Backblaze Aktie

Hi everyone. Welcome back to our quarterly drive stats report. My name is Stephanie Doyle, and I am the Senior Manager of Market Intelligence and keeper of stats. And joining me today is David Johnson, who's going to walk through this quarter's drive stats report as well as we're going to get into a discussion of some of the drive technologies on the rise in existence that are bringing the industry to what it is today and will be tomorrow. So David, do you want to introduce yourself?

Yes. Thanks, Stephanie. It's a pleasure to be here today. I've followed drive stats for a very, very long time. I've been in tech for a while now, so it's just like a backbone to me. I think we're in our 13th year of drive stats, I believe. So we're still kicking strong.

Yes, it's 13-plus. We started tracking the data and then we did our first report with a little bit of a gap. So if you look at the full data set I think we're not quite at 14, but if you look at where the reports started, we're at 13 solidly. So it's a data set that only keeps on growing and definitely keeps on giving, so always fun to do these reports.

A little housekeeping for folks who have joined us before or -- who are joining us for the first time today, webinar is being recorded. Of course, please ask questions. And there will be attachments that we add at the end, including the slide deck from today. You can always catch the blog and all that other good stuff on our website. We'll have links to those at the end of the slide deck as well.

So this is the project, of course, since 2013, David, you and I just talked about how long it's been around. We've collected, curated and published raw device metrics including -- and then transformed that into annualized failure rates and related statistics from hard disk drives and solid-state drives in our data centers. Each quarter, we publish this open-source data set, which can be downloaded on the Drive Stats page, check that out at backblaze.com/drivestats. And we do try to keep up an iceberg version of this data set on our GitHub, so you can find that as well linked.

This is a top line snapshot. So I always find interesting that drive count up and only going up, as I say. Drive days, of course, is how many days, all of our drives had to date. And then drive failures is just the raw count of numbers there. Drive population by manufacturer. Of course, we had WD purchase HGST. So from that perspective, we're at about 1/3, 1/3, 1/3 in drive manufacturers, which is always helpful when we're looking at numbers.

Certainly, not one to play favorites. We have a lot of -- there's a lot of great drives out there, and so we like to have everything nice and balanced.

Yes. And that's the benefit of our platform. It allows us to be able to use a good amount of these drives from different manufacturers and be able to have very efficient software and infrastructure that can help run it all.

Software layer on top of commodity hard drive, driving all the value there, of course. So this is the Q2 2026 data, as always, incredibly difficult to read the full table. We always encourage you to dig in on the blog itself. But you can see here at the bottom that our annualized failure rate for this quarter was 1.73%, and this is a market rise in quarterly AFRs.

Why? We did -- first thing that I always do when I see a quarterly rise is to run an outlier test. But before we get there, let's talk about drives that had zero failures. So clean sweep from the Seagate this month, we had zero failures on the 8-terabyte, 12-terabyte and 14-terabyte Seagates. These models identified here and one on the 16-terabyte Seagate. And we also saw the retirement of some very old drives in our system, so that would be the 4-terabyte HGST, 8-terabyte HGST as well, clocking in at 9 and 8 years old.

And then a little bit of a curiosity here. No new drive models. We certainly had new deployments, but they were all of existing drive models. I flagged last quarter that it's actually quite rare for us to not have a new drive model, I think, in the last 3 years or something like that. It's only happened once. So to get two quarters in a row, feels a bit odd but it is nothing to say there. It's really just so we're getting the same models deployed in our system at about the same volume.

Do you think that speaks to just efficiency of where we're at nowadays? I mean, it could be a number of things, but it seems like it's a good thing that we're sticking with the same models.

I think the biggest thing is that it's not that our growth of platform hasn't slowed down. I think it's the biggest reason I really want to talk about it, it's just that when you're deploying vaults as much as possible, you want them to be really identical. So if we can get our hands on the same models and those models are functioning well within our system, why wouldn't we continue with the drives that are already qualified?

Of course, there's like work on multiple ends of the spectrum there, qualifying drives. It also means you have more capacity to swap in and out if you have failures, all that good stuff. So in terms of creating a happy system, it certainly helps to have similar models, especially if they're functioning well.

So let's -- I think we did not pull our review of the outliers, but you can certainly go check out the blog for that. But let's go back to the table for just a minute, and you can see here that there's a couple of models. This one up here is 7.63%. And then you've got 8.26% here. And a third with -- excuse me, I'm losing it here, 9.33% on the Seagate 10-terabyte. So those three, from a statistical standpoint, outliers. And they did have very normal reasons for potentially reaching this high failure rate.

So in one case, you -- or actually, in two cases, you had drives that were quite old, so more than 7 years. And then in two other cases, so that would include one of those quite old drives, you had drives with a very low drive count. So that means that if you get a raw number of drive failures like looking at the Seagate 14-terabyte with the 8.26% failure rate, it's really only 25 failures, but when you only have 1,200 drives that makes more of a difference in your AFR at the end of the day.

So two of them that were quite old, two of them that had very low drive counts. And for that, those are all pretty normal indicators that they're actually already being cycled out of our system, so not a huge amount of worry even though the AFR did seem high on the surface.

Let's talk lifetime data. So again, you've got the full table here. The AFR down here at 1.41%. That's like very consistent to actually the AFR for the lifetime data for understandable reasons, tends to not fluctuate as much as the quarterly. So I think it's usually 1.39%, 1.40%, 1.41%. I feel a little bit like a broken record when we review the data on this one, which once again speaks to the fact that we are a managed system. Somebody is trying to make sure we're within good and healthy ranges there.

Yes, it's always a good thing when things kind of stay the same. It means things are working, nothing's going bad.

Yes. We want consistency, we want a low failure rate, we've got them, right? And there's always going to be some variability things you can't account for. But certainly, there's -- over time, it tends to come out in the wash.

Four models didn't make the cut to the lifetime table. This means that they've dropped below your total drive count. Interestingly, several of these were also our zero failures, David.

Yes. I always find that quite interesting because we've said for a long time, drives are getting better, lasting longer. And those aren't -- you still have to migrate things out, but I find it's always a question from our readership where it's like, why are you retiring the drive if it's still going so well. It's like, well, you have to do risk mitigation at a certain point. But if you're looking for longevity of drives, readership, sometimes these things happen, right, which is pretty cool.

Yes. It's very interesting to kind of look at all of the data that I've seen over the years where kind of see this trend moving where we're cutting out some smaller drives, putting in those larger drives and seeing how the failure rates track against those. And I know like just doing some math off the top of my head, I think it's like nearly 1/4 of our fleet now is approaching the 20-terabyte mark in drives. So I'm really interested to see that same kind of bathtub curve almost as we get into the [ larger ] drives.

I think it's important to note, and we did get this question on, I believe, our Reddit platform, like do we see inherently a larger failure rate for larger, smaller drives. I think the mix of capacity within our system really is indicative of industry trends over preference, although there is a little bit of a preference to maximize your -- how much density you have per rack.

So if you put 20 6-terabyte drives into one server, you can -- it doesn't store as much as 20 20-terabyte drives per server, of course, right? But we haven't really seen higher failure rates in the population as correlated to age, and that's something that we did most -- that analysis we did most recently in 2025, sort of separated out our drives by introduction point as well. But I will say that it's a bit early to make that call, right, because the 20-terabyte drives are still quite young, and I think our oldest one in the system is maybe coming up on 2 years. So certainly something to pay attention to over time, especially as we talk about new storage technologies, right?

So I brought this forward in our drive stats report this time around because I think that it's probably one of the biggest questions I've been getting at my last couple of speaking engagements. So let's just start with CMR, I'd say, and then talk SMR. So David, do you want to walk people through what CMR is and what it would mean to like a consumer in terms of what a hard drive is itself?

Yes. I mean CMR, it's kind of what we've been running with for a while now. I don't know exactly know how long we've been running with them, but it's kind of been the standard we've been with. I mean, they're...

Kind of your normal...

Yes, it's your normal. They're predictable, they're proven performance. It kind of led us to a lot of good innovations over time. But as we're kind of getting to the point where we're seeing more performance workloads with sequential things. We start to see some of the limitations of CMR drives. And I think that's where the industry is now moving into the SMR side of things, which starts to solve some of those issues and also give us some greater capacity as we move forward, too.

Yes. I mean I think my perspective is that the single biggest driver to SMR is what's been driving the drive industry for a long time, how do you store more data in the same physical drive, right? We've seen adding more platters. We've seen magnetically assisted reads and writes, all that other good stuff. And shingled magnetic recording is quite interesting, I think, because it actually uses your platter zones in a different way. and we have some handy-dandy visuals to make this clear because I think that it's a bit complex if you don't understand what's happening with the -- on the physical level.

So the first thing is the reason this is possible at all, is because the width of the head that writes your data is narrower than the width of the data that reads -- or the width of the head that reads your data. And we -- this is actually true in both CMR and SMR. So what you're looking at here is actually a CMR drive, how it would read and write the data, and these bands would sit side by side like traditionally, right, your widest, which means that you can write simultaneously and all other sorts of good stuff.

In an SMR drive, what you actually do is you overlap the tracks so that you can create more density, and you're write-with is going to allow you to write like this, whereas you're read-with is going to be kind of sitting inside that or overlapping that. So if you were to look at these side by side, you can really see the difference here. And this lets you make more use of this -- the existing space.

It does also have implications for how you can read and write? The biggest one is that you can't sequentially write things because you will have overlap, right? And so there's things that make it -- within a data center, things that you have to think about in terms of how you interpret it. There are also firmware implications there. So you have drive assisted -- or driveware SMR or you have firmware -- like firmware assisted -- let me go ahead and look at my notes because I feel like I'm butchering this now. But you have drive managed SMR, host-managed SMR or host-aware SMR. And that's just who makes the decision where, is the drive doing it, is the host doing it, or is it a combination of the two?

So all these things have implications for how we provision drives. It also let you get more data onto a single drive, which is very exciting, right? That's the same way we're talking about what we can do with bigger drives in a single server. Now we can do this in a single drive. We personally do not yet have SMR drives rolled out. But I think the biggest thing what we're hearing from all the drive manufacturers is that this is where the industry is going. So if CMR is our normal now, SMR is going to be everything in about like 18 months, 2 years, at least for enterprise. Is that what you're hearing too, David?

Yes. I think we're still a little ways out, but that feels like a long time, but trust me, it'll go quite very quickly.

Yes. So when you talk about why this matters. Number one, it matters because we're seeing drives get bigger and a lot of very interesting technology on the bottom of that. And number two, it matters because you have to sort of optimize your storage layer for the new way these things are working. So for Backblaze, that's kind of been our bread and butter this whole time. We're looking forward to it. You do have to interact with that both directly with the drive manufacturers and within your own system.

Stephanie, before we jump into HAMR, we had one question that popped up with the SMR drives. They were wondering, don't they usually have a massively increased rebuild time when they're used with a [indiscernible] drive?

They can, but there's things that reduce that. So when we're talking about provisioning, right, we're sort of looking at all the different ways that will matter. And I think this is a pretty astute question. I'm looking in the chat. I see Robert, you asked this, because when you're -- it's one of the biggest things that affects parity.

So we know that our parity schema has had to shift as we have more large drives, right? Because as you talk about what -- when you think about what parity is, it's a formula. It is, in fact, just the time it takes to rebuild data as one of the parameters completely without any data loss. So if you think about how your shards have to be distributed, the larger your drive, the more likely your shards will end up too close to each other, so you actually have to increase and reduce your parity shards accordingly.

Now if SMR takes longer to rebuild, that would also affect your parity schema. So when you talk about a single drive rebuild, again, we haven't gotten into provisioning these yet. So we're not -- we haven't thought about what it means for our testing purposes. There may be things you can do on the software layer.

But one of the biggest things is when you're talking about a ratio coating, it actually has a built-in mechanism so that we can control for this on some level, right? We might change the way our parity schema works in sort of in our distributed file rebuilds. But that does also mean that when you're rebuilding a single like a vault or if you have a drive that goes out, you might have to account for that.

Again, Robert, not something we have gotten into testing yet, so I couldn't give you real numbers on that. but certainly something our drive team will be on top of. So I throw in HAMR here, not necessarily because it's related to CMR, SMR. I know there was some confusion around that. But because a lot of times when you get into like the acronym soup of what is happening in drive technology, this one comes up to, they're like, well, but then what is HAMR if these are CMR and SMR?

And HAMR, I find interesting because it's one of those -- I'm calling it a co-technology, but that's kind of not exactly perfect, but it's another piece of how we write to drives that significantly moved us forward on drive density. So whenever you're writing on a drive, you can see this north-south layer here. And your biggest issue is that if you haven't written things well enough, those bits can flip -- and when those bits flip, you have data corruption, right?

So the first thing we did to help with this is heat fields or magnetically assisted writes. Those are two different ways you can create more durability. And HAMR itself is actually shoots a little laser straight down in there and gets a lot better write durability, which is very cool. And so it makes much more stable magnetic grades.

What's even more interesting about this is that these are typically written onto a glass substrate versus aluminum. And you can do that because you can heat it up to a higher temperature, right? So they're already standard in like 2.5-inch drives and increasingly common in your 3.5-inchers. And so when people say, CMR versus SMR, it's not really a versus HAMR. It's like, and also there is HAMR.

But if you look at, say, what is it the Mozaic drives out of Seagate, I believe all or almost all of those are glass substrate and HAMR drives. I may be speaking on the glass substrates specifically, but they're HAMR drives. That's what's getting them to those 40 terabytes and Western Digital is projected to get to 100 terabytes with HAMR technology as well.

So I did some leaking out in the blog article, in case you want to see what drive manufacturers are talking about with that. But just know it's not an either/or, it's an and. I think that's helpful for folks to know.

And it's also just a great shorthand way of saying HAMR is pretty good, easy to remember.

I did go heavy on the puns this time around with like building, considering that hard drives are the building blocks of our data platform and all sorts of stuff. It was great.

Got to love the acronyms.

Get right in there. So I'm seeing one more question in the chat here, so Tushar, I hope I'm pronouncing your name correctly. Do we also track data for SSDs? What a spicy question that we get often. I will tell you, we have put out data on the SSDs in our system before. It is deeply imperfect because when you compare what our hard drives do versus what our SSDs do, they're completely different workloads. So we don't really have any data storing SSDs.

We do have a layer of SSDs that are right through cash. So that would mean that they receive the input capture the data, tell you on the other end, hey, we've got your data because we do, and then they move that through to hard drive as in a queue, right? So it gives you better speed as a user, it gives us better ability to sort of control the file uploads I've considered restarting a report given that we do have some that are sort of data storing now. But if you compare even what the right through cash is doing versus what our boot drives and servers are doing, it's just not apples-to-apples.

And given that the population is a lot smaller than our hard drive population as well, the fact that we have far fewer drive models getting hit with very different data workloads and in a sort of much less uniform environment. So we've held off always a possibility we'll bring it back especially as our platform expands, because I know people are hungry for real data on that front, and so am I.

So we'll see what happens, no promises. It really, to me, it always depends on if we're giving the right context around the right data points and all that good stuff.

And if you're interested a little bit about how we use our right through cash, there's a lot of blog posts on our site around shard stash that go into depth about this. So if you're just a technical kind of person, those will be some great blogs for you to take a look at.

I'm seeing another question here from James. Do we normalize failure rate to drive size? Would you -- wouldn't you expect bigger drives to fail earlier or more often? What an assumption, James, and not in a bad way, I've heard it a couple of times. I think that the first thing to take into account, so when we talk about failure, we're not actually talking about failure to write, failure to read, failure to rebuild. We're talking about a true hardware failure here.

And we go on record about what that means for us, and I think it's the Q3 2025 drive stats. But this is really want to drive fails and can't be rebuilt or can't be reset. Have we normalized drive rate to failure size. There's a couple of times we're taking slices to look at that. But one of the limitations is that we have deployed in waves over time, almost correlative to drive side.

So if you look at early on, we were about terabytes. And then we had a period where 14 to 16 terabytes were quite popular for us. And now we're into the 20-plus like all of our -- almost all of our new deployments with the exception some 16-terabyte drives or 20-plus.

So what that does is that it introduces a variable that we know has significance. So if you say older drives fail more often, which is true. And then you try to compare apples-to-apples as far as the 14s and the 16s to the 20s that it currently exists, that's not really fair. We have done a bathtub curve analysis to say when drives are failing over time, how often do they fail at year 0, year 1, year 2, year 3. And within that, we took some slices of the 14 to 16s versus the 20s to 25s. So like at year 2, did the 20 fail more often than the 14 to 16.

But at that time, we had not a ton of -- we had maybe four or five models total of 20-plus terabyte drives in our system. So we'll likely refresh that analysis again. But the short answer is that you got to sort of correct for a couple of variables before we can make the question that you're asking very visible.

And if you ever want to do that analysis yourself, I will happily remind you that the full logline data set is published quarterly so that you could take that and run it if you want to, instead of waiting for our analysis that comes out whenever we can get around to it basically.

All right. Let's see if there's any other questions here, anything we didn't answer.

I think we had another question come in by Tushar. He left a reply that he's especially interested in the right heavy workloads like the right through cash. And he was asking would it be possible for us to share how we manage them at scale. That's a lot of what the blog posts talk about that we have, but Stephanie, what other thoughts do you have?

Yes, we haven't really taken a slice at like -- so especially interested in the write-heavy workload like you're write-heavy cache. So we don't -- when things get uploaded to our system, we don't like categorize by how many operations on which drives are happening here, right?

We don't have if you were in like a hyperscaler environment. Obviously, they have different tiers about access and all these other things, which means that you can get a different lens on that potentially or a different way to attack that. And we don't have that data correlated with our drives yet. And even with great heavy work we can't even say necessarily that the right through cash is accepting more right heavy workloads, I would say.

The biggest thing it captures really is small files instead of right heavy workloads. So as far as access patterns go, I don't think that you have a clear answer there. If we wanted to talk about how that works. Really, what we have -- where we have been talking about that is in the Performance stat series, where we do an external -- it's an external benchmarking report based on work standards, such as MinIO's published, open-source benchmarking of object storage. And there, we talk about -- we separate things by file size, thread count and run throughput standard throughput testing as well as upload download averages.

And I think you can get a better perspective on our platform there. It's just not correlated to drive failure rates. The other place you might want to look is that our network stats plug. It's imperfect as far as getting down to, I think, to your specific question, but it does have a perspective on how things are moving into and out of our system.

So with the triangulation of the three reports, you get a couple of different stuff, things in there. And Noriko, I'm so sorry, I didn't see your question. Also hello and I owe you an e-mail. So I'll be getting back to you shortly. We have to plan for SMR drive deployment and HAMR drive deployment. I don't know where those sit on the road map. I don't want to speak for engineering. But I think it is fair to say especially with what we just talked about, about how product road maps are going to drive manufacturers that near future is in the cards depending on how you define near future not tomorrow within 5 years, which I don't know if that counts.

It's a good way to put it.

And as far as system changes, that's something that we don't have the scope on yet. But what I wanted to lay the groundwork for in this report and the reason we're talking about it now is that I think that we did a similar thing when we saw our drive capacity start to raise. We said, "Hey, these things are coming down the pipe like let's make sure that we have a full understanding of what the effects might be and then let's track against it and see what actually happens." And I'm very excited to do that. with emerging drive technologies to SMR HAMR as well. And I think what we'll do is the same thing we've always done, which is say, "Hey, guess what, these drives are here now. pay attention to these models if you're interested in how those are functioning."

All right. I'm not seeing any more questions in the chat. Anything you can think of David, any final thoughts on your side?

No. I think this has been a wonderful look at kind of a very concise quarter. Things were consistent. We're seeing some new technologies come into play. And I think it just makes me more excited to see where we go next with our reporting because I want to see the day that we get to 100 terabyte drives and 120 terabyte drives and we'll be around to kind of report on how those are functioning when we get there.

David, I love it. We're going to be -- I feel like we just made the whole audience say when.

Yes. Yes, exactly. Won't be tomorrow, but it will be some time.

It is a very exciting time to be seeing all the leaps forward in technology in the hard drive manufacturing space. So always fun to have a front row seat. And just to wrap it all up, you've got your resources slide here, folks, if you need anything.

By the way, drivestats@backblaze is a monitored inbox, but you can also reach out directly to me if you'd like, I'm sdoyle@backblaze. And then we've got we just took some questions, and we'll upload everything else to this webinar page after we're through here. So thanks, everybody, for joining. We always appreciate it. We'll see you on the next one.

Thank you.

Good morning, everyone, and welcome to Backblaze's Investor Day for 2026. Today's session is being webcast live, and a replay will be available on our Investor Relations website following the event. My name is Mimi Kong. I'm the Head of Investor Relations. And on behalf of the team, I want to welcome everyone for joining us today here in person and those tuning in via live webcast.

Today, we will have a short break around 10:45 today just to let everyone know, and we are scheduled to wrap up by noon and some house cleaning things -- housekeeping. Today's presentation will include forward-looking statements, and actual results could differ materially from those statements. Please see the disclaimer on the screen for more details. Full materials, including today's slides, are available on our Investor Relations website.

With that, I'd like to invite Gleb Budman, our CEO and Chairperson, to the stage.

Good morning. So my name is Gleb Budman, Co-Founder, CEO and Chairperson of Backblaze. And today is our first Investor Day in the 5 years since we've went public. And a few people actually asked me, why now? And is there something specific about today that we wanted to share at the Investor Day? And the reason that we're doing the Investor Day today is because the company and the market have shifted dramatically over the last year roughly. And what I would say is over the 20 years that we've done this, right now is the most exciting time in our journey. So we wanted to share that with you.

Over the next 2.5 hours, this is the plan. I'm going to talk about how the market has shifted and what our role in that is. Dan Spraggins, our CTO, is going to talk about the platform and why it's uniquely situated for AI. Anuj Kumar, our CRO, is going to talk about the signals that we're seeing with AI and how we're building that into a repeatable and scalable engine. Marc Suidan, our CFO, is going to talk about how we're changing this opportunity into a great business. We're also going to hear from Hume AI, a leading multimodal AI company. And we're going to hear from WEKA, a new partner of ours in the high-speed storage space.

So listen to how those all come together and stitching together this ecosystem of AI. So we're going to cover a lot of information today. But if there's one thing that I want you to walk away with, it's that not only is AI creating a ton of data and using a ton of data, but the role of data and data storage is fundamentally changing. And that's how we're going to talk through it today.

So AI is creating this seat at the table for an independent capacity tier of storage. We're going to explain why that is and what we mean by an independent capacity tier of storage. But we believe that, that seat is open, and we believe that seat is for us. And part of the way that we are ending up there is we started this company 20 years ago. We started it solving a problem that was really important at that point, which was backing up computers. We plan to do that on Amazon S3 and store the data there. Where the economics were not going to work.

So we ended up building a full stack solution, servers, a cloud storage file system and the operational expertise to manage all that because we have to store massive amounts of data and do that very efficiently. So in doing that, we then spent 10 years optimizing that full storage stack. We then launched it out to developers and enterprises as B2 cloud storage.

Over the last 10 years, we've been optimizing and scaling that platform. And today, it's ideal for the AI use cases that are being needed in the market. So that's the foundation from which we believe we get to have this seat at the table as an independent capacity tier. And the rules of storage are being rewritten with AI. And here are the 3 rules that are being rewritten. Number one, data was growing steadily for years and years, basically always, right? But there's now this step change that is happening where the data is just exploding, both in terms of the amount of data being created and the way it's being used. And one of the interesting things with it is that it's not just the largest companies anymore. Even small new companies have massive data sets and massive data needs. And that is a significant shift from the past.

Number two is there used to be 3 hyperscalers and companies defaulted to one of them. Now there are about 200 AI infrastructure companies out there. And that is completely shifting how companies are building technology. And number three is that storage used to be just a thing companies use. But today, it's a critical performance layer of whether companies can innovate in AI and whether they can get an ROI from AI. And those 3 rules play perfectly into Backblaze's strengths. As these data sets scale massively, they need to go somewhere. And they need to be efficient in where that storage is put.

Backblaze's optimization of our platform for the last 20 years makes it possible to store massive amounts of data and to do that efficiently. The explosion from 3 hyperscalers to 200 neoclouds has 2 benefits where Backblaze fits in. One is that those AI infrastructure companies themselves need cloud storage as part of their workflows, and we're ideally situated to provide that to them. And the other is that all of the AI natives and AI builders using all of these companies need an independent capacity tier to store the data to be able to use them. And the third strength is that as companies care about the performance per dollar of their storage system because it is what's critical to get both the innovation out of AI and the ROI out of AI, Backblaze's optimization of that exact thing from hard drives over 20 years is exactly what they need to be able to do that. So this all supports us taking advantage of what is a large and growing data set.

The data explosion over the next few years is hard to fully wrap our heads around, right? In 2024, IDC said there was 173 zettabytes, zettabytes created. In the next 4 years, that number is going to quadruple to 700 zettabytes created in 1 year. Now not all of that data is going to be stored, not all of that data is going to be used, but an increasing amount of it is because AI is also making it possible to get value out of the data that's getting created in ways that wasn't possible before. So not only is the amount of data getting created going exponential, but the desire to keep and use it is as well. So that supports a large and fast-growing market that we get to participate in. We continue to serve the markets we've always served.

The core markets where we've served disaster recovery, backup, archive, media workflows, application storage, all of those markets still exist and are growing. That hasn't gone away, and that is still an important part of the trajectory of our business. But AI layers on a big market on top of that and a faster-growing market on top of that. And so that requires a company that can scale with that. And that is what we've built. We now have over 5 exabytes of data storage under management. That makes us one of the largest cloud storage companies on the face of the planet outside of the hyperscalers.

And we're planning to expand that capacity by about 30% this coming year. Now all -- obviously, that takes hard drives, networking, servers, data center space, power to do all of that. And having the ability to acquire that and the relationships to do that takes a certain level of expertise. But on top of that, it's the actual managed storage aspect of it that's really hard. It's the software platform that can scale like that. It's the operational expertise and rigor that's been built around that to be able to actually deploy and manage scale at this rate and this scalability. So we've talked about kind of the scale of the market.

Now let's talk about the second thing that we mentioned, how the rules are changing, right? The dispersion of these AI infrastructure companies. And this is probably the most important slide that I'm going to show you. And so I want to spend a little bit of time on it. For the last 20 years, technology was built inside of a hyperscaler, generally one hyperscaler. Even when there were 3 hyperscalers, companies would pick one and use it and build all of their technology stack inside of that one hyperscaler.

Maybe they were inside of Microsoft, maybe they Amazon, maybe they were inside of Google, but they picked one and they built everything inside of it. Today, there are 200 AI infrastructure companies, neoclouds, sovereign clouds, inference clouds. And that's not even starting to talk about the fact that all of these clouds have different regions, including the hyperscalers. And so a common customer has their base workloads inside of one of the hyperscalers still. They're using databases, maybe they're using networking, maybe they're using one of the 200 different services or 10 of the 200 different services that the hyperscalers provide. So they're still using a hyperscaler for a bunch of stuff. But then they go and they say, now I need to do model training.

And so the hyperscalers don't have access necessarily to the latest chips or they have availability of them or they don't have the price point of them. So they go to one of the neoclouds and they say, okay, I'm going to use you for the model training piece of it. But then that one NeoCloud doesn't have everything they need, so they go to another one for some of their other model training aspects. So now they've got base workloads in the hyperscaler, data they're collecting somewhere for the model training and then using 2 neoclouds for model training. Now once they train their model, then they want to do inference somewhere where it's optimized for that use case.

So they may be using 2 or 3 different inferencing providers around the world. And beyond that, then they may also be using different regions of these. What that means is that as that AI builder, you now have 4, 5, 6, 8 places where your data needs to go. If your data is sitting inside of one of the hyperscalers, you're paying massive egress fees every single time you send your data to any of these places. And so you really need this independent capacity tier, a place to keep your data so that you can innovate with AI. It's not just about saving money. It's about the fact that you want to use the AI infrastructure that's out there in order to innovate. One of the customers that I was talking to recently said, because they switch to Backblaze, they are now enabling their AI researchers to do their model building when they feel they need to, when they have some interesting part of data and some algorithm that they want to iterate. Before this, what they did was they told their researchers, you're allowed to run one model build per quarter.

So the pace of innovation that they're able to achieve has dramatically improved by switching to Backblaze because they're able to now go and actually move their data when they need to. So this is -- so this enables Backblaze to be the backbone of the AI ecosystem by providing this independent capacity tier where data can flow to wherever it needs to go. So that's for all the AI native builders. But the other part of this chart is the actual companies on this page are AI infrastructure companies. Now the hyperscalers knew how to build storage. But now you have 200 companies that are raising as fast as they can to become large AI infrastructure companies. Most of them know land, power and shell. Some of them know compute. Almost none of them know storage. And yet all of them, if they're going to be serious players long term in the AI infrastructure space, are going to need to be able to support the workflow their customers do, which means they're going to need storage. And so almost all 200 companies in that space are going to themselves need a storage platform for which we are ideally situated to provide them.

So like I said, this is probably the most important chart to understand because this is all a massive replatforming that's happening in the technology industry and has only really started in the last 2 years. We went 20 years of the cloud evolution, and we are now at the very beginning of this AI infrastructure revolution. So now that we've talked about the market shift, let's talk about the third part of it, which is the role of storage itself. And so in an AI workflow, there are different parts. You collect a bunch of data first, then you process it to prepare it for monitoring, then you build your model, then you run inference on it, then you monitor and log in and output it. Every single part of that AI process creates and uses data.

And every single part of that AI workflow needs a capacity tier. So that makes Backblaze a critical strategic partner to the AI builders who are building AI workflows. Now Dan is going to talk more about this in his section. But the platform that we built is what supports this. And the platform that we've built, we've optimized for scale and scalability, performance and economics. And it's taken 20 years of honing to get that platform to where it is today.

That is a significant moat that's hard to replicate because even if you were able to write all that software today, you wouldn't have the 20 years of operational expertise that it took to hone that platform and all of the systems that support that. And again, Dan is going to talk more about all of that. So I've talked about the market and how the market is evolving. Now we could say, well, some of this is thesis and this is, is this going to happen? But it's not that is it going to happen. It's happening. It's happened, right? So Backblaze is already winning in this AI ecosystem.

We have signed now 5 of these AI infrastructure companies. We've signed a $335 million deal with CoreWeave to provide them storage for their capacity tier, but we have also signed 4 other of these AI infrastructure companies. We also have 5 of our top 10 customers on B2 now are leading AI companies. And beyond that, we have the smallest and fastest-growing innovators through our Flamethrower program. So this is a program designed for AI start-ups, and we have hundreds of the leading AI innovators in that program now. And we've also signed on the other side of it, on the larger side of it, a leading frontier model developer to switch to us as well. So we are winning across the AI ecosystem, and we're doing that because these companies see the strategic value we provide them by offering them an independent capacity tier that allows them to innovate faster in AI and get ROI from AI.

So to summarize, AI is not only creating an explosion of data, but it's changing the role of storage. It's creating an open space a seat at the table for this independent capacity tier, and Backblaze is well situated for that seat. So with that, I'm going to let Dan Spraggins, our CTO, share more about the platform and why it's uniquely built for this.

All right. Thank you, Gleb. I'm Dan Spraggins. I'm the CTO at Backblaze. I lead our R&D group and oversee our product road map as well. All right. So today, I'll be talking about a number of things. I'll dig into our architecture and how we're differentiating with our platform. I'll talk about the ecosystem. NVIDIA recently made some changes to the reference architecture. And so I'll speak to that and where Backblaze fits in. I'll also talk about the storage spectrum.

So how SSDs, HDDs fit in the ecosystem and where Backblaze plays. Recently, I met with the Chief Strategist of WEKA, which Gleb mentioned previously. And so I have a conversation with him. So you'll see a short video there. I'll then get into what we're doing today with AI and how we're delivering features there and then our product road map. So I wanted to give you some insight into what's coming. Some of that isn't public yet, so that will sort of be fresh and new today. Okay. So first off, we have our architecture. And so this is something that's been in process for 20 -- roughly 20 years, and we continue to extend it. And so one of the questions sometimes I get is, okay, how are you -- you're combining commodity hardware. Why is this any different? You're just setting up racks of hard drives. And that's true, but there's a proprietary software layer on top of it that's very differentiating and hard to mimic.

And so this is essentially a high-level overview of what we're doing, which is we break this into building blocks. And so these building blocks are a combination of commodity hard drives, but then we layer on different pieces and eventually, we call this a vault. And then we have a cluster. And so these clusters can get up to roughly 1.5 exabytes. And so when we sign a deal like CoreWeave, this is what we're doing behind the scenes in order to ensure that we can hit that type of scale. And outside of the hyperscalers, there's essentially no other company that can do this. And so this is very unique and again, proprietary to Backblaze.

And so maybe one other piece to this is the way we're doing this is we're not relying on the individual drives for performance. Instead, we're breaking it up and in this case, into 20 separate drives and getting the performance we need by aggregating across those drives. So it's a unique architecture. It's a very valuable architecture, and it's something that we're extending even today, and I'll speak to that later. Okay. So this is probably the most important slide that I'm going to present today. It's complicated, but it's really important.

So this -- to the left side here, this is a screenshot of an NVIDIA reference architecture, their data center reference architecture. And they recently updated this and added a category for object storage. So this is what Backblaze does. And specifically, they added a capacity tier. And this is how we've been talking about this for years. So it's a really major and important validation of the category that Backblaze is in and what we do. And so I mentioned WEKA. WEKA is in this middle category, the high-speed storage, which is great. They're a great partner, and I'll speak to that a bit. But this object storage area is something to really focus on. And we -- there's no reason we shouldn't own this category. There's no one that comes close to the scale that we can perform at. So I keep that in mind, and I'll reference it again probably a few times. Okay.

So something else I wanted to talk about is I think sometimes there's a little confusion about what are the -- what is the storage spectrum? Where does Backblaze play? Who are our competitors? And so it essentially breaks down to 2 major media types. You've got SSDs and you've got HDDs. SSDs are extremely fast, very fast. And HDDs are very affordable, but still fast. And so the major difference is, if you want to get close to the GPUs, the SSDs are what you need. But if you want scale, what you need, you're going to use HDDs. SSDs cost about 6x what an HDD costs. And so this is where you get into that performance and cost, the economics. And so you can imagine a company that has a $10 million storage budget. It would cost them $60 million to store that data on SSDs versus $10 million.

So this is how companies are thinking about us and how they scale. But these 2 solutions are complementary. They're not competing. So going into -- so Gleb had showed this diagram, and this is just to click in, which is essentially where SSDs lay and where HDDs are as well. And so you think about the data life cycle. And so you go all the way through the process and you want the SSDs as close to the GPUs as you can get them because you need that speed for input and output, you don't want to bottleneck the GPUs. They're extremely expensive. And so you use your SSDs, but you get that data off into the HDD layers as quickly as you can, and that's where you store it long term. And so this is where that 85% of the data goes.

But the way this works is it's not like traditional compute or storage where you store it and you archive it and you don't really read it again. The way this works is you actually feed the data back in. So when you get into inference and monitoring, you're generating data that then goes to the HDDs. But when you train the next model, it comes back around. And so you feed it back into the SSDs into the model and then you go through that cycle again. So this is how this works. And again, these 2 are not competing. They're complementary, which gets into WEKA. So we announced this partnership. I think it's a great partnership and something I'm quite excited about. And so in this conversation, I'm speaking with their Chief Strategy Officer, and we'll sort of break down where each company fits and how we'll work together moving forward.

Hi, Nilesh, thank you for joining us today. I've been looking forward to the conversation. And I know you and the team at WEKA are doing some really interesting work. So I appreciate the time. And I guess to start, if maybe you could tell us about what you're doing, what's your role at WEKA and what the company is doing and what problems you're solving.

Great. Thanks for having me, Dan. Really appreciate the opportunity and looking forward to working with you. So this is Nilesh Patel, I'm the Chief Strategy Officer at WEKA. I lead alliances and partnership strategy globally. I've been with WEKA for almost 5 years now. WEKA is building a software-defined storage that delivers memory-like latencies on various workloads and addressing the needs of data layer from GPU clusters to even some of the edge deployments as well.

We, as a company, we serve AI-native companies, enterprises, neoclouds, sovereign clouds and providers who are really driving the GPU infrastructure business today. And anybody who needs to move data at the speed of the compute is what we are able to do. And that is actually the problem we solve. The traditional storage cannot keep up with the AI workloads and their needs, and they cannot keep the GPUs fed efficiently and WEKA is addressing that.

Excellent. So what are some of the changes that you've seen? So I know we're seeing a lot of change on the Backblaze side with new AI workloads. I'm curious what WEKA is seeing.

So what we are seeing is from a data layer perspective, there are a couple of interesting dimensions that the AI workloads are pushing storage. There are roughly around 2 axes we see both on the performance and capacity. And both are being stretched harder than anything we have seen before. GPUs need microsecond access to keep the -- sifting through the data Otherwise, they are sitting idle.

At the same time, the large amount of data sets from the actual enterprise data to vectorized data, we hear there is a 3 to almost 10x expansion of data in the enterprises. So that the checkpointing and then all the generated data are growing in an exabyte range, which is the hard part isn't like picking one axis. It's like customer needs both simultaneously, and they need to scale fast.

Now physical storage has lead time challenges and so on and AI demand doesn't wait for hardware to show up. So some of those challenges are really stretching the need today, and it's putting a lot of pressure in the folks are building either enterprise AI factories or building the neocloud infrastructure at scale. And that's where we see the storage requirements are growing in both dimensions.

Yes. Makes a lot of sense. We're seeing the same thing, throughput, API requests per second, just overall storage needs greater than they've ever been. So you hit on a term that's especially important for us at Backblaze on capacity. And so can you speak to how you and the team at WEKA are thinking about performance versus capacity with storage and AI workloads?

Yes, that's a great question. In fact, what we see is like 2 walls are hitting customers at once. Inferencing and agentic workloads are hitting the memory wall. Long context, multi-turn, multistep reasoning, all needing to sit right next to the GPU, faster than memory itself in many cases. So WEKA solves that. We deliver better than memory latency at scale. And then there is the capacity wall. Customers need more storage now.

