Hard Drive Failure Rates Climb to 1.73% in Q2 2026
Hard drive failure rates climbed to an annualized 1.73% in the second quarter of 2026, according to the latest Drive Stats report published by Backblaze on September 29, 2026. Stephanie Doyle, senior market intelligence manager, reported that the storage cloud monitored 359,101 drives used to store data. The company removed 3,881 boot drives and 705 hard drives that did not meet the inclusion criteria, leaving 354,415 hard drives to analyze. The annualized failure rate (AFR) projects how many drives in a group would fail over a full year if the current failure pace continued, and Backblaze noted that 1.73% was the highest quarterly AFR "in quite a while."
The quarterly table covers a reporting period from April 1 through June 30, 2026, and includes drive models with more than 100 drives and more than 10,000 drive days as of June 30. The combined fleet recorded 31,553,350 drive days and 1,498 failures. Three Seagate models had zero failures: the 8TB ST8000NM000A, the 12TB ST12000NM000J, and the 14TB ST14000NM000J. A fourth Seagate model, the 16TB ST16000NM000J, had exactly one failure, which the report called a clean sweep by Seagate. Two drives fully retired this quarter: the 4TB HGST HMS5C4040BLE640 at nearly nine years old and the 8TB HGST HUH728080ALE600 at nearly eight years old. No new drive models entered the fleet for the second quarter in a row, but new deployments continued to concentrate in drives of 20TB and larger.
A quartile analysis defined outliers as an AFR above 6.95%, and this quarter produced three: the HGST HUH721212ALN604 12TB at 7.63%, the Seagate ST10000NM0086 10TB at 9.33%, and the Seagate ST14000NM0138 14TB at 8.26%. The report explained that the HGST drive is nearly seven years old, while the 10TB Seagate is nearly 8.5 years old. Both Seagate outliers also have low drive counts: 965 drives for the 10TB model and 1,235 for the 14TB model, with 22 and 25 failures respectively, which pushed their AFRs upward. In total, Backblaze counted 10 out of 31 drive models with an AFR above 3.0%, a distribution the report called unusually high and worth watching. The full quarterly line item includes models such as the HGST HUH721212ALE600 (12TB, 1.08%), HGST HUH721212ALE604 (12TB, 3.34%), Seagate ST8000DM002 (8TB, 2.46%), Seagate ST8000NM0055 (8TB, 3.44%), Seagate ST12000NM0007 (12TB, 4.17%), Seagate ST12000NM0008 (12TB, 3.17%), Seagate ST12000NM001G (12TB, 1.35%), Seagate ST14000NM001G (14TB, 1.5%), Seagate ST16000NM001G (16TB, 0.65%), Seagate ST16000NM002J (16TB, 3.47%), Seagate ST24000NM002H (24TB, 3.3%), Toshiba MG07ACA14TA (14TB, 1.94%), Toshiba MG07ACA14TEY (14TB, 2.46%), Toshiba MG08ACA16TA (16TB, 1.01%), Toshiba MG08ACA16TE (16TB, 1.03%), Toshiba MG08ACA16TEY (16TB, 3.66%), Toshiba MG09ACA16TE (16TB, 1.07%), Toshiba MG10ACA20TE (20TB, 1.1%), Toshiba MG11ACA24TE (24TB, 0.58%), WDC WUH721414ALE6L4 (14TB, 1.07%), WDC WUH721816ALE6L0 (16TB, 4.07%), WDC WUH721816ALE6L4 (16TB, 1.16%), WDC WUH722222ALE6L4 (22TB, 0.55%), and WDC WUH722626ALE6L4 (26TB, 0.95%).
Lifetime figures, covering drives with more than 500 drives and more than 100,000 lifetime drive days, show a slight uptick to 1.41% AFR across 353,041 drives, 558,148,607 drive days, and 21,609 failures. Four models did not appear in the lifetime table because their drive counts were below 500: Seagate ST8000NM000A (239 drives), ST14000NM000J (497 drives), ST16000NM000J (171 drives), and ST16000NM002J (467 drives). Interestingly, the first three were also on the zero-failure list. The standard inclusion thresholds are: quarterly more than 100 drives and more than 10,000 drive days, annual more than 250 drives and more than 50,000 drive days, and lifetime more than 500 drives and more than 100,000 drive days. Backblaze noted that the complete line-item dataset is available for free download from its Hard Drive Test Data page.
Higher-capacity drives are becoming a larger share of the fleet. The 20TB and larger group now represents more than 25% of the quarter's dataset. The report also notes that Western Digital announced a 40TB UltraSMR drive this year and has a roadmap showing a path to 100TB by 2029. To explain the technology, Backblaze compared conventional magnetic recording (CMR) with shingled magnetic recording (SMR). CMR writes data in concentric tracks side by side with guard space between them, so any track can be rewritten directly. SMR overlaps tracks like roof shingles, which works because a drive's write head is physically wider than its read head; overlapping allows roughly 10% to 25% more capacity on the same platter. The tradeoff is that SMR organizes tracks into bands or zones, and updating data may require the drive or the software above it to reorganize a larger section. There are three main flavors: drive-managed SMR, where firmware hides the complexity but write performance can become unpredictable; host-managed SMR, where the operating system or application writes sequentially within zones and needs compatible software; and host-aware SMR, which supports SMR-aware commands while keeping some compatibility with conventional workflows.
In a data center, SMR offers more terabytes per drive slot in the same rack space at roughly similar drive-level power, improving both dollars per terabyte and watts per terabyte. But SMR cares about workload shape: sequential and append-oriented writes work well, while frequent small rewrites create extra work. Backblaze therefore says qualification of SMR is not just a drive swap; it requires planning around sustained write behavior, rebuild performance, parity or erasure-coding overhead, firmware, and the storage software above the drive. The report also covers heat-assisted magnetic recording (HAMR), which uses a laser to briefly heat a microscopic spot on the platter so data can be written to smaller, more stable magnetic grains. Glass substrates are standard in 2.5-inch HAMR drives and increasingly common in high-capacity 3.5-inch HAMR designs. Because SMR and HAMR solve different problems, manufacturers can combine them. Backblaze says there are currently no SMR drives in its fleet, and none of the drives in the tables are shingled, but the company will flag them when they enter the fleet and report their AFRs.
Backblaze will hold a webinar on October 8, 2026 at 12 p.m. PT with Stephanie Doyle and David Johnson, Director of Product Marketing, to review the trends. For teams that run storage at any scale, the same underlying lesson applies: automated backups, continuous health monitoring, and tested recovery procedures are essential, and a managed data infrastructure provider such as AEU Data offers those capabilities across its managed databases, encrypted storage, and high availability services. The complete dataset used in the report is available on Backblaze's Hard Drive Test Data page, with the requests that users cite Backblaze as the source, accept sole responsibility for how they use the data, and do not sell the data itself.