Every quarter, cloud storage provider Backblaze publishes failure rate data from the tens of thousands of hard drives running in its data centres — one of the few sources of real-world, large-scale drive reliability data that is publicly available. The 2025 year-end report, published in February 2026, covers 344,196 drives across 30 models and is worth examining closely by anyone making decisions about storage hardware.

The 2025 Annual AFR: Down to 1.36%

Backblaze’s annualised failure rate (AFR) across its entire monitored fleet came in at 1.36% for 2025, down from 1.55% in 2024. This means that across roughly 344,000 drives running continuously in a professional data centre environment, just over 1 in 100 could be expected to fail within the year. That improvement is meaningful, though the quarterly data tells a more nuanced story — Q4 2025 showed elevated failure rates on several specific models, illustrating that the headline annual figure masks variability across the fleet.

It is worth contextualising what this number represents for consumer users. Backblaze’s drives are enterprise-grade units running 24/7 in a temperature-controlled data centre, with professional monitoring and immediate replacement when a drive shows signs of distress. Consumer drives used intermittently at home or in small offices operate under different stress profiles, and drives that are older, have been through power events, or are running in high-vibration environments may exhibit meaningfully higher failure rates than the Backblaze average suggests.

Notable Models — the Good and the Concerning

Several models achieved zero or near-zero failures in Q4 2025, landing on Backblaze’s honour roll. These included the Seagate ST8000NM000A 8TB and ST16000NM002J 16TB, as well as the HGST HMS5C4040BLE640 4TB — the latter a drive that has consistently performed well across multiple years of the Drive Stats series.

What the Q4 Volatility Tells Us About Drive Aging

The HGST HUH728080ALE600 8TB is a telling case study. Despite broadly acceptable failure rates through most of its operational life, by Q4 2025 this drive — now approximately 7.5 years old — posted a 10.29% AFR. Backblaze investigated environmental factors including temperature and airflow and found no culprit there; the working theory points to vibration sensitivity combined with accumulated age-related wear. The drives were flagged for normal CVT (capacity, volume, type) migration.

The Toshiba MG08ACA16TEY 16TB is a different story. It posted a dramatic 16.95% AFR in Q3 2025 following a firmware issue, prompting collaborative firmware work with Toshiba. By Q4 it had dropped back to 4.14% — elevated but clearly trending toward normalisation. This illustrates that firmware quality can have a significant effect on failure rates independent of hardware age or design, and that firmware updates — or their absence — are a real risk factor in long-running drive deployments.

New Drives in the Fleet: 26TB and Beyond

Q4 2025 saw Backblaze deploy its first 26TB drives — the Western Digital WUH722626ALE6L4 — with 1,201 units entering the fleet. The model recorded just one failure across its debut quarter, which Backblaze noted positively. A new Seagate 16TB model (ST16000NM000J) also entered the fleet with 112 drives, posting similarly clean early performance.

The arrival of 26TB drives in a production fleet marks a continued push into higher areal densities. As drive capacities grow, the risk profile of a single failure grows with it — a failed 26TB drive represents far more data at risk than a failed 4TB drive. This makes reliable performance data like Backblaze’s Drive Stats increasingly valuable for anyone evaluating storage decisions at scale.

What This Means for Home and Business Users

The 2025 data reinforces a few durable principles. First, no drive lasts forever — the HGST example shows that even well-performing models accumulate risk past the 5–7 year mark. Second, firmware matters as much as hardware — the Toshiba Q3 event shows that a software update can push an otherwise healthy drive into elevated failure territory. Third, the annual AFR of around 1.4% is a fleet-wide average across professional conditions; individual drives in homes and small offices, particularly those that have experienced power events or heavy workloads, may be at greater risk.

For users in KwaZulu-Natal and the greater Durban area, SouthBit accepts drives via secure courier for professional assessment and recovery. If a hard drive or SSD is showing signs of failure — unusual sounds, slow access, errors, or simply not being detected — a professional assessment is the right first step. The Durban data recovery page has full courier details. View the pricing page to understand costs upfront.

Book a Free Assessment

Sources

• Backblaze — Drive Stats for 2025 — Year-End Report, February 2026