But hardware lead times are -- they don't move fast enough and they don't move at AI speed. And I believe Backblaze solves that. Instant capacity, no supply chain constraint, no location lock-in. Put together, WEKA extends its global namespace across that capacity, so you can get massive distributed scale without ever slowing down the workload sitting next to the GPU. So in a way, speed where it matters and scale wherever you need it is what I think together, we are able to deliver.

Yes, I couldn't agree more. And I think you're hitting on where the partnership makes a lot of sense between the 2 companies. I think they're very complementary. So maybe can you elaborate a bit on how you see our 2 companies working together with these performance, capacity, et cetera?

Yes. So I guess the partnership, in my mind, lets customers scale capacity instantly and cost efficiently without supply chain delays. I think that's a big part of it. And without all that without sacrificing the performance and the workloads, right? But it's not really about placing -- finding a place to park your data.

I think the object storage becomes almost like, in many cases, a secondary distributed data repository. And because it's part of WEKA name space, the way we integrate the solution that any capacity you add on Backblaze infrastructure could then very well be part of WEKA managed name space. Customers can now not only expand the capacity wherever they want, but they can also spin up compute and WEKA clusters that mount that secondary repository. And now they have an instant scale, not only on storage, but also on the compute and be able to, on the fly, get access to the data and so on.

So this is where combining the 2 companies together, the solution can not only address some of the capacity scale needs, but also be able to deliver data to the compute where it exists and the fact that Backblaze has distributed data access at a very cost-efficient and scale-efficient way, I think that really benefits the customers who are scaling their environment and the AI workloads.

Yes, I definitely agree. And I think it's just such complementary solutions. So I have one more question, and I know you're over strategy at WEKA. I'm curious about where you think things are going over the next several years. We've seen a lot of change. Where do you see this heading as you look over things at WEKA?

So I think over the last 2 to 3 years, we saw training and model training and so on really driving the demand for data. And I think that really pushed the burden on the performance vector, but mostly on the throughput and some of the capacity challenges. As we are seeing the inferencing use cases playing out, inferencing can happen anywhere. And also inferencing also pushes, as I mentioned earlier, memory limits of what can be stored in the cluster itself on the DRAM. So having the ultra-low latency petabyte scale storage will behave like a memory cache is becoming a critical need, and we are able to address that.

And that particular requirement is going to continue to go harder and stronger because the whole context windows are growing, the reasoning models are getting richer and richer and the complexity and the amount of context scale that everybody has to operate at is going to go high as well. And along with that, the need on the capacity and the bounds of where the data resides is also going to be challenged. So the AI workload needs to be inferred upon wherever data is. And so the data locality is not going to be something that AI workload will respect and they need to be served across the board.

So what we see from our perspective is a need for further expanding the inferencing at scale, which is driving the demand on memory latency and performance, need for data distributed everywhere, so a global name space and being able to leverage infrastructure like Backblaze, where there is a tremendous amount of capacity available to operate off of. And then finally, the data that is being stored in -- across the enterprise, old and new are going to be all current and required and processed. And so being able to access data wherever they decide is going to be another major trend that we'll see for the data layers to come up more and more.

That makes a lot of sense. Well, thank you so much for your time. You all are doing such good work at WEKA. It's a great company, really excited about the partnership. Again, I think the 2 technologies really complement each other. I have nothing but respect for all the great work you're doing. And thank you again for the time.

Thanks for the opportunity, Dan. I'm really looking forward to working with you and the team and see what we can do together.

I think that's an excellent partnership and again, quite excited about it. So Gleb touched on this earlier, which is we're working with a lot of great companies today, and we have an existing solution with some of the fastest-growing AI companies in the industry. And so probably one of the themes that you've heard is just this complexity demands around performance, scale, economics.

So we have a solution today that's working well for these companies. And one of the key points is throughput. So 1 terabit per second, this is one of the fastest in the industry. Retrieval time is about as fast as it gets for the HDD layer. We continue to have 11 nines of durability, and we scale at a level that no other company in the industry is scaling at, which gets into why did CoreWeave choose us. So they had a choice. They could have built and they seriously considered this or they could buy. They looked at the complexity and decided they wanted to buy.

And they looked at the competition, and they told us there were essentially 2 things that stuck out to them. So one is our ability to scale. So we've been doing this for 20 years, and I've mentioned scale a lot. Another is operational transparency.

And so quarterly, we release something called Drive Stats. And so they explicitly said this was one of the criteria that they looked at because it showed what we're doing. So essentially, what we have is we report on reliability statistics every quarter, and we've been doing this since 2013 for reliability stats for all of the drives. This is very unusual. Like no other cloud does this. And so we're showing everything we have, failure rates and how we're adjusting. And they explicitly said this was a major factor in their decision because it gave them insight into we knew what we were doing.

Another important piece of this is a managed storage offering. And so Marc will speak to this later. Anuj will speak to it as well. This is really important for Backblaze and a major strategic change for us and a new product offering. So there's a lot of talk around CapEx for obvious reasons. This plays to our strength, which is we don't have to have CapEx here.

So in this case, CoreWeave is handling the CapEx and our software is running inside of their data centers. So our software is portable. We're not making custom changes for CoreWeave, and we're using the same software for the long tail all the way up to the exabyte level customers. And so there's a lot of work going into this, but I'm quite excited, and I think it's really going to be a good business for us.

We'll sort of get then to what are we seeing now and what's the future. So one of the things that we're seeing is AI agents. So everyone is talking about AI agents. If you drive through San Francisco, every billboard you see is around AI agents, right? We're seeing the same thing. So there's this large uptick in the traffic we're getting. The most recent data I saw is that nonhuman traffic is actually growing faster than any other traffic we have on our site. And so we're adjusting for AI agents. And we have a whole team and their sole job is to focus on this. So they've been focusing on things like discoverability because in the past, things like marketing, brand, et cetera, still super important. That's not something that AI agents think about. It needs to be discoverable.

So it needs to be machine readable docs. We need to have integrations with the best open source tools. We need to integrate with AI platforms. We need to have what's called an MCP server, which is essentially a protocol for how these agents talk. So we're doing all of this work, and we're seeing a large uptick in the traffic we're getting. I think we're making some good decisions here. And we have an engineering team whose focus is just on this to ensure that we're getting this right.

And so last up, I wanted to speak briefly to what we're working on in addition to the things that I've already spoken about. And so this isn't public yet, and so now it is. These are internal targets that we have.

So again, performance, scale, economics. And this is what we're seeing from our customers. And so we're committing to internally and now externally, 2x increase on throughput, 5x increase on API requests and a 20% increase on drive capacity. So we're actively working on this, high confidence for new deployments moving forward, this will be -- these are the metrics that we're aiming for in addition to a lot of other interesting work we're doing around managed storage, AI agents, enterprise features, a number of things. But this is hard. And so maybe that's the thing that I would leave you with.

This isn't something that other companies can come in and just copy. We're extending our existing architecture. We're building on top of something we've been doing for 20 years with tens of thousands of optimizations, reducing bottlenecks. And so this is -- I get the -- what's the moat? This is the moat. And so we're working on really hard things and extending our lead in an area that we're already leading in.

So next up, I have my friend, our CRO, Anuj, and he'll cover where we're at with revenue.

Thanks, Dan. Good morning, everybody. Tough act to follow. The CEO set the stage for the market opportunity. Product guys telling you we have awesome things on the truck already and even better things to come. But I've been here about 4 months. What do I know? But I'll share that I saw this opportunity as a really market inflection point. And what I truly see is Backblaze is one of the very, very few companies that have really the scale, the performance and the capability to take advantage of this inflection that we have.

Before I go into it, I think my main section is about, as Gleb said, from signal to revenue. And I think it's important I define what the signal is. And for a sales guy at heart, I will say, I want to break it down into 4 simple things. What is the signal that we are going to talk about because what we need to do is where is the demand really coming from? Why is the signal really working and why you should believe that the signal is really, really strong. From there, I will take you to who are these people that are coming for the signal and what we can do to convert them finally into how this engine comes together to convert all this demand into something that we can repeat and scale.

So you're the room full of hopefully, math guys. I was a math major as I was growing up, and I think you'll appreciate this. On one of my long-time mentors said, Anuj, life is really simple. It's just a math equation, right? And if you break it down. So let's just start with the numbers since I'm sure you'll appreciate this more than anything else. What we look at in a go-to-market engine in a signal is what is it doing over a period of time? And what I'm sharing with you here is our own data over the past 12 months.

This is data from our AI customers. And when you see the conversion rates, we are winning at least 10 points higher than conversion rates of a non-AI customer. The deals are landing. This is really important for us because we want to make sure we are going after the most addressable opportunity. The deals are landing 7x. This is not an error in the slide. It's 7x the deal size of the rest of the customer base. But what makes this signal even stronger is once they land within a short period of time, they actually compound.

This is the beautiful formula that I would love to have with every piece of the business. This compounding happens within very, very few months. This is the total data in aggregate over the 12 months. Why don't I give you an example behind the scenes. And I'm truly fortunate, Olya is going to join us from Hume as well. She's going to share her story. This is another story because this is not just a unit of one. This is from hundreds and thousands of prospects that we see in our base. And what you see here is what we are coming back to the table of what we shared with you in the first quarter.

If you remember, we had said in our Q1 earnings that we had a customer that came in with AI and that deal converted in about 11 days. And what you see this is this actual customer came to us from another customer reference. And we were able to do from a trial to the first land within 11 days. Great. It's almost 1 million, life is good. But you fast forward less than 12 weeks later, the same customer compounded by doubling their capacity. And now we're in the third quarter, this customer has already grown to about 2.5, 2.7 discounting, almost 3x. But if you aggregate the total, we're looking at about a 4x compounding from this place that we started from the provider that we had.

Now why is this customer putting so much data on it? I think Dan alluded to, Gleb alluded to, AI really has a lot of ingestion, a lot of data collection, and this is a continuous stream as they are running inference models and want to have a capacity tier to take care of it as long as they can have this available with high throughput for the base that we have.

Now we've talked about Backblaze's history in terms of why it really, really works. This is the second big piece of what I'd like to share with you. The why that I would say has worked for Backblaze has always resonated. You know from the beginning of time, are we have really what we call disruptive economics. Our value proposition has been tried and tested. It's true against the hyperscalers. And we also offer a very, very predictive model. You get no egress, but -- and that gives you a way to make sure that you can put your data where it is, but also you want to make sure that you have something affordable that you can run the models from.

But in the AI space, it's slightly different. What has given us is it's not something new, but it's really given us an opportunity to expand the use case and the value that we have for our customer base. In AI, there's 3 specific things these customers are looking for. They're not just looking for low cost, but what they're looking for is sustained high throughput. That's what we call performance. At the end of the day, you're trying to do price economics, but you want to make sure that the performance doesn't suffer when your models are calling it for inference. And so high performance, high throughput is really, really important.

The number two thing is rapid scale. I think Dan talked about it in the WEKA conversation. It's not just the tiering of the data, but it's also instant availability of that capacity because you know GPUs are very, very expensive when they're sitting idle. I think there's a whole term token economics that's come around. At the end of the day, we don't want to keep these GPUs idle. Yes, the flash has its role to play, but hard drive and our managed storage service has a role to play to make sure that the capacity is constantly available at high throughput for the GPUs to be continuously working.

At the same time, what you do want is you don't want your data to be locked into any one particular location. I think Dan shared this -- the one slide that Dan you shared in terms of if you take away one thing, which is all these neoclouds that really didn't exist until maybe 5, 6, 10 years ago, and you have so many choices. At the end of the day, these AI builders really want to keep the data where they choose to, and they shouldn't be locked in, in any one place. So making sure that you have architectural freedom to keep the data where you have, make it available at scale and make sure that it's available at high throughput for this AI customer base is really, really important. And so what has helped us do is now we see an expanded use case in terms of our value proposition. That's probably the biggest why.

And now I'll probably get into the key pieces. When I think about a go-to-market team and a go-to-market engine, what do we do wake up in the morning and try to do from what it is? Because you clearly know there is a value. You clearly know that there is a strong signal. But then how do you make sure that you start converting that signal into something that you can build an engine from? The first step is really to make sure you're segmenting it right. A lot of sales organizations will show you triangles where there's strategic, enterprise, commercial at the bottom.

The way we look at the world is really, really simple. What's -- if you see on the left, what really is, is the core workloads, backup, disaster recovery, security, they have always existed. They will always exist, and there is a constant need for backup and data recovery and disaster recovery. And that stays true. We continue to grow in that market. We continue to grow above market rates, fantastic.

The other 2 adjacent is what is giving us the addressability of the use cases I just talked about, the high throughput, the performance the scale, the instant capability, the ability to make sure that we can move the data, keep the data where it is. And it gives us 2 clear segments in the market. One is what we call AI builders. These are GenAI media companies. One of them, Olya, is going to be here from Hume. We talked about a few others in the GenAI media space or the physical AI space that are really training insane in large amounts of data to kind of get to where they are.

And I would say the other big part is the neocloud, where we have about 200 of these neoclouds that didn't really exist about 4, 5 years ago. But what they need is instant capacity, but also the ability to offer managed storage as a service. That is the service that we can now deliver for them, either coming to our data center, which you can today, but also as part of a managed service, which is how CoreWeave leverages us. So both of these are now giving us new addressable markets to go after in this AI space.

And that's a clear segmentation of the who we really address when we go to market. From the who, the next logical question you will say is, well, how do you convert this who that you're trying to get to, to the how that we want to make sure that all the activities we are doing from the signals that we're seeing to convert into a repeatable, I would say, process and as well as a deliberate process of connecting with the buyers in all of those 3 segments. So to think about it, I'll keep it really, really simple.

There's 3 main parts. We have to make sure that we are educating the community. We want to make sure that we're educating the community. We are engaging them where they're going. And we're also making sure that they experience the product as they choose to. In education, you've seen us, we've talked about it. It's for the last, I would say, 13, 14 years, we've been producing quarterly Drive Stats. This is incredible information from a cloud storage company that doesn't no other cloud provider actually provides.

And this helps us build a community of engagement where they actually see us being really, really transparent, not only in our pricing model, but also in our development model and scale model in terms of what we do. The engagement then really gets to we want to meet where the buyer is. And today's buyer is not sitting necessarily in a physical, let's say, events. Of course, we do those, but they're really engaging more in social and specifically new social channels like Reddit, Stack Overflow, and we want to meet them where they are.

And that's the Flamethrower program working with hundreds and hundreds of start-ups on a daily basis to make sure that we can engage these start-ups and get them to meet them where they are and help them understand who we are and what we stand for. And finally, I'm the last guy that somebody like Dan wants to talk to when I call and say, Dan, I'm Anuj from Backblaze. I'd love to talk to you about this cloud storage thing. CTOs, CPOs, CIOs, they rarely talk to the sales guy in the first call.

And so we want to make sure that they can get their hands dirty in the platform on their own with no sales assist. And that's really important because we want to make sure that we are available in all these channels, whether it's Hugging Face, whether it's open source SDK tools, we want to meet them wherever they go and however they want to engage, making sure that they can touch and feel the product and the service before we ever call out and say, "Hey, we see that there's a signal you might want to engage with us."

This is a concerted continuous process. So think about this education, the enablement, the experience, it's something that we do on a constant basis. So we want to make sure the signal is not random, but it is something that is repeatable, something that is measurable, something that is instrumented and we can see the results and we can keep tweaking in terms of what it is. We're not trying to convert humans to robots, but we're trying to make sure that we can actually see what's happening and make sure that we are available in all of these channels.

Finally, as part of this, how do you make sure that all this wonderful signal we have and all the engagements that we are doing converts to something that, as we often say in sales, there is something on the truck, and I got to go and then convert it. Like the big thing here is I've got to convert from all this demand that I see into something that is a repeatable revenue engine. And this is the part where I would just say, in sales life, it's just 3, 2 and 1. I showed the 3 segments that we have. The demand comes through any of the 3 segments, and we have 2 motions to address this demand.

The 2 motions are self-service and then, call it, sales assisted or direct sales. The self-service, you choose as you come, you can just pay as you go, you get on to the platform and you can scale as much as you want. We've offered the service, and we continue to offer it as a viable option.

On the direct sales, specifically in the AI builder and neocloud, this is where we see the difference where the customer does engage a little bit on the PayGo, but very quickly, they want to have really high-value workloads that they want to do.

And that's what we want to make sure that when they test, they get the experience to convert in days and weeks and not months and years. And so the direct sales team is a dedicated coverage model to make sure that when we see the demand signal for high value, we attach the sales team. Both of these motions are powered by an ecosystem that's in 2 pillars, sell with and sell-through. I'll come to it in a quick second. But the goal is that these 2 motions are working in tandem to make sure that we are addressing the demand that's coming in.

To make sure that this actually converts from these 2 motions, what we have is a very strict, I would call it, sales process, and this is layers. It's -- the layers is just an acronym.

It sounds simple, but the goal is we want to make sure when you come in, the trial converts to something of an opportunity, that's landing. We want to make sure that what you committed for and what you want to consume as part of the data is what the adoption team takes care of. And each of these are dedicated teams to make sure that we can then take you through the journey of expansion. Once you've got a good workload on. I'm sure there are additional use cases, all about cross-sell and upsell and then making sure that the revenue line, the retention line stays strong.

And the glue that kind of pulls this all together is our support team because at the end of the day, you came in for a managed service. You didn't come in just for software that you were running on your own. So storage support glue really pulls it all together to make sure that we have this process, which is quite rigid, something that we measure, something we want to make sure that we are seeing the conversion rates and making sure that we can address them. But this process kind of keeps all those segments, the motions true in terms of where we want to go.

I'm sure the next logical question you'll have from me is great. You've got -- Dan said he's going to grow like 20% on the capacity, 5x the throughput, et cetera. Similarly, as we are going to grow in the go-to-market team, I would say it doesn't really come from just infinite headcount. I think that's a bad day. I started like Marc, I just want to go and hire 50 more people because I need to double the people that we have. So how do you think we're going to get there?

And this is, I would say, the final piece of that engine, which is to make sure that we are really building a; deliberate process and a deliberate motion with a partner ecosystem. It's really, really simple. There's 2 pillars. There's a sell-with. You saw an example with WEKA. The goal is we are doing validated reference designs, validated architecture. So it's actually tested and integrated at source. The goal is we don't want the customer to be spending the time to try to integrate 2 parties. And we know flash and SSDs -- sorry, SSDs and HDDs work in tandem.

And the cloud capacity is something that adds on to help the capacity in flash to make sure that you can run your models. And so we want to make sure that this is tested at source, so customers can have a validated design to go with. WEKA is just a prime example of that. And once we have those designs, what we do is we convert them into a downstream ecosystem of partnerships. These are resellers, distributors, cloud marketplaces, basically allowing the customer to have a choice in terms of whichever partner they choose, they can engage with us on. The goal is these 2 work together. The more validated designs we have, the more options we have for the customers and partners to engage.

And this is the deliberate motion that we are putting together. This is a relatively new muscle, I would say, for Backblaze. However, we've got a good part of this engine flowing already. And this to us is really the big motion for us to scale beyond, let's say, infinite headcount that we would have. To wrap it up, as I said, I'll try to talk to you about these signals that we believe are really, really strong.

And I hope I've given you some proof that the signals have some depth in them in terms of we really measure and instrument every part in terms of what's coming in and how fast it's converting. The second piece of that is really in terms of making sure that these signals don't sit in random, and we have a process to convert the signal to a demand that we can then finally convert with the process that we can then scale with the ecosystem.

So I just want to thank you for your time, and welcome Mimi on stage.

So we're going to take about a 10-minute break right now, and then we're going to welcome Hume to the stage to a fireside chat. So please take your time and get some refreshments.

[Break]

I'm going to bring on stage our guests for a fireside chat, and I'm going to let Anuj take it away, and you'll get to learn more about Hume.

Okay. Well, welcome back. I hope you guys had a good break. Don't worry, this is not a sales call. I'm not going to ask you, Olya, which I'm still learning how to pronounce her name properly. But really, really, first, thank you for coming in here. Great to have you. Thanks for obviously being a customer and a partner.

I'm sure we'll talk a little bit about Hume, but I just wanted to formally introduce Olya. I met her for the first time yesterday in person. We talked a few times before, but she's got an incredible journey in basically being and come up to the Chief Product Officer at Hume, and she is responsible for all of the strategy, all of the product. I think your portfolio has everything from strategy, products, cybersecurity, the current the future, it's a lot.

So thank you so much for spending a little bit of time and being with us here today and talk with our analysts. So why don't we begin a little bit about, tell me -- I show this audience. I talked about GenAI media, but that's really, really, really high level. You can talk a little bit about you, Hume, a little bit of the journey, I think it would be great, great.

Yes, absolutely. And thank you for having me. We were practicing names back and forth. So it's a -- we're getting there. So I have a background in machine learning, initially for health care and then transition through highly regulated sectors and then made my way more to machine learning product and then now more voice AI.

So Hume as a company is a really interesting organization because we have -- half of our team is AI research. And so a lot of what we do is we train models to improve other models. And we also have data solutions, which means we need to process a lot of data. And every time that we process data, we create new data.

So Backblaze has been quite helpful in our needs there. Additionally, what we've been doing in our goal to improve voice AI because we initially started out with 10 years in semantics doing research, really focused on assets, which is prosody. It is understanding of how you communicate. And so that would be emotion and how you can derive that from acoustic signals. And so we built models around that. We built the infrastructure around that, everything to process audio as well as the systems to evaluate and improve other voice AI models.

Yes. And Yes, when we were talking, I think one of the things that absolutely fascinated me with a couple of like key things that you mentioned. One, just the layers of complexity that a simple thing like voice has. I think when you think about it, it's not just the amount of data collecting transcripts and trying to put something in just to understand some of the tonality or how we are actually feeling when we are talking. We can hear it on the phone, but you guys are actually trying to pull all of this together. Can you tell us a little bit about that complexity and what it takes to actually put that model together?

So on the surface, voice AI seems like one dimension, right? Same as text. It's predominantly one dimension. But when you actually peel back the layers, it is so multidimensional. So think about us speaking right here. I have certain contacts that I came in to this conversation with as do you. I respond to your inflection changes. I account for the background noise, of which we have none in this room. And there are all these other components within that, right? There's the warmth of the conversation. And all of these things are actually very challenging for AI to understand.

So there are 2 core things that we solve for, which is, number one, does AI truly understand? And number two, does the user feel understood. And so right now, what we've been seeing is voice AI is at an inflection point. where previously you had predominantly text interactions, right? We type on our computers. And now the new modality that we're seeing for people to interact is voice. And so you have to solve for all of that complexity and you need to ground it in real-world use cases, which is quite challenging, too.

True. And I mean, that itself is one dimension. But the other dimension you also mentioned was just normal models like why wouldn't just OpenAI, Anthropic and all of these guys, they also have a model. But you said something really, really interesting that caught me yesterday, which is those companies are only interested in the research, I think, is what you said. They're really just doing it for raw research. But as you guys are building it in layers and layers of modality that really gets down to?

I think there's 2 parts to that. I think OpenAI and Anthropic and all of the large labs are fantastic. They're also large labs are customers, so we love them. But what's interesting there is they have a bit of a different focus, right? They want to push towards AGI. They want to make sure that they have the smartest, most capable models. Our goal is to improve all of the voice models, the whole ecosystem. And so you need models that can evaluate those models. And to do that, you need to be really, really good. I think there is the second half of it, which is as enterprises are coming up and sovereign environments are coming up, I think what we're noticing is that a lot of people recognize that they have a lot of data, and they want to build their own models as well.

And so we actually get a lot of outreach to us about, hey, like we want to do XYZ as this enterprise to build an AI model that optimizes for speech. And we want to make sure that it's performing well, and we want to make sure that it has all the emotional understanding within it, which is really interesting and perhaps relevant for you guys as well because as folks are now -- as we're thinking about this model landscape, we're not only thinking about OpenAI or Anthropic, we're thinking about sovereign clouds, enterprises who are now trying to leverage all the data that they have to build their own AI models.

It's like models, platforms, like they're doing both things at the same time, and it's constant growth. I think I called it compounding in my section. is basically just defines it in a way that's really, really, contextual, fantastic. No, I appreciate that.

Well, let's bring it a little bit into how you see what you're building and like some of the reasons you came to us, any of the things we talked about, which were the things that really resonated for you in helping you make the decision to Backblaze? And how do you see this relationship in the context of everything that you're building?

Yes. So I think Gleb was actually walking through the journey, and I was like, oh, that's our journey. That's right. So essentially, we go through and we collect a lot of data. Data is obviously very important for AI. It's a hungry, hungry Hippo. And so we collect that data, we process it and enrich it. And so we actually had a lot of disparate environments where we would store our data. So not only did we want to consolidate it, but it was really important for us to not get taxed for using our own data, which ties to Backblaze's egress, I guess, lack of cost.

So every time that we go through and process our data and we have petabytes of data and every time that we go and process audio data, what we create is transcript data, we have 600-plus emotional tags, expression tags that we put on top of that. We segment it out. And so every time we process data, we create new data. And we have to do that not just on a continuous basis, but we have client deliveries as well.

And so our clients come in and they want to process a lot of data, too. So when we have to do that, we need to be able to transfer all of that very quickly. And what was really important to us as well was not just our ability to transfer it or our ability to store it all in one place. But the third thing was how do you make sure that we don't have to have our own continuous compute that's up and running. And so we have serverless compute. And so what we do is we have this data store. And then when we need it, we spin up the GPU clusters and then we transfer it over to process the data accordingly, which gets used for our clients or for our models as well.

So the high throughput we talk about, the instant availability, the capacity, those are all good things for you to have that really available.

Yes. And honestly, the Backblaze team has been quite helpful. When we were processing a lot of data in one go, they really made sure that we had really great through there. So it was quite helpful.

Excellent. Excellent. Well, that's great. So how do you see this partnership? I also -- I don't want to leak the stuff that you just shared with me, but if you're comfortable, Great. I didn't realize that there was some commonality here, too, but how do you see this extended partnership and this relationship kind of continue to foster?

Yes. So I was just telling Anuj, that we actually also use WEKA. And so typically, what we do is we have our core data store with Backblaze and then when we need to use it as part of our models and processing, we have it in our GPU cluster, and we have the WEKA storage there as well. And so I think, as you can imagine, as the tailwinds of folks switching over to voice as a modality pick up, and they already are, we obviously anticipate processing a lot more data. And so it's really helpful to us that both Backblaze and WEKA work seamlessly together, and we are able to continue kind of growing and supporting this hungry, hungry hippo.

Well, Anna, we got to make sure that the integration really work. She's on our -- she's our Head of Partnerships and channels. So we just want to make sure we get all that validated design really cooking, so you don't have to do all the hard work. Great. Well, excellent. I think you gave us a lot of really fantastic valuable insights. Really appreciate you being here. Anything else you'd like to share before we wrap?

No. I mean it's been fantastic. Thank you so much to the Backblaze team, and we are excited for continued collaboration together.

Awesome. Well, thank you very much.

Thank you.

All right. Hello, everybody. Good morning, and thank you for being here. I'm Marc Suidan, the CFO, Chief Financial Officer. Gleb told us you got to show up looking your best. So I figured, listen, either got to do some biohacking, get more hair on the head or pay $40 to get the AI to do it for me. I think it went a bit too far, so I'll have to do a bit of refinement there in that picture.

Okay. Let's get rolling. I think it's a great day. We really wanted to get out adding new faces to the discussion. Generally, Gleb and I and Mimi have been in extensive discussions with a lot of you. So we really wanted to get a lot more of the extended team, so you could all meet them. But from a financial standpoint, what this all translates to is the 2 things we've always focused on, right? More growth, and operating leverage, right?

So I'm going to walk through how we've delivered on what we said we're going to deliver and how we're going to do more of it. So when you step back and you look at everything we promised 2 years ago, we set out a few goalposts, and we've delivered on all those goalposts. The first one we said we're going to do is strengthen the balance sheet. And back in Q4 of '24, we did a secondary offering that was oversubscribed, and we did a restructuring driven by zero-based budgeting exercise. And in that exercise, we reduced the OpEx, and we reallocated some of those savings to invest to accelerate growth.

And then we said we're going to reaccelerate B2 growth. So in Q1 of 2025, we started to reaccelerate B2 revenue growth. then we said we're going to get B2 revenue growth to over 30%, which we did in Q2 of '26, it was 34%. And we said for the rest of this year and next year, it will be 40% or higher. So we're delivering on all the things we said we're going to deliver, and we said we're going to do it in a profitable way. So under the capital lease model where our CapEx is financed by capital leases, we turned free cash flow positive exactly when we said we would. And we're well in that motion there.

So we always said, let's use B2 revenue growth and our free cash flow margin as -- to judge our Rule of 40 scoring. And in Q2, that was 34 plus 8, so 42. So we hit 42 in Q2 of '26, up from 18 a year before that. So delivering on the Rule of 40 score that we promised we'd deliver on. What's driving that? The accelerating revenue growth is translating to a lot of operating leverage. You get a healthy gross margin of 63%, which is really good for an Infrastructure as a Service company, combined with really disciplined OpEx management has translated to tremendous EBITDA margin improvement. So the operating leverage is well in motion and in action.

Underlying that is the mix shift. A few years ago, computer backup made up the majority of the business. Right now, B2 forms 62% of the business, and we will probably be around 75% somewhere around middle of next year. So the mix shift is well in action and the underlying fundamentals of the B2 platform is really attractive. ARR growing 39% year-over-year in Q2. really strong net revenue retention at 113%. And the gross customer retention at 89%, you're talking customers that stay with us with an average of 9 years. So phenomenal fundamental metrics.

The business for B2 has historically been consumptive, very pay-as-you-go. But as we're accelerating this growth, we felt it was prudent to start getting into longer-term contracts. So our revenue performance obligations, which are committed contracts are gone up more than 5x year-over-year. And in the marketplace, just given the supply constraints, a lot of customers actually prefer to be on the committed contracts because they know that we'll give them and guarantee them the capacity.

So that's really helping in giving us a lot more visibility in the growth and where to put our investment to fuel that growth. Unit economics. Over the lifetime value of a customer, we deliver 60% to 70%. It's currently 70%, right? But just to be conservative with the changes in hardware prices because it takes time to adjust the pricing models, we're saying it could be 60%. But that's the value we get out of all incremental dollars from our customers. They generally stay with us for 9 years. Every cohort since B2 launched in 2016 has grown their data consistently year-over-year.

You have to obviously build the CapEx upfront. It takes less than 2 years to pay back the CapEx. And then you've got the direct cost. Those are roughly -- I mean, they're step function variable costs, but it's effectively data center rent and power, telecom, data center technician, customer support, sales incentive comp. So that's pretty much our variable cost, right? That carries throughout, but the CapEx is upfront and then you recover heavily over the following 9 years. So you look at our EBITDA performance, we'll continue to improve our EBITDA margin and also operating margin, GAAP gross margin, all of those will continue to improve as operating leverage continues to feed the bottom line.

Looking at the debt. So we raised our convertible debt a few weeks ago. We raised $201 million, also oversubscribed, 0% coupon rate. So that strengthens our balance sheet for the next 5 years and helps us to invest. We're putting it all to CapEx. We have to put the CapEx as we have the committed contracts and we have the revenue coming in. So we got to invest in the CapEx. The difference is instead of paying 12.9% interest expense on leases, we'd be paying 0%.

So if you think about the cost avoidance over 5 years, that basically pays off half the principal of the debt right there. So we felt that, that was a prudent move to have a stronger balance sheet for the next 5 years to build up this cash position and deploy it on CapEx, and then we would spring load and come out of that with a much stronger free cash flow margin. So the metrics to watch for us over the next 2 years will continue to be EBITDA and operating margin. That's where you're going to see the operating leverage continue to improve. For free cash flows, we did turn positive in Q2, but now we're going to deploy this capital.

So operating cash flows will improve. But when you pay for the CapEx and cash, you got to deduct it on those operating cash flows. So for the next 4 to 5 quarters, the free cash flows will turn negative, and then we're spring loading the free cash flow margin to come out much stronger coming after that. So when you think about all the P&L key levers you got to look at and what the outlook is for each one of those, I'll start off with revenue. We said revenue will grow 40% or more for the rest of this year and 2027.

By the way, that 40% assumes the same existing guidance philosophy, which is no large deals, and we're defining large deals is greater than $0.5 million. We did 3 in Q1. We did 4 in Q2. So once again, this number assumes none of these large deals and assumes no customers are committing over their minimum. So we're just putting in the minimum committed contracts there. So it's a pretty prudent 40% is what I'm saying for the next 6 quarters.

Gross margin, as we're deploying the CapEx, and of course, it is more expensive, there's going to be a bit of a lag between when you deploy the CapEx or depreciating and when the revenue comes right afterwards. That lag could reduce gross margin by 300 to 600 basis points through the middle of next year, and then it starts recovering from there. And then as you get into 2028, Dan and Gleb spoke about the managed storage. The managed storage resembles a lot more a SaaS kind of gross margin. So that will start to be accretive into 2028 on the gross margin as that comes to feed in.

On the OpEx side, we'll continue to manage OpEx in a very disciplined way. We will make targeted investments in R&D and sales and marketing, but OpEx as a percent of revenue on the current trajectory we're on, should continue to be where it is or improve as a percent of revenue. So a lot of operating leverage there. And as it relates to adjusted free cash flows, like I said, for the next 4 to 5 quarters, you will see the adjusted free cash flow turn negative as we deploy the CapEx and then we spring load it so that it meaningfully improves coming after that in a very healthy way under the current construct if we move back to the capital lease model.

So key takeaways. We have better growth visibility supported by committed contracts and proven long-term earnings power, and we're going to keep powering that up. So those are the key takeaways for the financial section.

So with that, Mimi, we're going to turn it over to the Q&A session.

Yes. So I will invite executives up on to the stage.

Ittai Kidron from Oppenheimer. I appreciate the presentation today and great to see the acceleration in the business. Maybe a couple for me. First, on the CoreWeave transaction, clearly a landmark deal. Can you talk about the milestones that you have to go through in order to ramp the managed service in fiscal '28? I mean, clearly, it has huge potential from a margin standpoint, as you mentioned, Marc. But clearly, there's also a lot of work that needs to be done behind the scenes to get that going. So we greatly appreciate if you could provide some color on that.

And then on the go-to-market side, thanks for clarifying the 3 different kind of cohorts. It was very helpful to understand where the efforts are focused. Maybe a little bit more of a philosophical question here now where in those 3 segments, do you feel you're best aligned and where you have more work to do, number one. And number two, if you had an incremental dollar to put since Marc now has a lot of money in his pocket, if you had another dollar to put in the go-to-market organization, where is the next dollar going to?

Thanks, Ittai. So let me touch on the managed storage and then actually, I'll have Dan also maybe expand on it a little bit. So the managed storage has 2 minimums, right? And the first minimum is at the end of next year. So the -- we expect it to ramp basically starting in 2028. The very simple concept on it is CoreWeave comes to us and says, we would like you to deploy x amount of storage in this location, and then we hand them a bill of materials and say, please buy all of the following equipment per our specs and hand it to us. Our people manage the equipment, the racking stacking and managing it inside of that facility, and we deploy and manage all of our software. So that's just general concept.

And then maybe, Dan, if you wanted to expand on it.

Sure. So we are focusing a fair amount of energy on this managed storage concept. And so that includes things like deployments, releases, ensuring no downtime, telemetry, dashboards, things of that nature. So we're adding a number of features there. So yes, I think that's high level what we're doing now.

Yes. And maybe just one thing, just terminology-wise. So we're calling this managed storage, not managed service because it's not so much -- there are people obviously involved in racking, stacking their boxes, and there's also people involved in the -- sitting on our side of it, eyes on glass, right, managing the storage, but it really is a managed storage offering. And I think it's one of the key unique differences, right? So there are software platforms out there, including some open source and closed source platforms where you can use software.

But part of the reason why CoreWeave chose us for this and some of the other conversations that we're having with other neoclouds and other AI infrastructure companies is that it's one thing to just either buy a piece of software or take an open source piece of software. It's a whole different thing to actually manage and run storage at scale. And so that's why we're calling it a managed storage offering.

I am miced up. So I think you had 2 questions, just so I make sure I play the questions back so I got it. I think you said in the 3 segments that I showed, like which of them probably is growing fastest or like where we -- where do you best fit, okay? And then the second is if I had infinite money, where would I spend it, right?

So let's just go with the first part. I think the reason I shared those 3 is because the good news here is there's no one segment that's carrying the weight of the other 2 in any way. They just have slightly different velocity, slightly different motions, obviously, different buyers and different needs that we service from the same core platform.

Where -- like I also shared, the core workload, the backup, disaster recovery, it's been our most stable business. It's been there for a long time. It will continue to be a service. So I think that is probably the one that we've had the fit for the longest time. And I think something Dan mentioned that I want to have recall here is when we acquired these customers of AI builders or AI infrastructure, we actually didn't build anything new. So it's important because what we realize is the value that the platform already has is really good throughput and a really good performance per dollar. I'm inversing the equation. It's not price performance, it's performance per dollar because even when Olya was sharing, it's all about consistent throughput that they would like to have so that they can feed the GPUs when they need it.

And that combination is literally, as I would say, we have these things on the truck to be able to deliver right now. Would we like to have more performance? Sure. Would we like to have more economics at stage? Sure. But there's nothing impeding us from addressing this market right now. Does that help? I mean that's kind of the first one. And we obviously have dedicated teams to make sure we take this in and we convert the demand into the things that we'd like to do. If we had, let's say, more investments, and this is probably a tricky one.

But I would say, at the end of the day, what I want to do is the investment I shared with you was a little bit more on the ecosystem side. is we cannot grow infinitely by just adding more headcount, having more people call, making available on more social. I need the whole ecosystem to talk about the full value proposition. If customers have a journey that they go through, we're not the only game in town. I'm pretty cognizant of that. And so as they go through the journey, I want to make sure that from every angle, they actually hear of us in that combined value proposition. And that's really, really key. So what we stand for stays true, but they hear end-to-end.

