Windows 11 Aug 2025 Update Breaks Some SSDs on Large Writes — What to Do

Microsoft and several SSD vendors are investigating reports that the August 12, 2025 cumulative update for Windows 11 (KB5063878, OS Build 26100.4946) can trigger a storage regression that makes some SSDs disappear during sustained, large writes — a failure mode that, in a minority of reports, has left data corrupted or partitions unreadable. (support.microsoft.com) (tomshardware.com)

Background / Overview​

Microsoft released the August 12, 2025 cumulative update for Windows 11 version 24H2 as KB5063878 (combined SSU + LCU, OS Build 26100.4946). The public KB entry lists security and quality improvements and, at publication, did not list a storage-related known issue. (support.microsoft.com)
Within days of the rollout, multiple independent community test benches and specialist outlets reproduced a coherent failure profile: during sustained sequential writes (commonly reported around the ~50 GB mark), certain NVMe and SATA drives stop responding, vanish from File Explorer, Device Manager and Disk Management, and may return with unreadable SMART/controller telemetry or partitions shown as RAW. Reboots sometimes restore visibility but do not guarantee data integrity for files written during the event. (tomshardware.com) (bleepingcomputer.com)
Phison — a widely used SSD-controller supplier whose silicon appears in many consumer drives — has publicly acknowledged it is investigating “industry‑wide effects” linked to KB5063878 (and related reporting that mentions KB5062660) and said it is coordinating with partners to identify impacted controller families. Microsoft and other outlets report ongoing vendor and platform engagement. (tomshardware.com)

---

What users are actually seeing​

Core symptoms​

• The target drive becomes unresponsive mid-write and disappears from the OS view (File Explorer, Device Manager, Disk Management). (tomshardware.com)
• Vendor utilities and SMART telemetry become unreadable or unavailable, indicating a likely controller-level hang. (bleepingcomputer.com)
• Files being written when the problem manifests are often truncated, corrupted, or missing; in some reports the partition shows as RAW.
• In many cases a reboot restores the device, but in a minority of reports the SSD remains inaccessible and may require vendor intervention, firmware reflash, or RMA. (guru3d.com)

Typical triggers and reproducibility​

Independent reproductions from community labs converge on a narrow workload profile: sustained, sequential writes of roughly 50 GB or more, particularly when the SSD is already moderately full (community reports cite ~50–60% used capacity). Real-world triggers include large game updates, archive extraction, cloning, or bulk media copies. These are not universal; affected models and firmware revisions vary, which complicates definitive blacklists. (tomshardware.com)

---

Which drives and controllers are implicated​

Early community lists and test matrices show a pattern of over‑representation for drives using certain Phison controller families and some DRAM‑less designs, but the phenomenon has not been restricted to a single vendor or controller. Reported affected models include (community-sourced, investigative lists):

• Drives using Phison controllers (several generations). (guru3d.com)
• Some branded drives from Corsair, Kioxia, SanDisk, ADATA, and others in community reports. (tomshardware.com)

Caveat: these lists are investigative leads, not definitive blacklists. Firmware revision, drive capacity, host BIOS, and PCIe/driver interactions change reproducibility. Treat community model lists as indicators to guide caution, not as guaranteed failure rules.

---

Who has said what — vendor and Microsoft signals​

• Microsoft’s official KB for KB5063878 confirms the update was released on August 12, 2025 and initially listed no known storage issues; Microsoft has been responding to other August update regressions and has issued out‑of‑band fixes for separate WSUS/Reset-related regressions. (support.microsoft.com) (bleepingcomputer.com)
• Phison has issued a concise statement acknowledging it was “recently made aware of the industry‑wide effects” of the August updates and that it is investigating impacted controllers and coordinating with partners; publications relaying Phison’s statement also noted the vendor was taking legal action against falsified documents tied to the story. (tomshardware.com, notebookcheck.net)
• Multiple independent outlets and community researchers (Tom’s Hardware, BleepingComputer, Windows Central, TechRadar, and others) have validated the symptom fingerprint through hands‑on tests or aggregated user reports. These outlets repeatedly advise caution with large sequential writes after installing the update. (tomshardware.com, bleepingcomputer.com, windowscentral.com, techradar.com)

