Backblaze Drive Stats: Q2 2026, the Latest Quarter, and How to Read the Reports

Latest quarter: Q2 2026, published 29 September 2026

Backblaze's Drive Stats resource page, read on 9 November 2026, gives the Q2 2026 snapshot as a drive count of 354,415, 1,498 drive failures and 31,553,350 drive days. Beside it the page shows three AFRs: 1.73% for the quarter, 1.36% for 2025 and 1.41% lifetime.

Backblaze's investor-relations release of 29 September 2026 led on the rise, with the quarterly failure rate up to 1.73% from 1.24% in Q1 2026. The report records a slight uptick in the lifetime AFR as well, to 1.41%.

Backblaze reports that through Q2 2026 it monitored 359,101 drives used to store data, removed 3,881 boot drives and 705 hard drives that did not meet its criteria for inclusion, and analysed 354,415. That is the figure as Backblaze states it; its published components sum to 354,515, a gap of 100 in Backblaze's own numbers. Boot drives do not enter the failure-rate tables at all.

Between the two reports, which describe different fleets, the drives analysed went from 341,263 to 354,415, 13,152 more, and the printed quarterly AFR from 1.24% to 1.73%. Two long-serving models were fully retired during Q2, one of them the 4 TB HGST HMS5C4040BLE640 at nearly nine years old in total.

Backblaze explains two of the quarter's high model rates by fleet size: its Seagate 10 TB model had only 965 drives and its 14 TB model 1,235, so 22 and 25 failures respectively gave those models AFRs that were quite high. It also counted seven drive models that still had an AFR above 3.0%, said a casual look at recent quarters shows that share skewed towards the higher end, and added: 'That will, of course, bring up our average.'

A quick check on the snapshot figures shows how the rate is built. 31,553,350 drive days divided by 365 is about 86,450 drive years, and 1,498 failures over 86,450 drive years is 1.73%, the AFR the report prints. The rate is consistent with failures per year of drive running time, not failures per drive counted at the quarter's end.

Backblaze drive stats 2026 and 2025: the figures by period

• Q2 2026 (1 April to 30 June 2026): quarterly AFR 1.73%, from 1,498 failures over 31,553,350 drive days, on 354,415 drives.
• Q1 2026 (1 January to 31 March 2026): quarterly AFR 1.24% on 341,263 drives.
• Calendar 2025 (1 January to 31 December 2025): annual AFR 1.36%, from 4,317 failures over 115,638,676 drive days; the 2025 report sets it against 1.55% for 2024.
• Lifetime to 30 June 2026: AFR 1.41%, from 21,609 failures over 558,148,607 drive days. At the end of 2025 the lifetime AFR stood at 1.30%.

The 2025 annual report started from 349,462 drives monitored at the end of 2025. It removed 4,176 boot drives and 1,090 hard drives that did not meet its annual criteria, leaving 344,196 drives across 30 drive models; those figures reconcile exactly.

The lifetime record is the deepest of the three: 558,148,607 drive days is nearly 18 times one quarter's 31,553,350, which is why a lifetime AFR moves slowly. It stood at 1.30% at the end of 2025 and 1.41% at 30 June 2026, while the two 2026 quarters printed 1.24% and 1.73%.

The 2024 figure above is the one the 2025 report states. Each figure is a level for its own period, not a point on a trend line: the fleet changes between reports as drives join and retire, so two periods' rates describe two different populations.

Backblaze drive stats Q1 2026

Q1 2026 covered 1 January to 31 March 2026 inclusive, as the report's table states. After its exclusions, Backblaze analysed 341,263 hard drives, and the quarterly AFR was 1.24%. That printed figure sits below the calendar 2025 figure of 1.36%, though a quarter and a year are different windows on different fleets; Q2 then printed 1.73%.

If you came for the Q1 numbers, they have since been overtaken: Q2 2026 is out, at 1.73% on a fleet about 13,000 drives larger. The Q1 figures remain the right reference for a question about the first three months of 2026.

How to read a Backblaze Drive Stats table

Every report starts from the same raw material. Backblaze takes a snapshot of each operational drive each day, and the snapshot includes basic drive information along with the SMART statistics reported by that drive. The quarterly, annual and lifetime tables are summaries of those snapshots over different windows, after boot drives and some others are removed.