Jason Ader with William Blair. Gleb, I guess, for you, question is, is there any drawback to having your storage tier separate from where the compute is for a customer in terms of network latency, I don't know, things like that. And then that sort of begs the second part of my question, which is, do you envision CoreWeave? How you answer the first question is going to matter to the second question, which is do you think CoreWeave long term will be more on managed storage structure? Or do you think there'll be sort of still reasons to -- assuming they can get as much capacity as they need on their own, do you envision that would shift more towards managed storage?

Yes. Good questions on both of those. So because this is a capacity tier, latency is not as much of a question, right? So you heard him talking about the microseconds needed to feed the GPUs, right? So you don't want that being far away from the GPUs because you're talking about microseconds, right? As a capacity tier, the time in general to get from -- to get data off of the hard drive and get it out versus the time it takes to get from there to the storage, the latency doesn't play that big of an impact for the most part, right?

You don't necessarily want your capacity tier sitting on the West Coast in the U.S. and having GPUs in Japan, right? But in general, the whole idea of the ecosystem is that you're going to want to use multiple providers for your model building for your inferencing for all these different pieces. And so just by the nature of that, the data has to flow between. It almost doesn't matter where you put the data, it's got to go from one place to the other. So the first part of it is that, in general, as a capacity tier, having your data somewhere as long as it's, call it, like in the same country or in a nearby country, right, it's like that kind of thing, then it's fine.

As far as the CoreWeave part of it, so someone asked a version of the question and let me tweak it a tiny bit, and then I'll come back to it. Someone asked, why did they -- for the deal that we did, it's 2/3, 3/4-ish on our infrastructure and about 1/3-ish quarter on their infrastructure. And they said, why did they pick strategically that split? And my answer was they didn't, right? It's not that they said the right answer for us is 2/3, 1/3 or 3/4, 1/4. It was more that they looked at how much availability they thought they needed in what time frame we had the ability to deploy that on our infrastructure, and they looked at how much they're going to want in their infrastructure and when and that was that part of it for that.

So it's not about the mix shift. It's more about the timing of ramp. CoreWeave has dozens of data centers. I don't think that the likely scenario is that we're going to deploy a capacity tier scale in every single one of their dozens of data centers, right? I think that what they're going to do, just like what probably most of the 200 AI infrastructure companies are going to do at the capacity tier is pick a handful of major regions to deploy large-scale capacity storage to service the broader set of regions. And then they may have 20 data centers across the U.S. or across Europe, and they'll pick 1 or 2 locations where they're going to stand up a capacity tier that will service all of those locations.

Erik Suppiger with B. Riley. On the AI infrastructure versus the...

Builders?

Builders, yes. It seems like as you've described it, the -- an AI company would be inclined to buy storage or capacity from you rather than going through the neocloud. But you've described the neocloud market as $14 billion. So what is the bigger opportunity? Is it the AI infrastructure? Or is it the AI builders? And what kind of penetration do you expect to get within the neoclouds?

Yes, it's a good question, Erik. So there are about 200 of these AI infrastructure companies. I don't expect that, that number is going to go to tens of thousands right? There are thousands of AI builders, and I expect that number to continue to grow, right? So -- and I'll tell you that some of the AI builders, we see doing both. They use us directly and they use data on the neocloud to us. So there's different reasons for that, right? They may use the AI through the neocloud directly because they're part of the orchestration of their workflow inside of that neocloud, and that's easier through their dashboard for that.

But then they also use us directly because that neocloud is not the only one that they're using. They're also using hyperscaler 2 inferencing clouds, et cetera. And so they want data that's sitting independent of the neocloud so that they have easy access to the broad swath, but they also want to use the data inside of that neocloud because they can orchestrate what happens in that neocloud through a single pane of glass. So we see companies doing both. Hard to fully say which of the 2 sides is going to be a bigger one.

The AI infrastructure one is bigger chunks, right, because they're a smaller set of companies that bite off in larger chunks. The AI builders are ones that we see having a kind of more steady state, larger dispersion path to growth. So we talked about how we had signed -- we announced one in Q1. We announced obviously CoreWeave in Q2. We now just signed another one, right? So we're up to 5. We have talked about 2 before.

So my general belief is that for almost every AI infrastructure company, if they're going to be successful long term, they're going to need storage, right?

It's hard to just say you're going to be a pure-play compute provider and not play in any part of the workflow that your customers use, right? If you think about it, the customers, like you heard from Olya, right, the customers have data. It has to be -- if they don't use data, you don't have AI. So they're going to have data, they're going to keep that data somewhere. If they're not keeping it with the neocloud because they don't provide them that opportunity, they're going to keep it with a hyperscaler or with us. And if they're keeping it with the hyperscaler, they're literally using their direct competitor.

So I believe that the 200 of them are going to almost all either offer it or grow our business. long term, right? I think we're best positioned to be that provider for them. Would I love to say that we're going to get 100% of them? Sure. But I think that it will take some time as they go up the maturity curve of what they do. What we've seen from -- there was this interesting thing, which was counterintuitive. When we first went down this path, we said the ones that don't offer any kind of storage today are the ones that are going to be our best targets.

The ones that do offer some kind of storage are not because they already offer something they're not going to want to ship. What we found was it was actually inverse. -- the ones that offered storage, customers were coming to them saying, you can solve more of my actual need, but then they started feeling the pain of not having a solution that fully solved what they actually needed to solve. So they were more open to working with us.

The ones that didn't offer storage yet we're still just raising as fast as they could to try to figure out how do I get out of power? How do I stand up the GPUs? How do I make it so that I can orchestrate this for my customers? How do I get the workflow just the basics set up, right? So they're not yet at the maturity level of trying to service the broader issue that customers need. And so I think as they move up the maturity curve, we will have more and more opportunity to penetrate more and more of the 200. That's a hard thing to know at this point. But I think we're best positioned to do that.

Eric Martinuzzi with Lake Street Capital Markets. I wanted to follow up on your comment there. It's really a question around kind of back upstream on CoreWeave and maybe it's more a question for Dan. But you talked about the scale as one of the key reasons that CoreWeave went with Backblaze, but you also talked about the operational transparency. And I was just curious to know, to me, they have a pretty substantial scale. What was it about the Backblaze scale that was different? And then maybe it's more -- was it brand as opposed to transparency? I'd like to get a layer deeper on that.

Sure. So CoreWeave, I think, on the maturity curve is maybe the furthest along from what I've seen. I don't think brand was a big factor to them. They're very -- by the numbers, they're very technical. I'm quite impressed with working with them. They're a great partner. The scale that they're working at is extremely large. So in the -- we've spent 20 years to build 5 exabytes going on 6. Those are the types of numbers they're talking about. And so they were very serious about needing that type of scale. And even as they started to run their own pilots, they were finding that it's just very hard to run at the scale.

So you have open source solutions like CEF, they can't get anywhere near what is needed. On the transparency piece, that was also really important because companies say they can do this. but it's not a given. And with Backblaze showing what we've done with Drive Stats, et cetera, with our reputation with 20 years with very large customers, it was a real factor to them.

Was it then -- was the final decision that they were -- they had already made the decision to outsource and you were the leading candidate? Or was it they were still kind of on the fence of build versus buy?

They had already made the decision. And so they were looking for who to work with. And so I think another piece of this is, and Gleb sort of alluded to this, companies may initially think they're going to build it, but then they start to run into the pain of running at scale. And so this is where I think this operational experience has been very valuable. So I think I had mentioned some tens of thousands of optimizations. That's real. And each week, we're running into all sorts of complexities, bottlenecks, et cetera. And they saw that way ahead and thought, okay, this is going to be a distraction. So their business is around compute, and they wanted to work with a partner that could do this at the scale they need.

If I can add one other dimension to it. I think it's also -- you want to factor in that you're looking at data and storage, and you want it to be in proximity to the GPU compute that they have. If I'm CoreWeave, I want to put the data off my customers close to me, not just to monetize the whole thing, but to give them a whole experience. Just like Gleb said, they had storage, so they were ahead in as a service, but they wanted to give the full spectrum of experience from inferencing to model and everything else. Otherwise, the same customer is going to put the data in a hyperscaler and then part of the data with CoreWeave. And that's split.

Time to market to offer that full service is something they have to consider too. All of those play into the decision-making of making sure the customers that they're getting are asking for more. They want to be a full service provider. I mean I think at GTC last year, CoreWeave was Jensen's words, they're the fourth hyperscaler. So if you think of a hyperscaler, that's full service stack. It's not just GPU as a service or bare metal as a service. And so as they mature up, like we need to have the full service stack. And that service stack, if I'm CoreWeave I want to put it with me, I have the data split with 3 other hyperscalers.

The transparency that Dan is talking about also is what they told us directly what they said, the whole name of our game for CoreWeave is to get to scale and get to scale fast. And you're the only ones we could trust to be able to do that. And so it's one of those where you can imagine you go down a path, you start building something and then you're a year in and all of a sudden, their customers are demanding to get to 0.5 exabyte, 1 exabyte, 2 exabytes, whatever the scale is, and then you have a system that doesn't work, that's a pretty challenging place to end up, right?

So they needed to make a bet on someone they could trust. And a lot of the team behind the CoreWeave on their storage side was a lot of the team that built Amazon's S3 service. So they really -- in terms of the maturity curve piece of it, they know what this takes. And so they were able to say they could evaluate what choices were, pick Backblaze.

I'm [ Ben Bronston ]. I'm a private investor. I think one for Dan and one for Anuj. Dan, you mentioned NVIDIA changing its reference architecture. And I think we also had recently at a blog post on the benefits of HDDs. And it seems pretty significant them doing that. I would love to hear your thoughts on what prompted them to change the reference architecture and talk more about that.

And then, Anuj, it also seems like they'll probably open up some opportunities in the cell with movement on GTM. And I know NVIDIA has done some things with the neoclouds, either investing in them or having rev share. I'm wondering if there are opportunities like that on the horizon.

Yes. I think it's very significant and a very good thing for Backblaze. So I would say that why did they do this? It's similar to what CoreWeave is running into right now, which is, they sell a lot of compute. They have SSDs. It's very, very expensive. And so the customers don't want to store their data on something that's that expensive, but then they need to come back to it. So there was cost pressure just from customers on getting to a more affordable tiering option. And so I think that's a piece of the puzzle.

And then that leads to maybe a different trend, which is, okay, if you can't give me a more affordable price, then maybe I go with one of the hyperscalers or maybe I go somewhere else, which then puts pressure on NVIDIA because they're creating chips that are competitors. And so that's part of sort of the larger strategy. So yes, I think that's the main piece. And so NVIDIA is getting ahead of this. It's a category that we've been working in for some time. No company comes close to the scale. So yes.

And I think you're spot on. I mean it widens the spectrum in terms of the sell-with and the integrations -- because if you think about it, there was some confusion in the market that we might be competing with the WEKA. There's no -- as you can see, even like all your shared, like we are also a customer of WEKA, which is a great thing. And I think it complements our story. It complements the reference architecture that we want to build. And frankly, now that NVIDIA has put that capacity tier recognizable in the space, you can imagine that we have some good discussions ongoing. But at the end of the day, we want to make sure that we are working with these partners actively integrating as many as we can because the customers need it.

And it's a good point. So the blog post he's referencing just very, very recently, just in the last month, I think it was -- they published one where they said, hey, everybody thinks about SSDs as the answer. But by the way, HDDs are actually the right answer for some of the use cases, not just SSDs. And it was -- it was interesting that NVIDIA chose to publish this. And I think it speaks to last year, when we were at GTC, we were meeting with some of the new clouds. And one of them said, look, fundamentally, we understand that we're going to need something in this area. But in order for us to be successful, we have to follow the NVIDIA reference architecture.

So when NVIDIA says, this is the right way to do it, this is how we're going to build it. And because that allows for us to know that it's going to work, that allows for us to work well with NVIDIA, that allows us to work well with the ecosystem, that allows us to go to our customers and say, we support the NVIDIA reference architecture. So we are going to build the way NVIDIA suggests for the reference architecture. And so they said, look, we know there's a large data set out there, and we need to figure something out. But until NVIDIA kind of puts their stamp on how to use it, we have to be cautious, right?

And so the fact that NVIDIA is, I think, just fundamentally looking at and going, flash memory SSDs are incredibly expensive. The prices are through the roof, the volumes are sold out, et cetera. And for a lot of the use cases, that's not the best place to do it, right? NVIDIA is trying to foresee where the next bottleneck for the neocloud overall and all these AI infrastructure use cases overall and how to expand the reference architecture to enable that. So it is, I think, a pretty major step. And I think it's exciting that they kind of they're leaning into that.

A couple of questions. Could you segment the AWS marketplace in terms of how much is Glacier, how many other sort of SKUs they have and how you deal and how much of the market, $40-plus billion that AWS alone must be selling in it is in these other SKUs that you may not directly address? And then could you talk about your selling proposition has always been ease of use, ease of price, certainly transparency of pricing and radically lower pricing. Can you talk about where WEKA fits into that? And if they offer a very different sales proposition to the customer, which is performance at a big premium. Because I thought they were like 4x or at least that's the portfolio arm did told me they were 4x?

So let me try and touch on a piece of that, and I'll let others add to they'd like. So part of our story to customers in the past has been, like you said, it's ease, it's transparency, it's economics, right? And Anuj talked about like none of those things have changed. AWS has a bunch of different tiers of storage and the customers that we heard from said, "Oh my god, it's so complicated, right? I remember one customer said, if the only thing you ever do Backblaze is allow me to not have to navigate to those 14 things, I will be forever grateful, right? So it's knowing how and which one to use when and feeling like, okay, wait, okay. So I'm using this one, but my data just -- now I need to access it. So now I'm stuck here. We had a customer that switched to us. They were in the media space. they had a massive archive of media footage that they said, okay, I'm going to go ahead and use Glacier because this is all done. This is edited. I'm not going to need it anymore. They stuck in Glacier. And then what happened one day was SaaS came out. There was a media asset management system that was SaaS-based. They wanted to switch this different media asset management system.

Nothing changed about their data, but because they switched to a different media asset management system, that system needed to touch all the data to index it, which meant they had to pull all of that data out of Glacier. They said it was -- took so long and was so expensive. They're like, we are never going through that pain and suffering again. And AI is obviously like putting that on steroids, right, because you're often needing to retouch the data and relook at it. So one of our comments was that we are basically providing the S3 level of performance and availability and everything at closer to Glacier type pricing. And so it's just like why would you have to deal with all the complexity of that before.

So Amazon does not break out -- they don't even break out what technically what they make off of S3. There are some third-party estimates around it, but they certainly don't break out what they make for each of the different tiers. But I think we're able to support a large amount of the use cases that they add complexity around between the different tiers. We don't have an offering for like the coldest, coldest thing that you just store on tape, but we also -- it feels increasingly less relevant today because it's dangerous to put something assuming you're never going to access it again, right? So that's that piece of it.

I think the WEKA piece of it, and I would say, obviously, you ask WEKA more about their own piece of it. But I think they would talk about like the manageability, right? So it's not just about like the -- it's not about so much about tiers of storage, they're creating the name space for their customers and allowing for the high-speed storage. And you saw on the boxes on the reference that Dan showed, there's a lot of stuff going on there. But there was a big box kind of over here that's at high-speed storage. And then there was the capacity tier, which is this other box. And this other box with capacity tier had HDDs and SSDs even in it. WEKA was in this other box of high-speed storage. It's a little bit of 2 different parts of the data flow.

[indiscernible]

Yes. So WEKA has different interfaces. Obviously, part of it is S3 compatible, which is how we work together. Hopefully, that answers your question.

Matt Calitri sitting in for Mike Cikos over at Needham. On the 2Q earnings call, you guys spoke about a handful of other conversations you're having around managed storage offerings. I'm curious like how do you attribute the momentum there to like CoreWeave sort of being this great example of how strong your capabilities are versus this just being a new offering in general because you've never done the managed storage before. And sort of going off of that, like do you have a sense of these potential customers that you're talking to, are they definitely going to use managed storage, whether it's on Backblaze or a different vendor? Or is it more so they're definitely going to use Backblaze and they're trying to figure out if they want to use core or managed?

So maybe I'll touch on a little bit of the history. And maybe, Anuj, you can talk a little bit about some of the conversations we're having. So on the history side of this, so companies have come to us and asked us to do managed storage for a number of years in the past, not quite as far back as when we launched B2 10 years ago. But certainly over the last 5 years, it's come up as a question. In the past, our answer was no. And the reason our answer was no in the past is that it requires a certain amount of scale to make it interesting to do it, right? Because you're talking about the operational complexity of we're going to be in your data center.

We need to put people in your data center. We need to manage the rights and restrictions and SLAs and all that and the handoffs and stuff like that. So it requires a certain amount of scale to make it make sense. At our scale, we have the scale for it to make sense. But if someone comes to us and says, "Hey, we'd like you to be in our data center, manage it -- manage storage for 1 petabyte, the answer is like that's just not enough scale to make it interesting.

So CoreWeave was the first time when someone came to us and said, look, we have enough scale for this to be interesting, right? They're paying us about $100 million as part of the contract just for our piece of not including any of the CapEx just for the managed storage, right? So that was kind of enough scale to make sense. And so it's basically a launch pad for us to have an offering that now makes it where at the right amount of scale, we can go and do this for others.

AI is also creating the scale sizes of data sets that it becomes interesting at more places, right, whereas it used to be most companies didn't have the amount of data for it to make sense. Now that's become a more common thing. So that's kind of getting us up to today. And then, Anuj, maybe you can talk about some of the conversations that we've had.

Sure. I mean just to add, there's probably 3 things in it, right? One is scale in AI is just dramatically different than the scale in any other workload. And for us, I mean, the scale, but over a compressed period of time. So when they need so much capacity in a compressed period of time, in the past, you could build anything in infrastructure, no different decision like on-prem in the cloud, you could build it yourself. But if you need a compressed period of time and you had to offer an SLA to your own internal AI teams, how do you get that as fast as you can at the scale that you want. Those 2 -- just the scale and the complexity and the time are the 2 things.

And then running it as a service because now you are starting to think about, I got an inferencing piece, I got a modern training piece, I got a data ingestion. There's so much going on. Would you have the benefit of somebody that actually deals with it daily and understand what the service is about versus what the software is about -- so it's just those 3 things. If you just combine them, they're magically in the right spot where maybe 2 years ago, it was a petabyte or 2.

Now 50, 100 petabytes is a normal discussion, right? They're actually getting to that scale very quickly, and they can see that on the horizon. And they're also trying to get that compressed in time and try to offer it as a service. If you combine those 3, we are a managed storage service. We do that at scale. And so we become a viable option for them to consider.

I think the other question was, have they decided that they're going managed storage and then they're choosing who to use? Or are they choosing backwards and then deciding whether to do managed storage or using our infrastructure. So I haven't been involved in all of the conversations. The ones that I've been involved in, they basically have said, look, we understand that we need storage as an option. We're trying to decide whether we want to just leverage you for the infrastructure you provided to us or whether we want to do it in our data centers, but help us work through which choice makes sense. So at least the ones that I've been involved in is more. It's mostly that, yes.

Jeff Van Rhee from Craig-Hallum Capital Group on your Jeff's team. I wanted to kind of separate my question into 2 parts. Just first for neoclouds, why do you believe that can be an enduring solution looking like 3 to 5 years down the line? And then for non-neoclouds but AI customers that are coming to you directly, are you seeing any sort of shift in terms of the types of customers, the applications? It seems like anecdotally, earlier on, maybe a year ago, you were talking about really, really storage hungry applications like AI video generation. I'm wondering with Flamethrower and these kind of new things that you've put out, if there's maybe some more breadth in terms of the types of applications.

Yes. So 2 questions. So first of all, why would the AI infrastructure and neocloud side be enduring? I think the basics on that are, if you believe the various pieces, so do you believe AI is going to be a thing? I think that seems obvious that, yes, AI is going to be a thing. Do you believe that there are going to be AI infrastructure companies outside of the hyperscalers that are going to continue to operate and operate at scale? I think that it's clear that that's the case. Whether there's going to be 200 or 150 or 300, people can argue.

But clearly, there are going to be a significant scale of these other AI infrastructure companies around model building, training, inferencing, et cetera. And then the question is, do you believe that they are going to need storage as part of their workflow? And they have told us, yes. You can see CoreWeave and others leaning into it. And I think that it's clear that the future is that. You can see that the fact that NVIDIA has added to their reference architecture is NVIDIA believes that, that is also going to be a thing, right?

So if that's the case, then the only last question is, in the future, are they going to build or buy? And like you heard from Dan, if one choice they have is there are open source solutions out there. So they could try and cobble together and figure out how to scale open source solutions, which have never scaled to this size before. That's probably not the best path for them if they want to succeed, right? Or they can go with someone who's proven it and done it at scale. Their choices for that are Backblaze and then the list gets very short, right? So that's the path of why we believe that this is a long-term durable opportunity for us.

On the AI builders, your question was the types of -- yes, the expansion. So the -- we talk about video, but more broadly, it's multimodal, right? So AI generates and uses large volumes of data. The bigger the data set, the bigger it is, right? So the bigger the data format. So text takes up a fair amount of data, but audio, video and images are just much larger data sets, right? So you heard from Hume, you have Olya talking about how just the audio portion is dramatically larger than text, right? Video is dramatically larger than audio. So as we're looking at it, we see data providers.

So there are companies whose entire business is providing data to these model builders, providing data to you, providing data to Hume and others that were up on the slide. So those are companies whose business it is to collect data. Some of them are scraping the Internet for it. Some of them are generating it themselves. Some of it is synthetic data. Some of it you've seen maybe like some of these robotics movies where they have cameras on people's heads to watch what they're doing. All of this is about collecting data for model building.

And that entire set of data providers is a great set of customers for us because they're collecting large volumes of data. And then they need to get that data. Once they collected it, it's not useful if they have it. They need to get it to the companies that are using it. Because we have high throughput and free egress, it allows them to move the data to their customers. And you heard that even from Olya for their part, right? So that's a whole category. The physical AI companies are another large and significant group for us because the way the physical AI companies are teaching a lot of their systems, their robots, et cetera, is through video, right?

You've heard about world models and other things. It's all about like understanding how the world operates. And again, that's just a much larger data set than text, right? So this whole category of multimodal companies, whether it's GenAI media, physical, the data providers for it, like that -- all those categories are great targets for us because they're just large data sets.

Matt Smith with Halter Ferguson Financial. I'm wondering a little bit if you had any negative feedback with the price increase earlier this year? And if not, kind of how you're thinking about balancing either passing on cost increases in the future or maybe even taking some more margin?

Yes. I mean I can take the first part, Gleb, and you can answer the second part. So we introduced a price increase May 1. And we expected churn, right? Both from customer count as well as how much data per customer. None of that materialized. I mean, in fact, in Q2, we actually had more sign-ups and the average data per new sign-up was higher. So I'd say the demand in the market is so strong yes, we haven't seen any negative impact there, right, in terms of future plans to do.

Yes. So just in general, we want to provide a great service to the customers, right? So we want them to feel like they're getting great value from us. That's always been the case, and that continues to be the case. Now we've -- not only did we raise prices in May, but we also have higher-priced offerings, right? Our B2 overdrive is a higher-priced offering. It's just that we're getting -- we're providing a lot more value to the customers, and we're charging for that.

So I certainly don't rule out the possibility that we'll raise prices in the future. It's not the core of how we intend to grow, right? The way we intend to grow is a combination of getting more customers having more of their data with us and providing more value for them. That's the core. But depending on where the cost of infrastructure goes, depending on how the value we provide, it's certainly something we can consider again.

One more there.

Russ Kanga, Citizens. Regarding the upcoming platform enhancements, improving your throughput, API requests and drive economics, I appreciate these are challenging architectural optimizations to make. Should we just interpret these developments as continued efforts to improve the platform? Or Anuj, do you feel that once these products enhancements are on your truck, these will be an area where you'll feel a further unlock from customers in terms of being able to differentiate further from the competition?

Yes. I think from a technical strategy piece, this is a decision we made to double down on the platform. And we have signals that there are customers that are very, very interested in this. So it's -- we're seeing -- we need higher throughput, we need higher API requests, and we need more storage. And so we are making a strategic decision to double down on our platform. So that's what we're doing there. Maybe Anuj can speak to the.

Sure. I mean -- and it's -- I would say it's a symbiotic relationship. I mean, Dan builds and I sell is the simplest way you can think about it. But in the way we look at the business, I want to make sure that we have enough demand, not just the signal on the customers that we can then offer whatever high throughput and when we have it is, in a way, Dan is planning the quarters in terms of when it's coming out, and we start working with customers in advance in terms of this is the future throughput that you will get. These are the kind of API things that you could do.

So we have some good close relationships with, obviously, some key customers and we use that to make sure that it gives us the right demand for the services that we're building. So it's just normal working day. At the end of the day, we want to make sure what we have on the truck converts, and we are trying to build the demand for the functionality that we want to continue to build on and scale with. Yes. I mean that's exciting. We have new things to go and scale with.

Thank you, everyone. Thank you for joining. I hope it was a really informative day. We have lunch provided, so please help yourselves. And we also have some gifts to thank everyone for coming to Backblaze's Investor Day 2026. Thank you.

Thanks, everybody.

Thank you. Hello, everyone. Thank you for joining us and welcome to the Back Bay's second quarter 2026 financial earnings call. After today's prepared remarks, we will host a question and answer session. If you would like to ask a question, please press star one to your hand. To withdraw your question, press star one again. I will now hand the conference over to Mimi Kong, Director of Investor Relations. Mimi, please go ahead.

Thank you, good afternoon, and welcome to Backblaze's second quarter 2026 earnings call. On the call with me today are Gleb Budman, co-founder, CEO, and chairperson of the board, and Mark Sweetan, chief financial officer. Today, Backblaze will discuss the financial results that were distributed earlier. statements about our future financial results, the impact of our sales and marketing initiatives, cost savings initiatives, results from new features, the impact of price changes, supply volatility and pricing, our ability to compete effectively and manage our growth, and our strategy to acquire new customers, retain and expand our business, existing customers. These statements are subject to risks and uncertainties that could cause actual results to differ materially, including those described in our risk factors that are included in our most recent quarterly report on Form 10-Q and our other financial filings. All forward-looking statements that we make on this call are based on assumptions and beliefs as of today. undertake no obligation to update them except as required by law. Our discussion today will include non-GAAP financial measures. These non-GAAP measures should be considered in addition to and not as a substitute for our GAAP results.

Reconciliation of GAAP to non-GAAP results may be found in our earnings release, which was furnished with our Form 8K filed today with the SEC. You can also find a slide presentation related to our comments in the webcast, which will also be posted to our investor relations page after the call. also see our press release or a presentation for definitions of additional metrics such as NRR, gross customer retention rate, and adjusted free cash flows. And finally, we will be hosting an investor day on Wednesday, September 9th in New York City. Please reach out to IR at Backblaze.com to RSVP for the in-person event. A live webcast will also be accessible from the Backblaze Industrial Relations website.

Thank you for joining us. And I will now like to turn the call over to Gleb. GLEB BORCHARDT- Thank you, Mimi. And thank you, everyone, for joining us today. We had a fantastic second quarter. Revenue came in at $42.7 million, $2.5 million above the high end of our guidance range. Adjusted EBITDA margin was 30%, 700 basis points above the high end of our guidance range and B2 growth accelerated to 34% year over year. These results reflect broad momentum across the business. We also signed the largest contract in BackWiz's history, a $335 million multi-year agreement with CoreWeave, which I'll come back to in a moment.

We continued to move upmarket and ended the quarter with 235 customers contributing more than $50,000 each in ARR, up 57% year-over-year. ARR from this cohort grew 67% year over year. We signed numerous AI companies, including a leading frontier model developer, introduced the ability to run our cloud storage in customer-owned data centers to support regional and sovereign workloads. and expanded our AI startup outreach and agentic developer tooling. This quarter's results are proof that our AI strategy is working. The decision to lean into AI is translating directly into the financial performance you just heard and into the momentum I'll walk you through now. To understand our strategy, consider this. Every training data set, checkpoint, inference output, and Gen AI asset has to be stored and used.

Customers consistently tell us they have three needs to support that. Number one, the ability to scale with fast growing data. Number two, architectural freedom to use their cloud of choice. And number three, storage performance that optimizes their AI workloads. But all of that needs to be affordable so that AI scales efficiently. The combination of those three requirements are why they choose Backblaze. Let's talk first about how that plays out with Neo-Clouds and Inferencing Clouds.

Many of these initially focused on GPUs as a service, but quickly recognized that their customers also needed storage. Some of them began by building flash-based storage tiers to support high-performance workloads. However, as data scaled and flash prices spiked, it became clear that flash storage should only be used where it's necessary. As NeoCloud scale, they need a more complete storage stack and a hard drive based capacity tier for everything else. With five exabytes of storage and almost two decades of technical optimization, we believe Backblaze has built the most efficient hard drive-based capacity storage platform available. And NeoClouds, wanting to get performant scale efficiently, are choosing Backblaze. back place. We estimate that this NeoCloud demand for capacity tier storage represents a $14 billion market opportunity by 2031.

Our strategic agreement with Coreweave, a more than five year multi exabyte deal and the largest contract in Backblaze's history is the clearest proof this quarter that Backblaze can be the capacity tier for AI infrastructure. CoreWeave is recognized as the essential cloud for AI and runs many of the most demanding AI workloads in the world. As its platform expands, it is adding a variety of storage tiers that can scale rapidly and perform based storage infrastructure. CoreWeave is now the fourth major AI cloud infrastructure company to contract with Backblaze. and we're in conversations with many of the leading other ones. As these AI infrastructure companies scale to broaden support of their workloads, we become increasingly relevant to them. Part of the CoreWeave Agreement also introduces a new way for us to deliver that value. For Backblaze, this managed storage approach represents a capital light service model that brings our technology and operating expertise directly into a customer's infrastructure.

This approach expands our opportunity to service customers in their regional data centers and sovereign cloud needs. Now, beyond AI infrastructure companies and to the broader AI market, We continue to see strong traction with AI native companies like HeyGen, Hume AI, Mirage, and many more. And this quarter, we continue to add to that list. AI companies are choosing back ways for our ability to scale fast. Last quarter, we highlighted a training data provider that signed a nearly $1 million deal in just 11 days. Less than a quarter later, as its business grew faster than expected, it added another $1 million commitment. AI customers also choose Backvoiz for architectural freedom.

It used to be that companies were okay just building inside one cloud, but AI technology is evolving rapidly. AI native builders are choosing from an increasingly fragmented set of cloud infrastructure. That requires the ability to use and move data to whichever hyperscaler, neocloud, inferencing cloud, or other AI infrastructure they need. Backblaze enables that through a combination of free egress, high performance throughput, and optimized networking between us and these clouds. The need for architectural freedom resulted in a six-figure deal with a customer building conversational AI models. They needed a cloud agnostic home for their training data. Expensive egress fees from their prior provider kept their data captive and limited what they could achieve.

With Backblaze, they were then able to freely move their data to whatever cloud they wanted without the headache of calculating and worrying if egress fees will break them. And AI companies choose Backblaze for performance. We signed our largest B2 Overdrive deal to date, seven-figure ARR deal with a frontier AI model developer. At the scale of data they work with, performance is critical, and B2 delivers high throughput at efficient price points. Together, scale, architectural freedom, and performance, all at an affordable price, are why AI companies are choosing Backblaze. And while AI is making this need especially urgent, it extends to nearly every company using storage at scale. In addition to make companies start small and we're building for them, our technology advantage is one part of how we are strengthening our position.

We're also working to make Backblaze the natural platform for developers and their AI agents to build on. This quarter, we shipped our SDK for Type script, the emerging language of choice for AI coding agents, released GenBlaze, a generative media SDK, and built out a new set of tools after seeing developers turn to B2 to store their AI agent data. We also launched our multimodal focus generative media hackathon, which drove awareness of B2 as the storage layer for GenAI applications. In closing, we exceeded our financial expectations announced the largest agreement in our history, and delivered new wins and expansions across the AI market, including our largest overdrive deal to date. we also introduced a new managed storage approach that brings our software and operating expertise into customer owned infrastructure. But the bigger point is this, when AI scales, data grows, and that is good for us. When companies look to control AI costs, they come to us for that too. growth or discipline, either way, we are well positioned to benefit and continue building a durable growth business. With that, I'll turn it over to Mark.

Thanks, Gleb, and good afternoon, everyone. was a pivotal quarter reflecting both strong operating results and the significance of our strategic relationship with Corweave. Revenue was $42.7 million up 18% year-over-year and representing our strongest growth in six quarters. Adjusted EBITDA nearly doubled the over-year to $13 million, with the margin expanding by 1,200 basis points to 30%. Both results exceeded the high end of our guidance. These results demonstrate the benefits of our strategy. Based on our Q2 performance and outlook for the remainder of the year, we are raising full year guidance again. Turning to revenue, B2 accelerated to 34% year over year, our strongest growth rate in seven quarters.

B2's strong performance was broad-based with almost every route to market and GTM lever overperforming, with strong performance in direct sales bookings, continued self-service momentum, and increased usage from larger customers. We also signed larger and longer duration commitments, increasing RPO. The price increase implemented on May 1 contributed about 8 percentage points in B2 growth. While churn from the price increase was anticipated, it did not materialize. Excluding that impact, underlying growth continues to show strength. Sequential B2AR increased by $20 million, of which the price increase drove nine of the $20 million. B2AR reached $113 million, an increase of 39% year over year.

E2 net revenue retention was 113% compared to 114% last year. We also continue to make progress upmarket. Customers contributing more than $50,000 in ARR increased 57% the over year to 235, and we closed four deals valued at over $500,000 this quarter, including three AI-related wins. That progress is also showing up in the size and duration of customer commitments. We added approximately $320 million in RPO during the quarter, including $313 million from CoreWeave, net of the $22 million in warrant values. This increase provides greater visibility into contracted demand as more customers enter into multi-year agreements with committed minimum spend. We retain additional upside as usage above those minimums is built on a consumptive basis.

Computer backup revenue declined 2% year-over-year, better than expected, as turn initiatives, and targeted customer acquisitions help stabilize performance. The business continues to generate recurring revenue and cash flow, and our focus remains on retention, operating efficiency, and margin improvement over time. Moving on to total company gross margin, it was 63% in Q2, benefiting from the B2 price increase and continued operating efficiency, partially offset by higher hardware infrastructure costs. Operating expenses increased 6% year-over-year, well below revenue growth. As a percentage of revenue, operating expenses improved by 800 basis points to 73%, demonstrating continued operating leverage. We expect to make targeted investments in R&D while continuing to reduce operating expenses as a percentage of revenue. That operating leverage also translated into an adjusted free cash flow margin. of 8% for the quarter, even as we continue to invest in infrastructure for 2027's committed demand.

We ended the quarter with $50 million in cash and marketable securities, up from $45 million in the prior quarter. We also increased our available and unused capital lease lines to over $150 million, While our new lease lines are generally at lower interest rates, we will continue to look for ways to optimize our cost of capital. Before turning to guidance, I want to note that we filed an S3 today to register the warrants issued to CorWheels in connection with our agreement. The warrants reflect the strategic and mutually beneficial nature of the relationship and align both companies wrong the long-term success of the agreement. Moving on to guidance. For Q3, we expect revenue to be in the range of $44.4 million to $44.8 million. We expect adjusted EBITDA margin to be in the range of 27 to 29%. For the full year, we are raising revenue guidance to a range of $172 million to $174 million, up more than $10 million from our prior range of $161.5 million to $163.5 million.

At the midpoint, this revised guidance represents approximately 19% in overall year-over-year growth, up from the previous 11%. Consistent with our guidance philosophy, the raised outlook reflects Q2 actuals and greater visibility from contracted demand. guidance excludes variable usage above contracted minimums and potential deals greater than $500,000. We are also raising our full year adjusted EBITDA margin outlook to 27% to 29% from 23% to 25%. Looking ahead to 2027, based on the B2 underlying business fundamentals, Coreweave's minimum RAB, and the previously announced $15 million plus TCV deal, we expect B2 revenue to grow over 40% year over year. This is early directional commentary, not formal guidance. We will provide our full 2027 outlook in February. We wanted to give investors visibility into the contracted demand already supporting growth beyond this year.

Turning to capital investments, we are accelerating CapEx in the second half of 2026 and into 2027 to build the required capacity to support signed customer commitments. Using capital leases, we expect to be adjusted free cash flow neutral for the full year despite the increase in capex. Our capex track record demonstrates how we make use of these assets for well over six years and deliver healthy gross margins. Our capex break-even is less than 24 months. Moreover, the max The managed storage portion of the CoreWeave Agreement is delivered on customer-owned hardware, so we have no CAPEX requirements for the managed service. B2 revenue growth accelerated, larger customers continued to expand, and contracted demand increased our visibility. On a rule of 40 basis, we are proud of achieving a combined B2 revenue growth and adjusted free cash flow margin of approximately 42 from 18 a year ago.

We entered 2026 with a clear objective, demonstrate that our business can grow efficiently and generate profitable operating leverage. Q2 showed clear progress against that objective.

With that, operator, please open it up for questions. We will now begin the question and answer session. Please limit yourself to one question and one follow-up. If you would like to ask a question, please press star 1 to raise your hand. To withdraw your question, press star 1 again. We ask that you pick up your handset when asking a question to allow for optimum sound quality. If you are muted locally, please remember to unmute your device.

Please stand by while we compile the Q&A roster. Your first question comes from the line of Mike Sikos with NeedHEM. Mike, your line is now open. Please go ahead.

Great. Thank you to the team for the question here. Congratulations on the quarter. Mark, I was hoping to start with you on this outlook here. Help us think about to what degree the improved calendar 26 outlook is tied to the ramp for the minimum commitments from that contract or anything on the calendar 27 to support that 40% plus outlook we're putting out there for B2Cloud?.