Important nuance: Microsoft’s public KB entry initially reported “no known issues” for storage, while other Microsoft channels and third‑party reports show Microsoft engaging with partners and preparing mitigations. That mismatch is typical during active investigations where telemetry and vendor diagnostics must be coordinated before formal Known Issue entries or mitigations are published. (support.microsoft.com)

---

Technical analysis — what likely went wrong (plausible mechanisms)​

The current weight of evidence from test reproductions and telemetry signals points to a host-to-controller interaction exposed by a specific sustained-write workload rather than a simple file-system bug. The leading technical theories include:

• Controller firmware lockup: sustained sequential writes can stress SLC caching, metadata updates and wear‑leveling paths; a subtle change in host timing or command semantics can expose a latent firmware race that causes the controller to stop responding. The unreadable SMART telemetry reported by many testers supports a controller hang rather than pure file-system corruption.
• Host Memory Buffer (HMB) / DRAM-less interaction: some DRAM‑less SSDs rely on host memory allocations (HMB). If Windows changes how it allocates or times HMB resources under certain workloads, DRAM‑less controllers may lose critical working data or experience edge‑case failures. Earlier Windows 11 24H2 rollouts already revealed HMB-related issues with specific models, which makes this a plausible vector for the current regression.
• OS storage‑stack buffer behavior / memory leak: community testers have hypothesized a memory leak in the buffer used by Windows for storage operations that prevents the drive cache from being flushed properly. If host-side buffered regions are exhausted or mis‑managed under sustained writes, that could produce partial writes, sudden I/O errors, or controller-level timeouts. This remains a hypothesis and is not yet proven by vendor telemetry. (bleepingcomputer.com)
• Firmware and platform diversity: because drives from multiple vendors and controllers (not only Phison) appear in some test sets, the root cause may be an interaction between a Windows change and common firmware design patterns (e.g., SLC cache exhaustion, non-atomic metadata writes) that different controllers implement variably. That would explain why some drives fail while identical models or other builds do not.

Flag on uncertainty: none of the above mechanisms has been publicly confirmed by a cross‑vendor forensic write‑up at the time of reporting; the strongest public signals are vendor acknowledgements of investigation plus multiple independent reproductions that share a similar workload fingerprint. Treat the technical narratives as plausible rather than established fact until vendors and Microsoft publish coordinated findings.

---

Practical risk assessment​

• The issue is not reported as universal — millions of systems received the August update without incident — but the reproduction window and clustered device lists make this a high-impact, low-prevalence event: when it hits, the consequences can be severe (data corruption or permanent inaccessibility). (tomshardware.com)
• The risk is highest for users who:
• Perform sustained, large sequential writes (game installers, archive extraction, cloning, video exports). (bleepingcomputer.com)
• Have SSDs that are moderately to heavily filled (community tests often referenced >50–60% used).
• Use models or controller families present in community test matrices (particularly drives with Phison controllers in early reporting). (guru3d.com)
• The risk to typical light‑use desktop users (web, email, office tasks) appears low based on current reports, but the possibility of a latent edge-case still argues for conservative backup hygiene.

---

Immediate action checklist — what to do now​

Follow these practical steps to reduce exposure and preserve data integrity:

• Back up critical data immediately — use an external drive or cloud backup and verify the backup. The 3‑2‑1 rule (three copies, two media, one offsite) is prudent here.
• Avoid large, sustained writes on drives that received the August cumulative update until vendors or Microsoft publish targeted guidance or fixes. If you must move large files, split transfers into smaller batches (for example, <10–20 GB chunks) and monitor I/O health. (bleepingcomputer.com)
• Delay the update on unpatched systems where possible: pause Windows Update or hold KB5063878 off managed rings until you validate vendor guidance and test in a representative environment. This is especially important for production or content‑creation systems.
• Check SSD firmware and vendor tools: keep vendor utilities installed, check for firmware updates, and follow vendor advisories — but only apply firmware updates after making a verified backup. Firmware updates typically require careful vendor testing and are the most likely long‑term remediation path.
• For enterprises: use staged rollouts, hold at-risk fleets in pilot rings, and apply Microsoft’s Known Issue Rollback (KIR) or out‑of‑band mitigations when announced. Inventory drives and include representative storage hardware in test plans.