In our own words, four terms carry most of the meaning in the tables:

• Drive count: how many drives of a model the fleet held at the end of the period, the date the inclusion rules are tested on.
• Drive days: one drive in service for one day; a model's drive days add up every day each of its drives was running.
• AFR: failures scaled to a year of drive running time, which is why the Q2 snapshot figures reproduce the printed 1.73%.
• Lifetime: the same calculation over every drive day a model has spent in the fleet, up to the report's end date.

Not every model gets a row, and each table sets its own bar. Backblaze's Q2 2026 report states the three thresholds side by side, and the 2025 annual table repeats the annual one:

A model can therefore appear in one table and be missing from another, and rows from different tables are not directly comparable. Put in running time, the quarterly floor is small: a quarter of 91 days means 10,000 drive days is about 110 drives running for the whole quarter, so a row near the floor can swing a long way on one or two failures. The annual floor of 50,000 drive days is about 137 drive years, and the lifetime floor of 100,000 about 274.

Q2's 31,553,350 drive days over 91 days also give an average of about 346,700 drives in service, a little below the 354,415 analysed, which is what you would expect in a fleet that is growing.

Outliers are named by rule. For Q2 2026 Backblaze defined an outlier, from a quartile analysis, as an AFR above 6.95%, and three models exceeded it: the HGST HUH721212ALN604 at 7.63%, the Seagate ST10000NM0086 at 9.33% and the Seagate ST14000NM0138 at 8.26%. The cut-off belongs to that report's analysis, so an outlier in one quarter is a statement about that quarter's spread.

One kind of drive is absent altogether. Backblaze's Q2 2026 report says there are no SMR drives in its fleet today and that none of the drives in its tables are shingled, so its failure rates describe conventional recording drives only.

The fleet AFR is one number across every included model. It moves with the mix as large new drives join and old ones retire, whatever any single model does. Treat it as context and the model rows as the basis for decisions; the per-model rows for Q2 2026, with drive counts and average ages, are set out on hard drive failure rates.

Where the Backblaze drive stats data lives

The raw files are on Backblaze's Hard Drive Reliability & Test Data page, listed by period. On 9 November 2026 the list ended at Q2 2026: a 2.1 GB zip file that takes 11.7 GB on disk across 91 files. It listed no Q3 2026 data.

The files follow the method. All of the drive snapshots for a given day are collected into a CSV file with a row for each active drive, and the downloads are grouped into quarterly zips for recent years and annual zips for the earliest ones.

Backblaze adds a warning for anyone joining quarters together: schema changes from quarter to quarter do occur, so check for changes each quarter and align the data before you combine it.

Reuse comes with conditions. The dataset page asks you to cite Backblaze as the source and to accept sole responsibility for how you use the data, and says: 'You can not sell the data itself to anyone; it is free.' The Q2 2026 post words its version of the conditions differently, so quote the terms from the page you are citing.

Those files hold the daily snapshots the reports are built from, so they let you check a published figure or ask a question the reports do not answer, such as how one model behaved at a particular age. Most readers will want the published tables.

What Backblaze drive stats mean for a UK buyer

Backblaze reports on the drives in its own fleet, not on a UK range. A listing on Amazon.co.uk matches a Backblaze row only when the part number is identical. Maker, capacity and family name are not enough, because a single suffix can change the model.

A fleet AFR is not a buying signal on its own. A quarterly figure of 1.73% after 1.24% says something about Backblaze's fleet in one quarter, and nothing directly about a drive bought today. The questions a buyer can take to the reports are model-level ones: does this exact part number have a row, how many drive days stand behind it, and how old was the group when it was measured?

Because the fleet holds no SMR drives, the reports also say nothing about how a shingled drive fares. How Backblaze's measured rates differ from maker ratings such as MTBF and a rated AFR is the subject of hard drive reliability. This page does not rank drives.

When the next Backblaze quarter lands

On 9 November 2026 the resource page listed no Q3 2026 data. Backblaze published Q2 2026 on 29 September, about three months after the quarter closed on 30 June. If Q3 followed the same gap it would appear around the end of December, but that is our arithmetic, not a date Backblaze has given.

When it appears, the latest-quarter section here moves to Q3 2026, Q2 2026 joins the list of periods, and each figure keeps the period it belongs to.