Yes, sure, Mike. Can you hear me fine? Yes, I can. Thank you. Okay. Yes. So as it relates to the second half of 26, the $10.5 million raise, is benefiting from a broad base of things. The business performing better, the Q2B of 2.7 million, the price increase, and the core we've ramped. None of them have a dominant role in that. It's a healthy mix of all that. And just a reminder, the way we guide is remaining consistent, which is we're sticking to, you know, contracted minimum spend by customers, nothing over that. And, you know, given the large deals, even though we just won four greater than half a million, we're still not projecting more per quarter, just so we stay consistent in that approach.

And that continues through 2027 as well. And with that, I would say D2 should be well on track to grow at 40% or higher during the rest of this year in 2027. On your other part of the question, which is the ramp, it does ramp over the core, which basically does ramp over the coming year. Does ramp over the coming year and hits hits their minimum about mid 2027. And that's what we've got baked into these numbers.

I see. Thank you. Thank you for that. And maybe a question for Gleb. obviously you guys have made some pretty significant changes to the go-to-market in the last year. And I know we're starting to see that specifically in the, on the NRR front, or I remember last quarter, you guys were talking about, pipeline from existing customers. You also have the new CRO in place now for quarter to quarter or so. Status update on where we are with that go-to-market transformation that I guess, what are the findings for today versus 90 days ago? Thank you again.

Yes, thanks, Mike. So as you mentioned, we've been undergoing the GTM transformation. Anuj joined as our new CRO. He's been doing a great job as he's coming on board. We've also had a broader people, processes, systems rework. So in addition to Anuj, we have a new sales development leader and a head of RevOps and some of the other leadership functions ahead of an operational strategy under him that he's worked with before. So we've brought up, I think, the team in terms of the GTM site. We also, as you know, we're undergoing a big systems effort.

A lot of that is done. We are still continuing to invest, especially with some of the AI technologies that are out there. So we're leaning in on some of those. I think one thing that we look at is, obviously we're excited about the core we deal, but for as far as the GTM side of things, you know we have we we we have almost more customers that are in the $50,000 plus ARR group than we did last quarter. And that's almost as many or about as many as we add in a year historically. And so I think that that more broad-based repeatability is a good sign that the GTM is working now obviously i expected you know that'll fluctuate up and down but the general direction i think of that execution is showing up.

Excellent. Thank you so much. Thanks, Mike.

Your next question comes from the line of Itai Kedron with Oppenheimer and Co. Thanks, guys. Congrats again. Great numbers. Great to see the acceleration.

Gleb, I guess I have a little bit more of a bigger picture here. You know, the announcement of CoreWeave clearly is quite unique, I guess, in its size and its messaging. Does the deal with CoreWeave, does it generate more interest from customers or less? I'm kind of wondering if customers view that relationship as something that potentially ties you perhaps too closely to CoreWeave for people to do business with you. How do you think about that?.

Yes, thanks, Itai. So the announcement with CoreWave has been great. first of all, I would say the core team has been great to work with, but also the announcement has helped. I think elevate the elevate back ways of the key player in the AI infrastructure stack. I went to this AI infrastructure conference in Europe I think about a month and a half ago. And I'll just say that the conversations that I had were consistently... you know hey we're excited by this deal that you did you know we may not be as big as as core we but you know this seems very relevant to us very applicable can we talk to you about it how does how can we leverage your technology to do the same we're starting to see storage as a key need for us so um I think, obviously we provide a platform for a variety of customers, both in the NeoCloud and AI infrastructure space, and also to the AI natives themselves. So we have lots of startups and developers and larger size direct AI companies that see the CoreWeave deal and see that it's a stamp of validation. On the AI infrastructure side, there are certainly competitors to them, but it's a big market, it's a growing market, and a lot of people are trying to figure out how best to solve it. The other thing I'll just mention is, you know, we introduced this managed service approach or managed storage approach, right? So we do that for Corwe, but we're also doing that as an offering for others.

And a number of the conversations that we're now having with these other AI infrastructure companies is, them being interested not only in us providing infrastructure ourselves for them, but also providing this managed storage in their data centers and their sovereign cloud environments.

That's great, great to hear. And you kind of set me up for the next one, I guess. On this topic of managed storage, when you look at your pipeline, I know clearly part of core, this was part of the core of the transaction as well, but when you look at your pipeline, we look at the conversations that you're having with customers, Is this common that people are looking for this, or this is going to be more the exception rather than the rule? And then also when you talk about the two wins, for example, you have this score that you highlighted, the conversational AI company with six figures and the frontier model company with seven figures. Is there a way for you to have insights into their business to understand how much more opportunity you have within those organizations to kind of expand your use cases with them?.

Yes, so in terms of the managed storage side, it's obviously early, right? That's the newer approach that we're offering. What I will say is that in the past, we've had prospects that have come to us and expressed interest in us doing that. And in the past, we haven't done it. But with CoreWeave, we're doing this in partnership with them and offering it out to the broader market. And so we're having conversations now at CoreWeave. I would say there's probably half a dozen of these managed storage conversations that were fairly actively in discussions around the... This is not going to be every customer doing this in part because it requires a fair level of sophistication on the customer side and it requires a fair amount of scale to make it worth doing.

But for the larger AI infrastructure organizations and even frankly, not just AI, but anybody who needs large scale capacity capacity storage, I think that the managed storage is a good approach. So I think we'll see number of those. It's not going to be the predominant number on a volume basis, but I think that those will be larger opportunities.

I'll add to, I mean, if you want to jump in, Gleb, on a second question, whether we have visibility. I'll let Gleb answer on how much visibility we see into the large type of customers. But generally speaking, what we notice in Q2 is a lot of our, we said, four deals greater than half half a million. A lot of those were expansions. So we're seeing their appetite and needs. We said that before that AI companies generally grow a lot faster in their data appetite. And so we're seeing that profess itself.

Yes, and maybe just the one thing to that is, I mean, in the world that we were in at IPO, we were almost entirely self-serving. So we had very little forward insight into what was happening with the customers because they would just sign up and pay on a consumption basis. Now, as we're heavily investing in the sales-led side of the business and working with these larger opportunities, the team is actually actively in discussions with them, working around what their needs are, what their plans are, and kind of co-planning together. So we do have better visibility as well as the commitment.

themselves. Great stuff. Congrats. Thanks. Thank you, Dave.

Your next question comes from the line of Jason Adder with William Blair. Jason, your line is open. Please go ahead.

Yes, thanks. Good afternoon, guys. First question, just on the core weave deal, could you give us a sense of what the gross margins are going to look like relative to your traditional B2 business?.

Yes, hi Jason, this is Mark. For our gross margin, I mean, for the time being, we're pricing everything. on all deals to keep it in and around where it is. So even with CorWee despite the lot of scale, there shouldn't be that much detriment to the current 63%. What I would say is, between the core we've deal, the previous deal, the $15 million plus DCV deal, all of these committed contracts we're signing up, it does increase our CapEx needs. And so our CapEx for the year, will be between, you know, between 55 and 65% of revenue. And the reason why I mentioned that is, we're putting a lot of CapEx out, so depreciation will start, and then we'll, you know, we'll start wrapping up the customers in terms of that revenue spend. So that, that, to, quarters or so before they get ramped on that capacity, you'll have more depreciation.

So there could be a few hundred basis coins set back to our gross margin.

But then it should recover after that. Got you. And then if over time you end up doing more of the managed storage option with Corweave, is that, I'd imagine that would be a significant boost.

To the gross I think. Yes, I think I think that both of those are currently similar ish in gross margin. Although, as we think about our capital light approach with the managed storage overall, we think that that certainly has the possibility of being a higher margin offering.

over the longer term. Okay. Maybe I'm not clear. I thought you were just selling software there. What are you selling there in the managed storage approach of their if it's delivered on their hardware?.

Yes, good question. It's a managed storage offering. So the way that it works is they provide us the data center space, they provide us how much storage they would like, we provide the bill of materials, they buy the equipment, they hand it to us, but it's actually our people in that part of the data center, which is cordoned off for our purposes. are people racking, stacking, managing with our software that part of the storage stack. So, you know, CoreWeave will have multiple parts of their storage stack. We'll have the part that we're managing, but they will own the equipment. We will.

have the people. Got you. Okay. So the main cost for you is people in that scenario. Exactly. Okay. Got you. All right. And then one last question for me, for Mark. And for GLAB, I guess just as we think about going forward and the coral weave situation and the CapEx needs that you're going to have over the next few years, I think it's What are you contemplating in terms of capital needs? Do you have enough capital today to be able to meet the needs of this build-out, or are you going to have to raise more capital?.

Yes, Jason, with the. I'd say between our cash balance, We have and we have over $150 million of lease lines and and our operating cash flows have become really healthy due to operating leverage. So between all of those three things, we're well set now to do it the way we're doing it via capital lease lines. I mean, we're always going to evaluate all options and see what's best for shareholders. But we're set to proceed as is now.

Okay, thank you. Good luck. Thanks, Jason. Your next question comes from the line of Jeff Van Ree with Craig Hallam Capital Group. Jeff, your line is open. Please go ahead.

Great, thanks for taking the questions. Congrats guys, just breath, really impressive here what you guys did. Maybe spend a second on the Frontier model and the win there. I'd love to hear a bit more color competition, maybe a little more particulars around use case, duration of deal. And then I think you just said seven figures. I mean, any sense you can dial that in a bit? Are we talking mid single digit, seven figures, upper, lower?.

Yes, call around the front-term model would be great. Yes, thanks, Jeff. Good to chat with you. What I'll say is that this one, similar to some of the others, right? The pattern is the same, but I'll talk to this one, which is they have storage that they use and the data sets that they use to build their models. that they were having a couple issues. One was that the level of data that they were, the size of the data increasing, they were actually hitting quota ceilings. So the cloud provider that they were working with, they were having trouble providing them the amount of storage that they needed. So they were actually hitting certain ceilings there. They also had performance requirements. So the need to move that data at high throughput over to places where they would be building the models themselves.

And so between those things, they needed both the ability to scale and they needed the performance of V2 Overdrive. And so they were on another cloud provider previously. So they were familiar with, obviously, the model. They switched to us. The other thing I'll say is, like the other use cases that we've always said, they're not running the GPU model training directly off of the data on Backblaze, they're using it Backblaze as the place to store the big dataset, the capacity tier dataset, and then they're moving it when they're ready to actually do a training one, they're moving it to the flash tier next to the GPUs, but they're keeping it long term and at scale on back ways. The other thing I would say is the pricing is B2 overdrive pricing, so it's higher than what you see as a $695 range. price for the self-serve business on the website. And maybe the last thing to say there is it's you know, it was a it was a it was a large commitment from the outset, but it was also what was called, you know, what they referred to as an opening commitment and a building block to start from, but one that they expect to actually expand significantly.

Yes, I would think so. And then maybe just to follow up back to the managed storage offering. So when you're selling to a neoclide or one of these larger AI players, what is the delta between when they want the managed offering versus a white label offering? I mean, I understand there's sort of some geo, you know, data sovereignty issues, a lot of things probably come into the play, but like why one over the other, traditional P2 versus managed storage?.

Yes, the main reasons why most people want us to take care of it on our own infrastructure. And the reason for that is because it's fully taken care of, right? They don't have to worry about it. They don't have to think about it. And even for NeoClouds, the NeoClouds obviously range in the level of sophistication and their level ability to operate a full platform. You have CoreWeave on one extreme of a company that is very, very good at managing the whole infrastructure and technology and stack and everything else. And you have others who are just brand new. They have data centers, they have GPUs, but they're still building out all the other pieces, right? And so for many of them, they prefer to just have us fully take care of it. the managed storage side comes into play when they want to have the physical data in their own data centers and that that has sometimes the conversation has come up because they have data centers in regions that we aren't and so they would like the data there sometimes because They have a sense of they would like a more sovereign experience with their data.

And then for some of them, the conversation has simply been that they would like to actually own the assets their balance sheet. So those have been the reasons why they go one way or the other.

Great. Okay, I'll leave it there. Thanks so much. Congrats. Thanks, sir.

Your next question comes from the line of Eric Sepiger with B. Riley Securities. Eric, your line is now open. Please go ahead.

Yes, thanks. Thanks for taking the question and congrats. Great quarter. Couple questions. One just on the go to market. Have you hired most of the executives across the go to market team that you need at this point and then secondly uh i think You talked about the core we business reaching a minimum level, meaning meeting the minimum commitment level in mid 27. Does that mean that we can assume that one you're kind of at the one fifth of the three thirty five million, which is about sixty five million run rate? Does that imply that you're reaching about a sixty five? 5 million run rate by mid-27 on that CoreWeave agreement.

Yes, thanks Eric. So the on the GTM side, yes, we've hired the chief revenue officer, we've hired the head of sales development, head of robots, we've got the head of customer success, you know, we've got the head of GTM, the strategy operation. So we are, you know, obviously there may always be additional folks, but we are in the process. In practice, I would say we've got the team in place. And one thing I'll mention too is, I think when I was reflecting earlier on our journey, You were with us when we went public. And so you remember that when we went public, our average customer was a self-serve customer that paid us less than $500 a year. And we said our goal is to move up market, become more of this core infrastructure for startups, for companies, for enterprises. And we started signing companies that were paying us tens of thousands, 50,000, at some point we signed our first million dollar deal.

Then we started highlighting roughly a $1 million deal per quarter. Then we had our $15 million deal that we announced in February, and then this 335 million deal that we just announced. So, you know, obviously it was quite a journey to go from a primarily self-serve company doing a dollar a year deal to a company that is able to service million, 10 million, and multi hundred million dollar deals. But I feel like we're now in a great place with a great team and process and systems to go and execute against this opportunity.

Yes, and on the second question, Eric, let me dive into the second question. So the CoreWeave deal has two components. As you know, there's working off of our platform. That's one component, and that was all disclosed in the June 23 deal. And then there's the managed of service component. So roughly speaking, it's like almost a 70-30 split. So when I said we would reach the minimum it's relates to that 70%, not the 30.

The 30 would come afterwards. because the managed service has a different kind of ramp. And then the other thing to keep in mind is the warrants are a contra revenue. So that's why you shouldn't take the just a 335 times 70. You also have to deduct the $22 million value of a warrant.

Can you just expand on that? You deduct the $22 million for the warrants. Is that across the five years? Is that a straight up division?.

Yes, exactly. So the warrants just they follow the revenue. So there's five years for both components of the deal. So if you take that 70%. of the 313 million, which is net of the warrants. That one ramps up over the first 12 months, and then once it's up to 12 months, then it starts operating at that minimum.

Okay, perfect. Thank you very much and congratulations.

Thank you. Your next question comes from the line of Eric Martinuzzi with Lake Street Capital Markets. Eric, your line is now open. Please go ahead.

I wanted to revisit the upward revision to the 2026 guidance. The way I understood it, Mark, you talked about three reasons for the upward revision, and they were all equally weighted. The business outperformance to date and the price increase. As the self-serve product-led growth, we did that price increase on May 1st, so we anticipated some churn. We really haven't seen any churn.

In fact, what we've seen is an acceleration of people signing up and the art group sign up is higher. So we're, yes, so I think we're seeing, you know, good momentum on almost all route to markets and all go-to-market levers. So it's not, it's pretty broad-based. And Gleb mentioned the, you know, the 50 customers that went over 50,000 in ARR, you can see our RPO every quarter goes up. We're getting customers committing into either one year or multi year contracts. So it's broad based the general business health. So we're seeing really healthy acceleration.

And then on the CVU, I think you said last quarter that you were expecting it down to What was it, 3% or so or low single digits? Is there any change to that expectation for the year in the new forecast?.

Yes, that business is kind of like we talked about, Eric. It's a good business, people like it, the customers like the experience. It's cash flow generating and helps fund some of the B2 growth. you know, it's an area that we're spending some time and some investment on, but it's still a business that has overall market head and so it's, you know, we still think it's kind of single digit declines. It did perform better this quarter than expected by a little bit. And that was in part because I think we've been doing some some efforts on trend mitigation and customer acquisition, but it is still like a kind of a single digit it's declining business. Got it. Thanks for taking my questions.

Thank you. Your next question comes from the line of Rustem Kanga with Citizens. Rustem, your line is now open. Please go ahead.

Great. Thanks, Mark, and glad for taking the question. Great to see the sustaining momentum here. My question is just around CoreWeave. Can you help frame the extent to which that recent win is helping accelerate discussions with other NeoCloud providers evaluating HDD-based storage tiers? And are you finding that the best conversations are those who have already experienced challenge challenges with the costly slash storage approach or is it better or more effective to cut them off at the pass and approach those who are even yet to begin a DIY approach?.

It's a good question, Martha, and I'll tell you, it was actually counterintuitive for us when we went down this path and you know, we we saw that this was going to be an opportunity, you know, about a year and a half ago, right? We launched B2 Overdrive, we launched B2 Neo, you know, we leaned into this idea that the Neo clouds were going to need and AI infrastructure companies were going to need a a capacity layer for storage. Our assumption at the time was that the best path would be to go after those that did not have storage yet. we've found is that generally speaking the more engaged conversations are from those that do and the reason for that it seems is those that have storage are feeling the the pain of only having um the flash-based storage or trying to do it themselves. Meaning they have customers coming to them, expecting them to be able to service them, but struggling with dealing with the scale or the expense of those things. the ones that don't have storage yet, their customers are going somewhere else for those workflows. And so they're missing out on those workflows and they're not feeling the need for them yet. But as they start having those conversations with customers where they're talking to them about like, servicing their broader need, then they're also starting to feel that demand. So, So I think the short of it is that. The Neal clubs are generally heading down the path where they're going to need this capacity of tier of storage.

They also will need the flash based tier of storage. It's not it's not I don't see it as an either or I think it's they need both. as part of servicing their customers and we're a great solution for the capacity tier.

Perfect. Thanks. And then, Mark, just you gave the caller on the CapEx of 55 to 65% of revenues. Was that comment more for the back half of this year, or do you expect that to hold through 2027? Just help us think about if it would ramp from that level, or if that comment.

was applying to this year and next year. Yes, Ross, that's for this year. That's for 2026. Too early to give 2027, frankly, mainly because the prices of this hardware changes pretty quickly and obviously our growth outlook keeps accelerating. So that that number will for both those reasons will change the 27 number. So for the time being that 55 to 65 is for the revenue as a percentage of revenue for 2026.

Thank you very much. We have reached the end of the Q&A session. I will now turn the call back to Gleb Budman for closing remarks.

Thank you. So AI is reshaping the entire infrastructure market and Backways has built exactly what this moment demands. Storage is that durable layer beneath AI. I'm really pleased with how our team has stepped up to capture this generational opportunity. I want to thank all our backblazers for leaning in and to our customers and partners and investors for joining us on this journey. Finally, we look forward to seeing many of you at our Investor Day in New York City and on our live webcast on September 9th. Please RSVP to ir at backblade.com if you'd like to join the in-person event. Thank you all for joining today's earnings call.

Operator, you may now end the call.

This concludes today's call. Thank you for attending. You may now disconnect.

This live transcript is auto-generated without human intervention or review.

[Call has ended.]

Hi, everybody, and welcome back to the Q1 2026 Drive Stats Report. Today, I am joined by Laquie Campbell. I'm Stephanie Doyle. Let's do our meet the team jumping right in. I'm Stephanie Doyle, Senior Manager of Market Intelligence and affectionately called internally as the keeper of stats. So I'm here for Drive Stats, here for performance stats and here for network stats. And today, I've got Laquie Campbell with me. Laquie, you want to talk about yourself a little bit, let everybody know who you are?

Sure. Hi. I am Senior Product Marketing Manager focused on media entertainment. I've been in the creative technology world for a while. I'm also part of SMPTE, and I'm the Director of Education for Programming with them. And right now, we're focusing a lot on AI and media workflows and infrastructure.

And I'm super excited to have Laquie with me today because I feel that media and entertainment has always been a really important industry through which to view hard drives. We were talking about it before we went live today. And it is a space where both the individual drives very much matter as well as this top-level orchestration of data workflows. So you really see multiple lenses on a hard drive when you're [ M&E ]. So thanks for joining me today, Laquie.

Thanks for having me. Drive Stats are super, super popular when I go to shows.

I love that. So a few housekeeping notes for you every time. Webinar is recorded. Please feel free to ask us questions. We'll leave some time at the end and don't miss the resources and attachments. We'll drop them in the description section of BrightTALK once this webinar is over. But you should see links and all that good stuff as well.

So what is Drive Stats? If it's the first time you're joining us today, Backblaze has been collecting our raw device metrics for over 13 years at this point. And we publish the failure rates of hard drives. If you go to our website, you'll see a maintained data set that has daily logs from each of the drives in our pool. And then we also filter those out and look at them on a tabular level, so we can talk about annualized failure rates.

What happened this quarter, very exciting. So we've got about over 340,000 drives now. We had about 1,000 drive failures and equivalent drive days, which is how many drives existed per day in a single quarter. So you can see the drive population by manufacturer. It's about 1/3, 1/3, 1/3, which is interesting. And we'll talk about some of the quarterly data here.

Now this is the big eye-crunching table that you can -- if you go to our blog, you'll see that we actually have a much more friendly version of that this quarter with responsive tables. We're very excited about this. But top level here, you've got 1.24% annualized failure rate for the quarterly stats this time around. And if you want to just look at what's been happening here quarter-on-quarter for about the last year, you can see that it's up from last quarter, from Q4 2025, but down year-over-year.

We've got new drives. So the 26-terabyte [ DC ] is now in production. And you can see some pretty interesting stuff. This last quarter, we deployed a little over 10,000 drives. And of those, 9,400 or so were more than 20 terabytes. We've been keeping an eye on that population because it's -- of course, as you deploy larger drives, it's got implications all the way around. And so this quarter, the AFR for that population specifically was 0.85%. Don't over-index on that. These are pretty young drives, and we know for sure that younger drives fail at fewer rates or fail fewer times.

So there's that. And then you got a couple of drives here with 0 failures, which is pretty exciting when you look at the age of those 4 terabyte drives. Those guys are -- I don't know, they've been around for a real long time. If you look at the average age in months, the 4 terabytes drives are 106.1 age in months. We do measure our drives like toddlers in months. That's a little under 10 years, which is pretty cool. There's only 186 of those remaining, which says to us that they're on their way out, and you'll see that they've dropped off the lifetime table this time around as well.

So lifetime data, if you -- these have slightly different exclusions here. So it will be -- they need to be hanging out for more drive days and have more drives in the population. So you'll see it's slightly different. What's interesting this quarter is that our annualized failure rate is 1.39%. We've been holding pretty steady actually for the last several quarters at right around 1.29% or so. So this is a bit of a jump. But the jump is actually because some of those much older drives came out of the pool. So the 4 terabytes have dropped off the lifetime table. And like I said, those had 10 years of data behind them. So as those exited, you'll see a little bit of a fluctuation because things have changed. And then we've got 3 new drives entering this as well. So you'll see the 22 terabyte and the 26 have made the exclusion criteria to be tracked in the lifetime drives.

So what's interesting this quarter is actually the data is always interesting, but it's a little bit something that we were able to flag when a drive was not feeling that crazy. So this is -- the disturbance in the Drive Stats is what I'm like affectionately calling this. But I think what's really important is that and something we've talked about before is that all systems that we talk about in the reporting that we're doing here, they represent a managed system, right? We're monitoring our drives at all times. So we can actually affect whether things have higher and lower failure rates. If we see something, we can take like proactively or reactively change it.

So this is a really interesting instance of this because what happened is that we had a pretty acceptable drive model. We started seeing failures climb. And we actually saw -- it was a little bit of a confusing investigation because once the investigation was complete, at the end, there ended up being 2 separate issues that were happening that were very different -- distinct from each other. And the drives themselves could have experienced one or the other or both of these failures, right? But we identified the power cycle issue early on. So we noticed that if we shut them completely off, they would have trouble coming back online or maybe not come back online at all.

So what we did is we took a mitigation step that we just reduced the power cycle frequency to the vaults of those affected drives. And once we did, the failure rate came down to a reasonable level. So this is obviously not real numbers here that you're seeing. But if you can imagine, what we did is reduce the risk of one of these types of failures, and that left us totally fine. And then once the investigation completed, we saw that there were multiple issues happening.

So from our perspective, when I went to Drive Stats, I was like, hey, would you look at that? That failure rate is not very high, but I know something happened. So I had to do a bunch of investigation on my end to make sure that we were properly capturing failed drives because I didn't want there to be a situation where we were underreporting where we should. And what that led us to is a really interesting lens on how we define a failure.

We've talked about how we define a failure in the past. And the quick and dirty is essentially that there's a C++ job, the custom program that collects the SMART stats for drives each day. And then there's some things about the exclusion tables or inclusion tables that are maintained by humans that affect these things. But at its most basic, it's conditional logic. Was the drive there yesterday? Is it gone today? If it is gone and it was here yesterday, then we log it as a failure. And if we look at -- if it's a capital F failure versus a lowercase f failure, let's say it that way, there's -- that's where those exclusion or inclusion tables come in. So you can imagine if you swap a drive out for normal routine maintenance, we're not going to call that a failure because the drive did not fail. We just swapped it out. So -- and then the other thing is that if the -- if the serial number comes back online before the end of the quarter, then it's no longer a failure, like we took the drive offline for some reason or another and brought it back.

So there's a little bit of squishiness there. The quarter end cutoff means that like if the drive comes back after quarter end, potentially, you've got 1 or 2 false failures in there, but it's really not -- you got to cut it off somewhere, right? But what this does mean and what this -- what came into play for this specific incident that we were tracking is that if you have day 1 failures, you actually don't log those, right, because it has to exist before it cannot exist the next day. And so from our perspective, I think it's an interesting caveat to make on the data set. Lots of folks look at the data set as a source of truth. So if we're undercounting day 1 failures, we want to make sure that's well understood.

On the other hand, how often do you get a day 1 failure? In this case, with this specific drive, there actually weren't that many, but it was something that I wanted to double check on before I came out with failure rates. But on the other hand, it's pretty rare for drives to have day 1 failures like there's qualification drive manufacturers do. We personally do qualification on the drive. So it's not something that happens all that often. But again, because this is used so widely, there you go. We've got a new caveat to the data set. If you're using it, make sure you're checking logline data and see what's happening.

All right. I think that brings us to resources here. We got the report, of course, follow the series. These are many ways you can interact with us. We've got the Drive Stats homebase on the website, join the insider's newsletter or reach out to us directly. [email protected] is a monitored inbox. I'm there. So I'm always happy to respond to real humans. And then we've got socials in the comments section.

So yes, let's take this time and talk questions. Laquie, anything on your end that sparked while we were going through the data?

Yes, absolutely. From my POV, I told you many times, I always get stopped at shows. And when I bring up Backblaze, they're like, "Oh, yes, Drive Stats. I look at that every time." And it's because in media -- and when I say media, I don't mean just film and television. Every company these days is a media company, whether we're talking about corporate marketing, bio. I mean, it doesn't matter. Everyone is trying to tell a story visually and those files typically are larger.

In a traditional media library, a video file might carry maybe a few hundred metadata attributes. And so now that everyone is working with AI, trying to figure out how they can efficiently use it in their workflows, that same asset can generate 3,000, and that's just the metadata. So every stage of the workflow is producing more data, more iterations, more outputs.

Another kind of tangible example would be AI upscaling. So AI has made the ability to make an older show, remaster it and make it look great quicker-ish. So if you have a show that was shot natively in 2K and a decision is made, let's make it 4K for OTT or streaming or even 8K, let's say, there's a premium tier or you're trying to future-proof it. That 2K original isn't going to disappear. They're not going to trash it. It's the source of truth. It's the legal master.

So what we're seeing, obviously, is now you have that one asset with 3 retained versions, the 2K, 4K, 8K. The 2K, let's say, it's a 1-hour show, it's ProRes. Let's go with that. That might run near like 100 gigs. So that 4K upscale typically is going to need its own render pass, going to need its own mezzanines. It's going to need its [ own ] QC deliverables. So that new version is realistically going to bring 3 to 4x to your footprint per title and that's before 8K. So all of that is happening, and that's great. I mean it's great that you can be able to do that faster.

However, if your infrastructure wasn't great before, you're going to start to see the cracks. And so I think Drive Stats is becoming even more important to this audience because they're having to bolt together really smart systems and infrastructures. The story is not whether we do cloud or on-prem anymore, that's dead. It's more how do we put all of this together to work efficiently, to be performant, for capacity's sake because obviously, budget is a factor. So how much can we really put on this drive and not make all of our originals disappear. Those are always going to be very front of mind thoughts. So this is why Drive Stats is really important in my domain. So...

I love it. Yes, it's always good to hear. And it's interesting you mentioned AI because we've got a question from [ Raghav ] here in the chat. How is AI impacting Backblaze? How do we think it will affect small and medium businesses when it comes to storage based on what we're seeing? And just based on what you are saying here and sort of the conversation we had before the call, like I totally agree, right? Like people in M&E space have to consider individual drives, but also all these other moving parts of their architecture.

So as far as AI goes, I think it's a very interesting conversation. And I can talk about it from a data center perspective, but I'm interested to hear what you have to say in the M&E industry or even in just like content generation at all, like thinking of content broadly as multimedia files, I should say.

Yes. I mean it's just that continuous derivatives that are being created that people are thinking, "Oh, I can be more creative. Our team can deliver more." But again, if you don't have that IT team, if you don't have that technical person that is thinking about, that's awesome. Like I love everyone being creative and wanting to grow and develop, but how do we store all of that efficiently? How do we get to it when we need it? How do we monetize it and be able to find it quickly. So that's -- all those things start to come into play when you are doing all this iteration.

Another example would kind of be different grading. I don't know if people think about -- when I say grading, I mean color grading. Sometimes -- and we've all seen it, the warm grade for a certain type of movie versus a blue color grade, there's a lot of testing. Directors, DPs, the directors of photography, they do a lot of testing. And so now we're maybe doing AI color grading upscales and upscales. So now we have the upscale and we have the color grade. And maybe there's 4 different ones. And so -- and then let's add on HDR to SDR grades or A/B test audience variant.

So it's amazing. AI is great because you have the ability to create these things possibly faster. But again, if you don't have the infrastructure built out and the processes to like manage it and figure out where everything is and make sure everyone is getting all of the files that they need to create, that's when it can become an issue.

Yes. It's just like an explosion of data on the most basic level and managing data, particularly in active archives has always been a conversation in this space. I think when you're talking about how it's affecting -- if you're saying how is AI impacting Backblaze, I think that's a pretty complicated question because it kind of depends on what lens you view it through.

So we certainly see this huge demand for data. But on the flip side, we also see access patterns for these things moving in really different ways. If you look at the network stat series that we work on too, you'll see just a high volume of data moving all at once and then being processed a lot of times through -- sometimes even through traditional CDN providers who are converting to or are already neoclouds. And that in conversation with like traditional media tooling and where a lot of that data lives, you're seeing a lot of these integrations become super, super important to your point, Laquie, where it's like you have a lot of moving parts, and they all have to work together.

Absolutely. And that's why I love the ecosystem that we've built at Backblaze. We have amazing partners that are doing just mind-blowing things with AI and technology. And so I think that's another way that Backblaze is being impacted and also being part of the story is that the openness of Backblaze. And so yes, it's an amazing time, but it's definitely one of those like firehose moments. So like how do we manage all this? I think the infrastructure was built so well here that I think that we're just excited about what we're seeing and able to ride the ride with people and our partners. So...

Yes. Yes, agreed. So next question, what's -- 2 kind of related ones. About how to use the stats to buy a drive or what sort of brand we'd recommend. And I'll give our standard language here. We really do not play favorites when it comes to recommending drives because -- and I think this is an important part of the conversation. In some ways, what we do is sort of the ultimate litmus test for a hard drive, right? They are running at max capacity for their entire life until they die. That is what we do in a data center, right?

But on the other hand, your personal use case might look quite different, right? And we also have redundancy, both on the software layer and with secondary drives to be able to manage potential data loss. So the way that I would recommend using the stats is to identify what size of drive you need first and then see with what you're comfortable with, look at reviews and then build out for backup.

As always, I always recommend having at least, at least a bare minimum a 3-2-1 backup strategy, right, with parts of it in the cloud, parts of it off-site. And then if you want to have your on-site for a traditional air gap, that makes sense too. All of that gets complicated. Data management on a personal level is something that I have a lot of respect for. So really, you can get as durable as you want to on a personal level and your drives are certainly part of that conversation.

Yes. And I think these drives really help trying to help people figure out what reliability at scale looks like for them. When they're feeding data into whether it's automated pipeline or we're not just talking about the human editor aspect of it. People are doing a lot of modeling and training against their footage and their archives. And so undetected drive failures, mid-data set, it's not just, you lost a file, it's like silently corrupting the whole training model, right? So yes, this is, I think, a great report for people to look at and just try to assess accordingly what's going to work for them in their current workflows.

Yes, totally agree. All right. It looks like we've got another question from Hans -- I'm sorry, [ Hank Gunns ]. There we go, guys. So excuse me, for mispronouncing your name there, [ Hank ]. So a member of a group of 200 photographers, awesome, love it. You are the techie and you showed them how to back up onsite, but you want to have a presentation on why they should use Backblaze to back up off-site.

[ Hank ], with no personal motivation, I direct you to our blog. And I say that because I've written on the blog historically for quite a long time these days. But we've got quite a few articles on the benefits of off-site backup. I think when you're talking about off-site, it's really important when you think about disaster recovery and that sort of thing. But the other piece that I think is important specifically for this audience is a little bit what Laquie was talking about where you have lots of things everywhere and you might want to be able to access them remotely or with different tools or in different ways. So you might want to -- I think, just simultaneous and distributed access is an important part of this conversation, too, beyond backup, those active archives that folks need to work with. Right. Anything else you're thinking about? Laquie, anything top of mind for you?

No, I don't think so.

Yes. Yes. I think there's a lot to look at here and many different lenses to look at the impact of hard drives and AI. I think what's particularly interesting to me is sort of people have started to look at their tech stack and understand that it's -- how much they need to move with them, right? So those active archives are always part of a conversation in my neck of the woods.

No one wants single points of failure. Like we're always trying to figure out how to make a workflow smarter. And your archives, I mean, they are literal gold. Each frame has such high value. And so it's just really important to make smart decisions at this point as you're growing and growing and growing as you iterate.

Yes. I totally agree. All right. Well, if we don't have anything else from the comments section, I feel like we had a great time here today. I appreciate everybody for showing up. And as always, feel free to reach out if you've got additional questions. We're happy to answer them. Thanks, everyone.

Thank you.

Hi. My name is Frantz Famie, and I cover technology and investment banking for BofA. And today, we have the pleasure to receive Gleb Budman, who is the CEO of Backblaze. And Gleb, I think not that many people know who Backblaze is, right, and especially the story around like being able to the history of you, the way you founded your company and everything.

So can you maybe walk us a little bit through the history and where you are today?

Yes. And thank you, everybody, for coming and joining and chatting. So it is painful to say not that many people have heard, I mean, we have 0.5 million paying customers. But I don't think you're wrong. I think that our opportunity to be better known is quite large. So we started the company almost 20 years ago in 2007. We initially started it in the cloud backup space, but we ended up designing and building our own cloud infrastructure for that purpose. And so today, we are essential infrastructure for AI.

And I certainly talk to more of how that's the case. But it started from a need of we needed to build storage infrastructure that was high performance and very efficient for the purposes of providing it to our cloud backup service. And our original plan was to store that on AWS. We did the math, realized we were going to lose money on every single customer, started from first principles, wrote our own file system, built our own infrastructure, built the whole stack and offered that to our cloud backup service.

Over time, companies came to us and said, "Love your cloud backup service, but you built this great infrastructure as a service, give me access to that for all of my other storage needs." And so we launched Backblaze B2, and that is now the dominant part of our business and what we offer to the market. We still have the cloud backup service, but the focus is the Infrastructure as a Service.

And what are some of the pain points that the legacy hyperscalers cannot address that bad place is able to like feel?

Yes. So for a long time, one of the key things that people were coming to us for was simple. It was -- it's a 1/5 the price of Amazon S3, and that was simply a compelling value proposition, right? You need storage, it's less expensive, that's easy, right? With AI, there's been some interesting new shifts, right? There's this big replatforming happening. We just heard from one of the neocloud companies. There are a variety of neoclouds that are out there. And they are -- and customers are choosing to use them either because of availability or performance of GPU infrastructure or because of the software stack that those neoclouds are focusing and providing, right?

And in order to use any neocloud, your data can't be locked inside of a hyperscaler. And the hyperscalers all try to keep your data locked up inside of them. And for many, many years, the last 2 decades, most customers have felt that, that was fine. Most customers have felt, "I use one of the hyperscalers and I use the services they provide me and it's fine."

And what's increasingly happening now is the customers are saying, "Sure, maybe I use one of the hyperscalers for X, Y and Z services. But in order to stay innovative in order to do the AI workloads I need, I also want to use that neocloud, this neocloud and that neocloud. I might want to use this inferencing provider. I also may want to use one of the different CDN providers that are out there. I need to use infrastructure that's not part of that one single hyperscaler. And to do that, I can't have my data locked inside."

And we had one customer I remember, they developed their whole own infrastructure and they were using one hyperscaler. That hyperscaler didn't have the availability of the GPUs they needed. So they literally did a whole engineering effort and migrated their entire system over to another hyperscaler. A few months later, that hyperscaler didn't have the most efficient GPUs that they wanted for this. And so then they paused because they were thinking about moving it again, and they paused and said, wait, this can't be the way we operate.

We have to free our data so that we can use whoever we want to use. And Backblaze provides you this independent location to put your data efficiently in a high-performance, low-cost way, but one that provides you free egress and good connectivity to all of the places you would want to use. And so that's been a key benefit to the customers that the hyperscalers just don't provide. They can't because they don't want you using all these other services. They want you using their services. So that's been a key part. So that's one.

And then the other I would say is for the neoclouds themselves, initially, most of the neoclouds started just renting GPUs. And many of them are trying to go more full stack and offer additional services, but be part of the way that AI gets built. And to be part of the way AI gets built, you have to be part of the customers' workflow. And all AI workflows require data and generate data. And so the neoclouds increasingly realize they need to offer cloud storage as part of their offerings.

