KubeCon + CloudNativeCon NA 2025 | Gleb Budman, Backblaze

Gleb Budman, Backblaze

Exploring Cloud Storage Solutions at KubeCon + CloudNativeCon NA 2025: Insights from Industry Leaders Gleb Budman, Chief Executive Officer and co-founder of Backblaze, joins Savannah Peterson, Principal Analyst and host from SiliconANGLE Media, Inc., and Rob Strechay, Director and Principal Analyst from theCUBE Research, at KubeCon + CloudNativeCon NA 2025. This engaging discussion focuses on advancements in cloud storage, data handling, and their pivotal role in the realm of artificial intelligence (AI). During the conversation, Budman shares insights about Backblaze's journey as a key player in the cloud storage domain. They delve into Backblaze's role in evolving data storage solutions, from foundational backup services to AI-focused data management. Budman emphasizes the impact of cloud platforms on AI development, sharing Backblaze's initiatives such as B2 Overdrive for rapid data throughput and highlighting their cost efficiency compared to hyperscalers. The dialogue, hosted by Peterson and Strechay of theCUBE Research, sheds light on how Backblaze is paving the way in facilitating data movement and storage for various industries. Highlighting key insights, Budman illustrates Backblaze's expansion into AI, stressing the importance of robust and cost-effective data storage solutions tailored for modern AI demands. According to Budman, Backblaze supports crucial AI workflows, significantly lowering storage costs and eliminating data egress fees. Both Strechay and Peterson highlight the growing significance of open, collaborative ecosystems, affirming that Backblaze's approach enables seamless interoperability with multi-cloud environments and supports diverse use cases across enterprises. This discussion provides valuable insights into the evolving landscape of data management and AI, spotlighting Backblaze's critical role in this progression.

Share this session

Clips
More from KubeCon + CloudNativeCon NA 2025

Gleb Budman

CEO & Co-Founder

Backblaze

play_circle_outline Backblaze addresses backup needs, ransomware protection, and evolving AI solutions.

play_circle_outline Backblaze's B2 Overdrive service allows terabit per second data throughput.

play_circle_outline Customers using Backblaze for diverse AI applications like image libraries and drug discovery.

play_circle_outline KubeCon as a central hub for AI development and Kubernetes integration emphasized.

Info
Transcript

Gleb Budman, Backblaze

Gleb Budman

CEO & Co-Founder Backblaze

In this KubeCon + CloudNativeCon North America 2025 interview, Gleb Budman from Backblaze joins theCUBE’s Rob Strechay and Savannah Peterson in Atlanta to unpack how Backblaze’s cloud storage platform is evolving from backup mainstay to AI data backbone. Budman explains how customers are using Backblaze to store massive datasets, push them at terabit-per-second speeds with the B2 Overdrive service and move a growing share of traffic – now roughly a quarter of outbound data – to neocloud GPU providers such as CoreWeave, Nebius and Lambda for model training and... Read more

explore Keep Exploring

What types of problems does a large cloud storage platform solve for its customers? add

What are the stages involved in processing data with Backblaze, and what recent service enhancement has been introduced to improve data throughput? add

What are some ways companies are utilizing large amounts of data for new business opportunities? add

What role does Kubernetes play in the development and scalability of AI systems, and how does it interact with data storage platforms? add

bolt Powered by CUBE AI

Gleb Budman, Backblaze

search

Savannah Peterson

>> Good morning, nerd fam, and welcome back to Atlanta, Georgia. We're here kicking off day two of KubeCon. My name's Savannah Peterson. Delighted to be bringing you the best and the brightest from the show floor with the one and only Rob Strechay. Rob, this is our first time in Atlanta together.

Rob Strechay

>> It is. You've been here a bunch lately, but I haven't-

Savannah Peterson

>> I've been here three times this year....

Rob Strechay

>> I haven't been here in a while, so I'm excited to be here and sharing it with you.

Savannah Peterson

>> Yeah, thank you. I've been enjoying the South. It's a good little touch. Speaking of nice, warm, fuzzy feelings. Our next guest is somebody I've known for a decade. Gleb, welcome to the show.

Gleb Budman

>> Thank you very much for having me. It's great to see you. It's great to be here with you guys.

Savannah Peterson

>> Yeah, it's so fun. Love keeping it in the Backblaze family. Before we really dig into the meat of this. Backblaze, publicly-traded company, y'all have been on quite a journey over the last 14, 15 years and I mean I've gotten the front row of the evolution. Tell us a little bit about where you are now and the types of problems you're solving for your customers.

