Marc Suidan

AI is redefining how we think about storage, shattering old norms. post.dsmn8.com/q18-AtNHh0

Scaling AI Data Storage and Delivery with NVIDIA AIStore and B2 post.dsmn8.com/F8j-vKREp5

Blockchain analytics depends on keeping a lot of data available, not just storing it. Lukka continuously ingests data from more than 50 blockchains and preserves complete transaction histories. With Backblaze B2, it moved nearly 3PB of active data to a shared storage layer serving compute in AWS Frankfurt and its on-premises data centers. The result: Lukka can run jobs wherever compute capacity is available. More than 9PB of its data is read outside AWS each month, and those reads fall within B2’s included free-egress allowance. post.dsmn8.com/EDUYPN2P0f

Neoclouds: Your Storage Layer Just Got a Public Grade post.dsmn8.com/CmD3PnNy-1

Code42 Is Shutting Down. Is CrashPlan the Safe Choice, or Just the Convenient One? post.dsmn8.com/fxrb4mgtES

Why Your Computer Backup Strategy Determines Whether You Pay the Ransom post.dsmn8.com/rFXsFKEspm

The fastest-growing AI platforms trust B2 Cloud Storage. So can you. Moving data shouldn't cost more than storing it. B2 Cloud Storage means no surprise egress fees eating into your AI budget. post.dsmn8.com/NvRQ6GBmjC

When scaling #MachineLearning models, teams usually face two massive hurdles: delivering data fast enough to keep expensive #GPUs fully utilized, and finding a cost-effective way to store those massive datasets long-term. That's why we're excited to announce our new ecosystem collaboration with @WEKA to deliver a seamlessly integrated, high-performance storage solution for your most demanding workloads! 🤝 By combining the predictable performance of WEKA’s NeuralMesh with the scale of Backblaze B2 #CloudStorage, joint customers can experience a fully validated integration right out of the box. Why build, test, and troubleshoot a custom storage architecture when you can deploy a proven solution? Keep your compute clusters fed, scale your retained AI assets effortlessly, and get back to building. post.dsmn8.com/mFmSWJC4CM post.dsmn8.com/-fnEvqHTgw

The fastest-growing AI platforms trust B2 Cloud Storage. So can you. Moving data shouldn't cost more than storing it. B2 Cloud Storage means no surprise egress fees eating into your AI budget. post.dsmn8.com/RKNpxxwcmD

Congratulations to Barbara Nelson, independent director and chair of our Nominating and Governance Committee, on being named to the 2026 NACD Directorship 100™. The honor recognizes directors whose work has shaped boardroom practice and performance. Barbara's record fits. She's led global P&Ls as large as $3 billion, twice as CEO. She's chaired boards across AI, cybersecurity, data storage, and health care. And she brings that experience to Backblaze as we deliver the object storage layer on which today’s AI innovation depends. Congratulations, Barbara. NACD (National Association of Corporate Directors) post.dsmn8.com/C6ysahdrM4

The failure rate went up this quarter. We're publishing the number along with the context that matters. Q2 2026 AFR came in at 1.73%, the highest in a while. Here's what's driving it: two of the three outlier drives are seven-plus years old, and the remaining are small-population Seagate models where a handful of failures produces a big-looking percentage. None of it comes from the drives doing the real work of this fleet. Meanwhile: 20TB+ drives are now more than a quarter of everything we run, and two of our longest-serving models — a 4TB and an 8TB HGST, both close to a decade in production — just retired. That's the story this quarter: capacity is climbing fast, and the drives replacing the old guard are holding up. No other cloud publishes this level of model-by-model detail. Thirteen years in, that's still the point of doing this at all. Full report, tables, and the complete dataset are in the comments. We're also walking through it live on Oct 8. #DriveStats post.dsmn8.com/qt5n8SHpvK

Backblaze B2 x Suite S3 Native File Streaming is available today 📂 The integration turns a Backblaze B2 bucket into a high-performance working drive. Teams can mount an existing bucket through Suite and work with cloud files as if they were stored on a shared NAS. Behind that experience: 👉 One file in Suite maps to one object in B2. 👉Files remain standard, whole S3-compatible objects. 👉Your bucket stays the system of record. 👉 Suite streams only the data each application needs. For media teams, that means editors can work in applications such as Adobe Premiere Pro while MAM, transcoding, automation, archive, and distribution tools access the same underlying media. Read more on how to get started, on the blog ➡️ post.dsmn8.com/AqxvxS4DPj post.dsmn8.com/n3tTpY4K5a

What IT Teams Get Wrong About Immutable Backups and Object Lock post.dsmn8.com/GkJ0-fDGtU

Zero-Touch or It Doesn’t Scale: The New Standard for Mac Fleet Backup post.dsmn8.com/DC-96BMyG0

The fastest-growing AI platforms trust B2 Cloud Storage. So can you. Moving data shouldn't cost more than storing it. B2 Cloud Storage means no surprise egress fees eating into your AI budget. post.dsmn8.com/wG39aEmjde

AI agents crave a rock-solid, reliable spot for their creations. Enter the open-source Backblaze B2 MCP Server. It offers AI agents a secure, standardized path to cloud object storage. Leveraging standard B2 application keys, it ensures least privilege access. Plus, it directs hefty file transfers straight to storage, keeping your model running smoothly. Check out the 1-click install for Claude Desktop and dive into all the launch details here: post.dsmn8.com/nSvuJeXCw5 post.dsmn8.com/TB-gFyGdvD

Four Platforms, One Timeline: Bringing TAMS Workflow to Life on Backblaze B2 post.dsmn8.com/yuGmB0Ab62

Is your costly #GPU fleet just cooling its heels? At NVIDIA's GTC, Adobe lit up the room with a revelation: in their initial generative #AI training phases, about two-thirds of GPU time was wasted waiting for data. That's like tossing 60 cents of every dollar spent on compute. 🤯 The villain of the piece? A jammed #DataPipeline. Conventional storage and networking aren't cut out for the relentless, petabyte-scale parallel data movement needed for training cutting-edge models. If your storage lags, even the slickest data loader won't stop your compute from twiddling its thumbs. Don't let storage bottlenecks drain your compute dollars. Discover how to keep your clusters bustling: post.dsmn8.com/0a4TNf8e9K post.dsmn8.com/fkRSaSKsA1

Thrilled to see the magic the data science and tech communities create with our open-source Drive Stats data! 📊 Economists just released a peer-reviewed study via IEEE, diving into 12 years of our public data, spanning over 443,000 hard drives and 1.66 million drive-years. The study went beyond simple manufacturer rankings, uncovering some intriguing insights: 🌡️ Temperature impact: A 1°C rise in average drive temp increased failure rates by 2.1%. 📈 Capacity advantage: Larger drives had lower failure rates, thanks in part to those newer helium-filled designs. 🏆 Brand champions: HGST and WD led the pack in long-term reliability. Get the full scoop and analysis here:🔗 hubs.ly/Q04vqFkn0 post.dsmn8.com/JVAFrTBGgc

#B2Overdrive won SiliconANGLE's 2025 #TechForward Award in #CloudNetworking less than six months after launch. 🏆 That's because it's an unusual solution in the market: one predictable monthly price, unlimited API calls, unlimited egress to other cloud providers and on-prem, and up to 1Tbps throughput. hubs.ly/Q04rJCJ40 post.dsmn8.com/QbJjjHPY49