And so we have a white label version of Backblaze B2 for the neoclouds, which, again, they don't really get from the hyperscalers because they are their competition. So if you think of it as we're serving kind of 2 sides of the market. We're serving the actual builders using AI in order to enable them to use whoever they want. And that's critical for them to innovate. And we're also serving the picks and shovels of AI, the neoclouds and others who have realized they need to offer storage as part of their workflows for their customers.

Okay. Can you maybe expand a little more and go over like a case study of specifically for AI on how you're able to -- what was able to transform a company going through like the AI transition?

Yes. So I mean I'll give you, I guess, examples on both sides of that, right? So on the demand side of things or on the side where the builders are, right, so I was actually talking to a company just a week ago. They are in the GenAI media space, right? So they build a model in order to then use that model for inferencing where the customers can prompt and create video. There are now hundreds of these companies out there, right?

And so they are collecting lots and lots of data sets in order to build their models. Those data sets were previously stored on the hyperscaler. They were finding it increasingly painful because not only are they -- was the expense of storing the data becoming prohibitive for them for their business, but they were using multiple neoclouds as well as one of the hyperscalers actually for the model building for the GPUs. But getting the data from to these different places was just crushing them, right? And so they moved all of that data over to Backblaze.

So now they're able to have one single large data lake with tens of petabytes of data just for them, just for their model building. And they can use it and send it wherever and however they want, it's free to do. It allows them also to innovate faster. It's not just about the cost. It allows them to innovate faster because in the past, because it was so expensive to get the data to the different GPU locations, they were doing fewer training runs fewer iterations on their model because they have to balance it was too expensive to get the data to the GPUs. Because they were able to send it when they need it, they could do more training, more frequently, innovate faster. So that first part.

Then the second part goes to the inferencing part, right? A lot of our customers that came to us in AI came to us for the model building because that's the first part. That's where they have these large data sets to start. But they are a GenAI media company. The whole point is to create videos. So now they're talking to us about the model itself, the inferencing is creating videos and they're storing all these videos out there. Once created, these videos live on at this point forever. Maybe at some point, they'll decide to not, but at this point, they keep them forever.

And so that data volume just grows and grows and grows. They need to send those videos to their customers. They want to send them as quickly as possible. It's very important for them to get fast. We've developed something called Sard Stash. It's a patent-pending technology, allows you to ingress the data as quickly as possible. One of the things they told us is that ability to get it out of the GPU and into storage as quickly as possible allows them to free up GPU time.

So again, it allows them to be faster and more efficient with their business that way. And then our free egress and connectivity to all the CDNs means they can use the fastest CDN available. So it really is almost an end-to-end workflow that enables their business to not only be much more cost efficient, but actually more innovative and faster. That's on the demand side.

On the supply side, on the infrastructure side, we've talked about how we've signed multiple neoclouds. So if you go to neocloud A, B or C at this point and they say they offer storage as part of their product set for a handful of them, it's actually Backblaze underneath white labeled for them, right?

And so how has that changed for them? In the cases of 2 of them, they had storage before. What this is enabling them to do is offer storage that is better, faster, higher throughput. It's a better offering that scales more. For one of them, it was a completely new product offering. So they were able to expand the value they provided to their customers and have customers be stickier to their entire workflow. So those are 2 examples of how we're helping the 2 different sides.

Okay. Yes. And I think during your last earnings, you disclosed a lot of traction in AI, right, including, I think, 76% growth AI customer. Can you maybe discuss more your recent traction in AI? And how large you think that can become as a growth vector for the company?

Yes. I think the other thing that we talked about is that 1 in 3 of the new bookings came from AI, right? I mean, it's a material change for the business, right? And as we look at it, I think we are still scratching the surface of it. The urgency that we're seeing from these companies, one of the examples that we gave was that we signed a customer that was roughly $1 million of ARR in 11 days, right? So they clicked on a link and they signed a contract like an ad click on an ad and signed a contract within 11 days for about $1 million of ARR, right? So that is a very fast close cycle, right? And it just speaks to the urgency that these companies are feeling around the Infrastructure as a Service storage need that they have.

So I mean, how big can this be? I mean the market is huge. The opportunity is huge. When I think about just taking this one single GenAI media company, right? So this one single GenAI media company that I'm talking about, it's a good sized 6-figure, high 6-figure type deal. That's only the training data, and that's only the training data today. So one of the things that they're doing is they're continually adding more to their training data sets, right?

So one of the things they've talked about is, well, now they want to create these videos not only in English, but they want to create them in all the languages that are available around the world. That's more training data, more training data and more training data to improve their model. So the training data grows. But the training data, as far as their business is concerned, is just cost.

The inferencing is where they will end up getting all their value. So the inferencing has to be orders of magnitude bigger in data than the training. Otherwise, the model for them doesn't work, right? And so just thinking about is we signed a customer in GenAI media, high 6-figure deal, fairly short sales cycle and that training data is growing. And then the inferencing side has to be orders of magnitude bigger. For that one single customer, there are hundreds of these GenAI media companies today already. That's just on that one single segment.

On the neoclouds, we estimate our opportunity for it, right? So neocloud Infrastructure as a Service, just storage, just storage at the data lake layer, right? So not talking about high-performance flash sitting with the GPUs, which is not what we offer, right? There's like half a dozen companies doing that. There's very few choices for large-scale, high-performance data lake. Just that part of it, we estimate about $14 billion, right?

So we are leaning in heavily into this AI opportunity because of all the opportunity we see. We've built a business that is growing and steady and great supporting ransomware protection, backup, archiving, media workflows, that continues to be a good, reliable, steady growing business. But on top of that, we're leaning in heavily to the AI side because of all this opportunity.

Yes. And can you maybe go into some of the innovation that you -- like recent product releases or new features or products in the pipeline that you're excited about specifically in that data lake, high-performance data lake market?

Sure. Yes. So I mentioned one, which is shared stash. It's not that new. We released it a little while back. But it's an example of, I think, how you can innovate on the storage layer, right, which is one of the things that we heard from people was that the ability to quickly upload small files was important. And we analyze every single step of the way when you upload a file, what is every single step that it takes to get that file actually uploaded.

We found that the slowest part of it was the acknowledgment itself from the hard drive back to the user, and we were able to build technology to optimize that piece to make it both faster and cheaper than other providers, and we have a patent pending on that. So it's an example of technology, not a feature to the customer, but a way that we've built as one of the people said, the bottom part of the full stack, like the part where, how do we optimize the technology stack to make it as efficient as possible using software.

On the release side, 2 things that I'm excited about. One is B2 Overdrive. B2 Overdrive was our extremely high throughput offering to the market. And it came from conversations that we actually had around GTC and NVIDIA's GTC last year, not this year, where we talked to customers and they were talking to us about this need of saying, I've got this huge data set, when I've got it all assembled and when I've paid for the GPUs to start being available, I need to get this from here to there as fast as I can because otherwise, those GPUs are sitting idle and it's extremely painful and expensive, right? So I need to move that data quick.

But until I'm ready, that data needs to be in a durable and affordable place. And that's an ongoing thing. So we built this B2 Overdrive, which is an extremely high throughput up to terabit per second offering. The other one that we launched more recently is B2 Neo. And B2 Neo is the dedicated white label offering for neoclouds that provides the high throughput, the durability and everything else, but also provides manageability controls for the neoclouds to be able to manage rate limits and users and all that.

Okay. Yes. And as a public company, what do you think investors are missing about like bad place, whether on business model, products, team, like what would you like to highlight to...

I mean I think investors I think one of the things that gets missed is sometimes people go, isn't it just a bunch of hard drives, right? We have 300,000 hard drives. We have 5 exabytes of customer data, right? So I think some of the perception of like what's the value beyond the cost of the hard drives, right? And I look at that a little bit of it's kind of the same as saying, well, AWS is just a bunch of servers. Like sure, they provide infrastructure with a bunch of servers, but it's all the software and infrastructure experience and the technology stack on top of that, right?

But I think that, that's often a misperception, right? And I think that the fact that we have been able to -- since we launched B2, it's been about a decade. Our price for B2 has gone up, not down. over the course of a decade. And I think that, that speaks to the fact that we have technology that we have built and customers value above and beyond the cost of the hard drives, which obviously have gone down in cost over the last decade. So that's one thing I think sometimes gets missed.

I think another thing that we sometimes hear is, what if AWS just lowered their price, right? And what I would say is that if you look back, it was the same question that we got almost 20 years ago. At the time, we designed our own infrastructure because they were too expensive. We built all the technology and people said, well, but what if AWS lowers price. It has been 20 years. We are still about 1/5 of their price point, right? So I think that it still comes up as a thing, but I would say that, that's -- and I think sometimes it's just the I guess maybe another thing I would say is some of the people don't understand that there's flash and then there's this data lake layer.

And there are at least half a dozen companies providing different types of flash storage. And flash storage is great. It's important. It's valuable for a number of use cases, but it doesn't solve all the use cases. And the use cases that it doesn't solve are very big, material and important. And there's a very small number of choices that customers have for that data lake layer that we, in some ways, I would say, got lucky to be in a place where it is incredibly valuable with AI. It's something we started building before AI was a focus, but the technology today that we've built over the last 20 years is just becoming increasingly important.

Okay. Okay. Yes. You can't come to a research conference without us talking numbers. So we saw that Google raised $80 billion to fund the AI infrastructure. So what are your views on the amount of capital that will be needed for you to fund that like the growth in AI?

Yes. I mean there's definitely -- the world is interesting right now with billions being thrown around here and there and everything, right? What I would say is Q4 of 2025, we had said about a year leading up to that, we had said Q4 2025 was going to be the quarter -- our first free cash flow, adjusted free cash flow positive quarter since being a public company. We hit that milestone.

The next milestone we said for ourselves was that 2026 is going to be our first adjusted free cash flow positive year, right? And we expect to hit that milestone, right? So we don't intend to need to raise capital. Having said that, we signed -- as we leaned into offering the neocloud storage, we've signed multiple neoclouds. We signed a 6-figure deal, a 7-figure deal and most recently announced an 8-figure deal, right? That requires capital build-out to support those.

Again, for all of those, we can do them under our own steam. But we are talking to a broad range of neoclouds. We are talking about large-scale opportunities around those. If some of those materialize at enough scale, it is possible we will want to support those. But we have about $100 million on equipment lease lineup that we have, and we generally fund equipment out of equipment lease lines that align with revenue as opposed to raising capital. So we don't need money to run the business. if the business materially accelerates, that's always a possibility.

Yes. And when you look at long term, next 3, 5, 10 years, given the change that Bay is going through, what would you define as success?

Well, I think that the industry is going through this replatforming. And for the first time in 2 decades, right, we can see that the hyperscalers are not the only game in town. And I think that our opportunity is aligned with that, right? Our opportunity is aligned with the fact that part of the replatforming that's happening is these neo clouds and the semi analysis talks about almost 200 neoclouds out there, right?

The part of our opportunity is to be an essential infrastructure provider to a number of those neoclouds, while we, in parallel, directly support the thousands and thousands of AI builders out there with the essential infrastructure that allows them to use all of the providers that they want. So I mean, I think we have a very, very big opportunity here. We built this fairly unique storage platform and outside of the few hyperscalers who've also built impressive storage technology.

But if you want a white label offering as a neocloud or you want a provider that provides you independent access to all the things you want to use as an innovative builder, the choices are very few. And so as we look at it, the just the Infrastructure as a Service storage market alone was forecast to be approaching $100 billion, right? That's just the Infrastructure as a Service storage market, and that was not really fully accounting for the AI impact of all of it into the future.

So as I look at it, I think success is us being essential infrastructure for the neoclouds for storage, us being the preferred provider for the storage platform for AI builders who want to be the most innovative and use the different services out there and having that be reflected appropriately in the value that we're providing and the value we're receiving.

I think I'll open it to the room for questions. Anyone has a question? All right. So maybe what kind of like message would you like to leave to some of the investors that are here specifically about like that, please?

Yes. I mean I think that for many years, we were providing a service that was call it backup in nature, small business focused in nature. I mean, even when we went public in 2021, our average customer paid us about $500 a year. On our last earnings call, we talked about a customer that paid us -- that signed a contract for over $15 million, right? So as I think that the market perception in some ways, hasn't caught up with the reality and certainly not the opportunity, right?

But thinking of the reality of it is, we went from servicing customers paying us $500 a year to ones that are signing $15 million contracts and where that can head from there because, again, that an initial contract. That's not a total opportunity capture. That's going to land before and expand, right?

And so I think that the good news about having 0.5 million customers and being around for 20 years is there is a certain level of awareness for the company. The bad news is that, that awareness is -- the perception around that awareness is tied to the rearview mirror, not today and not the future. And I think that today, we already are providing a central infrastructure to a number of the old clouds. We already have 1 out of 3 new customer bookings coming from AI natives. That's today. And I think that, that only scratches the surface of where our opportunity is tomorrow.

And how do you think about managing basically 2 businesses, right, which is the incumbent portion of the business, computer backup and SMB, mid-market cloud storage. At the same time as you're also transitioning to that new cloud storage story, right? How do you manage the way you're positioning the way the Street is going to view the company? And what are your views whether or not you need to keep both?

Yes. So the computer backup business is still a great part of our business. It's a durable cash-generating great part of the business that also provides access for us to go to those people and sell our other services, right? So that's a good solid ongoing part of the business. B2 is the growth engine, right? And so the nice thing is we've got a good predictable business on that part. And frankly, we have a good predictable business on B2 in the use cases that we have served for a long time with ransomware protection, our cloud backup and the like. And then the AI opportunity with B2 is the inflection point and the inflection engine for us to drive business.

Question?

You touched earlier that your inventory that you have and the age of the infrastructure that you're dealing with. I think you recently announced that you extended the useful life of hard drive, I think, from 5 to 7 years. Just generally how you think about if there's a longer runway on this and the memory supply for you straight and all?

Yes. It's a good question. And certainly, the supply side of the business is interesting just globally for anybody in dealing with supply today. One thing that I would say is -- so your first question was just about the ages of the inventory. So we are regularly cycling inventory out, right? So obviously, hard drives that we bought 20 years ago are no longer in our system, right? And we migrate data on a somewhat continuous basis as the drives age out. We did that useful life study where we found that we were depreciating much faster than the actual lifespan of the drives.

And in fact, we published these very popular hard drive reliability statistics where we show what the failure rates of drives are over time. So in general, the drives still tend to last longer than what we have as useful life for them. So I think we're being conservative on what the reality there is, but it's more reflective than where we were in the past.

In terms of the memory side of things and the hard drive and just supply constraints, so the good news for us is that while supply is constrained across almost everything technology based today for infrastructure, hard drives are not as constrained or the prices are not as significantly skyrocketing as memory and some of the compute prices. And we do buy some of those other components also, but they don't make up the bulk of our CapEx. So in general, it's something we work very hard on to keep good relationships with multiple vendors and have longer purchases and have pre-commitments in the whole bit, but it's not as tight as if we were in some of the other spaces.

It's actually been one of the benefits, frankly, for us. Some of our customers have said they were previously buying and intended to buy more flash-based systems as a way of storing data. And because those systems and the underlying components are becoming increasingly expensive and an increasingly short supply, they are looking for ways to offload those systems and only keep what is absolutely necessary to keep on flash, but get a data lake for a layer for what they can, and we're a great offload to that.

Okay. Thank you, Gleb.

Thank you, Frantz. Appreciate it.

Ladies and gentlemen, thank you for standing by. My name is Abby, and I will be your conference operator today. At this time, I would like to welcome everyone to the Backblaze First Quarter 2026 Earnings Call. [Operator Instructions]

Thank you. And I would now like to turn the conference over to Mimi Kong, Head of Investor Relations. You may begin.

Thank you. Good afternoon, and welcome to Backblaze's First Quarter 2026 Earnings Call. On the call with me today are Gleb Budman, Co-Founder, CEO and Chairperson of the Board; and Marc Sudan, Chief Financial Officer.

Today, Backblaze will discuss the financial results that were distributed earlier. Statements on this call include forward-looking statements about our future financial results, the impact of our go-to-market transformation, sales and marketing initiatives, cost-saving initiatives, results from new features, the impact of price changes, our ability to compete effectively and manage our growth and our strategy to acquire new customers, retain and expand our business with existing customers. These statements are subject to risks and uncertainties that could cause actual results to differ materially, including those described in our risk factors that are included in our most recent quarterly report on Form 10-Q and our other financial filings.

You should not rely on our forward-looking statements as predictions of future events. All forward-looking statements that we make on this call are based on assumptions and beliefs as of today, and we undertake no obligation to update them, except as required by law.

Our discussion today will include non-GAAP financial measures. These non-GAAP measures should be considered in addition to and not as a substitute for our GAAP results. Reconciliation of GAAP to non-GAAP results may be found in our earnings release, which was furnished with our Form 8-K filed today with the SEC. You can also find a slide presentation related to our comments in the webcast, which will also be posted on our Investor Relations page after the call.

Please also see our press release or presentation for definitions of additional metrics such as NRR, gross customer retention rate and adjusted free cash flows. We will be participating in the Needham Technology, Media and Consumer Conference on May 12 in New York. I hope to see many of you there.

Thank you for joining us, and I would now like to turn the call over to Gleb.

Thank you, Mimi, and thank you, everyone, for joining us today. Q1 was a strong quarter. We beat revenue and adjusted EBITDA guidance, ending the quarter with $38.7 million in revenue, up 12% year-over-year with B2 growing 24%. We more than doubled our average sales deal size and drove 72% year-over-year growth in our $50,000-plus ARR cohort as we continue to move upmarket and are on track for our first full year of free cash flow positivity as a public company.

What excites me most about Q1 goes beyond the numbers. AI is making storage increasingly important, and our organization is gelling and executing better than ever to capture that opportunity. This is evidenced by more than 1/3 of all new bookings coming from AI and the number of AI customers using our platform growing by 76% year-over-year. We entered 2026 saying we would build a more scalable, more predictable growth engine that serves the AI opportunity. Q1 started to show what that looks like.

In AI, we are seeing demand from 2 parts of the market. One is companies building the infrastructure and tools that enable AI. The other is companies using that infrastructure to bring AI into products and workflows. We are winning in both. On the infrastructure side, there is a major replatforming happening in the market. For the first time in about 2 decades, the traditional hyperscalers are not the only place companies are building.

They're also building on the neoclouds. Synergy Research estimates that the neocloud market was $25 billion in 2025 and growing to about $400 billion by 2031. In order for these neoclouds to support their customers' AI workflows, they need to offer cloud storage. Some neoclouds have offered cloud storage built on Flash. It was fast and it worked. But as these platforms have scaled and AI workloads have grown, the economics have become increasingly difficult. Flash is now about 10 times more expensive per terabyte than hard drives. It works well for use cases requiring the lowest latency for smaller data sizes, but it becomes unsustainable at exabyte scale. As a result, neoclouds are now actively looking to introduce a cost-efficient hard drive tier time patch to manage both performance and economics across their infrastructure.

At Backblaze, we built an Internet scale file system to optimize performance per dollar out of hard drives and thus believe Backblaze is ideally positioned to provide exactly what these neoclouds need. We've seen support for that belief not only from the multiple signed neoclouds where we provide this for them already, but also the active engagement we're having with many of the top neoclouds. We estimate our opportunity to support neocloud at $14 billion by 2030. And with the success we're seeing, we are aligning resources internally behind that opportunity.

In addition to neocloud, we're seeing a significant opportunity for us supporting other AI infrastructure. For example, we are also seeing strong demand from companies supplying large data sets into the AI ecosystem because they need a place to store large data sets efficiently, but also be able to move them where they need to go rapidly.

One recent example is a train data provider serving AI use cases that selected B2 to store large volumes of video data. A hypergrowth company, it was experiencing rate limits and bandwidth constraints with its existing provider and needed a solution that could scale quickly. Backblaze won on both economics and technical fit. The deal closed in just 11 days at nearly $1 million of ARR, underscoring how quickly these companies move when infrastructure becomes a constraint and how well Backblaze is suited to the infrastructure side of the AI opportunity.

The other part of the AI market we're seeing is companies using infrastructure like ours to bring AI into their products. As AI models move from text to multimodal, incorporating video, audio and images, the volume of data required to train and run those models grows by orders of magnitude. This is not a future trend. It's happening now and is creating significant and growing need for storage that can handle it economically and at scale.

With the Generative AI customers we have today, we are finding that price and performance get us in the door. But it is the experience that keeps them and grows them, transparent pricing, responsive support and a team that works with them rather than just selling to them. These customers are scaling fast, and they do not have time to manage infrastructure problems. With Backblaze, they don't have to.

A good example from Q1 is an AI-powered video creation company that selected B2 to store data used to train its models. The customer had been running into cost and performance issues with its existing provider. The platform was difficult to manage and the economics were not working at its scale. Backblaze offered the best performance per dollar and a platform that was easy to use and easy to scale. The initial deployment represents nearly $0.5 million of ARR and creates a clear path to expand into higher-performance workloads over time. These customer wins are just examples of where we won in Q1 and are reflective of opportunity we have in pipeline going forward.

It's clear that whether customers are building AI infrastructure or using AI in their products, they are scaling fast, their data is growing exponentially and they need infrastructure that is performant, open and cost efficient at scale. That is the moat we have spent 19 years building and AI is making it more valuable, not less.

To be the leading storage platform for AI, we are also meeting developers where they already work. We are embedding Backblaze into the AI ecosystem by integrating directly into the tools developers already use. For Hugging Face, which has 13 million users and over 2 million models, we shipped a tool that lets teams store and share model caches on B2. For ComfyUI, which recently raised at a $500 million valuation, we built a plug-in to support Generative AI workflows. For CVAT, which is used by tens of thousands of computer vision teams, B2 is now integrated as a back end for training data. And for MLflow, the most downloaded tool for taking AI projects from lab to production with 16 million monthly downloads, B2 has now been added as an integrated artifact store.

So the AI opportunity is making what we do increasingly critical. We're also stepping up to meet it. One year ago, we began a meaningful transformation of our go-to-market organization focused on 3 things: increasing awareness; driving greater pipeline consistency; and expanding revenue within our installed base. In Q1, we delivered progress on all 3.

On awareness, the Flamethrower start-up program is gaining real traction. We have now welcomed approximately 100 companies in under 3 months, half the time it would typically take. We've been added to the a16z Founder Resource Program, the Launch Startup Showcase, and the Startup Grind conference, all of which expand our reach with venture-backed start-ups.

On pipeline consistency, we have completed our core go-to-market systems upgrade, giving our team better visibility and a stronger foundation for a faster, more disciplined revenue motion. And within our installed base, pipeline sourced from existing customers has nearly doubled year-over-year, reflecting our growing ability to land and expand with our customers.

To accelerate this next phase, we welcomed Anuj Kumar as our Chief Revenue Officer. Anuj has scaled go-to-market for cloud infrastructure and enterprise storage at NetApp, VMware, Red Hat and SUSE. He brings the pipeline discipline and execution rigor this phase of our growth requires, and we believe his leadership will be a meaningful complement to the upmarket momentum we have already built.

We also saw encouraging new customer momentum during the quarter across a range of data-intensive use cases. That included a health care data company who selected us for disaster recovery, a cloud gaming platform that chose B2 to store video across multi-cloud environments, and an audio streaming platform migrating from self-managed infrastructure to B2. These wins reinforce a broader point. Backblaze is winning where data is valuable, active and operationally important.

And this is why I am excited about the opportunity ahead. The shift to multimodal AI is driving exponential data growth and the need for high-performance, yet cost-efficient storage has never been greater. The customers who are choosing Backblaze are exactly the kinds of customers that compound with us over time. We are stepping up to this opportunity with an up-level team, a go-to-market transformation well underway and a platform we have spent nearly 2 decades building and optimizing. AI is making everything we have built more valuable, and we are becoming the storage infrastructure that powers the AI economy.

With that, I'll turn it over to Marc.

Thank you, Gleb, and good afternoon, everybody. Our first quarter results reflect the strategy that we have been executing against. We exceeded the top end of both revenue and adjusted EBITDA guidance. The Q1 outperformance reflects stronger sales execution and the EBITDA beat demonstrates the operating leverage in the model.

Let me walk through the quarter and then cover our outlook. We finished Q1 with revenue of $38.7 million, above the high end of our guidance of $38 million. The beat was broad-based across both B2 Cloud Storage and Computer Backup, with B2 remaining the primary growth driver. B2 Cloud Storage grew 24% year-over-year to $22.4 million and ARR grew 28% year-over-year, reflecting the underlying strength and momentum of the business. The Q1 revenue outperformance was driven by higher customer data consumption on the B2 Cloud platform and Computer Backup coming in slightly more favorable than our forecasted decline.

On bookings, which primarily affect revenue in future quarters, we closed multiple large deals for a strong quarter. We made several updates this quarter to improve the calculations of our ARR and RPO metrics. I will briefly walk through those changes as I cover the results.

ARR increased by more than $5 million sequentially to $158 million with B2 growing 28% year-over-year. This quarter, we updated our ARR methodology to improve comparability across periods, and the change is defined in the earnings presentation posted on our Investor Relations website. Under both the new and previous methods, the sequential ARR improvement is approximately $5 million. We ended the quarter with 187 customers contributing over $50,000 in ARR, up 51% from a year ago, reflecting continued strong progress upmarket.

We also updated our RPO methodology this quarter and described the change in our earnings presentation. The change is aligned to our peer group and RPO is now a more important metric as we continue to move upmarket, signing both annual and multiyear customer commitments. Under the updated methodology, RPO increased by $6 million sequentially and by $31 million from the prior year period. Our gross customer retention metrics remain very healthy with customers continuing to use both our B2 and Computer Backup solutions for 9 years on average.

Beginning this quarter, our reported net revenue retention reflects an in-quarter methodology, which we believe provides a more current view of our customer expansion and retention trends. In B2, net revenue retention was 110%, up from 105% a year ago, reflecting continued expansion within the customer base. As a consumption business, B2 benefits from both the organic customer data growth and the cross-sell, upsell sales motion.

Q1 gross margin was 61% versus 56% in the prior year. The year-over-year improvement shows strong operating leverage continuing to kick in as we tightly manage costs and also from the extension of the useful life of our fixed assets. Total operating expenses were $29 million in Q1, roughly flat compared to Q4 and improved by approximately 600 basis points from the prior year as a percentage of revenue, reflecting strong operating leverage as we maintain our focus on cost management.

Q1 adjusted EBITDA was $10 million or 26% of margin, up from $6 million or 18% in the prior year, reflecting strong operating leverage as revenue scales. Sequentially, margin declined modestly from 28% in Q4, primarily reflecting the onetime benefits we referenced in our last earnings call.

Adjusted free cash flow was negative $1.8 million in Q1, reflecting earlier payments in the quarter. We are also pulling forward a portion of 2027 CapEx into 2026 in response to strong demand signals. Even with that pull forward, we continue to expect adjusted free cash flow to be positive for the full year with improvement weighted towards the second half of the year.

We have the capital in place to support the growth that we are seeing. We currently have more than $100 million in capital leasing capacity with approximately half of that utilized. Based on our current operating plan, we expect to fund growth through operating cash flow and capital leases, and we do not anticipate the need to raise additional capital through follow-on equity offerings. In fact, we plan to continue to focus on reducing our dilution through our modest stock buyback and our net share settlements for RSU brands.

Looking ahead, we introduced updated B2 pricing and packaging effective May 1. The change reflects the investments that we have made in our platform performance, our effort to further simplify pricing by removing API transaction fees and the rising cost of hardware and data centers. On a net basis, we expect the pricing update to be accretive to revenue and margins, and that will be reflected in our guidance.

So moving on to our guidance. For the second quarter, we expect revenue to be in the range of $39.8 million to $40.2 million. On our last earnings call, we said B2 growth in the second quarter would be 12%. Based on this new midpoint, the B2 growth in Q2 will be closer to 20%, which is a big improvement. The Q2 outlook includes a partial quarter benefit from the May 1 pricing update, along with variable usage from customers that we have already actualized in April. We are not assuming the same level of variable usage in the second half of the year. Adjusted EBITDA margin is expected to be in the range of 21% to 23% for Q2. The sequential step down from Q1 reflects the timing of investments as we continue to build for growth.

Turning to the full year. We are raising our full year revenue guidance to $161.5 million to $163.5 million, up $5 million from our prior midpoint of $157.5 million. That increase reflects 2 factors: stronger first quarter performance impacting the rest of 2026; and the benefit of the new B2 pricing and offering, each contributes to approximately half of the raise. We are also raising our full year adjusted EBITDA margin guidance by 400 basis points to a range of 23% to 25%, up from 19% to 21% previously.

As a reminder, our guidance philosophy excludes individual deals greater than $500,000, high variable usage above contracted minimum and incremental upside from our go-to-market transformation. As these elements become more predictable and repeatable, we will incorporate them into our forward guide and communicate that transition clearly.

In summary, Q1 was a strong quarter across the board. Revenue beat, adjusted EBITDA beat, B2 growth accelerating and bookings improving. We remain focused on executing on our AI opportunity by driving forward our go-to-market transformation and scaling our B2 business.

We look forward to your questions. With that, operator, please open up the line.

[Operator Instructions] Our first question comes from the line of Mike Cikos with Needham.

Congratulations on the strong start to calendar '26. First question, I guess, is more for Gleb. But I just wanted to get more on the success that you guys are seeing with the AI customers following the go-to-market transformation initiatives we put in place. Can you just talk to the improved visibility you have for those AI customers in the pipe? And as you have more of these customers, I guess, begin to season, are you noticing -- is there a significant departure as far as cohort behavior or sales cycles? And then I just have a follow-up.

Mike, thanks for the question. I'll use actually the 2 customers that I referenced in my prepared remarks as a good example. So one of the customers came to us through the GTM motion that we're building, right? So the machine that we're building, the combination of outbound targeting, better systems, going to the AI events. And so we found them through that outbound process. The other one actually came to us as a referral from one of our existing AI customers who said that they were having a great experience. And specifically, they were saying that the combination of the performance that they were getting from our platform at the price point that they were getting was unmatched. And so they were -- referred them over.

And so we're seeing more AI companies coming to us. We kind of feel like the market is coming our way, and it's really from both of these. Part of it is from the work that we're doing, part of it is from the referrals in the market coming to us. So that -- maybe that answers the first part of that.

In terms of the cohort part of it, maybe you can answer question if I didn't completely get it. But one of the things we mentioned on the prior call is that we're seeing the AI companies growing much faster, about 3 times faster than our average customer, and that's just a function of their inherent data growth driven by their AI use cases. Did that answer the question you were asking?

It does. It does. And then for the follow-up, I think it might be more geared towards Marc, but I just wanted to double check on the B2 with the NRR of 110%, I know you said, hey, we drive that between 2 factors, right? You have the data consumption which grows each year, and then you also have the cross-sell, upsell. And I just wanted to see, can we unpack that a little bit more to get evidence of the go-to-market actually driving adoption, whether it is the cross-sell, upsell motion? What are the drivers behind that B2 NRR today if I'm trying to unpack consumption growth versus go-to-market initiatives to expand wallet and drive additional offerings into the installed base?

Yes. Mike, I'll start off by saying I think the best evidence of the GTM working is the RPO disclosure of committed contracts that change quarter-over-quarter. So for commitments of less than a year, it's up $3.4 million. You could see it on Slide 19 of our earnings deck. So I think that's the best evidence of the performance. Now that is made up of both, new logo as well as expansion sale.

The expansion sale realistically does fluctuate. On the NRR, we did move to in-quarter reporting versus the trailing 4-quarter average, specifically to give you more visibility and to hold us accountable to explain what's happening. So there will be more fluctuation there from that perspective. So it's up to 110% from 105% a year ago because a year ago was a quarter where we did talk about one customer -- one large customer going away. I mean, since I joined, that was the only time we've had to reference that. So that's what drove that improvement year-over-year. But it generally fluctuates around 110% on a stable basis, but there's going to be some ups and downs and the expansion changes will be the biggest driver of that because the organic growth tends to be incredibly stable and predictable.

Congrats again on a strong start to the year.

And our next question comes from the line of Ittai Kidron with Oppenheimer.

Congrats, great solid numbers. I had a couple of things, maybe starting with you, Marc. Can you be a little bit more -- can you give us a little bit more color on the pricing update, the magnitude of this? How much of this you think you can capture? I'm just trying to think about your growth without the pricing update, how would you -- your outlook would have looked without it? I'm just trying to get my hands around that.

Yes, absolutely, Ittai. So in the $5 million raise for the year, half of it is from the pricing and packaging change. Half of it is from the strength of the business that we observed in Q1, so the booking -- the strong bookings in Q1. So if you look at that RPO number I referenced just a few moments ago, that kind of roughly equates to the raise from the organic health of the business for the rest of the year because that has no price increase in it. We continue to guide very prudently for the rest of the year. So the same philosophy we laid out last time, which is no large customers, no go-to-market benefits and not accounting for large -- variability of the -- of that large customer.

What I would say also within Q2, I can give you a bit more color there, last time, we said Q2 would grow by -- for B2, we grow by 12% year-over-year. Now it's 20%. That difference is more anchored on the organic health of the business because the price change took effect May 1. So it's not a full quarter. And we obviously actualize some of the things we saw in April in the business.

And on the price, I mean, we could elaborate a bit more on that price change. It's not a flat price change. It's a pricing and packaging change. So for instance, we are including now transaction API fees. So in the spirit of being the simplest billing model out there, we further simplified by no longer billing customers for transaction fees.

Got it. Okay. And then as a follow-up, maybe one for each of you. Marc, for you on the Computer Backup, the net retention rate is now well below 100%. So is this a model -- is this a business we should model towards decline or going forward? And for you, Gleb, on the go-to-market side, great to see the progress there. What else is left here? What is it that between now and year-end still needs to kick in, that hasn't from your perspective?

Yes. So Ittai, just to reiterate, the -- we're still thinking of Computer Backup as declining year-over-year 5%. And the NRR is going to be tightly tied to that because it's a subscription business, not consumptive, which would mean that B2 would grow 24% year-over-year. So the change in outlook is pretty much all on B2 and Computer Backup remains at a decline of 5% is what we're forecasting and guiding.

[indiscernible], thanks for the question about the GTM transformation, what's done, where -- and what we have to do still. So what I'll say is I think we've made great progress this quarter, and there's still a variety of things that we want to get further, right? So we hired Anuj Kumar to run that organization. I've asked Jason to -- who's with us to take on and focus most of the time on the neocloud opportunity. So Jason works for Anuj, and we see that as a $14 billion opportunity. So we're putting focus and resources on that specific part of the opportunity with Jason focusing on that.

The awareness generation is off to a good start with Flamethrower, but it's only been a couple of months in. And -- so we've been moving faster than, I think, expected on that, and we've been invited to participate in some great organizations and partnerships with a16z and Startup Grind and Launch. But it's -- there's a lot of opportunity there still between that and the open source developer efforts that we're doing.

There's still a lot of opportunity to make sure that everyone thinks of Backblaze as their first spot for their price performance storage. So there's a lot that we've done. There's still, I think, a lot of opportunity that we have. There's -- I am excited that we're seeing pipeline growth stronger than we've seen in the past. We're seeing more of our sales team hitting their quota than we've ever seen in the past. So a lot of the right things are happening, but we still -- we're always going to keep working on it.

And our next question comes from the line of Erik Suppiger with B. Riley Securities.

Congrats on a good quarter. Can you speak to what portion of the neocloud market you're either servicing or at least engaged with? And then where are you in terms of the hiring on the sales front? Are you adding additional salespeople at this point? Or where are you on that -- from that perspective?

Yes. Thanks, Erik. So there are about 200 neoclouds. We are -- we went to GTC, the NVIDIA's premier conference and had just a host of great conversations there at GTC -- there's some noise on the line. So the -- what I'd say is we're engaged with most of the top neoclouds as part of it. The part that we are servicing for them is this data lake layer, right?

So if you think of the AI workflow, the GPUs themselves, there's the very low latency, high-performance Flash that you want adjacent to the GPUs. And then what you need is the place where you store all of the data, right? So you can almost think of it -- if the whole AI workflow was a laptop, you've got your compute, your CPU, you've got the RAM and you've got the hard disk or SSD. We are basically providing that hard disk layer. There's about half a dozen companies that provide that RAM layer. And then the base neocloud part is that CPU, GPU part.

So we're providing that large scale, high performance, not the highest performance, but high performance per dollar data lake layer for them. And so we're the -- we're a white labeled provider for them. We're doing that, as we talked about in the last call, we've got the 6-, 7- and 8-figure deals that we signed for that. We have others that are in the works, and we're engaged with a bunch of the neoclouds at this point.

Do you -- Can you give us a sense of what portion of the neoclouds out there -- of the 200 that are out there that you're speaking to? Do you think it's a quarter? Any gauge on what penetration you've had?

So I think in terms of the conversations and engagement side, probably somewhere around that number. But I would say we're engaged with pretty much all of the top ones at this point and having different levels of conversations and some in POCs, et cetera, with them. On the sales side of it, we talked earlier in this year that there were a number of different roles we wanted to fill. At this point, I'm excited to say we filled the CRO role with Anuj, we filled the Rev Ops role. We filled the sales development role. And so we're -- we've got a really strong kind of build-out of that team now.

And the next question comes from the line of Jeff Van Rhee with Craig-Hallum.

A couple. First, just maybe, Marc, help me with the guide and the outlook. I'm trying to understand the progression here. So in -- at the end of February, Feb '24, you took roughly $4 million out relative to the consensus, and now we're putting $5 million back in. I'm trying to understand, in the Feb '24 call, was the May 1 price increase in B2 already contemplated in the guide?

Jeff, no, that was not contemplated in the guide. And in the $5 million increase we just did, half would be from the pricing, half would be from the organic momentum and health of the business we saw in Q1. And the change is really -- a lot of this is guidance philosophy we spoke about, just a lot more prudent going forward. That's what drove the change.