Gleb Budman

>> Sure. So, we have a large cloud storage platform, one of the largest, as you know, and we're solving a variety of problems. Some of them start as simple as companies that have a backup need and we are a destination for their backup software.

Savannah Peterson

>> You have always owned that space.

Gleb Budman

>> That has been a core bread-and-butter space. It's evolved over the years because with ransomware protection and the like, it's evolved. But increasingly, we're now solving customer's problems around AI. Everybody is trying to help. And what we're seeing is customers building models, customers doing inferencing and needing just a place to put a lot of data and be able to move that data quickly to wherever they need it to go.

Savannah Peterson

>> And having the safety and security of that with a trusted partner like you, I can imagine is critical. Yeah, go ahead Rob.

Rob Strechay

>> Yeah. No, I mean I think that ties off with a lot of what... We were talking a little bit beforehand and you guys are really doing some interesting stuff where you've looked at your network and what's going on because I think a lot of people are saying, "Is it a bubble? Is it not a bubble?" But I think people need to get the ROI right, and I think they have to use the data from where it is and use it. So, what was some of the stuff that you were seeing with different destinations for the data and things like that?

Gleb Budman

>> Yeah, so you followed this for a long time, so you know we've been publishing these hard drive liability statistics for over a decade, where we take hundreds of thousands of hard drives that we have and we monitor them and then we publish a blog post about it.

Savannah Peterson

>> It's always been really cool. It was tangible and real, not just some hypothetical 10 people we surveyed and here's what they said.

Gleb Budman

>> Yeah, we do periodically get someone saying, "Oh, but my one hard drive didn't perform this way." And we're like, "Okay, that's true-"

Savannah Peterson

>> Congrats.

Gleb Budman

>> "... but we've done about 300,000 hard drives and here's the data that we see." Now, every so often we've gotten people complaining to us saying, "You're publishing these reports. I want you to do this report analysis a different way." And we're like, "This is not our business to publish these reports. We're just sharing this with the community to put it out there."

Savannah Peterson

>> To be inclusive and to let people know what the landscape is.

Gleb Budman

>> Exactly. So, what you were talking about is one of the things that we've just started doing is publishing network statistics. So, as a storage backbone for the internet, we see a lot of the data and where it flows. And what we used to see is a lot of the data would flow from our customers to an ISP or to on-prem or to a hyperscaler. And one of the things that's been fascinating is that with the one that we just published yesterday, we've saw a quarter of all of the data is going to the neoclouds.

Savannah Peterson

>> Wow.

Gleb Budman

>> A quarter. So, these are-

Savannah Peterson

>> That's significant.

Gleb Budman

>> It's significant. And these are organizations that didn't exist really a few years ago. They're all the GPU rental companies, CoreWeave, Nebius, Lambda, et cetera. And so, that, that's been fascinating to see. So, what that says to us is customers are storing a lot of data on Backblaze and then they're using it for AI purposes, sending it to these neoclouds.

Rob Strechay

>> Yeah, which is great because I think a lot of them are here. A lot of the neocloud guys are here, some are on the main stage and they're talking about just how the entire stack is progressing. Where do you see Backblaze really fitting in that entire process? Because again, data is core to AI in a big way. I mean, you can't-

Savannah Peterson

>> Well, there is no AI without data.

Rob Strechay

>> There's no AI without data, to put it mildly. So, where do you see your fitting and how customers are using you?

Gleb Budman

>> So, we think about the five stages of an AI workflow. So, first, you need to grab all the data. It's got to come from wherever it comes from, whether it's scraped data, synthetic data, your own data, wherever you have to assemble all the data. Then, you have to make sure that it's labeled, so that you can organize the data. Then, you send it to a place where you can create a model. From there, you've got your model, then you use it for inferencing, and then you're running your application, you're doing inferencing, and then you log the data. So, there's the five stages and we help customers in each of the five. So, we'll help customers collect all the data and store it on Backblaze. We'll enable them to label it there, then send it as fast as they can. We launched a service called B2 Overdrive earlier this year. Customers can do up to a terabit per second of throughput. We think it's the fastest throughput offering per dollar on the market full stop worldwide.

Savannah Peterson

>> And that's what everybody wants right now. At the end of the day, I mean, especially as we're in the messy cutting room floor era of AI, as people figure out what their MVPs are, that cost is really critical. You have some data in here about AI tooling and competition looking at retailers using AI report 5% to 15% annual growth and operational cost reductions of 30%. You're on that operational cost reduction end then too.

