Phison Acknowledges Windows 11 KB5063878 NVMe SSD Failures, Promises Firmware Fix

A single flawed Windows 11 cumulative update is bricking NVMe SSDs mid‑operation, and Phison—the dominant controller maker behind many of the affected drives—has confirmed it’s investigating. Users worldwide report that after installing KB5063878 or KB5062660, their SSDs disappear from File Explorer and Device Manager during sustained large‑file writes, often around the 50 GB mark. The issue, first spotted by Japanese users and replicated by independent testers, points to a controller‑level firmware lockup rather than a simple driver glitch.

“Phison has recently been made aware of the industry‑wide effects,” the company said in a statement, pledging to work with partners to identify affected controller models and deliver firmware fixes through SSD vendors. This is not just a performance regression—it’s a genuine data‑integrity risk that has already corrupted files and rendered drives temporarily or permanently inaccessible.

The Updates in Question

Microsoft released the August 2025 cumulative update for Windows 11 on August 12, 2025, as a combined servicing stack update (SSU) and latest cumulative update (LCU) under KB5063878 (build 26100.4946). A related update, KB5062660, also triggers the failure. The public KB article initially listed no known issues related to storage, but community testers quickly reproduced a consistent pattern: during sustained sequential write workloads, some NVMe SSDs stop responding, vanish from the OS, and sometimes return with corrupted or inaccessible data.

Symptoms and Trigger Profile

The failure is alarmingly reproducible. Affected drives exhibit a clear failure signature:

• The SSD disappears from File Explorer, Device Manager, and Disk Management while a large write is in progress.
• SMART data and controller telemetry become unreadable to any diagnostic utility.
• A reboot may restore the drive, but file integrity is not guaranteed. Some users report volumes that remain inaccessible even after power cycling.

The typical trigger is a sustained sequential write of roughly 50 GB or more, often coupled with elevated controller utilization (above 60%). Real‑world tasks that hit this threshold include large game installations, massive file extractions, disk cloning, and video exports. The failure is workload‑dependent, meaning a system might work fine for days until a heavy write pushes the controller over the edge.

Phison Steps In

Phison, a Taiwanese controller maker whose chips power SSDs from Corsair, SanDisk, Kioxia, and many others, acknowledged the problem within days. In a brief statement, the company said it is “reviewing controller families that may be involved” and will provide “firmware updates and partner advisories as applicable.” Phison’s approach is partner‑centric: it will not release direct‑to‑consumer patches but will instead distribute corrected firmware to SSD brands, who must then validate and push updates through their own tools.

This is standard industry practice—firmware must be tested against each vendor’s specific hardware configuration, factory setup, and update mechanism. However, it means end users will have to wait for each SSD vendor to release a tailored fix, a process that could take weeks.

Technical Analysis: What’s Really Happening

Two plausible, non‑exclusive root causes have emerged from community analysis and independent engineers:

1. NVMe Driver Regression in the Windows Kernel

The cumulative update likely altered low‑level NVMe command ordering, DMA timing, or buffer allocation. If the new host driver issues commands in an order or with timing that the SSD firmware never anticipated, the controller can stall. Because Phison‑based drives are so widespread, they are over‑represented in failure reports, but other controllers (Maxio, SanDisk in‑house) have also been flagged, suggesting a common host‑side trigger.

2. HMB and DRAM‑less Design Flaws Exposed

Many budget NVMe SSDs are DRAM‑less and rely on the Host Memory Buffer (HMB) feature to borrow system RAM for mapping tables. HMB behavior is sensitive to OS memory management; even small changes in allocation or timing can provoke race conditions or resource‑exhaustion bugs in firmware. Windows 11 24H2 already exposed HMB fragility in some drives earlier this year, and this update may have further stressed that fragile interaction. Sustained sequential writes are a worst‑case scenario: they hammer mapping tables, controller queues, and HMB usage simultaneously.

Both hypotheses align with observed behavior. The failure occurs under heavy write stress, when the controller is juggling large amounts of metadata. Whether the primary fault lies in Microsoft’s driver (requiring a hotfix) or in the SSD firmware (requiring a controller patch) remains to be determined, but Phison’s investigation implies at least some firmware‑side fixes are needed.

Which Drives Are Affected? A Community‑Sourced Warning

Community lists and independent lab reproductions have flagged a range of SSD models. The most frequently cited controller is Phison’s PS5012‑E12 (from the E16 family), but E21T and E31T variants also appear. Specific models mentioned repeatedly include:

Important caveats: These lists are based on user reports and not every drive with a listed controller fails. Firmware revisions, motherboard UEFI versions, and specific NVMe driver stacks all influence whether a given combination triggers the bug. Treat the lists as investigative leads, not a definitive recall inventory.