Did you -- if you take the final month of the quarter, March and then April, was there -- I don't know if radical is the right word to use, but did you see substantial improvement in close rate? Because it sounds like you're saying your conviction is coming both from improved bookings as well as usage. So I'm trying to understand how Jan, Feb bookings were weakish and then all of a sudden, March, April really killed it? I know you've made some process change over time to sales, but it was just such a quick snap. Is that -- maybe you can just help me dial it in there a little bit?

Yes, Jeffrey Van, this is [indiscernible]. Maybe I'll touch on and then Marc can also weigh in. So we certainly had a more back-ended quarter in Q1, and we've started off Q2 strong. So there's definitely [indiscernible] feel from the numbers that we're seeing, right? And then I think -- and we talked about like the $1 million-ish deal that closed in 11 days, that started and closed towards the end of the quarter. But it wasn't the only deal, right? The pipeline itself has been building strongly this April. And I think we're layering that on along with the execution that we're doing on our own side. So I think that that's kind of the, I guess, the conviction and emotion side of things based on the data and the execution. And then I'll let Mark, if you want to add anything on -- beyond that on the guide side of things.

Yes. I mean Q4 bookings -- back in Q4, bookings were good, but we wanted to hit that 30% growth, Jeff. To hit 30% growth, we'd have to be booking like $5 million a quarter. okay? So we weren't at that rate-rate yet, but it's been improving pretty much every quarter. And this latest Q1 is a further improvement and probably the closest we've gotten, frankly.

And the demand signals are really strong. So the demand signals being really strong. We're feeling good about the outlook, but we're still guiding with that prudence. And we'll use some of that price change to also fund some additional CapEx, so we could have further capacity in place to handle that demand because we don't want to be in a position where we're declining any revenue opportunities.

Yes. Yes. Got it. Got it. And just to follow up on that last piece then in terms of the outlook for the year for CapEx for '26, I heard you referenced it, but can you just give us a number there, what are you expecting? And then also on the stock comp?

Yes. On the CapEx side, we're probably going to be around mid-30s as a percent of revenue. I would say there's 3 factors there. One, last quarter, we spoke about that large customer we got to service next year. So we need to get that CapEx in place now. Two, all the strong demand signal. And three, the general equipment cost is 30% higher than it was on a per unit basis from a year ago. So for those 3 factors, we're beefing up our CapEx plan for this year, accelerating it from '27 into this year.

Yes. And your thoughts on stock comp?

I'd say pretty stable. If you look at our headcount, I mean, generally speaking, year-over-year, our headcount is actually coming down. So we're continuing to drive more efficiency out of the business. So stock comp should be pretty stable in dollar terms. And as a percent of revenue, it does improve over time.

Jeff, the other thing -- one thing I would also just mention since you bring up supply chain and supply chain constraints and all that. What's interesting is we have to buy the equipment, right? So we have to spend more on some of that side of things. But the interesting thing is also we get 2 tailwinds from the supply chain being constrained.

On the GPU side, when -- because the supply chain is constrained on the GPU side, customers are saying, "Well, I need to go and have access to wherever the GPUs are available." And so we regularly talk with customers who say, "I have to have my data somewhere, that I can send it to whichever neocloud has the GPUs available." And on the memory side, which is also obviously heavily constrained, the neoclouds that offer cloud storage have been building out often on Flash, and that becomes really expensive, especially now with the constraints there. And so it's driving additional interest from the neoclouds in working with us on that data lake here. So on the one hand, we have to deal with prebuying ourselves on the equipment side for CapEx. But on the other side, we have these 2 tailwinds to the business.

That's helpful. Congrats...

Jeff, just to step back on stock comp. I mean, if you look at the statement of cash flows, Q1, obviously, stock comp is higher as we settle some of our annual bonuses in equity as well. But you'll notice this year's stock comp was actually lower than last year's.

Congrats on the turn, guys.

And our next question comes from the line of Jason Ader with William Blair.

Just wanted to get a better sense on the neocloud. What are the size of some of these deals? I know you talked about the 8-figure deal that's coming in, I believe, next year. But maybe just some more detail on some of the other deals that you've landed or are in the pipeline? Are we talking about kind of household neocloud names that are contracting with you for potentially further kind of 8-figure deals? I mean, just, I think gauging kind of how significant an impact you might have from some of these neocloud opportunities would be helpful.

Yes. Thanks, Jason. So first of all, the -- we estimate that our opportunity in the neocloud market by 2030 is $14 billion. And that is just the data lake tier that we provide, right? So that's not the entire storage footprint. The deals that we have signed, the 6-, 7- and 8-figure deals that we signed, I'll say 2 things. One is you would recognize them, right? They are companies that you would know. And two is that all 3 of those are initial deals. So all 3 of them are ones where companies -- the companies look at it as the way to start, not the total opportunity. So I think there's -- frankly, I can -- I could see a path where the 6- and 7-figure deals could become 8-figure deals themselves. The 8-figure deal can certainly scale from where it is once it's ramped. So that's kind of a little bit of that side of the opportunity.

The other conversations that we're in, many of them are -- assuming they move forward, are of that same scale. Some of the conversations are -- we may want to start with a 6-figure or 7-figure deal, but many of them, the scale of the opportunity is 8 figures at ramp.

Okay. Helpful. And then just on the -- I guess, the risk potentially that the neoclouds add a lower cost storage tier and then you guys are helping them for a little bit, but then they kind of in-source it?

I mean it's always possible, but it's a little bit like -- for the first almost 2 decades of Backblaze, one of the questions we were always asked was what happens if AWS ends up lowering their price to match you? And we're 2 decades in and that hasn't happened. I think that the challenge is it's not easy to build the type of IP that we have built up over the last 2 decades. It requires scale and expertise and a focus over a long period of time to get it really honed and right.

And the thing for the neoclouds is they have a lot of things they need to do, right? There's opportunities around GPUs and GPU scaling and optimization and how do you make tooling better for inferencing and all kinds of things. Spending all their resources to try to replicate what we have built over the last 2 decades is probably not the best place for them to invest their own resources when they can -- when time to value is so much faster by using Backblaze.

Okay. Great. And then, Marc, for you, just a couple of quick ones. So the -- with the higher CapEx, are you still guiding for free cash flow positivity this year?

Yes. So the second half of the year, we're still guiding for that to be free cash flow positive. For the whole year, I mean, Q1 was minus $1.8 million. Q2 should be somewhere around neutral and the second half of the year should be positive. So net-net for the year, we should be neutral or very 1% of revenue, free cash flow positive despite the acceleration of the CapEx.

Okay. Great. And then just on the gross margin, just last question for me, sorry. It's sort of -- as I look at the last few years, I'm looking at just the -- not the adjusted gross margin, but the reported non-GAAP gross margin. It was like mid-50s for a few years. And then last year was 62% roughly, 62% in Q1. Can you just remind us of like what caused the kind of the significant increase in the gross margin and then maybe some puts and takes going forward on that gross margin line?

Yes. Sure, Jason. So if you recall about 1 year ago, we reviewed the estimated useful life of our fixed assets. And it turns out we're using all our fixed assets for typically 6 years onwards. So we moved all the depreciation to 6 years. So that drove a big benefit to gross margin.

Second, all other lines, like if you think about all the labor or payment fees or everything else that fits to our cost of sales, we've managed really tightly year-over-year. So it's kind of staying flat in absolute dollars pretty much and improved as a percent of revenue. We're looking at everything within our gross margin now to further drive optimization. But I would say between the price increase which benefits gross margin, but the accelerated CapEx, which will push our gross margin down, it should stay flat around where it is now. We're not seeing -- we're not guiding to any major changes in gross margin through the rest of the year.

And our next question comes from the line of Eric Martinuzzi with Lake Street Capital Markets.

Yes. I was just curious about the timing of the price increase. I went back and looked it up, I guess it was, October of 2023 was the last time you raised the price on B2, and you really hadn't raised it since you rolled out the product back in 2015. So we're at about the 2.5-year mark here with the price increase. Was this something that you felt like, hey, we're delivering more value, we need to capture more value? Or was there competitive issues where competitors were raising price and kind of provided an umbrella for you to do the same?

Yes. Thanks, Eric, for the question. So we periodically reevaluate what the pricing and packaging of the offering should be. When we were looking at it, there were a few things that came together. One is that we've been investing more into the performance of the platform. More of our customers are using us in these hot use cases where we're driving high throughput, high IOPs. We've been -- we made egress free before. And one of the things that, that's enabled is not just that it's less expensive for the customers, but it allows them to actually run more frequent training of their models in the AI use cases. So it's actually unlocking their ability to innovate, but that makes it -- it's free for them. It costs us money to provide that.

So we've been working to increasingly provide more and more value to these higher performance, more active use cases. And we also wanted to simplify the pricing by removing transaction fees. So the pricing and packaging combination along with, as Marc said, the underlying costs of the components, have been increasing. So taking all of that together, we decided this was the right time to do that.

Okay. And with regard to competitors, do -- I mean, historically, you guys have thrown out that, hey, we're 80% cheaper than Amazon. Does this -- obviously, you're raising, I think what I said was about a 15%, 16% per terabyte per month. Does that shrink that gap now? Or do you still feel like there's a big delta?

There's -- we are still dramatically more cost efficient than the alternatives out there. I was literally actually just talking to one of our account execs a couple of days ago, who was talking about a customer who has been ramping on our platform. And they said that they moved over a lot of their data and they're continuing to move over more of their data, more of their use cases because on the one hand, we're more affordable on the storage side, right? So just at the base level of storage. But where they were getting hit dramatically at their prior provider was that each time they egress the data out from their provider to one of the other neocloud providers, they were getting hit with massive egress fees, one. And two, the transaction fees were actually costing them 3 to 4 times more than the cost of the storage at their prior provider.

So when you put it all together, they were more than -- 5 times more expensive at their prior provider, and they were literally wondering whether that was going to even be affordable for them to stay in business. So the scale of total cost of ownership that we provide on a benefit basis is still quite dramatic.

And our final question comes from the line of Rustam Kanga with Citizens.

Nice, clean set of results here. Just one on B2 neo. As workloads begin to shift more towards inferencing from training, will that lead to improving predictability and visibility? And then to that end, could you potentially share what percentage of neo business on average or even directionally represents inference versus training workloads?

Sure. It's a good question. And the first part of the answer is yes. As things move towards inferencing, it does make it easier to be more predictable. Today, more of the use cases that we're seeing are related to model building. And that makes sense because a lot of the data sets right now that are very large and that need to be moved, are related to the model building, and we're a great service for that.

But I'll give you an example. In the Gen AI media space, I was talking about -- to a customer about their data flow. And the data flow is they accumulate a lot of data. They store that data, they annotate that data, then they find a GPU provider that is available and then they run iterative model building on the different GPU price. So they use us to store that large data set. And as they acquire the data sets, they use us to store more of those large data sets. They love the fact that they can store those efficiently and then send them quickly and for free to the GPU provider that they want. So they're using us in this whole model building process.

Now as they do that, the other side of their business, the actual thing that they offer and they charge for is generating videos. That's all the inferencing side. And so they're looking at us for the outputs of all that video because every single time a user generates a new video, that video then gets stored basically forever and each version and each iteration gets stored forever. And so we become a great place to store that. And that inferencing side is a much more smooth and predictable side.

So the short answer is, yes, it will be more predictable as we get more inferencing. Today, the bigger workloads that we see are related to model building because we're great for that. But we are seeing more inferencing start-up on our platform.

And that concludes our question-and-answer session. I will now turn the conference back over to Gleb Budman for closing remarks.

Thank you, everybody. Q1 was a proof point. We beat on revenue, beat on EBITDA, B2 is growing 24%. The deal size has more than doubled. The AI customer is up 76% year-over-year. We are not just riding the AI wave, we're building the infrastructure that supports it. We are key for the neo clouds, key for the AI builders, and we've had nearly 2 decades of optimizing performance per dollar at scale, which makes us ideal for the needs of AI. We raised guidance. We're on track for our first full year of free cash flow positivity as a public company, and we're picking up steam. Thank you to our customers, our partners, and thank you to our amazing team that's making all this happen.

Thanks for joining our Q1 call, and we look forward to connecting on the next one. Bye-bye.

And ladies and gentlemen, this concludes today's call, and we thank you for your participation. You may now disconnect.

Hi, everyone, and welcome back to Network Stats where we track quarter-on-quarter a bunch of different metrics and heatmaps and all sorts of fun stuff about the network traffic happening at Backblaze.

I'll take a minute to introduce ourselves here. I'm Stephanie Doyle. I'm the Technical Narrative Content Manager and lovingly called the keeper of stats since I'm on many of these webinars. And I'll let Brent go ahead and talk about himself.

Hi. Hello. I'm Brent Nowak. I'm the Manager of Network Engineering here at Backblaze. Our group is responsible for the connectivity inside the data center, so all the copper, the fiber and then also the connectivity external. So all of our Internet connections, our IX connections and also PNIs to our various partners.

Definitely, a mighty team there. You guys do a lot of great work. So just to talk about the agenda today, we're going to review this quarter's highlights and insights, talk about some of the quarterly data that we're tracking, take a few questions if we have the time. We've got some resources. I'll also be putting attachments on the webinar page itself after this.

And just a couple of little housekeeping things. This is being recorded. Please do ask us questions. We'll try and take some time at the end, but you can also drop them in the chat. We are monitoring it. And of course, don't miss the attachments.

All right. Highlights this time around. So Brent, I think we saw some pretty interesting things, and we actually transformed the data in some new and interesting ways, too. So...

Yes. Our data set historically for the last quarter showed us who we're talking to, which is the networks, and then what and when we were looking at the TCP conversations, the length of the TCP conversations and the size, how many bits were transferred. What we added this quarter was the where. So we added some geo information that allows us to know where certain types of traffic are going. And it led to some really interesting results that we're going to go over.

Yes. Great. So let's jump in. This is sort of our full picture here. Take it away, Brent, tell us what we're looking at.

What we're looking at is a total amount of traffic that we have sent and received through the Backblaze network per month. What we use here for a metric is called bits 95th. So this is the 95th percentile of the traffic. This is a common metric that we use in the ISP world to measure traffic. It is a better approximation than using average or mean because the 95th is the value of 95th that lies where 95% of the numbers fall below and 5% fall above. It's a really good way to reduce your outliers and get a better picture of what we look at from a traffic perspective.

And this is a month-over-month view. So we've got history going back from May of 2025. We saw a large amount of traffic that we were sending into the winter season. We saw a little decrease in December, January and an uptick again in February and March for total traffic. And each colored slice here represents a different type of network classification. We have our CDN, our hosting, hyperscalers and neoclouds are very interesting for us to monitor. And then also our ISP connections such as Tier 1, which are global reach partners and then regional and ISP regionals in this darker purple color.

And I think what we saw defined here and elsewhere in the report, but we might as well introduce this term now, Brent, is really the prevalence of elephant flows as we're talking about our network these days. So if you want to describe for folks who are unfamiliar the term.

Sure. In the networking world, when we talk about traffic, there's 2 kind of colloquial ways to classify traffic. We call them mice flows, and we call them elephant flows. And they're very descriptive towards traffic. A mice flow is sort of a small amount of information, maybe a kilobyte, a megabyte, 10 megabytes of a file being transferred here or there. There's many different participants. There's maybe 1,000 of these. For example, when you go to a website, you load many different small assets from many different locations.

What's moving and driving a lot of innovation at Backblaze is what we are defining as elephant flows. And these are large single transfers between 2 parties. These could be at line rates of gigabit, 10 gigabit or even higher. And when we see partners try to scale their workflows, they're sending multiple of these elephant flows, which can add up to 100 gigabit, 400 gigabit in total aggregate of connectivity between 2 partners.

And I think it's interesting because we see it showing up in different ways through many of the graphs. So I'll let them speak for themselves, but you'll hear us using words like bursty or how we're tracking for these things. And you can see that you've got spikes and then you also still have a higher baseline as well. So there's an interesting traffic move that's happening here. So these are our Sankey charts. And these track, if I'm not mistaken, traffic to different workloads. Is that correct, Brent?

This is a way for us to visualize how we are transferring our data over various connectivity types. For us on the network engineering team, it's cheaper for us to transmit traffic over PNIs. These are often 0 settlement costs for us. So if we have partners that are geographically located close to us, we like to initiate conversations to see if we can have a fiber run to them, which allows us, again, to have very cheap transit. We also have connectivity over different transit networks such as our ISPs, and we also have what we classify to our cloud partners, where we deliver CDN content.

And this is a sample for Q1, and it shows the total aggregate traffic, the same as that previous graph, but just in a different way to view kind of how it's transferred via the different transport methods.

And just for reference, we went ahead and pulled up Q4 2025 as well just because we did see a shift in these slices. One thing you'll find us harping on throughout the reporting is just that it's really hard to define patterns when the data set is young. So we try to keep things in a description space. But to that end, you can see the difference here between Q1 2026 and Q4 2025. So let's talk magnitude here. I believe that's where we're at.

Yes. This is the amount of total traffic that we've sent over our different types of networks. based on the different regions that we have. And when we start to add this regionality information, us being Backblaze employees, we kind of have a little more history. We understand why this layout looks the way it is. But just a little insight here. Our U.S. West infrastructure was our first deployment, and that has the most amount of historical content. And that shows up here where we see a heatmap in the U.S. West CDN, very colored, very deeply red. And additionally, we also see that in the ISP regional traffic.

This makes sense for us from a business perspective because U.S. West was our original deployment for the Backblaze network. As we expanded, we also added EU Central and EU West, and we're starting to see a more heatmap concentration where we have neocloud activity in those different regions, ISP regional traffic, hyperscaler and also CDN traffic. And interesting here, CDN traffic is pretty well spread for us across our U.S. East, U.S. central locations or EU central locations. So this is a very interesting graph that lets us know how the amount of total information is transferred over our network by region.

Now we're on magnitude. So that's this one here.

So magnitude for us is a metric that we coined. This is a measurement of the amount of bits transferred per IP address. So rather than looking at the total amount of traffic, we're adding a 2-dimensional metric here, where it's the amount of information transferred per speaker. And that to us has been very insightful for us. You can see here that it's very different to the amount of total traffic, that orange graph previously. We see a very deep green color concentration in the U.S. East for neocloud. And this is a location for us where we see a lot of neocloud activity in our U.S. East cluster. This is because of the geographic location of a lot of hyperscalers, GPU providers, and it just makes sense, and it's really good to have data that sort of validates what we're seeing on the business side.

Totally agree. And let's talk about unique addresses and how that differs a little bit.

Much like the total traffic graph where I spoke to U.S. West being our most original and the first implementation of the Backblaze network. This graph also is very boring, but it helps validate our assumptions here about how we run our business. A lot of the content in our U.S. West clusters is sent over ISP regional traffic. So this is traffic to Comcast, Verizon, Google Fiber, and it shows up here where we talk to many, many, many different unique IP addresses out of U.S. West, mainly on ISP regional networks.

So for us, this drives decisions on where we want to put connectivity. It means that in our U.S. East locations, we may want to augment with higher gigabit ports, whereas in U.S. West, we want to partner with more Internet exchanges to get more local to consumers.

Makes total sense. And that brings us to a really fun question, where in the world is the neocloud? So I found your explorations here, Brent, to be very cool this time around.

So we did add geo data to our data set for this series. And when I started to look into the data set and produce some heatmaps of where we are sending traffic by country, you can see here the United States for us is very deeply shaded. So that's deeply shaded for neocloud activity for hyperscaler activity and also CDN content. And what I -- go ahead.

Yes. We call out in the report that that's somewhat unsurprising because what is it, 40% to 60% of data centers are located in the U.S. as of right now, I believe, is the common metric. So you might see this on any network provider, really...

Right?

Yes. But now we get a little more granular.

The next question I asked once I saw the first heatmap was, if we exclude the U.S. data, what does the heatmap look like? Because this will give us more differentiated results on a per country basis, and it definitely shows in this heatmap. So what we see is concentrations in hosting CDN activity and neocloud activity for countries like Germany, the Netherlands, Singapore, Finland and the U.K. And again, this also helps us inform our business as we want to grow and expand where we place things, how we want to connect to people. So this was a really nice visualization of the data, excluding those U.S. numbers that were sort of skewing us and not showing a lot of differentiated results in this heatmap.

Yes, absolutely. And the alternative is [indiscernible].

Then the next question was, if we just look at the U.S., what does that look like from a footprint standpoint? And we see a lot of neocloud activity towards California addresses. And this makes sense for us based on our partners, our connectivity. We also have a lot of CDN activity that ends in California. And also, we see -- do see hyperscaler activity in Virginia, which makes sense for us because there are a large amount of hyperscalers there. And new for us, which is interesting, is Illinois, Georgia, New Jersey are also showing up with areas of concentration. We haven't dug too deep into that yet, but there's always room for improvement in our data set.

I think it's important to note here, too, just to sort of clarify what we're looking at. We're not saying that like we're saying the data moves back and forth from there. So what can that look like, Brent? Is that coming from like a regional exchange? Or is that physically coming from the endpoint? What are we talking about when we say where it's going to and from?

We see a lot of activity over Internet connections. And for us, that means that where we were deploying 100-gigabit links, we're now increasing that to 400 gigabit or multiples of 400 gigabit. One of the offerings that we have is a product called B2 Overdrive, where we allow you to have S3 compatible object storage that can scale from 100 gigabit up to 1 terabit. And that's been a driver for us as we've been choosing port capacities where we deploy links and expanding the network.

Yes. Very cool. So let's get into our next slicing. So talking here about just comparing different types of traffic coming through the network.

Back to our magnitude metric. And again, this is a measure of how many bits were transferred per unique IP address. And this is a diagram of that metric over time for our neocloud and our hyperscaler operators. What we see here, what we saw on that graph, we had a lot of activity into August, September, October, November, a sort of lull in traffic January, February and a resurgence again in March is that the magnitude, the bits of -- the amount of data transfer for IP address still is coming in very high. And this graph is orders of magnitude. So those 2 green dots on the upper right for neocloud are very impactful to our network infrastructure.

These are flows that are not 1 gigabit. These are 100 gigabit to 400 gigabit flows. And as network operators, it's really interesting because it means that our network on a Tuesday is performing very differently on a Thursday. These flows that we see from hyperscalers and neocloud partners happen at any time. The magnitudes are very great. They don't follow our typical pattern of people are working during the day. There's a lot of content generation, a lot of backup happening overnight, people sleep. Our network is kind of sleepy just like people sleep. These are workflows that are populating GPU infrastructures with as much data as quickly as possible because time is money when you are renting time on a GPU cluster, and we see that show up here in the magnitude of the workflows.

Yes, absolutely. So let's talk how dynamic are these traffic patterns?

The next series of graphs dive into each traffic type over time per region. And this is a new set of data that we're publishing to kind of take a look at how our different regionality changes over time. We want to start tracking if the U.S. East is always an area of concentration or if we see shifts in the U.S. West because, again, this speaks to how we improve our network from a network operations standpoint. What we see is, again, in October, there was a lot of activity with neocloud activity in the U.S. East, a smaller period in January, February and a resurgence again in March. So this is telling us that the U.S. East is still a concentrated spot where we're sending a lot of traffic to our neocloud operators.

Yes. Very cool. And we've sliced for our other use cases as well. So we've got hyperscaler here.

Yes. Hyperscaler follows the neocloud activity. We see a lot of pairing of the data between our neocloud and hyperscaler data set. And this is because the data that's stored with B2 can be sent to neocloud #1 or neocloud #2, hyperscaler #1, hyperscaler #2, depending on what serves the client at any given time. There may be periods where the rental time on a GPU cluster is advantageous on partner A versus partner B. And that freedom of choice means that we have to sort of plan for our data to be sent to any of these partners at any time. But we generally see a very tight coupling between the neocloud and hyperscaler activity.

Yes. And then CDNs, I think these next 3 you also sort of grouped and I find them to be very interesting as well in a different way than neoclouds and hyperscalers.

When we speak about the neocloud-hyperscaler activity, we are talking about first traffic that can happen any time. It could impact our network at any moment. When we look at the next 3 graphs for our content delivery, hosting and ISPs, these are more steady state for us. These again follow traffic patterns that are steady. People are working during the day. There's a lot of content being generated backed up at night, we see our bandwidth graphs get a little lower. So when we look at these 3, CDN, hosting and regional ISP, these are very much a different mode of operation where it's just steady-state network. We can plan for this growth. It's very easy to track. It doesn't keep us up at night. It doesn't -- it keeps us challenging...

So this is CDNs here. But then I found the difference between hosting and then the ISP regional to be the most interesting from a comparison point of view.

Yes. If you want to advance to the hosting.

So this is ISP Regional and this is hosting here.

I don't see a slide advancement.

We may be having some network issues, but I've got ISP Regional up on there, Brent.

Okay. The ISP Regional for us, again, is pure steady state. This is easy to plan for. This is where we have the most ease of operation. As you can see here, U.S. West being the original source of our or the original source of the Backblaze network has the most content that's being delivered to ISP regionals. So it may not be the most exciting graph, but for us, this speaks to very steady-state planning.

I think the biggest takeaway I had from those 5 different slides was really just how different you get by use case slice. And it really reflected back to me why we slice and dice the data in the different ways that we do. And the challenges of the workflows that we're balancing when you're talking about planning for network engineering growth and all that good stuff.

So I do see -- well, before we go there, so we're on to questions. I love this. And I do see there's one question in the chat from James Kurtz. He says, I noticed a lot of demand from neoclouds in Finland. How is latency and throughput from Amsterdam to GPU clusters in Finland? How are we using these trends to plan your future expansions?

So I would say, number one, this is actually a question for a different series of ours called Performance Stats in which we actually track latency and throughput. And I think in general, what I would say about using trends to plan future expansion plans is that all data is good data, right? All these things are trade-offs in what we invest in and how over time. So it's an interesting question because a lot of that -- a lot of what happens for us is we bring data and other people make business decisions that trade off a lot of different things, including what we see here.

But Brent, if you want to speak to some of the things that we have talked about in ways we've expanded our network to handle some of those elephant flows, I think that would be interesting.

I think the key takeaway is that the neocloud landscape is very dynamic. There are facilities being built that are coming online. There are facilities that are being built. It's a very dynamic space. And we're open to pursuing any connectivity if it sort of suits the growth pattern. But we are in a state where what we assumed last month doesn't really make sense the next month. So this discovery of GPU clusters in Finland could be a new discovery for us. It is a very dynamic landscape, and it's driving a lot of innovation for us, which is great.

And what we do with it, I think, is always where sort of the magic happens. We don't necessarily have an on-demand answer for that. But it's -- I think visibility -- more visibility is always better.

Agree.

Yes. Any other questions coming from the chat? I see a lot of that in here. I do -- I wanted to surface, I think, one thing that we got when we were doing the report in other channels, Brent, was a question about how you can really see how different these look from these elephant flows look from a networking perspective, like how persistent are they, all that good stuff and what they say about your customers. And I remember you talking a little bit about how we sort of -- obviously, we never see our customers' data. So we don't know exactly what they're doing with it, right? But we talked a little bit about definitions and tracking these things. And I think that might be interesting to talk about here.

Yes. We take a sampling of network traffic, and we use a protocol called sFlow, which is where we sample every x number of packets. So every 64,000 packet, we take a sample of and aggregating those together, we can get a picture of what a TCP conversation is between 2 partners over time. And that lets us know what the IP addresses are involved, the length of the transfer, how many bits were transferred, what protocol is it, IPv4, IPv6. And again, it doesn't tell us the content. We can't see the content of the stream that's encrypted.

But it does let us know the networks that we're talking to, whether it's a Comcast network for Verizon, whether it's a neocloud operator or a hyperscaler. We then enrich that data with some additional fields in a database where we classify different networks as different types, and that's where our data set comes from.

And so James, I think that kind of leads into your second question here. Can that large flow in October be a sign of movement of data off the Backblaze network and on to other storage networks? What are you seeing in terms of migration of data lakes into homegrown neocloud storage?

So our visibility is really activity, right? We don't know whether it's coming on, going off, coming on and going off, which often happens when you think about training data sets and how they move. So there's kind of no way from a networking perspective to talk about that, except for that I would say, if you look at our total baseline, that very first chart, we can see that we do have our peaks and valleys, but our total baseline is up from an activity perspective, which again doesn't speak to total volume of data.

And as far as migration of data lakes into homegrown neocloud storage, Brent, is there any way that we would have visibility into something like that?

I think we can't get that granular into the specific application or service. We can merely tell what network that we're talking to and then what magnitude. So I don't think we can answer that question today.

And James, I'd be interested to know what you're defining as homegrown neocloud storage because we're certainly seeing just from an industry perspective, a lot of different experimentations to serve a storage demand, right? I think people focused quite quickly on the GPU demand, but we know that there's a massive storage demand there, too.

So if you want to reach out to us in comment section or anywhere really and let us know sort of where your head's at, I'd love to hear what trends you're tracking because I'm liking your questions here. And we contact information, [email protected] as the email or you can jump into the comment section of the report. We're pretty active on paying attention, of course, socials. So wherever you want to find us. Thank you, James, for your questions. We always appreciate it.

Yes. So with that, if there's no more questions, I think we can say thank you very much, Brent. I always appreciate all the wonderful analysis you bring to bear, and we'll see everyone on the next Network Stats.

Of course. Thank you very much.

Good day, everyone. Welcome to the Backblaze Fourth Quarter and Full Year 2025 Earnings Call. Just a reminder, this call is being recorded. I would now like to hand the call over to Ms. Mimi Kong. Please go ahead.

Thank you. Good morning, and welcome to Backblaze's Fourth Quarter and Full Year 2025 Earnings Call.

On the call with me today are Gleb Budman Co-Founder, CEO and Chairperson of the Board; and Marc Suidan, Chief Financial Officer.

Today, Backblaze will discuss the financial results that were distributed earlier. Statements on this call include forward-looking statements about our future financial results, the impact of our go-to-market transformation, sales and marketing initiatives, cost savings initiatives, results from new features, the impact of price changes, our ability to compete effectively and manage our growth and our strategy to acquire new customers pertain and expand our business with existing customers.

These statements are subject to risks and uncertainties that could cause actual results to differ materially, including those described in our risk factors that are included in our quarterly report on Form 10-Q and our other financial filings. You should not rely on our forward-looking statements as predictions of future events. All forward-looking statements that we make on this call are based on assumptions and beliefs as of today, and we undertake no obligation to update them, except as required by law. Our discussion today will include non-GAAP financial measures. These non-GAAP measures should be considered in addition to, and not as a substitute for, our GAAP results. Reconciliation of GAAP to non-GAAP results may be found in our earnings release which was furnished with our Form 8-K filed today with the SEC. You can also find a slide presentation related to our comments in the webcast, which will also be posted to our Investor Relations page after the call. Please also see our press release or presentation for definitions of additional metrics such as NRR, growth customer retention rate and adjusted free cash flows. And finally, we will be participating in the Citizens Technology Conference on March 2 in San Francisco.

Thank you for joining us, and I would now like to turn the call over to Gleb.

Thank you, Mimi, and welcome, everyone, to the call. We finished 2025 with solid fourth quarter results. Revenue came in line with guidance and adjusted EBITDA margin reached 28%, doubling over the prior year. We also delivered adjusted free cash flow profitability for the first time as a public company, a major milestone demonstrating the inherent operating leverage in our business model. For the full year, total company revenue grew 14% year-over-year with B2 Cloud Storage growing 26%.

Today, I want to focus on 3 things: first, the strength and durability of our core business; second, an update on the meaningful progress of our go-to-market transformation; and third, how we're positioning Backblaze to take advantage of the AI opportunity. Let me start with the core of our business. As data creation accelerates exponentially, Backblaze addresses a large and growing market where long-term demand for a scalable, cost-effective storage compounds over time. Our business compounds within that market as we add new customers and retain them for an average of 9 years. B2 net revenue retention of 111% reflects consistent expansion within our installed base, reinforcing durable, long-term growth.

We've proven our ability to grow in that market, delivering an annualized growth rate of 21% since IPO, being a cash-generating business is an important financial milestone. Year-over-year, we meaningfully improved profitability, demonstrating how we are building a sustainably durable company, one that can invest in growth, while maintaining financial strength.

Now let me talk about our investment in growth and the progress on our go-to-market transformation. While we didn't achieve our budgeted Q4 B2 growth rate, we made meaningful progress and have positioned ourselves for success. More importantly, the underlying fundamentals of the business remain stable and the investments we've made position us for durable growth going forward. Excluding the highly variable growth of the large AI customer we previously mentioned, we stabilized on a baseline of around 20% B2 revenue growth in each of the last 5 quarters.

Now we've shared our goal of moving up market. We ended the year with 168 customers generating more than $50,000 in ARR each, up 35% year-on-year. The ARR of this cohort increased 73% year-on-year to $26 million of ARR. We're very proud of this upmarket progress. We've also launched 3 key initiatives: number one, increasing awareness. We launched Flamethrower, our Startup Program designed to engage high-growth companies early and established Backblaze as their long-term storage infrastructure partner; number two, driving greater pipeline of consistency. We're upgrading our top-of-funnel systems and scaling demand generation programs to drive higher velocity sales motion; number three, expanding revenue within our installed base. We are implementing processes to proactively identify and capture additional share of wallet across our more than 119,000 B2 customers.

People are the cornerstone of our success and we continue to strengthen our leadership bench to support these initiatives. We have already hired the co-founder of an edge compute company to drive our Flamethrower program, a business systems leader for our systems work and a head of customer success to build out that expansion effort. We will keep up-leveling our leadership and talent. For instance, we are also in the final stages of hiring a sales development leader to drive pipeline and a revenue operations leader to drive tighter coordination and accountability across the entire go-to-market organization.

Scaling into this next phase requires even greater execution discipline to support that, Elias Mendoza joined us as strategic transformation leader. He previously served as partner and COO at private equity firm, Sirius Capital and held leadership roles at IBM and Morgan Stanley. In these roles, he has helped companies drive strong strategy to execution. Under his leadership, we also established a go-to-market advisory committee of operators who have scaled enterprise and platform businesses to a billing and revenue and beyond at companies such as Okta, Snowflake, ZoomInfo and Carda.

Their role is to bring pattern recognition, pressure test key decisions and provide external perspective as we scale. We have made meaningful progress in our go-to-market transformation, and I'm excited about the team we're putting in place to drive it forward. Now let's talk about how we're positioning Backblaze to take advantage of the massive AI opportunity ahead. We all understand there's a lot happening in AI today. But sometimes the scale is still hard to fully comprehend. I saw a report recently that capital spending on AI as a percent of GDP by just the hyperscalers in 2026 is forecast to be 5x larger than the entire spend to create the U.S. interstate system, 10x larger than the Apollo Space Program. AI CapEx spending accounted for 92% of all U.S. GDP growth.

It's hard to hyperbolize AI. With AI, a big focus is who's disrupting and who's getting disrupted. We believe Backblaze is one of the disruptors, participating in this infrastructure replatforming as a storage backbone for the next wave of cloud infrastructure. So while like any major new innovation, there will be market volatility. We are firm believers in the long-term growth opportunity and are leaning into it. We're doing that with 2 growth vectors. Number one, on the supply side of AI, Neoclouds and other AI tooling companies are building the platforms for AI workflows.

Our opportunity is to be the storage backbone of those platforms. And number two, on the demand side of AI, companies are using AI to build everything from a anomaly not only detection to zonal forecasting. These companies are using and generating large data sets. Our opportunity is to be the storage of choice for their developers and use cases. And we are uniquely positioned to be the glue between these, creating a virtuous cycle, developing a platform that can deliver massive performance with large-scale data sets, while providing that cost efficiently is a significant technical challenge. Backblaze has done that, and AI is driving an increasing need for this technology.

On the supply side, roughly 200 Neoclouds have sprung up and industry estimates project that market to reach $237 billion within the next 5 years. These companies provide GPUs as a service. And most will need Cloud Storage to fully serve with their customers. We've already signed multiple of these multibillion-dollar Neoclouds with not only 6- and 7-figure deals, but our company's first 8-figure TCV deal and over $15 million deal. And we believe all of these have material upside potential, and we're in discussion with half a dozen others. By our estimates, Neoclouds storage for our solution alone represents a $14 billion opportunity by 2030 -- for Neoclouds.

Developed in collaboration with our new cloud customers, B2 Neo allows Neoclouds to offer a top-tier storage solution without the massive capital costs or years of engineering required to build a storage back end from scratch. On the demand side, the growth in AI developers is exponential. GitHub disclosed they were adding, on average, a new developer every second. Hundreds of AI companies and countless individual AI developers already use B2. For example, one of our customers uses AI to generate audio. They just launched a year ago and already have multiple petabytes with us signing a 6-figure annual deal with this. As they add new users, and those users generate more audio, that data grows exponentially.

Our self-serve platform, where we added 12,000 customers this year alone is a great enabler for this class of AI developers who just want to get going. We launched our start-up program called Flamethrower and a Developer Relations Initiative to ensure developers are building with Backblaze. To drive our road map forward for the AI opportunity ahead, we strengthened our product and engineering leadership. Dan Spragans joined as SVP of Engineering and Ret Dillingham as SVP of Product, bringing deep experience in AI and high-performance cloud infrastructure. We also added Russ Arps, Co-Founder and former, Head of R&D at Computer Associates, as an adviser.

Together, this team strengthens our ability to scale the platform for larger, more complex AI-driven deployments. We entered 2026 with a strong and growing business, a rapidly improving go-to-market motion and a tremendous AI opportunity with a targeted B2 Neo offering and a strong product team. AI is reshaping how data is created and scaled and storage sits at the center of that transformation. Across Neoclouds platforms and AI native developers, we are building the foundation for the next generation of data infrastructure. Durable growth and massive AI potential are the hallmarks of our opportunity.

With that, I'll turn the call over to Marc. Marc?

