NVMe SSD firmware bugs cause drives to disappear from the OS under sustained write loads

Certain NVMe SSDs — particularly DRAM-less and HMB (Host Memory Buffer) designs using specific controller families — stop responding and vanish from Device Manager and Disk Management during sustained sequential writes, typically around the 50GB mark on drives at 60%+ capacity. The drive becomes invisible to the OS: no error message, no graceful degradation, just sudden absence. A reboot may temporarily restore the drive, but the same workload reproduces the failure. So what? A video editor rendering a 4K project to their NVMe drive mid-export loses the entire render progress and potentially the project's temp files — hours of render time wasted, plus the risk of project file corruption. So what? Because the drive 'comes back' after reboot, the user cannot get warranty service — the drive passes all diagnostic tests when not under the specific sustained-write load that triggers the bug. So what? The user is trapped: they own a drive that fails under their exact workload but appears healthy to every diagnostic tool, and the manufacturer denies the claim. So what? This disproportionately affects budget-conscious professionals who bought DRAM-less NVMe drives (which dominate the $40-$80 price segment) because they were marketed as 'NVMe speed' without disclosing the sustained-write limitations of HMB designs. So what? Trust in the NVMe SSD category is damaged, and users over-spend on premium drives not for performance but for reliability anxiety. This persists because SSD vendors race to cut costs by eliminating the onboard DRAM cache and relying on HMB (borrowing system RAM), which works for bursty consumer workloads but fails under sustained writes. Firmware QA testing at manufacturers rarely covers the specific sustained-write scenarios that trigger controller lockups. Windows driver updates (e.g., KB5063878 in August 2025) can also change NVMe driver behavior in ways that expose latent firmware bugs.