung and Seagate, for instance—appear largely absent from the initial wave of reports, but this does not guarantee immunity. Firmware revisions, specific branded SKUs, and even motherboard BIOS versions can influence whether a particular drive triggers the bug. For now, any model list circulating on forums should be treated as a triage aid, not an authoritative inventory.

Vendor and Microsoft Responses

Phison was among the first to acknowledge the issue, stating it was coordinating with SSD brands and OEMs to identify affected controller families and develop remediation. Because firmware updates must be validated for each branded drive configuration, Phison itself is unlikely to distribute fixes directly to end users; instead, SSD vendors will release them through their own support channels.

Microsoft, for its part, confirmed it is working with hardware partners and investigating the reports. The company initially marked the KB5063878 support page as free of known issues but later added release-health advisories for a separate enterprise deployment regression that affected WSUS and SCCM. As of now, no Known Issue Rollback (KIR) or host-side mitigation has been issued for the storage regression, though Microsoft is collecting telemetry and coordinating with controller designers. The delay in host-side action suggests the root cause may lie more in firmware than in the OS storage stack itself.

Technical Analysis: A Complex Host-Firmware Interaction

The NVMe protocol relies on a delicate dance between the operating system’s storage stack, PCIe drivers, and the controller firmware. When an SSD disappears entirely from Device Manager and vendor utilities can’t retrieve SMART data, it points to a controller-level hang or crash, not a simple file-system glitch. Essentially, the drive stops responding to NVMe commands at the bus level.

Why would sustained writes trigger this? SSDs use an SLC (single-level cell) cache to buffer incoming data before background garbage collection and wear-leveling kicks in. When the cache is overwhelmed by a continuous streaming write of tens of gigabytes, the controller must juggle write commands, metadata updates, and internal housekeeping simultaneously. A nearly full drive has less overprovisioning space, which compounds the stress on these background tasks. This aligns with the observation that drives above 50–60% utilization are more prone to failure.

DRAM-less designs introduce an extra variable: the Host Memory Buffer. If the August update changed the timing or manner in which Windows allocates or reclaims HMB memory—perhaps through updated NVMe drivers or memory-management tweaks—that could push controller firmware past a previously untested edge case. There is precedent: previous Windows updates have inadvertently broken HMB-dependent drives, only for fixes to emerge from firmware patches rather than OS rollbacks.

At this stage, it’s critical to avoid jumping to conclusions. No public telemetry has yet confirmed that KB5063878 contains a deliberate change to the NVMe host driver. The bug may be a latent firmware issue that only manifests with the specific combination of driver versions, memory pressures, and I/O patterns introduced by the August update. Both Microsoft and the SSD vendors are analyzing crash dumps and diagnostic logs to pinpoint the exact failure mechanism.

What’s Verified vs. What’s Speculation

Given the high stakes for data integrity, it’s important to separate confirmed facts from community conjecture.

Verified and corroborated:
- KB5063878 (OS Build 26100.4946) was released on August 12, 2025, and has been linked to the issue by multiple outlets and independent testers.
- Phison publicly acknowledged an investigation into “industry-wide effects” tied to the Windows August update.
- The failure is consistently reproduced using sustained sequential writes of roughly 50 GB or more, especially on drives with over 50–60% fill.
- DRAM-less SSDs using HMB appear disproportionately affected, though not exclusively.

Provisional or unverified:
- Any specific list of affected models is not authoritative; community-compiled lists vary and lack vendor validation.
- The early geographic clustering of reports from Japan likely reflects where testing first occurred, not a regional root cause.
- Claims that the update includes a definitive NVMe driver change that crashes Phison controllers remain speculative until forensic analysis is published.

Immediate Mitigation and Recovery Steps

While the investigation unfolds, the safest approach for consumers and IT administrators is to protect data and avoid the known trigger workload until validated fixes arrive.