Thank you, Gleb, and good afternoon, everyone. We grew revenue while achieving adjusted free cash flow profitability in Q4. This is a significant milestone and an important step forward in our profitability journey. This progress was not driven by short-term cost actions, but by the inherent leverage in our operating model as revenue scales. For the quarter, total revenue was in line with guidance at $37.8 million and adjusted EBITDA exceeded the high end of our guidance by approximately 600 basis points. In the fourth quarter, B2 revenue grew 24% year-over-year, up from 22% in the prior year. This is modestly below the range that we outlined last quarter. We delivered record bookings this quarter. As Gleb noted, we closed our largest contract in the company's history with over $15 million in total contract value. We're excited about this 8-figure deal. This deal validates the product market fit at scale. .

We don't expect to see meaningful revenue in 2026 as we complete certain development work. In 2027, we expect this customer to contribute over 300 basis points to B2 revenue growth. This customer helped drive our RPO, up 60% year-over-year to $66 million. In quarter B2 NRR was 111% compared to 116% in the prior quarter. The sequential decline reflects variability from the large customer that we mentioned in our past 2 earnings calls. Factoring out that one customer, the underlying retention and expansion trends remain stable.

Moving to the income statement. Q4 gross margin was 62%, flat sequentially and up from 55% in the same period last year. Adjusted gross margin was 80% compared to 78% last year. Margins remained stable despite higher data center costs, reflecting continued efficiency in our infrastructure and disciplined management of our operating model. Looking ahead, we anticipate some pressure on gross margins driven by increased costs. In response, we are proactively launching a gross margin optimization initiative focus on structural improvements across pricing, packaging and infrastructure. Our Q4 adjusted EBITDA margin was 28%, doubling year-over-year. The adjusted EBITDA outperformance was primarily driven by nonrecurring items, including variable compensation alignment and office restructuring savings.

Excluding those onetime items, adjusted EBITDA would still have been above the 22% high end of our guidance. Adjusted free cash flow was positive $4 million in the quarter, representing a margin of 11%, exceeding our outlook of being adjusted free cash flow neutral. We ended the quarter with $51 million in cash and marketable securities. Based on our current operating plan, we expect to fund our growth through operating cash flows and capital leases. We do not anticipate a need to raise additional capital. We will continue to evaluate opportunities to optimize our capital structure over time in a disciplined manner.

To improve accountability and further align management incentives with shareholders, we are shifting part of the compensation to performance-based stock units. These awards are tied to clearly defined performance objectives. Turning to our guidance for the year. Our objective is to provide a clear incredible baseline that reflects the most predictable portions of our business. While pipeline activity remains healthy, larger customer wins in usage-driven workloads can introduce variability in timing and revenue recognition. To maintain forecast discipline, we have derisked our outlook by excluding large swing deals and the anchoring guidance on opportunities with more predictable demand characteristics.

For our customers with high variable usage patterns, our assumptions reflect contractual minimum commitments rather than potential upside consumption. Our outlook is, therefore, based on continued expansion within our existing customer base and steady adoption of B2 across core use cases consistent with recent operating trends. We believe this approach provides a prudent and reliable foundation for the year, while preserving upside as deployment timing and usage visibility improve.

For the first quarter of 2026, we expect revenue to be in the range of $37.6 million to $38 million, with adjusted EBITDA margins in the range of 18% to 20%. For the full year, we expect revenue to be in the range of $156.5 million to $158.5 million. Full year adjusted EBITDA margins are expected to be 19% to 21%. We expect adjusted free cash flows to be roughly neutral for the year with normal quarterly variability. Due to the difficult comp from last year's large variable customer, we expect B2 year-over-year growth in Q2 and Q3 to be in the range of 12% to 19% and approximately 20% for the full year.

To wrap up, over the past year, we made meaningful progress towards becoming a Rule of 40 company, with our combined B2 revenue growth and free cash flow margin improving from $9 million to $35 million. As we look towards 2027 and beyond, we believe Backblaze is well positioned to grow efficiently. Our platform is already built. Our infrastructure scales with discipline and incremental revenue increasingly translates into profitability and cash generation. This capital-efficient model allows us to pursue the massive AI-driven opportunity ahead, while maintaining financial discipline expanding margins over time and building a durable self-funding business.

With that, operator, let's open it up for questions.

[Operator Instructions]

We'll take the first question from Ittai Kidron from Oppenheimer.

Solid numbers, and thank you very much for derisking the outlook for the year. It's hopefully a very smart move. Gleb, I wanted to dig of course, into the Neoclouds and the large deal First of all, just from a big picture standpoint, are demand patterns any different? Can you explain how Neo -- your B2 Neo Cloud solution? How is it different in B2? And what way are the demands different the pricing difference, the margin different. And if you could elaborate also why this deal is going to take a year before we started seeing revenue, I would appreciate that.

Yes. Thanks, Ittai. All good questions. So one thing I'll say, first of all, is our pursuit of the Neoclouds is one part of the business pursuit. There are about 200 of these new clouds. We do think it's a large and important opportunity for us, right? The -- just our part of the Neocloud opportunity, we view as about $14 billion, so it's important. And we are really well suited for it. The hyperscalers are not key competitors here because they are competing with the Neoclouds as opposed to being vendors for them the way that we are. So it's a good opportunity, which we're well positioned for.

In terms of what B2 Neo is it is a white label offering. So B2 is generally sold directly to the end customer. B2 Neo is a white label offering that they can build in directly into their service. It is -- it provides a lot of the same functionality that B2 provides. It's high performance, it's low cost, it's durable, it's scalable. But it also provides them the ability to manage that storage on behalf of their customers through APIs, with API integration, single sign-on, et cetera.

So it's really leveraging all of the technology that we've built over the last years for the company and then layering on top of that technology to make it simpler for them to integrate natively and make it easy for them to manage and offer that storage offer. So that's what we. Now in terms of why it's going to take a year for this one Neocloud provider to start seeing the benefits of it. It's a combination of work we need to do and work they need to do. So they have an existing storage offering that they're going to be switching to use B2 Neo instead.

And so it's basically, we have some work to do to make it so that it's even easier and more robust to automate and natively integrate for them. One thing I'd like to make clear is all the work that we're doing for them is useful for other new cloud providers and also other companies but not required for most. So we have multiple new clouds that have already signed up that don't need this work, and we think that there's a large number of them that won't need any of this work. But the work that we're doing is broadly useful for others as well.

Okay. Appreciate it. And then I guess, first of all, the TCV $50 million, that's great, but can you tell us the duration of the contract? And is the margin profile of this business? As you ramp up the Neocloud, Gleb, is there a potential upfront costs hit to you as they ramp before margin normalizes on these businesses?

This is Marc. I can take that question. We do have to accelerate some capital expenditures that would impact that and other things happening in the market would impact our gross margin by a few hundred basis points to help us prepare for this because it's obviously a large deal, you need to have the capacity in place.

Okay. And then lastly, on computer backup, Mark, can you comment on the expectation? I mean this business is -- the number of customers is now declining here I guess, help me think about the framework for this business for '26. How should I think about the quarterly cadence and the annual cadence of this business? Is there a different long-term outlook for this? .

Yes. I mean, I'll start off by the coming year, Ittai, we see this business declining 5% year-over-year. Currently, in Q1, that's more like a minus 3%, that builds up throughout the year and makes an -- averages out for the end of the year at a minus 5%.

Okay. And longer term, is there any reason we just continue to expect this business to slowly decline?

What I would say you, Ittai, on that one is we have programs that we've put in place and are putting in place to stabilize the business. We would like to get it to a place where it is flat and possibly even slowly growing. We don't think this is a fast-growth business, as you know, but it would be good for it to not be a declining business. But it's a little too early for us to have confidence in those programs getting into that place. So for this point, we're estimating it at that shrinking rate, but we are putting effort into getting that to be flat to slightly growing.

The next question will come from Jeff Van Rhee from Craig-Hallum Capital Group.

Congrats on the free cash flow, great to see it. A couple for me. Maybe if you could just start in terms of B2 coming into Q4 came in a bit below expectations. Just expand a bit more on what missed there. And then as you're looking at the annual number, I didn't catch what you had guided it for in Q1. So if you could just fill in the gap, I think we can back into it, but maybe you could just share it. So what happened in Q4 and what do you think in Q1.

Yes, Jeff, this is Marc. So on the Q4 '25 we were expecting when we set our guide quite a few deals to close in November. They came in very late in the quarter, so they didn't benefit Q4 that's why we've adjusted our guidance philosophy going forward, where we said going forward, we're going to factor out the swing deals because they're less predictable in timing of closing. So that feeds into the guide going forward. And we said for B2 year-over-year, it will be 20% in 2026. The ranges that we provided of [ 12 to 19 ] A lot of that has to do with the comps of that high variable customer in 2025.

So Q2 would be the low end of that range, and Q3 would be about the higher end of that range. And overall, the year would average up to 20%. Does that answer your question?

Yes, I think it does. And so the growth is, if I do the quick maybe in Q1 looks like it's 9%, if I vet, right, on year-over-year and you're decelerating to 8% for the overall year. So it actually looks like maybe you're assuming some deceleration in the year. I'm sure there's a little bit of lumpiness from the large customer. But generally speaking, you had some pretty good momentum in sort of Phase 1 of the sales build and build out. And it sounded like you felt like you had some early signs on Phase 2, but the numbers are painting a picture of deceleration. So just help me reconcile the 2.

Yes. I mean the the deceleration that you're seeing is largely driven by the one monthly customer. If you go to Slide 21 of our earnings deck and you factor out that one customer, you could see that it pretty much movement stable around the low 20s. So if you recall, factoring out any price increase, B2 growth rate has always been growing but decelerating for 5 years. We've managed to stabilize it in the low 20s. So now with this new guidance philosophy, we're seeing 20% year-over-year, and that includes the lumpiness that I described in Q2 and Q3. But -- then with all the Phase 2 changes we're doing, Gleb could elaborate on that, that will then afterwards come drive benefits.

I didn't go out to Jeff, I think you were talking about the whole company, not just B2, right? And so part of what's driving that is that computer backup was growing in part of the price increase before and it's -- as Marc said, we expect it to shrink about 3%. So it's putting some downward pressure on the overall company in Q1. But On the GTM transformation, I think some of the things that we look at is in terms of progress, there is progress that we're making in terms of actions, things like we've hired the VP of Revenue Operations, we've made material progress in moving the systems forward and expect that work to be largely completed at the end of this quarter.

We've gotten pretty far in the past with some sales development leaders to bring in. We've made a number of kind of improvements. And then you can also see some of the outcomes like the 73% growth in ARR from customers over $50,000 and the 8 figure deals. So I think we've made progress on the GTM side. Obviously, we all want more work to be done there.

Great. Maybe just one last one, if I could. On the large Neocloud win, can you just expand a bit on what the competitive landscape looks like there? Maybe the finalists, the kind of 2 or 3 that it came down to at the end of the day. And if there were specific features, capabilities that were the deciding factors for your win there? .

Yes, it's actually -- it's interesting because this new cloud, they have the own storage. They started realizing from their customers that the source that they had wasn't going to provide what they needed for this next phase of evolution. And so they started thinking about how to handle that. A number of their internal engineering and business leaders were actually familiar with Backblaze from prior roles in other places, and they needed Backblaze a really strong reputation for providing a great storage platform that it was trusted.

Basically, we built a moat around this idea of high-performance but predictable economics and low-cost storage. And so we were at the top of their list for consideration. Now when they went and evaluated, they wanted to make sure because they were going to be basically placing their brand on the line for saying they're going to use us for this underlying platform for all of their customers. So they wanted to make sure they absolutely worked. They did detailed technical due diligence and then chose us. So the why it came in part because had established a lot of credibility over many years that we are a great storage platform, and then we met their technical requirements for both performance, scale, affordability and openness.

Your next question today comes from Mike Cikos from Needham

If I could just come back to the gross margin comment this expected headwind that we're up against. I guess it's a bit of a 2-parter here. But when I think about the headwind we're facing this year, is that really tied to your success initiatives or deployment in advance of recognizing revenue from this large Neocloud agreement that we're talking to today? Or is there potentially an ongoing presence or multiyear factor we need to consider when evaluating corporate gross margins on a go-forward basis.

Yes. Mike, it's Mark. There's a few factors in there, right? First of all, data center cost and equipment have gone up That, combined with us needing to accelerate some CapEx does reduce our gross margin this coming year by a few hundred basis points. That's why we said we're doing that gross margin optimization initiative to look for opportunities to offset that. Now in terms of business model, when you go after a white label, large-scale solution like that, generally speaking, the gross margin will be a bit lower and the OpEx will be lower as well because you have to spend less on sales and marketing. So it nets out to the same economic model for us, but that's the P&L benefit, if that makes sense.

It does. And then I just wanted to come back again to the risk guide that we're talking to here. And I appreciate the commentary in the prepared remarks. But just to better understand these sweet factor deals or the idea that we're only going to underwrite minimum contract commitments from customers. Is that really tied to in the Neoclouds when thinking about those swing factor deals? Or is it maybe the move up market? Anything else you can provide that's creating that dynamic? And then second -- go ahead, go ahead. I just have a follow-up.

Okay. I'll answer this one and then you could ask your next question, if you want. So moving upmarket, I mean there's different sizes above markets. But when you look at the average deal size of those 168 customers it has grown quite a bit. But I think the even larger ones, and let's call larger ones, $500,000 in ARR and greater they do take longer to close. So there's less predictability for us to factor that into our guide. So that's why we factored them out. Doesn't mean they won't happen. It's just harder for us to guide on that. So I think it's less around the new cloud. I mean the new clouds are big deals, too, and they have similar attributes, right, where you got to take longer to do the technical feasibility and make sure you went over the POCs. So that's what's driving that side of it.

And then I guess the final follow-up on my side. But for those, let's say, $0.5 million plus deals that you're signing, can we start bifurcating the extent to which those sales cycles are longer versus a more typical run rate business? And then final fees, but for the calendar '26 guide, is there any way you can give us some pointers as far as the NRR that you're thinking about when we look at this calendar '26 guide? And that's all on my side.

Mike, in terms of the bifurcating the size of the deals on the length of time, when we look at those, they certainly are a longer sales cycle ones, but it's interesting because they're not dramatically longer. So some deals like the 8-figure deal that we talked about, that did take the better part of the year, in part because they had to look through their own systems they have to understand what it would take to switch out to a different system, what integration that would require, et cetera. A number of the other Neocloud didn't take anyone near that long and many of the other larger customers, especially ones that are 50,0000, 1000,000, 200,000, we actually moved quite quickly. But certainly some of the largest of those deals they did take, call it, so some of them took 6 months or so to close, whereas we've seen a lot of the deals close in sub-90 days.

Yes. And then I could jump in and discuss the NRR outlook. Due to the lumpiness of that large customer in '25, we factored out any usage above their minimum commitment level and our guide for '26. So assuming that, that's what materializes, the NRR, just like the revenue growth rate for B2 and just like the overall growth rate of the company will be lower in Q2 and Q3. NRR could go down to closer to 100% for 1 or 2 quarters. But our overall growth rate of 20%, which is where we should be finishing the year and year-over-year overall should equate to an NRR that's closer to 110%. So pretty much where we are now plus or minus to 300 basis points.

Mike, one thing actually, I'll mention also on NRR I think I find quite exciting. The -- we have a broad base of customers, but we're leaning in heavier to the overall AI customer type, not just the Neoclouds. And we have hundreds of those customers that are using us for AI workflows specifically we've seen a growth rate of 75% in the number of those AI customers. But one of the things that I find even more exciting is that the growth rate of those customers is about 3x faster than the growth rate of our average customer. So as we sign up more of these AI customers, -- we see the opportunity for NRR to go up over time as well because they are generating data at a faster rate than your average customer.

Next, we'll go to Jason Ader from William Blair.

Wanted to first ask about your comment, Gleb, that most Neoclouds don't have storage. I think that's what you said. I just wanted to understand why that might be. And then also your comment that the 8-figure win was with the Neocloud that did have storage, but the storage wasn't going to handle what they needed. Maybe just if you could elaborate on why you wouldn't be able to handle what other customers needed.

Yes. Thanks, Jason. Both good questions. So with these 200 Neoclouds that have come up, they almost all started with GPUs. Right. So the need that happened was for these AI use cases, they needed the GPUs first. The second thing that they need is they need a place to keep the data to feed these GPUs. So initially, they setup data centers, a lot of them setup data centers that were more specifically designed for GPUs, which are very power hungry, oftentimes they were in liquid cooled environment. They don't need nearly the square footage in the data centers that they need, they need more power in the space, et cetera.

So they built these -- focused on the GPU opportunity. What they realized then is customers who want to use the GPUs need a place to keep the data. They needed the place to keep that data to build the models. And then they needed to place to keep the data when they're doing inferencing for the outputs. And so what some of them have done, many of them have not done anything on that front yet. They've just stood up the GPU side of things. But what some of them have done is said, okay, well, we can do something, and they -- some of them have used open source projects to stand at their own infrastructure where some of them have set up storage infrastructure using flash systems.

The problem is what they found is the flash systems are incredibly expensive to operate. And so for large-scale data sets that becomes very quickly unaffordable. The open source tooling is difficult to manage. You have to have experts ongoingly working to tune it and operate it, et cetera, and they're really not designed to scale to exabyte scale. Most of those open source projects were designed for potentially handling a single enterprise scale. And so once they start seeing some movements of success, they start reaching the limitations of those projects.

So the opportunity for us is that there are these 200 providers they've built up the GPUs. They're starting to realize that they need storage. They're not going to get that from the hyperscalers for the most part because those are their direct competitors and the solutions that they have are either really expensive, really complicated or don't scale.

Got you. Okay. And then the Neocloud that you announced that you talked about the 8-figure one. Can you say if that is a publicly traded company?

They are a publicly traded company, yes.

Yes. Okay. Great. And then last one for me. Just, Gleb, what's your confidence level that you could win additional deals like the one that you announced on the call today?

I mean, I'm very confident that we can do additional deals. The timing is obviously always uncertain, but this is -- it's not like this Neocloud is the only new cloud that we have won. We've got others that are 6 figures and 7 figures already. Those that we have already signed at 6- and 7-figure deals I think they themselves have the opportunity to become 8-figure deals because as they roll this out to more of their customers and more scale, they're big enough that they could be coming 8 figure deals for us themselves. And we're currently in discussions with about half a dozen other Neocloud providers that are somewhere in this same scale of size of organizational opportunities. So timing is obviously a question for us, but our ability to be a good fit for these kind of customers and the discussions we're in, give me a lot of confidence.

And I may have missed it, but did you say how the duration of that 8 figure went was.

That was a 3-year deal. .

Eric Martinuzzi from Lake Street Capital Partners.

Yes. You mentioned the revenue impact from the 8-figure transaction really doesn't start to until 2027. Is that -- based on your answer about the 3-year duration and over $50 million, is that to say then that we're a small amount, maybe the end of 2026 and the bulk of it split between '27 and '28.

Yes, that's correct, Eric. And for now, honestly, we're not factoring anything into 2026 for that.

Eric, you -- I just want to make sure that it sounds like you said $50 million, it's $15-plus million, 1-5. I look forward to a $50 million view in the future, but we're not there just yet.

The other thing I wanted to ask about was your comment regarding the adjusted free cash flow, you talked about it being neutral for the year. And I'm just wondering, given the investments you're making, you have the infrastructure in place here, it seems like it's sort of front half loaded. Is that to suggest then that the adjusted free cash flow positive. We're Q4 for sure, and potentially Q3. Is that the right way to think about it quarter-by-quarter.

Yes, Eric. I mean, generally speaking, the first half of the year is our cost base increases. It starts kicking into -- and our OpEx lines, honestly should not be really increasing that much other than maybe around 500 basis points, not as a percent of revenue, just off the dollar baseline from last year on a non-GAAP basis as it relates to just basic inflation, salary raises and so on.

Other than that, we're keeping our OpEx model pretty tight. I spoke about the gross margin being set back by a few hundred basis points. So when you combine all those factors and accelerating some of the expenditures to prepare for these customers, that's why we're free cash flow neutral for 2026. It is lumpy during the year, usually Q2 is also where we have the least of our computer backup renewals. So Q2 is usually the worst set in the second half of the year is in better shape. And that would be a nice improvement from the minus $5 million for 2025 as a year and the minus $20 million in 2024. So I think we're pretty well set on exiting the phase of cash burn and our aim is to stay here and get better.

[Operator Instructions]

Up next is Zach Cummins from B. Riley Securities.

Ethan Widell calling in for Zach Cummins. I guess Guess to start with Neocloud with their being a high portion of leverage there to AI and HBC. How would you define, I guess, the incremental revenue opportunity or overlap, whether it be like customer base or function or revenue opportunity versus B2 Overdrive.

Yes. Thanks, it's a good question. So B2 Overdrive was initially actually developed because we heard from customers saying they wanted to use high-performance storage, high throughput storage that would enable them to send their data to the Neocloud when they meet them or to other hypers, for example. So B2 Overdrive is not a white label offering. It's designed for end customers to actually use themselves. Neo is specifically designed as a white-label offering for the Neoclouds to them themselves offer storage to customers. So they're largely serving different sides of the market, but both serving the needs of AI and HPC type use cases. .

Understood. That's helpful. And then the large TCV deal, can you clarify whether that was from an existing customer? And generally, is the revenue upside from existing customers there based on increasing usage?

So the $50 million-plus TCV deal is a new customer, completely new to us. However, what I would say is if you look across the $1 million-plus deals that we've had over the last year is it's roughly half-half. Half of them are net new customers to us that came in, evaluated considered tested then signed a 7-figure deal with us. And the other half of our customers that started off small, some of them started off self-serve, some of them came in at just smaller sales deals, got familiar with the platform, like the platform and then expand it into for deals.

Ethan, this is Marc, What I would add. If you look at Slide 17 of the earnings deck, it breaks down the new versus expansion from the existing and it's at half and half. So it's pretty well distributed because the self-serve product line growth is about half of that as well. And then the larger direct sales customers have. And each one is kind of breaks out into a half by itself of what is expansion versus new logo. So it's basically that's why if you look at the stacked bar, it's like 4 quarters, it's pretty well diversified in terms of how it comes through.

Yes. Maybe one other piece of color just to add in terms of -- so one of the things we look at is as a forward-leading indicator is pipeline. And in 2024, we generated about $15 million of pipeline. And in 2025, we roughly doubled pipeline to about $30 million. Our aim with our continued GTM transformation is to get to a run rate of about double of that. So with our industry-leading win rates, pipeline transfers into AR quite efficiently. And so we're not there yet, but that's -- we made, I think, meaningful progress in '25 and aim to make more meaningful progress on that in 2026

The next question is from Rustam Kanga from Citizens.

Marc and Gelb congrats on the RPO acceleration. Just building on another question that you answered, Marc -- Gleb where you kind of mentioned that to B2 Overdrive versus B2 Neo are serving 2 different sides of the market. And as we sort of think about the build-out of the pipeline for B2 Neo, is it fair to say that these opportunities are going to be anchored towards larger deals, albeit maybe not as large as this one that you've just put it up in the quarter, but is it fair to say that this is kind of the larger opportunity? And is that likely to sort of lead to higher ASP engagements as you look towards this opportunity?

Yes. It's a good question, Russ. So one of the ways I would look at it is the market for the Neoclouds, if you take just the hard drive-based storage opportunity inside of those 200 providers. That market is estimated at about $14 billion in the next 5 years. So with 200 players representing $14 billion of opportunity. Every single one of those deals on average is going to be a large deal. So the short answer to your question is, yes, the B2 Neo deals, we see as large opportunity deals. The ones that we've signed so far are 6 and 7 and now 8-figure opportunities on those. Some of those, I imagine, may start smaller just as they start getting familiar with it. But I think all of them have the opportunity to get quite large.

Great. That's helpful. And then just kind of thinking about the investment cycle for next year, is there any sort of relative color that you can share with us in terms of the level of CapEx investment that you guys are thinking about for '26?

Yes, Russ, this is Marc here from you. Our our CapEx will be higher next year. As a percent of revenue, when you look at our PP&E at the end of the year, it should be in the high 20s percentage of revenue. We typically finance our CapEx through capital leases, and we're fully set up to do that. And that would be the principal lease statement on the statement of cash flows, which is around mid-teens of revenue, right, because you're buying today but financing over 5 years over a growing revenue base. That mid-teens, I mean, over the past few years, has actually improved from our side as we continue to optimize our cost of capital.

And everyone, at this time, there are no further questions. I would like to hand the conference back to Gleb for any additional or closing remarks.

Thank you. We have a strong and durable core business, made meaningful progress in our go-to-market transformation and have a tremendous opportunity in AI. We drove growth while becoming adjusted free cash flow positive, we launched B2 Neo and signed multiple Neoclouds, including this $15 million-plus deal. We also launched Flamethrower our program for high-performance start-ups. In just the last few days since the launch, it's exceeded expectations, growing faster than the kickoffs at other leading companies that are a leader for that has driven. We had about a dozen start-ups that have applied, been evaluated, accepted and given credits, including ones from Andres and Horwitz and Y Combinator, and we've bolstered our team overall to take advantage of this tremendous opportunity. I'm really excited about the year that we have upcoming together. I want to thank our employees, our customers and our investors for taking this journey with us, and we look forward to chatting with you next quarter. Thank you.

Once again, everyone, that does conclude today's conference. We would like to thank you all for your participation today. You may now disconnect.

Hi, everyone. Welcome back to Drive Stats. Today, it's our special edition once a year where we get to talk about the full year of 2025. So as a reminder, I'm Stephanie Doyle. I'm a technical storyteller here at Backblaze, and I get to carry drive stats forward with my partner in crime over here, Pat?

Yes. I'm Pat Patterson, Chief Technical Evangelist. And yes, I run the queries and wrangle the databases. And yes, Stephanie and I work as a team producing all of this data every quarter. So before we get started, just some quick housekeeping. The webinar is being recorded. So if you need to drop off for whatever reason, you can come back, pick it up tomorrow after the recording is posted online. Please do ask questions at any time during the webinar. [Operator Instructions] And please don't miss the attachment. So we've attached what we got PDFs, I can look over here, actually. We've got the year-end report link, the archive link to subscribe to newsletter and the drive stats home base, where you can find statistics going back for 13 years.

13 years?

Yes. So what is Drive stats? If you're joining us for the first time, you might be thinking, what are they talking about? Well, since 2013, we've collected, curated and published raw device metrics. So what we're doing is for every drive that we are running in production, we are collecting every day, the drive serial number, model ID, the smart attributes and all important, whether the drive failed that day. And that's what lets us put together these annualized failure rates and other statistics to present here in our report. So every quarter, we do this, but this is our annual report. So you get a little bit more data, we look at 2025, the whole year as well as just the quarter snapshot.

So where are we? We're currently spinning over 330,000 hard drives. So in the quarter, we saw nearly 1,000 drive failures. And if you can do mental math, you can get to the number that's down on the bottom right there. We saw an annualized failure rate of 1.13% in the quarter. And that was -- we love this, drive days. Drive Day is a drive running for a day. So 1,000 drives run for a day is 1,000 drive days, 1 drive runs for 1,000 days is 1,000 drive days. So that's how we measure the volume of statistics we're collecting.

And you can see there that the drive population, it's actually quite interesting because we've got Seagate and Toshiba with about 1/3 each. And then if you look at HGST, that was Hitachi Global Storage Technologies. That's like a legacy brand that was absorbed into Western Digital. That actually makes up another 1/3. So really, this drive population is pretty much split in 1/3 between Seagate, Toshiba and Western Digital, if you count HGST.

And we also have the annual stats and the lifetime stats we'll be digging into, and you'll see those annual failure rates are a little bit different. Now Steph, this is just because we're looking at different time periods, right? The quarterly one is -- bounces around a bit more.

Yes, yes. Each time you -- when you think about creating boundaries around these things, quarterly, annual and lifetime, it's not just that they ran for a quarter. It's also that we have different exclusions set up. So your quarterly, that's 250 drives minimum to make it on to the chart as well as a certain amount of drive days. Annual, it's 500 drives. Am I -- actually, I think annual is -- we've got the exclusions listed on each. I do this every time. But the point is that we try and give you more confidence with each additional interval because if you create a higher minimum standard, then what we actually see is less volatility in those AFRs, which is something we've visualized in this presentation throughout. So we'll be able to talk a little bit more about that when we get into it.

And naturally, there's more drive days incorporated in the data. So we're spreading that -- information is more kind of spread out.

Absolutely.

So Steph, take us into the numbers. What are we looking at here?

Yes. So this is the quarterly data. And like we said, this is your lowest bar to reach on to a chart that we track here. So you'll see that this quarter, we had an incredibly low annualized failure rate that was 1.13%. That's the lowest I've seen in well over a year. But it does represent a pretty solid fluctuation. If you look at the quarterly AFRs for this year, we've bounced around a little bit. So 1.42%, 1.36%. Last quarter, we were pretty high, and this quarter, we're pretty low.

We've talked previously in other reports about how we define a failure and what that means from a data engineering perspective. But the important part here is that every time we're talking about quarterly failure, actually any kind of failure rate, we're talking about real failures. So if you go back and you look at some of our older drives up here, anything that's an age in months of 100 months or 90 months, some of those are actually being filtered out of the drive population through our normal migration process without ever failing. So we only classify a true failure as a failure, and that means that the drive has stopped working for whatever reason. And we end up with some really cool highlights because of that.

So first of all, we've got our first 26-terabyte drive, always really cool to see those high-capacity drives coming on. There's implications for the drive fleet and what that means for parity, but that's a conversation for another day. But as we've seen over the years, drives have gotten bigger and the cost per terabyte has gone down. So that certainly is something you're seeing in buying trends for our data centers as well. And then we've got the sort of the Honor Roll as we're calling it this quarter, where you see 0s and 1s, so no more than 1 failure.

And I'm going to kick it over to Pat to talk about whether this matters or not.

It's kind of a bit of fun. Actually, if we go -- you're there already. So yes, it's kind of a bit of fun. So if we look 0 failures by definition means you have 0 annualized failure rate. But if we look at the drive counts here, for some of these, they are very low. I mean, 187 drives, 247 drives, that's like almost nothing compared to some of our hard drives in operation. In fact, if I go back a couple of slides, we can see we've got 40,000 of the 16-terabyte Toshibas here in -- towards the bottom of the table. So really, these are a bit of fun, but it lets us call these out.

And in some cases, like the Seagate 12-terabytes and the Western Digital 26-terabytes, we're getting towards 100,000 drive days, over 1,000 drives in operation. So we're starting to see some significance there. 1 failure out of 1,200 drives. We're hopeful that, that drive will continue to perform, but we're literally in the first quarter of operation there. You see we've -- 1,200 drives is a vault. So if you're wondering what I'm talking about, our storage is split into storage servers, which we used to call pods. I'm not sure they're officially pods anymore, but each pod has 60 drives and 20 pods chain together in what we call a vault. So that's our deployment unit. So we've deployed 1 vault of these 26-terabyte drives. And we've seen 1 drive failed out of that vault. So that's pretty encouraging news, a 0.4 annualized failure rate is pretty good.

Yes. I think we both -- when we saw that 1,201, we were really excited to contextualize the stat because when we're talking about normal migration process, you try to deploy a full vault of drives at once. So we were like, oh, look at that, you can see very cleanly 1,200 full vault, 1 failure, 1,201.

Yes. It's always quite gratifying when the numbers fall out like that. Over time, we replace drives as they fail. It's not always possible to replace them with the same exact model. So you do get into this situation where there's like 465 dives of this model and 112 drives of that model, these 16s they're only different in an ultimate character of the model ID. So those are mixed up together in a vault. But it's really nice when you see, okay, we deployed 1,201 and 1 failed, so we still got 1,200.

Moving forward. So what I always like to do is we talked about our Honor Roll, so to speak. But there's also -- any time you see a significantly high failure rate, we want to talk about that, too. So what you see when you're looking at this HGST, which I do every time, this 8-terabyte drive, we saw a 10.29% failure rate. That's pretty high. And you can see from the historic numbers like this actually isn't a vault that's performing badly traditionally. So that says that something happened.

And now, of course, any time I get into investigation, I'm actually looking back at the quarter before. So somebody has already done this. Somebody saw that there was an issue and looked into it. And when I went ahead and checked it out, what happened with this drive was it was a little bit hard to identify. There weren't any issues with temperature we could see. There was a potential that maybe because this is all actually 1 vault of drive, so there's something going on with the chassis or vibration. But frankly, with the age of the drive, it's over 7.5 years old. We just said, all right, this drive is telling us it's time to rotate out. So instead of spending a lot of time to really get down to what happened or even doing something like replacing the chassis, we just said time to migrate. So we'll see that one come out of the population as well.

And the other interesting one was this Toshiba 16-terabyte drive, which if anybody was here last quarter, we actually had a huge spike in that drive here, the 16.95% and had done some investigation and saw that there was some very normal updating going on in one of our data centers that read as a failure. And at that time, I said, this is going to come back down as we see this work settle out. So we always like to report back in. It's still a little high, 4.14% is a little high, but I wouldn't call it an outlier, although I didn't run a quartile analysis this time around. But it's definitely coming back down. We're probably seeing the tail end of that work and then it will be coming [indiscernible].

Exactly. Yes, I guess the work spilled over into the beginning of the quarter, and that's what we're just seeing as it settles back down.

Yes, absolutely. All right. So moving on, let's talk annual data. Pat, do you want to take it away from here?

Yes. So with the annual data, we're looking at essentially the same set of drives, but spread over 12 months instead of 3. So what you'll see is a little bit of variance in the numbers, obviously, around 4x as many drive days, around 4x as many failures. I don't think, it's -- I don't know if we've ever had a 0-failure drive model across an entire year.

I'm not sure. We certainly didn't this year, but that would be a fun query to run. We should check that out.

Right. And what we -- again, like very small drive numbers here, we see 1 failure. We see this -- since this 26-terabyte drive was deployed in Q4, this row was exactly the same. That's the quarter snapshot. But all of this is across the whole year. And yes, we see much more -- much less variance in the AFRs between drive, it's kind of smoothing out over time. So I think the highest there is just about 6% for this one and that's the drive that we were just talking about. I think that's the one with that elevated rate. So yes, I mean, that's the annual data, not a lot to pick out, I think, in the table. But we do have drives that averaged less than 2 failures a quarter. So this HGST, 4 terabytes, some of these Seagates. So those are our -- those are the ones that we like to look at and say, hey, that's great. These drive models are where we're seeing the least trouble.

Yes. But it is also just like we're talking about how sometimes the numbers can be a little deceiving. It's worth saying that it's not evenly distributed. So I believe that top model there, the HGST, the 4 terabyte, they had like three failures 1 quarter and then the rest of them 0, 0 and 1 or something like that. So the history of each drive, I think it's something that we bang this drum quite a bit where we're saying that models have a lot of model to model, you really can't compare because they have a lot of like very internal variance and sometimes it's hard to tell what actually is creating a failure. And so that's, I think, the most fun part when we slice the data differently with different exclusions or look at things like this because you end up always coming back to the fact that the more precise you can be with the line item data, the more you really understand a single model.

Right. And really one character difference in the model ID, it can be -- it can behave as a completely different drive. There may be firmware changes. There may be -- may even be like, I don't know, a different motor or something like that as they went from one model ID to another. So it's even hard to generalize across same capacity, same manufacturer and extensively the same drive. When you go buy it, it just says BarraCuda 16 terabyte or whatever. But it's -- really, you've got to look at that model ID to get the whole picture.

Absolutely. And I think what's fun about the annual data, too, is that we actually compared this too. We wrote another article at the end of the year last year that was based on some internal metrics about where our data center techs were spending work. So we were able to sort of compare not just like this, you can see the total time spent where people were working for data center, this is in hours. And one note there, Canada, we actually deployed that in March. So it's looking like it has low hourly metrics. But in fact, we actually tracked the launch a little bit separately from our normal work center activities. So Canada is -- we love all our data centers, but I think that Canada is getting a fair read here or an unfair read here.

But kind of no surprises here. U.S. West is our largest footprint, right? And then you've got a lot of ongoing work. You can see where things were happening. And then comparing that to what the most replaced drive was by capacity, I thought was very interesting. And you'll note that in this data, you're going to see some drive sizes that don't show up on those annual and quarterly, and lifetime charts.

And that's because this is inclusive of all of the drives in our data centers. So this would include boot drives and SSDs and all that other stuff and drives or there's too few of them to make it into the quarterly data yet. So if you see that, that is why. But being that we are interested in who's doing what, you can see that this sort of makes our 12 terabyte and 14 terabyte and 16 terabyte drives look bad. And that's not really fair because these are raw numbers, and we know that we have a solid concentration just by quantity of those drives, but also that they're aging. So they look like they're doing bad, but in fact, it's probably the fact that we just, by the numbers, have more of them in our drive fleet.

And I'm going to say, if we just go back for one second. I think -- I mean, I'm going to guess. I haven't analyzed it specifically, but I'm going to guess that a lot of these replacements -- these replacements because of failure or just replacements?

Just replacements. So this could be normal migration activity we're seeing here.

Exactly. Yes, that's what I was going to say on the 8-terabyte drives, that is what we're seeing there is likely the migration just as vaults age, and we do these planned migrations, what we call CVT - Cluster, Vault, Tome. I think we've got a whole article about that.

We do. Great article.

But we migrate your data from maybe, whatever, 5-, 6-year old vault, whatever, onto a new vault. And that frees up rack space. We migrate from an 8, say, to a -- let's make the math easy. We migrate from an 8 to a 24 and we can have that same mass data in 1/3 of the footprint in the data center. So that's one of the motivations as well as drives aging and not wanting to get into like huge numbers of failing drives. It's just more space-effective to have the data on higher capacity drives.

Yes. Storage density, definitely something we're interested in. And I think that when we move forward, looking at what our drive capacity is versus the population, it's funny because this sort of shakes out almost like a bell curve, right? So we did a number of drives here where it's 26.1%, 54% or 51% or so on the 14 terabytes to 16 terabytes and then 20 terabytes. So validating our earlier claim that like by the numbers, you just have more drives in the 14 terabytes to 16 terabytes, that's definitely sure. But the other thing I think that's interesting here is if you think about the age of the drives and compare the age of the drives by the percentage, you're going to see different -- it may look like a bell curve in terms of the numbers, but we know that we're shifting to those higher capacity drives because the 20 terabytes are younger.