---

If a drive disappears mid‑write — do this (do not panic)​

• Stop writing to the machine. Continued writes risk overwriting residual on-disk metadata or complicating recovery.
• Do not reformat the drive or initialize the partition table unless you have a validated forensic image or are prepared to lose the data. Formatting can destroy recovery evidence.
• Capture logs and vendor telemetry: collect Event Viewer records, NVMe driver logs, and any vendor tool output; that information may be critical for vendor diagnostics and RMA.
• If the data is valuable, image the drive (use write‑blockers or read-only imaging tools) and work with vendor support or a professional data-recovery service. Imaging preserves the current state for forensic analysis.
• Contact your SSD vendor for guidance and RMA if the drive remains inaccessible; vendors are the distribution channel for firmware fixes where controller issues are implicated. (notebookcheck.net)

---

What a fix will likely look like​

There are two typical remediation paths for these kinds of host–controller interactions:

• Controller firmware updates distributed by SSD vendors after partner-level validation. Firmware fixes are common when the root cause is a controller-side race, cache management bug, or metadata-handling edge case. Distribution is usually via vendor utilities or partner dashboards.
• OS-side mitigation or patch from Microsoft if the root cause is an unintended host-side change (for example, a timing or buffer management regression). Microsoft can deliver servicing updates or targeted mitigations through Windows Update once telemetry and vendor diagnostics align. A Known Issue Rollback (KIR) or out‑of‑band patch is a known tool in Microsoft’s playbook for this class of regression. (bleepingcomputer.com)

Both paths are often required together: a firmware change on affected controllers and an OS update or configuration change to avoid re‑exposing latent conditions on older firmware.

---

Timeline (concise)​

• August 12, 2025 — Microsoft publishes KB5063878 (Windows 11 24H2 cumulative). (support.microsoft.com)
• Mid‑August 2025 — community researchers and independent test benches reproduce SSD disappearance during sustained large writes and publish reproducible test details. (tomshardware.com)
• August 18–20, 2025 — Phison acknowledges it is investigating and coordinating with partners; multiple outlets (Tom’s Hardware, BleepingComputer, Windows Central) report vendor engagement and advise caution. Microsoft responds to other related August regressions with release‑health advisories and OOB fixes for separate WSUS/Reset issues. (tomshardware.com, bleepingcomputer.com)

---

Strengths, risks, and final analysis​

Strengths in the response so far

• Rapid community triage and reproducible tests accelerated vendor triage and focused investigation on a realistic workload profile, enabling better vendor diagnostics and quicker pathing to firmware or OS mitigations.
• Vendor acknowledgement (Phison) and Microsoft’s engagement signal that the issue is being treated seriously at platform and component levels — the right institutions are involved to deliver a coordinated remedy. (tomshardware.com)

Clear risks and gaps

• Public documentation is still incomplete: Microsoft’s KB initially reported “no known issues,” producing a mismatch between official notes and community/press findings. That gap slows automatic operational responses for administrators. (support.microsoft.com)
• Community model lists are useful but not definitive; relying on incomplete lists to block updates can produce over- or under‑reaction without vendor-validated inventories. Firmware, SKU, and host firmware (BIOS) interactions materially affect exposure.
• Data‑loss potential is real and immediate for users who continue to perform large writes without backups. The single most effective mitigation remains verified backups and staging updates for representative hardware.

---

Practical recommendations (summary)​

• Back up now. Make a verified external or cloud copy of essential data before doing anything else.
• Avoid heavy bulk writes on patched systems until vendor guidance or firmware fixes arrive; split big transfers into smaller batches if you must. (bleepingcomputer.com)
• Hold the update in production rings if your workflows include sustained large writes or content production; test KB5063878 in a pilot with representative storage hardware.
• Monitor official vendor advisories (SSD vendor firmware pages, Phison partner notices) and Microsoft Release Health for coordinated mitigations and KIR updates. (notebookcheck.net, support.microsoft.com)

---

This remains an active, evolving incident. The best defensible posture for both consumers and IT administrators is conservative: prioritize verified backups, avoid stress‑testing patched drives with sustained large writes, inventory at‑risk hardware, and await coordinated, vendor‑validated firmware and platform updates before resuming normal high‑volume I/O operations. (tomshardware.com)

---

Source: Techzine Global Microsoft investigates issues with SSDs after Windows 11 update