A SATA II SSD from roughly 2010 has accumulated 1 petabyte of writes — 25 times the manufacturer's rated TBW (Total Bytes Written) limit — and continues to work reliably, according to a report published on 20 June 2026. The finding challenges a widespread assumption in the industry and among users: that when an SSD exceeds its TBW rating, it is nearing failure.

That assumption needs clarification. TBW is a warranty boundary, not a cliff edge. Kingston's endurance documentation is clear on this: TBW defines how much data a drive is designed to handle during its useful life and sets the point at which the manufacturer's warranty ends. It does not predict the moment the drive will stop working. NAND flash — the memory technology in SSDs — degrades gradually as cells are repeatedly written to and erased. Manufacturers set conservative TBW figures to limit their warranty obligations, not to mark an exact failure point.

The real endurance of a drive often exceeds its rating by a significant margin. That gap depends partly on the age of the NAND technology used. Older consumer SSDs, especially those built with single-level cell (SLC) or multi-level cell (MLC) NAND — the standard during the SATA II era — store fewer bits per cell and therefore tolerate many more write-erase cycles before the transistors inside begin to fail. Newer drives using triple-level cell (TLC) or quad-level cell (QLC) NAND pack more data into the same space but can withstand fewer total writes per cell by design. A 25× overrun is striking but not physically impossible when the drive in question uses older, less-dense NAND architecture.

SATA II (3 Gb/s) itself is a relic. It was the standard interface for consumer drives through the mid-to-late 2000s, well before SATA III (6 Gb/s) and years before NVMe over PCIe became the norm. A drive still running on SATA II after sixteen years has outlived the interface's relevance — most modern motherboards include a SATA port for legacy hardware, but using SATA II as a primary interface is firmly the exception.

For anyone managing aging storage systems, the practical lesson is straightforward: reaching a drive's TBW limit is not, by itself, a reason to retire it. Hard data comes from S.M.A.R.T. attributes — standard health metrics that track reallocated sector count, uncorrectable error rates, and wear leveling activity. A drive that has exceeded its TBW rating but shows clean S.M.A.R.T. readings and stable error rates is in a different condition from one that hit the TBW limit with rising sector reallocation. The former has usable life remaining; the latter is genuinely degrading.

This does not diminish the value of TBW as a planning tool. For procurement decisions, capacity planning, and warranty management — especially in write-heavy applications like database logging, video surveillance, or caching — TBW per day per drive (DWPD) remains a sound basis for specification. The distinction is important: reaching the rated limit does not mean the drive is about to fail, but it is also not advice to disregard endurance ratings when sizing infrastructure.

The broader context matters here. Consumer SSDs typically deliver five to ten years of service under normal use, and that expectation holds up well across the market. A drive running for sixteen years at 25× rated endurance is an outlier — but a useful one. It reveals how conservatively manufacturers have historically drawn their endurance figures, particularly on older, less-dense NAND. TLC and QLC drives aging into their own long tenures will likely show different endurance curves. These denser geometries are rated for fewer total program-erase cycles and operate closer to their statistical limits by design.

At this point, the drive is a curiosity as much as a benchmark. But curiosities can reframe thinking. The assumption that exceeding TBW means your SSD is failing has been holding back pragmatic decisions about storage management — and it deserves reconsideration.