Phison’s confirmation that it is “reviewing controllers” suggests that model‑specific advisories will emerge. In the meantime, if your SSD appears on a suspect list, assume it is at risk.

Microsoft and Vendor Response Timeline

• August 12, 2025: KB5063878 and KB5062660 released.
• Within 48–72 hours: Community testers replicate the issue, forums light up with reports, and specialist outlets begin coverage.
• August 14–15: Microsoft deploys a Known Issue Rollback (KIR) for an unrelated enterprise deployment regression (WSUS/SCCM error 0x80240069) but does not address the storage regression.
• August 16: Phison publicly acknowledges the investigation.
• Current: Major SSD vendors are collecting telemetry and deciding whether to issue interim guidance or wait for firm root‑cause determination. No vendor has yet released a corrective firmware.

At the time of writing, Microsoft has not added a Known Issues entry for the NVMe failures to the KB articles. Historically, the company has used both KIR and targeted out‑of‑band fixes for hardware compatibility problems, so a future update may address the host‑side component.

Practical Mitigation: What to Do Right Now

The priority is data protection. Whether you are a home user or an IT administrator, these steps are critical:

For Home Users and Enthusiasts

• Back up immediately. Copy critical data to an external drive or cloud service before performing any large write operations.
• Pause updates. If you haven’t installed KB5063878/KB5062660 and your system relies on an NVMe SSD for critical work, delay the update until vendors publish guidance.
• Avoid large writes. If you’ve already patched and your system is stable, steer clear of activities that involve writing more than ~50 GB at a time—game installs, large decompressions, video rendering.
• Check vendor firmware tools. Use Corsair iCUE, SanDisk Dashboard, Kioxia SSD Utility, or your drive’s management software to look for firmware updates. Apply only vendor‑recommended firmware after backing up.

For IT Administrators and Fleet Managers

• Inventory your SSD fleet. Identify models, controller families, and firmware versions. Prioritize DRAM‑less and Phison‑based drives.
• Block the update. Use WSUS, MECM/SCCM, or your MDM to withhold KB5063878/KB5062660 from at‑risk machines until validated fixes are available.
• Reschedule heavy IO workloads. Defer large imaging, cloning, or bulk file operations to unpatched systems or hardware not on the suspect list.
• Handle failed drives carefully. If a drive becomes unreadable, power off the system, capture NVMe logs, and, if the data is critical, engage a data recovery specialist. Do not repeatedly power‑cycle a failed drive—it can worsen corruption.

Registry and HMB Workarounds—Proceed with Caution

Some community members have reported success with registry tweaks that limit HMB allocation, a technique used in past HMB‑related incidents. However, such tweaks can severely degrade performance and are not official fixes. Treat them as temporary emergency measures only if you have full backups and understand the risks.

Why This Matters Beyond a Patch Tuesday Headache

This incident is a textbook example of the complex interplay between modern SSDs and operating system updates. NVMe drives are miniature computers with their own firmware, DRAM (or HMB‑dependent caches), and error‑correction logic. A seemingly trivial change in how the host OS queues commands or allocates memory can expose latent firmware bugs that remain dormant under normal usage. When those bugs trigger, the consequences can be catastrophic—silent data corruption or total drive inaccessibility.

The Windows ecosystem has seen similar events before. Past cumulative updates have caused SSD slowdowns, boot failures, and even worn‑out write endurance on some drives. Each time, the resolution requires coordination between Microsoft, controller makers, and SSD vendors—a process that is often slow and opaque. The August 2025 episode underscores the need for more rigorous pre‑release testing with a broader range of storage hardware, especially given the proliferation of DRAM‑less designs and diverse firmware branches.

Looking Ahead: What to Expect

• Firmware updates from SSD vendors: Phison will deliver corrected firmware to its partners, who will then validate and publish updates per model. This is likely days to weeks away.
• Microsoft Known Issue entry: If host‑side behavior is confirmed as a factor, expect an official acknowledgement and possibly a KIR or targeted fix in a future update.
• Community vigilance: The list of affected models will continue to evolve. Users should monitor forums and vendor announcements for confirmed lists and mitigation steps.

Final Takeaway

The KB5063878/KB5062660 update is an urgent compatibility warning, not a universal recall—but the risk of data loss is real. Until Phison and its partners deliver validated firmware, every user with a suspect NVMe SSD should assume the worst. Backup, postpone heavy writes, and keep a close eye on your SSD vendor’s support page. In the modern PC, the old rule has never been more true: trust your backups, not your hardware.