Gleb Budman

>> We are. It's interesting because we definitely save companies money. I mean, we're about one-fifth the price of the hyperscalers for the storage.

Savannah Peterson

>> Go ahead and dust your shoulders off on that one.

Gleb Budman

>> There's a lot of work over 15 plus years that the team has worked into really making the storage platform very efficient. So, we're about one-fifth the price for the storage, but then one of the huge parts is the egress, right? Because there are these neoclouds, there's about 200 of these neoclouds now around the world, and customers are saying, "I want to use that one," or, "I want to use this one." They need to get their data over to them, and the hyperscalers charge these crazy fees, these egregious egress fees to send your data out. So, we make that free. And so, we do. We save people a lot of money. But I think the stat you were referencing about the growth is that companies are also finding that they drive more revenue from using AI. And so, it's not only a cost savings, it's also an innovation and a revenue growth opportunity.

Rob Strechay

>> Yeah. And I think again, to your point about why does speed and efficiency matter? I mean, these neoclouds have to keep those GPUs busy. When somebody's looking at it, they need to know where their data is and where to move it to. And to the point of egress, that's how people... I worked for one of the hyperscalers. Even me building a service inside of one of the hyperscalers, that was always a worry of mine was because people wanted to move data out. And I was in the hybrid side of things. But when you look at this, are you seeing people coming to you with new use cases about, "Here's why I want to use you in this workflow, because again, I need that performance to keep and feed those GPUs"? But to your point, that other part of the five steps, hey, a big conversation here is inference. And it's being-

Savannah Peterson

>> It's been exciting, actually. It's been such the focus... I'm loving it. We're finally making it real.

Rob Strechay

>> Right, it's about the real use cases that actually drive return. Is that a lot of what you're seeing is people coming to you with new, "Hey, we have some AI now. We're going to want to expand it, but we're looking at inference and other things that we need to put in other places and have good data to feed that"?

Gleb Budman

>> Yeah, so we see both and we see actually at all the stages. So, one customer that we have, they have been building an image library for decades, frankly. And so, they've got tens of petabytes of images, and now they've discovered a new revenue stream, which is selling that catalog to other model builders. And so, what they needed is a place to store the data and then be able to send it to model builders very quickly. So, they're in the business of selling that data. We have have another customer that is actually collecting this... A company comes to them and says, "We need the following data. Can you go get it?" And they go out and they scrape the internet and find the right places, and they collect that data and then they store it on Backblaze, and then they send it. So, there's the front-end of that. But then, we also have lots and lots of companies, hundreds of AI companies doing all the inferencing stuff. So, I know one company I was looking at, they actually are helping build robots that build physical buildings.

Savannah Peterson

>> Cool.

Gleb Budman

>> It's very cool. Whether it's the housing shortages that we have or commercial businesses, having it more efficient to build buildings faster is awesome. And they're doing robotics to do that while they're using us for the data platform, for the AI for those robots. We have another customer that is accelerating drug discovery using AI. So, we're seeing it both on the front-end, which what we're calling the supply side of the AI world. And then, we're also seeing it on the demand side, which is the, like you said, putting into real-world use and actually getting outcomes for medicine, for driving, for video, for content and the like.

Savannah Peterson

>> On that, just because those are wonderful customer examples. And we're getting what I wanted when we talked in our analyst segment yesterday, I was like, "I want to hear what people are actually doing. I want to hear the out-in-the-wild stories." What's the most exciting application that you've seen within your customer base that you can talk about for you personally?

Gleb Budman

>> For me personally?

Savannah Peterson

>> Yeah.

Gleb Budman

>> I mean, I will say that it's not the ones that are the most applicable for me today, but the things that I am most excited about it are the ones that are around medicine, right? Because it does feel like the idea that in a year, three years, five years, 10 years, we will have potentially many more medical solutions than we have today is amazing. And so much of medicine a is analysis-based. It is trying to correlate, well, we use these things. We saw these... The idea that you could evaluate tons of MRI, CT scans and everything, cross-correlate that with all of the statistics and biomarkers and everything and have more personalized medicine, I think for all of us might be the biggest thing that comes out of it.

Savannah Peterson

>> I really agree with you. I think we're going to be able to do so much more preventative medicine as a result of that, rather than trying to figure out the mystery when everything's going on and it's kind of scary.

Gleb Budman

>> Yeah.