And the other thing here, too, is speaking to your point about storage density, Pat, is that 22% of our drive fleet is 20 terabyte plus drives. That represents a lot more capacity than the 26% of 0 to 12 terabyte drives. So just some interesting notes there when you think about how we're presenting this data to you. We're giving you a raw number of drives, but there's other things to think about as well.

Yes. There's quite a lot of nuance in the numbers.

Certainly. And then this is a comparison of our last 3 years of annual data. I always love this as a top-level view. Pat, do you want to take us through this?

Yes, absolutely. So what you're seeing here is like from the highest level view, our annual failure rate has been dropping quite nicely. So 2023, 1.70% and then to 1.57% in the next year. And then for 2025, 1.36%, which is great news. We're having to swap out many fewer failed drives. And you're seeing -- this is an interesting view as well as you see these new drives pop up. So you see this 26 terabyte drive there, the 24 terabyte coming in last year. So what is that 4,000 sets, 4 volts of the Toshiba 24. And then interestingly, you've got -- well, we've got Seagate 24 terabyte as well, but you've got lower capacity new drives coming in. And this is just -- okay, why are we buying a new model of Toshiba 16 terabyte. Well, the older generation just ages out of the inventory. And we buy the closest to swap in, which is the '09 Variant rather than the '08. And that's why you see these new drives joining the fleet. So a couple of those, yes, just if you're interested, why is this showing up? Well, that's the Seagate 14 and Toshiba 16 that we're just buying the current drive model.

Yes, which I think we haven't done this, but I'm actually now thinking that would be an interesting comparative analysis for that model-to-model stuff you're talking about because what we're saying there is basically they're the next generation of the same drive. So if anybody wants to -- I see this come up in the comments section all the time, like I can't find this one with that one character difference or how are these two 16 terabytes distinct? And when you see something like this, that might be a good way for you as a person buying drives to sort of compare drive model over time with those slight differences and see what works for you because obviously, a data center use case, very different than a consumer use case, which will…

Very, very. We spin our drives 24/7.

They never stopped working. All right. And I think one of the cool things just reflecting back on Pat's point about how we've been seeing the drive or the AFR change over time as this is the 4 years. So it's a little bit of a cheat because I gave you 3 years on the first chart previous, but this is 2022 to 2025, and we can see really that we're returning to this 1.37% rate. What does that mean for the quality of the hard drives, probably not much, right? Like we're just seeing same drives performing differently over time. And I like to use these moments to always reflect that like when we talk about these numbers, just like I said, we can remove drives without failure. What we have here is a little bit of a curated number in some ways, right? Like low failure rates in our data centers are reflection of all the work that those data center techs are doing on an ongoing basis and that our storage engineering team is doing on an ongoing basis to make sure that we have a healthy drive population.

And it's impressive as well that the annualized failure rate reduced despite those couple of drives where we kind of saw incidents that raised their failure rate temporarily. Even with that in the numbers, -- it's come down from the previous year. So that's quite gratifying.

Yes. So shout out to the folks doing the work. We love it.

[indiscernible]. We love you.

So we've got about 5 minutes here. So we'll go through the lifetime data somewhat quickly, but we don't have to be superfast about it. But I do see there's a question or two, so we want to get to that if we can. So lifetime data, 1.30% is our failure rate here. And just like with the other tables, you'll see maybe fewer models here, and that's just because they haven't reached the minimum standards to be on this table yet.

To explain what we mean here, this one is a little bit different in that rather than being like a quarter or a year, what we're saying is for these drive models that we're currently using, if we go back over the lifetime of all of those drives like within those models, what annual failure rate do we see? So this is like trying to really bring together the most data. And you'll see some of these like with 80 months or 90 months or even 100 months here that we've -- that's the count of like all of those drives that we've ever deployed.

Yes. Yes. If I were to make an analogy, I feel like this is the -- it goes down on your permanent record number.

So yes, lifetime AFR is 1.30%. It doesn't tend to change much because it is such a large data set. In fact, let me just go back. 557 million drive days. It's pretty significant data. And we got a new entry, that 24-terabyte. We deployed another 2 vaults bringing us to like 4 vaults total, as I mentioned, 4,800 drives and some change, some failed drives there. So we see new drives coming into this lifetime table every few quarters.

Yes. And I think it's also worth calling out that like we're talking about how volatile the AFRs can be depending on how we shape the numbers, right? The lifetime AFR has been extraordinarily consistent. I think the last 4 quarters, we've basically gone up or down by 0.1%, which speaks to the fact that you -- when you collect enough data, you can really sort of find the sweet spot in the numbers.

Yes. Absolutely. We've got a bunch of links. So as I mentioned, there are attachments below the screen there, below the video output. So the year-end report, you can see all the blog posts in the series, join our newsletter and visit our home base. If you want to contact us, you can e-mail us at [email protected]. We're on the socials, just get in touch with Backblaze and the comments section on the blog post, we love to see your comments on the blog. So with that, we do have -- we do have a couple of questions. So Abraham is a student sharing his information with this class. Awesome. Since we're covering RAID and -- okay, so Abraham's class is covering RAID and its level, how does the RAID level used affect its statistics? Does it affect them at all?

That's a really good question. Now RAID, we're not actually using RAID in our data centers. We use an algorithm called Reed-Solomon erasure coding, where we split the incoming data into so many chunks and add a number of parity chunks that is variable. And so we split, say, 15 data chunks, 5 parity into 20. So the number that we're seeing there can be -- is what we're reporting on here. But RAID tends to be smaller numbers of drives linked together. And I don't know, would you see any difference, I don't think so. I mean I think it's more in the usage patterns, the volume of data being going back and forth from the drives maybe that's moving the drive heads across the data rather than the RAID level itself.

And I went ahead and linked one of Pat's articles that he [indiscernible] previously about NAS -- specific to NAS, but RAID levels, right? Because I think it does a really great job with the visuals in that article. But one thing about RAID drives and specific to NAS is that you're really just talking about different redundant storage patterns, right? So when you think about creating a full backup or what it would mean to restore, this is really giving you a logical pattern to how to store your data. And to Pat's point, we use Reed-Solomon at a code level for that. But if you're talking about applying hardware failure rates to what you would want with a RAID configuration. Really, you're just talking about how many drives do you need to achieve, whatever rate configuration you would want and how much redundancy do you need to build in for the 1 drive failing. So you can always add another drive, as I say, and create a different configuration if you want more confidence.

And one last question. Well, it's more of a suggestion than a question. When we report on those 0 failure drives, we should order them by the number of drives deployed because then you see like the real stars, if there's 1,000 drives, 2,000 drives with no failures, that's kind of more significant, more interesting than 200 drives. So yes, I think we'll probably do that and start reporting on those drive numbers as well as the 0 just so you've got a little bit more context there.

Yes, I totally agree.

With that, I think we have…

I think we've got one more in here from [indiscernible], and I know we're about a minute over. So if anyone needs to drop, feel free. But just to answer this question here, how do we achieve which mechanism and means that we do to replace or migrate the data from the drive when it crushes without -- I'm assuming that crashes without the user noticing that something is wrong.

Well, the first thing is we're a heavily surveilled environment. But I think that, that gets into actually to Jose's question about redundancy, right? We compared with what a consumer does or needs, you're usually talking about having 1 drive that may or may not fail. Maybe you have a few in connection. But within our data centers, we have data so that we can rebuild a file from as few as 1 parity shard, which -- and we store that 20 times across the data center in different configurations, which is that Reed-Solomon erasure code we're talking about. So if you want to check all those things out, I linked to those resources above as well. And the other thing is that what we mentioned earlier about the CVT framework we use to be able to migrate data effectively, be proactive about it, it's a really interesting process to look into.

Yes. And also, I think we've talked before about what makes a failure. And we'll watch statistics like -- so those smart stats I mentioned. We'll watch statistics like unrecoverable errors. And if those start to show a pattern, we don't wait for the drive to actually crash and then say, I can't give you the data. We swap it out before then so that we get ahead of that event. And often, we can clone the drive, which is a lot faster than rebuilding it from the erasure coding and get it back into -- get that data back up to its reliability level much faster.

Absolutely. So thanks for the questions, everybody. We're going to go ahead and jump off since we're a little bit over. But we always appreciate the interaction. We always appreciate you showing up, and I'll see you next time.

All right. Bye for now.

Hello, and thank you for standing by. My name is Tiffany, and I will be your conference operator today. At this time, I would like to welcome everyone to the Backblaze Third Quarter 2025 Earnings Call. [Operator Instructions].

I would now like to turn the call over to Mimi Kong, Investor Relations. Mimi, please go ahead.

Thank you. Good morning, and welcome to Backblaze's Third Quarter 2025 Earnings Call. On the call with me today are Gleb Budman, Co-Founder, CEO and Chairperson of the Board; and Marc Suidan, Chief Financial Officer. Today, Backblaze will discuss the financial results that were distributed earlier. Statements on this call include forward-looking statements about our future financial results, the impact of our go-to-market transformation, sales and marketing initiatives, cost-saving initiatives, results from new features, our ability to compete effectively and manage our growth, our strategy to acquire new customers, retain and expand our business with existing customers and the impact of previous price changes. These statements are subject to risks and uncertainties that could cause actual results to differ materially, including those described in our risk factors that are included in our quarterly report on Form 10-Q and our other financial filings. You should not rely on our forward-looking statements as predictions of future events. All forward-looking statements that we make on this call are based on assumptions and beliefs as of today, and we undertake no obligation to update them, except as required by law. Our discussion today will include non-GAAP financial measures. These non-GAAP measures should be considered in addition to and not as a substitute for our GAAP results. Reconciliation of GAAP to non-GAAP results may be found in our earnings release, which was furnished with our Form 8-K filed today with the SEC. You can also find a slide presentation related to our comments in the webcast, which will also be posted to our Investor Relations page after the call. Please also see our press release or presentation for definitions of additional metrics such as NRR, gross customer retention rates and adjusted free cash flows. And finally, we will be in New York to participate in the Craig-Hallum Alpha Select Conference on November 18 and the Needham Tech Week one-on-one event on November 20. Thank you for joining us. And I would now like to turn the call over to Gleb.

Thank you, Mimi, and welcome, everyone, to the call. We delivered strong results this quarter. In Q3, revenue and adjusted EBITDA margin both came in above the high end of guidance, and we continue to be on track to be adjusted free cash flow positive in Q4. Overall company revenue grew 14% year-over-year and B2 Cloud Storage delivered strong results, growing 28%. Now I'd like to take a step back and share how we see the AI industry evolving because AI is becoming central to both our customers and our opportunity ahead. AI is built on models, compute and data. While just a couple of years ago, models were the domain of only a few large providers. Today, they are widely available with literally millions of open source options. So models are no longer a bottleneck for AI innovation. The next component of AI is compute. The GPUs that make up compute have become increasingly available, but access at scale remains challenging, and the market has also become quite fragmented with approximately 200 Neo clouds, such as CoreWeave and Nebius offering GPU capacity for it. That brings us to data, the key differentiator for AI success and where Backblaze helps companies win. Data sizes are exploding as AI expands beyond text to images, audio and video, driving massive storage and performance needs. The teams leading in AI aren't just the ones with the most data. They're also the ones who can aggregate, clean and organize it for today's workloads, move it to whichever Neo cloud they choose, all while preparing it for tomorrow's architectures. Backblaze has already built one of the largest, fastest and most cost-effective storage clouds on the planet, offering throughput of up to 1 terabit per second and priced at about 1/5 that of traditional cloud offerings. Add to that, our free egress and universal data migration that allow customers to easily move their data from other cloud providers where they may feel locked in and then freely send their data to one of the 200 Neo clouds or anywhere else they needed to go. Then layer on our team actively supporting customer success, and it becomes clear why hundreds of AI companies are choosing Backblaze. Highlighting this, Backblaze recently received industry recognition in both cloud security and technology innovation, including an award for B2 Overdrive. These honors reflect the strength of our technology and the pace of our innovation. Now let me share a few customer examples that bring this to life. This past quarter, we signed a new 6-figure deal with an AI start-up focused on large-scale vision language models. The team was small but scaling rapidly and racing to finish key projects. They couldn't predict how their data volumes might grow and needed a high-performance and cost-effective solution that could scale with them. They started self-serve and later engaged our sales team to access even higher performance. B2 Overdrive gave them the performance they needed, and our team made it easy, freeing them up to focus on their critical projects. Another AI customer in the surveillance space expanded their commitment by almost 10x to a 7-figure TCV deal. We won this customer away from a hyperscaler whose platform limited how they could automate their workflow. Our platform offers a valuable multi-cloud upload that streamlined their workflows. They expanded their commitment with us because we offered a predictable and cost-effective way to scale easily. Finally, we had another 6-figure win is powered by Backblaze's white label deal with an app developer in the media space. This was another competitive win from a hyperscaler. The customer was growing fast and needed storage that could support millions of users. After being hit with a surprise roughly $100,000 charge for moving data with the hyperscaler, they came to Backblaze for predictable pricing and scale. Backblaze is now their storage platform, and they will join our 100,000 customers who collectively serve hundreds of millions of end users to our platform. Across all these examples, the reasons customers choose Backblaze are consistent, performance that rivals the biggest clouds, predictable and fair pricing, and people who make it easy to get things done. Whether it's a start-up training models, immediate team moving petabytes, or an enterprise scaling to millions of users, the story is the same, faster, simpler, more affordable, and supported by a team that actually helps. Now I'll share how we see our growth opportunity. In Q3, we grew B2 28%. And for Q4, we now expect B2 to grow in the range of 25% to 28% year-over-year. We're proud of our growth. But this is below the 30% target we set for ourselves at the beginning of the year. To help reach our growth goals, we're launching Phase 2 of our go-to-market transformation, focused on accelerating the velocity of both our self-serve and direct sales motions. For self-serve, we're proud of the consistent growth engine we've built. Now we're going to scale it by making it more frictionless for data-heavy AI use cases and by kicking off robust developer relations. These steps are aimed to deepen our connection with the developer community and make it even easier to adopt our platform. For direct sales, we're building on the progress we've made over the past year. We're adding talent and upgrading our core systems in order to better target customers and improve sales efficiency. To accelerate this, we're also partnering with advisers and a consulting team who has helped both Snowflake and Databricks in their go-to-market execution. We're reinforcing our sales and marketing engine and remain confident in our ability to deliver consistent, durable growth over time. With that, I'll turn it over to Mark to take us through the financials. Marc?

Thank you, Gleb, and good morning, everyone. We delivered a strong third quarter with results coming in above the high end of our guidance for both revenue and adjusted EBITDA. We have solidified our balance sheet, improved our path to be free cash flow positive, and accelerated revenue growth. We're focused on getting B2 to a Rule of 40, and we are on track to roughly triple our score this year. Now, let me walk you through the details of the quarter. Starting with revenue. Total revenue exceeded expectations. It came in at $37.2 million compared to the high end of guidance, which was $37.1 million. This represents a 14% year-over-year overall growth. B2 grew 28% year-over-year compared to the organic growth of 19% in the same period last year. This represents an acceleration of about 900 basis points this quarter. This improvement was driven by the first phase of our go-to-market transformation. As mentioned last quarter, usage from one of our larger AI customers has been variable as the customers' data storage needs have fluctuated for their business. Overall, industry-wide demand for data storage is expected to grow rapidly, and our platform is well-positioned to support those expanding needs. We're also seeing diversification within B2 across our core use cases, which are live application hub storage, backup, media, and AI-related workloads. In Computer Backup, revenue was flat year-over-year, reflecting the final roll-off of the price increase implemented in 2023. Turning to net revenue retention. Overall, the trailing 4-quarter company NRR for Q3 was 106% compared to 109% in the second quarter. Going forward, we believe it's clear to show in-quarter NRR versus the previously reported trailing 4-quarter average. As an example for Q3, in-quarter NRR for B2 improved to 116% from 109% in Q2, and this was driven by a large AI customer we discussed last quarter. You can see that dynamic on Slide 11 of the earnings presentation. Moving on to gross margin. It was 62%, up from 55% a year ago, reflecting operating leverage and the benefit from our recent useful life study. Adjusted gross margin was 79% compared to 78% last year. Overall, margin performance remained stable. Data center costs are generally rising. However, they are being offset by scale in labor, greater code efficiency, and lower amortized R&D as a percentage of revenue. Operating expenses were 71% of revenue, an improvement from 92% a year ago. This reflects the structural changes that we made last year through our restructuring and zero-based budgeting process. R&D spending held steady in dollars, but declined as a percentage of revenue from 33% a year ago to 30% this quarter. When you combine the R&D expense and capitalized R&D, the improvement is even more pronounced at 35% of revenue, down from 43% a year ago. Sales and marketing came in at 24% of revenue, down from 36% a year ago. We're focused on improving efficiency in our go-to-market model, including reallocating funds to strengthen our top-of-funnel programs and investing in operational go-to-market talent, all while maintaining the same cost discipline you've seen from us over the past year. G&A was 17% of revenue, down from 23% last year. We continue to drive efficiencies across our corporate functions and general and administrative spend. Operating results. GAAP net loss was $3.8 million, a 70% improvement from a loss of $12.8 million in the prior year. On a non-GAAP basis, net income was $1.9 million compared to a loss of $4.1 million last year. Adjusted EBITDA margin reached 23%, almost double the 12% from a year ago and a quarter ahead of our outlook. That performance reflects continued financial discipline and the operating leverage we've built into the model. In Q3, adjusted free cash flow was negative $3.5 million, improving by roughly $0.5 million year-over-year. As we discussed last quarter, we used $2.5 million of our line of credit to fund capital expenditures outside of the U.S., so we can cut our international borrowing rate in half. Balance sheet. We ended the quarter with $50 million in cash and marketable securities, largely unchanged from last quarter. We believe the balance sheet remains strong and provides flexibility to support continued growth and investment. As we announced last quarter, we initiated a modest share repurchase program. In Q3, we repurchased $1.2 million of shares as part of our ongoing work to manage equity dilution. Guidance. For the fourth quarter, we expect revenue in the range of $37.3 million to $37.9 million. We're slightly widening this range to account for the variability we see in certain large customers. B2 growth in Q4 is expected to be between 25% and 28%. We continue to focus on driving operating leverage and remain on track to be adjusted free cash flow positive in Q4. To close, our financial transformation is progressing well. We're reducing equity dilution through our repurchase program, on track to achieve positive adjusted free cash flow in Q4 and continuing to build operating leverage towards GAAP profitability. While we haven't yet reached our 30% B2 growth goal, we expect the next phase of our go-to-market transformation to drive stronger growth. Together, these efforts are building a stronger, more efficient company, one that's on path towards operating at a Rule of 40 profile. Just looking at B2, we started the year with a Rule of 40 score of 9 and are on track to roughly tripling that in Q4 of this year. Operator, please open it up for questions.

[Operator instructions]. Your first question comes from the line of Jeff Van Rhee with Craig-Hallum Capital Group.

Just a couple here. In terms of the sales evolution in Phase 2, talk to me about how you were envisioning the sales sort of evolution. I think you commented last quarter that most reps are more than halfway through the ramp. Are the existing reps ramping the way you had expected? Or was there something you experienced in the sales process that led to you sort of implementing the second phase?

Jeff, this is Gleb. Thank you for the question. So what I would say is that in the first phase, one of the things we talked about was that our goal was to move upmarket. And so you could see that we've definitely moved upmarket with the multiple 6- and 7-figure deals that we've consistently announced over the year. The Phase 2 for us is about driving the velocity -- of the execution velocity. So it's a slightly different focus. We want to continue to move upmarket and support those larger deals, but we're spending more time focusing on moving things through the funnel more explicitly and more actively. In terms of the reps themselves, most of the reps were hired at the beginning of the year or earlier. So they are largely through their ramps as part of their onboarding.

Okay. And on the -- just revisit, if you would, on the B2. You had the 30% goal for Q4, and I heard a little bit of commentary there, but just expand on that a bit more. What's explicitly -- what specifically was it that you thought was going to play out that didn't play out? And when do we get to that 30% plus?

So on our -- from our last earnings call, basically, 2 things happened. One is we had a large customer that we talked about in the prior call that in Q2, the outperformance was in part driven by the variability of a large AI customer. And as we look towards Q4, the variability in part is driven by this same large customer trimming more than we expected. The other thing that drove that is just the larger deals as we move upmarket, a number of them have taken longer to execute. And so the combination of those 2 things have changed our expectations for Q3 somewhat. That's why we're doing Phase 2 of the GTM transformation. We love that we've been able to move upmarket. It's proven that the platform supports these larger customers, it supports these larger use cases. And we also want to have more core predictability in the business. So it's interesting because before we used to have lots and lots of small deals and having lots and lots of small deals provided really great predictability, and we were able to go public on lots of $300 and $2,000 type deals. The challenge with those was it was harder to significantly outperform, significantly rapidly accelerate growth. The bigger deals in moving up market allow us to do that, but we want to focus on increasing that core base of smaller deals to drive that consistency also.

One last quick one, if I could, on the data variability, Gleb, you mentioned several times on the call. Just talk about how that variability is maybe different than what you had expected from these customers, obviously, as you're getting AI workloads, some new experiences here as to how people are going to use you both in that moment as well as over time. Just a bit more about the variability on the data usage would be helpful.

Yes. The AI use cases are obviously all evolving very rapidly, and we're trying to support the customers where they are. So the different use cases that we see across AI are things that we are supporting and learning along with the customers. This particular customer that we talked about, as their business goes up and down in terms of the needs for their data, that we support them with that. So the way that we work with customers is many of our customers are pay-as-you go. They enter a credit card and they just sign up and go. Some of our customers, we sign contractual commitments with. And for a number of the contractual commitments, they'll often sign a contractual commitment and then have the ability to scale up above that. And with some of these larger AI customers, they'll sign a contractual commit and then they'll ramp up even faster than that because their business is growing faster than they expected and then sometimes they need to do some pullback. So that's what we're seeing with some of these. The nice thing, obviously, is we have 100,000 customers on B2, we're distributed across a variety of use cases. So it's not it's not all variable AI customers. We still have a lot of core predictability in our business. But we do want to lean in with the AI use cases because we do see that, that is transformative for the industry and where we see a lot of opportunity ahead.

Your next question comes from the line of Eric Martinuzzi with Lake Street Capital Markets, LLC.

Yes. I'd like to touch on - I know you haven't given a guide for 2026, but I just wanted to learn your expectations for the 2 sides of the business going forward. Obviously, we've got good growth, but not the 30% you want in B2. And then we were flat here in Q3 on the CBU as we got through the last of the price increases. But to put it in broad strokes, we're looking at a year in 2025, where combined, we're at about 14% growth. I know I'm looking for that 14% growth to persist into 2026. But what are you thinking about for longer-term growth rate for B2 and CBU?

Yes. Eric, this is Mark. So 2025, right, for B2 is on track to be in that mid-20s year-over-year growth rate. 2026, we don't want to give guidance yet, but we feel pretty comfortable it's going to be in that range given what we're seeing now as well as that second phase of that go-to-market transformation. And those larger deals Glen spoke about, if those come in, I mean, it could bring that number higher when they come in, in those specific quarters. On the computer backup, just like we said last time, it would be low to single, mid-single digits, the contraction of that business. We're obviously taking action to try to stabilize that. We probably need more time to come back and report a different outlook there. But for now, those are the right assumptions for B2 and computer backup.

Okay. And then the restructuring that you announced this morning as well, just curious to know where are we - you talked about some facilities, but it also talked about severance. Where are we cutting back on the headcount that we had versus the end of September?

Yes. Listen, that is how we're funding that next phase of our go-to-market transformation. Gleb talked about needing that mid-market, high-velocity deal volume to improve. It's done really well over the past year, but we think it should be way better than where it is. And that's the key formula to get to that 30%. So a lot of that restructuring cost is to transform our go-to-market practices. It's to bring on new talent on board. We have started that. So it's a lot more around getting talent that's a lot more operational in their go-to-market skills, know how to use the technology, the data science behind it. So I would say that it's more of a reinvestment versus a cost-cutting exercise.

Your next question comes from the line of Ittai Kidron with Oppenheimer.

This is [ Nolan Genvine ] on for Ittai. First, I just want to sort of double-click on these 7-figure customers. You're clearly showing traction in selling to enterprises now, and you've added yet another 7-figure customer this quarter. How many 7-figure ARR customers do you have now? And then how has their expansion dynamics evolved in recent quarters versus sort of the smaller cohort? And then I have a follow-up.

No, thanks for joining us, and thanks for the question. One clarification I would say is we have these 7-figure deals. The interesting thing is they're not all enterprises. They're heavy users of data. right? And so one of the things that we've seen is that customers are generating large volumes of data and needing high performance for those data sets. The AI-powered video surveillance customer that I mentioned, they're not an enterprise company, but they're a 7-figure deal. So, and that actually, I think, bodes well for us for the future because these smaller mid-market companies that have high data needs are generally faster moving than the traditional enterprises, and there's more of them out there. So that's one thing I would just say in terms of outlook and opportunity for us. In terms of the number of them, we've generally announced 1 or 2 per quarter over the last year. So it's not a large number of these 7-figure deals, but it's considering that in our first roughly 15 years of the company, we signed 1. And in the last year, we've highlighted at least one a quarter, I think is a great proof point to say that when we said we were moving upmarket, we moved up market.

Yes. And what I'll add, Nolan, is last quarter, we shared the number of customers with $50,000-plus ARR. It was up in Q2, 30% year-over-year. In Q3, it was up 41%. So we're maintaining good momentum on that front as well.

Got it. That's very helpful. And then I know you're not giving '26 guidance. But how should we think about the potential catalysts heading into the next fiscal year, things like overdrive ramping, this sort of Phase 2 of the go-to-market transformation? Could you just maybe put a finer point on what you think could maybe drive upside or downside, just to the sort of current growth rate?

Yes, it's a good question. So the go-to-market transformation Phase 2 is certainly something that we're leaning into heavily, right? We're putting a lot of focus on them in terms of talent system upgrades, leveraging advisers, and the consulting team. We believe that there's a lot of opportunity in driving the ability and execution for us to get to market, faster with the products that we have. In terms of B2 Overdrive, we announced that product line fairly recently. We've already closed multiple 6-figure deals on there, which is, I think, a great sign that it has product-market fit. It's we believe it's the highest throughput per dollar offering on the market, and that's a great fit for customers that really need that performance. One of the other things that we've seen on Overdrive, which has been interesting, is that we've seen customers come to us because of Overdrive. They heard about Overdrive, they heard about the performance availability, and they came to us for that. And then when they showed up, they realized that our B2 standard offering is actually quite a high-performance offering, and it was sufficient for them. So I think it's changed some of the perception around what we offer. And so even customers that aren't signing up for Overdrive are often coming to us with these higher performance needs, which is a great direction for us as we become a more strategic partner for them. Overdrive provides up to 1 terabit per second of throughput. So it's a blistering fast offering for them. So I do think that the GTM transformation is the thing that we look at as key to the overdrive offering. The other thing that I would just mention is I don't want to undersell the Phase 1 transformation that we did, right? The Phase 1 transformation was all about moving upmarket, and that is continuing to be an important catalyst for us, including for 2026. Mark mentioned that the big deals are something that can accelerate us even further, and that's something that we're excited about. The last thing I'll just mention is that the GTM Phase 1 transformation, I mean, it doubled pipelines, it doubled bookings, doubled channel. So it did help quite a bit. And so we're excited for Phase 2 of that.

Your next question comes from the line of Simon Leopold with Raymond James.

I wanted to ask about this Phase 2 initiative, in that I don't imagine there are free lunches and have to assume that there's some investment involved. I know earlier, you talked about sort of reallocating. But if we think about it, your sales and marketing have been running between roughly 21% and 23% of revenue. What are you budgeting for next year? How should we think about either a dollar basis or a percent of revenue basis, but some metrics to try to get an understanding of what investment you're making for this Phase 2 initiative?

Simon, it's Mark. The percentage of revenue for sales and marketing should stay stable as a percentage of revenue. The funding is going twofold. The restructuring allows us to fund the one-time cost that's not in that percentage of revenue. That allows us to drive the transformation work, which is really to rejigger all our go-to-market systems to make them work better together, cleanse the underlying data, and drive that data science capability I was talking about. The talent refresh we're doing is within existing OpEx budget, right? As one of you highlighted earlier, yes, there are severance costs that are one-time charges that fit into the restructuring. But on a recurring basis, when we finish this as a percentage of revenue, it should be pretty stable.

Great. And then in terms of the outlook for B2, slight downtick, nothing to be embarrassed about, high 20% growth. But what changed in your mind versus your expectations when you set the goal? What's different?

Well, we set the goal all the way back when we launched our first phase of that transformation, which was exactly a year ago. That was a while back. That is our aim. Frankly, it remains our aim. And our Phase 1, as Gleb said, doubled our pipelines, doubled our bookings, doubled our channel business. What we noticed candidly is our inbound motion is really strong, right? Like we addressed a lot of our technical marketing content, or blogging. We started doing more events and webinars, and all that really improved our inbound. I mean, our inbounds are up substantially. The inbound pipeline is up 100% year-over-year. Our outbound motion is newer to the company. So that muscle is newer. That's the muscle we're working on fixing now in this next phase of this transformation. And it's unfortunately, one of these things, like until you get into it, you don't know how much you don't know. And we're realizing it's a muscle that needs to be a lot stronger to make it successful, and we're hard-charging at it now. Our aim is that, when we do our Q4 earnings release in February to give a lot more details around what makes up that Phase 2 go-to-market transformation. When should you expect a change in the outcome from it, an improvement, and how that would address the year-over-year B2 revenue growth?

Your next question comes from the line of Zach Cummins with B. Riley Securities.

Gleb or Mark, either one. I was wondering if you could give a little more context around some of the larger deals in the pipeline. It sounds like maybe those were pushing a little to the right when it came to the B2 side of the business. So any additional context around just the pipeline that you're seeing with some of these larger deals and maybe the reason that some of them pushed to the right?

Yes. Thanks, Zach, for the question. So one thing I'll say is when I looked at some of the deals and dug into it with Jason on the sales side, frankly, there wasn't any pattern in them. They were kind of random reasons. signer got sick, a different project internally came up that has to take priority, et cetera. So there wasn't any kind of clear pattern for why, but it just spoke to us about that in the small deals, a lot of times what happens is it's one person who is the person that's interested, the decider, the buyer, the purchaser, the user, and they go in, they make a decision and they go. In these larger deals, it's just more complex where you have a buying committee, you have the various different oftentimes security compliance reviews. There are sometimes different departments that need to be involved for how it's going to get used, how the migration is going to happen, et cetera. So it's just one where as we're moving upmarket, we're closing these bigger deals, and that's fantastic. And at the same time, we want to be realistic about that some of them take longer than I think we were seeing because we also have seen some large deals close very quickly. The BT Overdrive deal at the beginning, the first one that we announced last quarter closed incredibly quickly. Some of the customers that I talked about in my prepared remarks, when I was looking at them, the 6-figure deals that from start to finish were three months. And that's quite quick for a 6-figure deal from the first conversation to fully onboarded. But we're also realizing that not all deals move that quickly.

Understood. And in terms of the Phase 2 of the transformation, I'm sure we'll get much more detail in your Q4 earnings call. But can you give us a sense of some of the things you're looking to improve within the self-serve motion? Is this largely targeted towards some of these faster-moving higher data usage customers and reducing the friction there? Or anything you can provide there would be helpful.

Yes. It's interesting because we built the company basically on the self-serve motion, right? When we went public, almost the entire business was driven by self-serve. We had a very, very nascent sales motion at the time. And so the great thing was that we built a blog that a few million people a year would read that drove a lot of inbound interest into the company. We had people will be able to sign up, enter their e-mail just enter a password and go and then try it and then get a credit card and sign up. And as we moved upmarket, we had said concretely at the beginning of the year that our focus was going to be on that upmarket move, and we were not going to be doing a lot to change the self-serve motion. But at the same time, our team was focused on that the whole SEO world was changing, right, with the way that companies and individuals were searching was changing and moving to AI use cases where people would look on ChatGPT or Anthropic or whatever to find information. And so, they actually leaned in on making sure all the content was updated and positioned well to be read by these applications. And one of the things we saw was that our self-serve accounts account creation was -- is actually up 56%. And if you -- as you look to a lot of the companies out there for whom they're self-serve or driven by inbound type content, a lot of them were down quite a bit because they hadn't adjusted to the new AI chat type of search algorithms. So the team leaned in and supported that quite well. Going forward in this Phase 2, one of the things that we're doing is really focusing on how these new AI native start-ups and developers build themselves and ensuring that we're well integrated through the whole life cycle of those workflows and making sure that it's really easy for them to learn about and then adopt the platform. So it's a lot of work in terms of making sure that the content, the guys, but also the flows and the integrations are all there to support those data-heavy use cases. One thing I'll mention is one of the customers I gave on the call, they came in as self-serve. They started just by themselves. And then as they wanted even higher performance, even larger deals, they requested to reach out to the sales team, and they had a conversation with us and then they bought B2 Overdrive. And that's a motion that we love to see.

Your next question comes from the line of Mike Cikos with Needham.

This is Jeff Hopson on for Mike Cikos. I just wanted to see if there is any more info on how the power -- the Powered by white label solution is going. Obviously, NeoClouds is coming more popular trend, and it seems like that could be a good place for that or any sort of partnership to offer their customers more flexibility. So maybe just any info on that opportunity.

Thanks, Jeff. Good to have you on. So we're actually pretty excited by the Powered by. As you saw in my prepared remarks, one of our largest deals this quarter was a Powered-by deal. We also have seen some of our channel partners actually adopting Powered by where instead of trying to resell and integrate at the customer level, they actually build it into their own offering and offer it directly. And then, same like you said with the NeoHubs, we have various discussions that are in progress around that front. It's certainly an exciting opportunity. There's about 200 of them out there. So they all have GPUs, and they all have storage needs. So we engage with them today in a variety of ways where we service our customers using the NeoCloud by providing this open platform, free egress, high throughput, but also are in conversations with some of them to help them with their storage needs directly.

And maybe hardware storage has been on the top of investors' minds recently as AI video generation comes into the forefront. And I know you guys called out some AI video wins with customers. Just curious if you're seeing an actual uptick in AI video companies just in the past 6 to 9 months, as those models have kind of have become more popular.

I'm sorry, can you repeat the very first part of the question? Did you say hardware storage?

Yes. We've seen in the market hardware storage has become..

So initially, AI was all about text, right? It was generative AI for text. Then it became generative AI for images, then audio, and now video. Obviously, each of those is in order of magnitude bigger in terms of the data sizes. So, video is a very data-heavy cloud format. So we absolutely are seeing customers signing up for that. I was looking, we had one of the interesting AI, Gen AI, video companies was a recent customer this quarter. There was another large one in the prior quarter. And we have hundreds and hundreds of AI companies. So obviously, I don't know all of them, but just as they come up, I see ones that catch my attention. So it's something where they create models for the video that requires a lot of data on the front end, then it requires the data to generally get sent to one of the Neo cloud providers for the model creation. And then the inferencing itself, where they're generating the video, that video needs to go somewhere. And so we're supporting customers on the various fronts. One thing I'll tell you is that AI in general is a space that we're seeing significant adoption. Today, about 1/4 of all of our new business is coming from AI companies. Over time, it's going to become harder to say what is an AI company and what is just a company using AI. But today, about 1/4 of that new business is actually coming from companies that are specifically in the AI space. So it's an area that we're excited to lean into.

Your next question comes from the line of Rustam Kanga with Citizens.

Great to see the outperformance and new high watermark on the net income and adjusted EBITDA there. Mark, given your prepared remarks, it sounds like there's going to be a lot more honing in on the operational go-to-market skills in Phase 2. And given the success from Phase 1 moving upmarket, getting these larger deals, perhaps you have some better picture on the talent most equipped for what you're looking for in Phase 2. Is that reps with a prior focus on cloud computing and storage or more well versed for lack of a better term, the language of AI? Ultimately, any insights you can share on what kind of reps have been the most successful on the direct sales front and that you might be looking to replicate in Phase I?

This is actually Gleb. I'll start, and then Mark can join in if he wants to add as well. So first of all, what I'll say is in terms of talent, a lot of the talent is actually on the systems and operational side of it. So we're looking for a top-tier Rev ops person, the consulting firm that we're working with, we're doing a large project around both the systems transformation as well as the sales enablement and execution side. So Mark mentioned that we have a success with inbound. At the same time, what we see is there's a lot of stuff that comes at the very top of the funnel. And then there's a lot of opportunity to be more efficient and effective in having it go through the funnel. And then on the outbound side of it, one of the things that we realized is that we can do better with targeting the right types of customers who are ideal customer profiles, both in terms of identifying them and also in terms of how we reach out to them. So a lot of the work in the transformation isn't even specifically about the different reps. It's about the infrastructure of people and systems around the reps to help them.

This is Mark. Our reps have done well. Our win rate is 30% from opportunity to close, which is a very healthy win rate. So we just want to be flowing more opportunities through that team. And then as that opportunity flows through that team, you start expanding capacity and optimizing capacity within. So as Gleb said, there's a lot more around making things available to every stage of the process in a way where people are following up at the right time. They got the right content, the right messaging, the multimodal approach to the customer, when you place advertising versus e-mail versus text message. So there's just an incredibly scientific way of doing that these days. And I mean, the people helping us are people who help companies like Snowflakes and Databricks. So they bring the best-in-class on that process. And we want to adopt it and drive a lot more volume through there.

That concludes our question-and-answer session. I will now turn the call back over to Gleb Budman for closing remarks.

Thank you, everyone, for joining us. With the double beat this quarter and how well we're positioned to help companies with AI workloads, we're enthusiastic about our opportunity ahead. I want to thank our employees, our customers, our partners, our investors for being on this journey with us as we build this core storage backbone of the Internet. For our investors, we look forward to seeing you at the Needham Conference and at the Craig-Hallum conferences this month and chatting with all of you next quarter. Thank you.

Ladies and gentlemen, this concludes today's call. Thank you all for joining. You may now disconnect.