For all users:
- Back up critical data immediately. If your system has received KB5063878 and you use an NVMe SSD—especially a DRAM-less model—create a verified backup now.
- Avoid large, sustained sequential writes. Postpone huge game installs, mass file copies, and disk cloning operations if the target drive is more than 50–60% full. Splitting large transfers into smaller chunks (e.g., copying 10 GB at a time) has been reported to reduce the risk in some test systems.
- Check your SSD vendor’s support page for firmware advisories or updates. Apply any recommended firmware only after backing up and following the vendor’s instructions carefully.

If a drive disappears mid-write:
- Stop all write activity immediately to prevent overwriting in-flight metadata.
- If possible, create a block-level image of the drive before attempting recovery; this preserves the state for forensic analysis and may aid vendor support.
- Collect system logs, Event Viewer entries, and any vendor diagnostic output.
- Do not reformat the drive until imaging is complete; reformatting can complicate recovery and invalidate warranty processes.
- Contact both the SSD vendor and, if necessary, Microsoft support to file a case and contribute telemetry.

For enterprise administrators:
- Stage KB5063878 in a test ring that includes representative SSDs from your fleet. Run heavy write workloads—such as large SCCM package deployments or scripted file copies—to see if any devices trip the bug.
- Hold back the update from high-risk endpoints, such as workstations used for video editing, large software compiles, or data migration, until vendor firmware guidance is available.
- Leverage WSUS or SCCM to defer the update using servicing controls. Microsoft has already mitigated a separate enterprise install regression for this KB; monitor Microsoft’s release health dashboard for any upcoming Known Issue Rollbacks or admin-targeted mitigations for the storage issue.

The Enterprise Angle: Install Failures in Managed Fleets

In parallel with the storage regression, some organizations reported that KB5063878 failed to install via WSUS or SCCM. Microsoft acknowledged this separate glitch and rolled out servicing controls to mitigate it. While unrelated to the SSD vanishing act, the install failure highlights how a single cumulative update can harbor multiple pain points for managed environments. IT admins should treat both issues seriously when planning patch deployment cycles.

Why This Regression Matters

Though the scope of affected systems appears narrow, the impact can be catastrophic. Permanent data loss is a real possibility for those who trigger the bug while writing critical files. The financial and reputational cost of corrupted SSDs is high for content creators, IT shops, and even casual users who don’t maintain current backups. Conversely, the bug is manageable because it requires a very specific, reproducible workload—meaning organizations can proactively test and block exposure.

This incident also underscores a recurring weakness in modern storage ecosystems: tightly coupled host-firmware interactions mean that even a minor OS update can flush out latent SSD bugs that had been dormant for years. The lesson for IT departments is to maintain diverse test hardware and to treat storage validation as a first-class part of update staging, not an afterthought.

What to Watch Next

The path to a fix is likely to come from SSD vendors in the form of firmware updates. Phison’s partner-coordination model means each brand will need to test and release its own patch. This process could take several weeks, depending on the complexity of the fix and the number of affected SKUs.

Microsoft may issue a host-side mitigation if the investigation reveals an OS change that can be safely rolled back. For example, a Known Issue Rollback could temporarily revert a specific driver or configuration without requiring a full update uninstall. However, if the root cause is purely firmware-level, a host-side fix may not be possible.

Users should watch for:
- Official advisories from their SSD manufacturer, typically posted on the product support page.
- Updates to the KB5063878 support article on the Microsoft website, where the company may add details once the root cause is confirmed.
- Independent reproduction matrices from tech outlets and labs, which can help narrow down affected firmware versions and controller models.

Conclusion

The Windows 11 KB5063878 update has opened a rare but serious front in the eternal struggle between OS patches and hardware firmware. For the vast majority of users, daily computing will continue without incident. But for those who rely on large, fast writes to NVMe SSDs—gamers, content creators, and data professionals—the risk of suddenly losing a drive is too great to ignore.

Until vendors publish validated fixes, the only defense is preparedness: back up, avoid the known trigger workload, and monitor support channels. The coordinated response from Microsoft and Phison signals that the industry recognizes the severity. How quickly they can deliver a remedy will determine whether this regression remains a cautionary tale or becomes a chapter in the annals of infamous Windows bugs.