Rob Strechay

>> And I mean, I look at it and go, I think the timing for you guys being here is so right. I mean a lot is going on within KubeCon and CloudNativeCon as we expand out. And there's a lot, I mean the focus on AI, and I think they're starting to get with some of the new projects that are coming into the CNCF and this. But from your perspective, how's the fit with Kubernetes and what you guys do? I know, but why don't you tell people why?

Gleb Budman

>> I mean, we're sitting in the heart of where AI is getting created here. So, I think for us, what we see is obviously Kubernetes is the technology that people are using to build the systems to allow AI models and processing and inferencing and everything to happen, scale and be more efficient. All of that then needs a data platform to be supported. It's not inside of Kubernetes itself, but Kubernetes speaks to Backblaze into all the storage layers. So, it's a very natural fit of enabling all those workloads.

Rob Strechay

>> When I look at it, I think the fact that you guys are multi-cloud friendly, you've got S/3 compatibility and it makes it easy to use and bring to that. Are you seeing that the folks that are using Kubernetes are just the barrier to entry's got to be so low for them in that way?

Gleb Budman

>> Yeah, and it's an accelerating path. And one of the things is everybody is figuring it out. So, there's all these new tools that are popping up all kinds of new open source projects that are popping up to make all these workflows happen. And each time that there's a new breakthrough, there's a new problem because you break through and then there's a new bottleneck of what's the next thing you want to solve? And so, the nice thing is most of those things require data, that's nice and fresh, they require data or they create data. And that data is generally increasingly being put in object format. So, having us be an easy-to-use object storage solution that is compatible. And one of our core beliefs that I think is unique from the hyperscalers is we want to enable this whole ecosystem. We want this all to exist. We don't want to say, "You have to stay inside of Backblaze. That's the only thing out there," right? You're not-

Savannah Peterson

>> And siloing everyone in a little data jail.

Gleb Budman

>> Exactly, right. There's so much innovation happening here, and so we want to facilitate that. And the way you facilitate that is by saying, "The data is yours. We are here to provide you a really durable, fast, easy-to-use platform, but it's got to be open. You've got to be able to send the data and use it how you want."

Savannah Peterson

>> Well, I'm sure that resonates particularly in this room. The folks who've been building transparently and open are really starting to reap the benefits, I think, as we see AI get to scale, which is very exciting. I'm curious from a education perspective, because you've worked with a lot of your customers for an extended period of time, probably got a whole new pipeline coming in with this AI offering. How are you and the team approaching educating your customers, so they know the availability of all of this and they know what to do? I mean, you've educated us just sitting here on the news desk.

Gleb Budman

>> So, we have a blog that's actually quite widely read. A few million people a year read our blog. Historically, it's been very storage-focused. And so, for storage enthusiasts, it's been a popular blog.

Savannah Peterson

>> Oh, yeah.

Gleb Budman

>> And we're increasingly trying to add more information there for practitioners and people who are using it. So, we share information not only on drive statistics and network statistics, but also, how do you do things with your data? How do you build? Giving use cases and customer examples. So, a lot of what we're trying to do is share that information with the community through evangelism, through the content on the blog, and through going through shows like this and sharing information.

Savannah Peterson

>> I think it's so important. It's something I've always known you did well, which is why I wanted to call that out. I think there's the quiet whisper conversations, and then there are other folks who are saying, "Hey, here's how you put the engine in the car and put the wheels on. Up to you how fast you drive, but we're here to help you out with that. We'll be the nitrous." Final question for you, Gleb. Everything's moving pretty fast right now. When we are hanging out at KubeCon in Salt Lake City in 2026, what do you hope to be able to say then that you can't yet say today?

Gleb Budman

>> I think what I'd love to be able to say is, "Look at all these incredible companies that have been able to exist and to innovate because of what AI is enabling and what we have helped them enable with Backblaze."

Savannah Peterson

>> Perfect. Well, hopefully, we'll get one of those stories on the stage with us next year and we can make that magic happen.

Gleb Budman

>> That's right.

Savannah Peterson

>> Thanks, Gleb, for being here. It's so awesome.

Gleb Budman

>> Thank you for having me. Thank you for having me. Good seeing you guys.

Savannah Peterson

>> Of course. Yeah. And thank you, Rob.

Rob Strechay

>> Thank you.

Savannah Peterson

>> And a quick shout out to the entire Backblaze team because they've always made me feel like family. I hope you're all having a beautiful day, wherever you might be. We're here in Atlanta, Georgia at KubeCon. My name's Savannah Peterson. You're watching theCUBE, the leading source for enterprise tech news.