Backblaze Q3 2025 Drive Stats: Rethinking Failure, Celebrating High-Capacity Drive Strength

Four models achieve zero failures across 328,000+ drives; new failure analysis on outliers prompts dive into failure definitions

SAN MATEO, Calif.--(BUSINESS WIRE)--Backblaze, Inc. (Nasdaq: BLZE), the high-performance cloud storage platform for the AI era, today released its Q3 2025 Drive Stats report, analyzing the performance and reliability of 328,348 hard drives deployed across its global data centers.

Since 2013, Backblaze has provided the industry’s most transparent and widely used dataset on hard drive reliability, offering insights to IT leaders, researchers, and data professionals worldwide. This latest installment continues that transparency while exploring a fundamental question for modern infrastructure: What does it truly mean for a hard drive to fail?

The quarterly failure rate increased from 1.36% in Q2 to 1.55% in Q3, aligning with the 2024 annualized failure rate (AFR) of 1.57%. Four models recorded zero failures this quarter: Seagate HMS5C4040BLE640 (4TB), Seagate ST8000NM000A (8TB), Toshiba MG09ACA16TE (16TB), and the newly added Toshiba MG11ACA24TE (24TB).

Meanwhile, high-capacity drives (20TB+) grew by nearly 8,000 units and now represent 21% of the active drive pool. The lifetime AFR remains steady at 1.31%, consistent with the previous two quarters.

This quarter’s report examines how Backblaze defines and detects drive failures across its infrastructure. Using a combination of SMART monitoring, drive monitoring tools, and automated data enrichment, Backblaze distinguishes true mechanical failures from temporary removals. A drive is classified as failed when it either appears in data center work tracking systems or hasn’t re-entered the active population by quarter’s end.

The outlier Toshiba MG08ACA16TEY (16TB) recorded an unusually high 16.95% AFR. An investigation revealed that the spike was caused by one of several ongoing upgrades. The project required temporary drive removal and was not a result of mechanical issues; however, because of the reporting period and failure definitions, failures did show up in the current dataset. Backblaze expects the AFR to normalize in future reports.

"After more than ten years and billions of drive days, we've built one of the largest collections of drive reliability data in the world," said Gleb Budman, CEO of Backblaze. "This quarter, we're going deeper—showing exactly how we spot drive failures and turning that insight into something anyone can use."

The full Q3 2025 Drive Stats report is publicly available at www.backblaze.com/blog/backblaze-drive-stats-for-q3-2025/. The Drive Stats dataset and other supporting materials are available at www.backblaze.com/drivestats/. Since 2013, Backblaze has invited the global IT and data science communities to analyze, visualize, and learn from its openly available reliability metrics as it continues to set the standard for cloud transparency.

About Backblaze

Backblaze (NASDAQ: BLZE) gives businesses the freedom to innovate without limits by removing the barriers of lock-in, complexity, and cost. Our high-performance cloud object storage accelerates AI workflows, powers data-heavy applications, streamlines media management, and protects critical data. As an award-winning independent cloud, we provide unparalleled levels of interoperability that enables over 500,000 of our customers to reach and serve hundreds of millions of end users in 175 countries around the world. For more information, please go to www.backblaze.com.