Good? Perfect. All right. Thank you for joining us for the, at the end of the day here for the Needham, Made Tech Conference. This last one's going to be a fireside with Gleb Budman, CEO, founder over at Backblaze. Thanks for joining us. Looking forward to this. I have questions on my side, but it would be great if you guys wanted to lob them in too. Maybe we can get a little bit lively at the end of the day, but we'll see. With that, just to level set, for Backblaze, can you give the background for overview of, who Backblaze is, what is it they do, what was the initial vision for this company when it was founded?
Yeah, absolutely. And also appreciate being here. And also, hopefully we can end here on a high note. Backblaze is a cloud storage company. We provide inexpensive, high-performance cloud storage to a broad swath, but especially businesses. We compete with the traditional cloud providers, Amazon, Google, Microsoft, specifically focused around object storage. We started the company 17 years ago. My co-founders and I started originally focusing on backup as the use case, but then we built a high-performance cloud storage platform to support the backup use cases and then got pulled into serving many other use cases for customers, including supporting media, application storage, and most recently AI.
Just for context too, object storage, what is it and then why start, or why the focus on object storage versus other types?
Yeah, so object storage is basically the default underlying storage of the internet. If you have an app on your phone, if you visit a website, if you're an enterprise, chances are the underlying storage for that is object storage. It's increasingly the direction of the future for that because it allows you to have large-scale, unstructured data that is generally cost-efficient to store. It's an ideal form of storage for large-scale over-the-internet type storage.
From a positioning and pricing standpoint, that's probably one of the more unique pieces of the Backblaze story. Talk about the incumbents and then how is it you guys are able to enter this market and compete as effectively as you do from a pricing standpoint?
Yeah, so compared to the traditional cloud providers, Backblaze is about 80% lower cost. We provide free egress, which sounds subtle potentially, but it's actually a huge impact on customers because customers need to move their data to different places. Egress is a form of locking people in to stay inside of a single walled garden provider. We make it free and easy to move your data around, and it's high performance. We launched a patent-pending functionality called Shard Stash. It enables us to be up to 30% faster than the traditional cloud providers for uploads. We actually just launched, a couple of weeks ago, something called B2 Cloud Storage Overdrive, which is an incredibly high throughput offering for large data sets, large performance needs.
Great. Customer use case, like we're tapping into the initial founding was for backup. Then based on your needs, you expanded into storage. Now you're, I guess, I don't know if you want to call it product and packaging, but you have Overdrive, which you guys just recently announced. Can we start talking about the throughput-intensive workloads that Overdrive is meant to handle, the customer pull in that direction to launch that for the market and what initial customer feedback's been like? I know it's early, but.
It is early.
Yeah.
What happened was we've always had a platform that's high performance. We've had customers use us for all sorts of different use cases. I mean, our customers use us for streaming, streaming video, streaming audio, for surveillance, for a variety of application storage use cases for analytics. These are already all high-performance workloads. What we were seeing though was customers saying, especially because of AI, I'm collecting this large data set and I need to move it around rapidly. Can you help me with that? We found that our platform was able to support them in those use cases.
What we did was we then launched Overdrive, which did not require a lot of additional R&D or a lot of additional capital expenditures because we were able to tune the platform to support those use cases for those customers at a higher price point. I'll give you, you know, one example. One of our customers has an application where they are scouring the entire internet. They're collecting basically a copy of the internet. They're doing that in order to package it up and sell that data to companies building LLMs. Right? That's a large data set. They need to put that data somewhere. Once it's ready, and they've sold it to a customer, they want to move it to that customer as quickly as they can.
What we're able to do is allow them to upload all of the data to us, store that data inexpensively at 80% less than they would be storing it with the traditional cloud providers, and then very rapidly egress it to one of the providers. That's an example of a use case which is ideal for an Overdrive-type customer.
When I think about like the throughput intensity, and this is, sorry, this is probably a basic question for you, but I'm trying to think through, like, I have all the, the volume of the data understood. It's, it's, you can't deny just how much data is going through these AI workloads. The second piece is, like, is it also just the delivery of it is almost like instantaneous to a, a greater number of endpoints versus what it's historically been or not necessarily?
It's about, so there's different kinds of performance.
Mm-hmm.
Right? One example is on the upload side. On the upload side, the reason we launched Shard Stash, which was a, it's patent-pending technology to get data uploaded faster, is because one use case is backup. And enterprises who want to back up a lot of data, they use a variety of different tools to do that. You know, common tools are Veeam and Rubrik and Cohesity and, you know, Commvault and other things. Almost all of these tools, they take that data set, they chop it up into small files, and then they upload it to the cloud for safekeeping. In order to support large enterprise data sets getting protected quickly, we optimize the platform with Shard Stash to enable lots of small files to get uploaded quickly. That's one form of performance. Overdrive is all about throughput.
It's about moving large data sets rapidly in and out of the environment. It's less about smaller or large files and more about just the mass volume of data transfer. That requires the actual storage infrastructure to be able to move the data quickly. It requires the pipes and all the networking to be able to move the data through them rapidly. We've effectively optimized that entire flow from starting at the hard drive level, up through the server level, up through the network level, and out of the data centers out to wherever they want to send it to. A common location that they want to send it to is all the GPU providers.
Mm-hmm.
Right? One of the things that's changed in the industry is it used to be that a customer would say, "I'm inside of Amazon or Google or Microsoft, and I'm going to stay inside of that walled garden, and I'm going to use the services they provide me." That was fine for many years. Increasingly, customers are saying, "I'm okay maybe with using some of their services, but I also want to use that service, that service, and that service over there." They may want to have a service that they use with Amazon and a service they use from Google, or they may want to have a service with Google and use one of the new GPU providers like Nebius, CoreWeave, Vultr, OVH, you know, Lambda. There's a whole list, right?
Being able to get their data to those places requires you to have the data somewhere where it is free to travel to these different locations. If you have a lot of data and you need to get it there quickly, especially because GPUs are very expensive, you don't want to have them sitting idle. You need that kind of mass throughput as well.
I know you've spoken about, I guess some of the differentiated IP that you guys have in-house, that was able to underwrite this Overdrive launch, right?
Mm-hmm.
Can you just walk us through what you guys did from your basic B2, basic V1.0, if you will, and then what was needed from either a software or hardware capability standpoint to get Overdrive out the door?
Yeah, sure. First of all, the core platform, right? We've been working on the core platform for 17 years, like focused on this one singular problem with a lot of very smart technical people focused on building, hopefully, the world's best storage platform, right? That focus enabled us to build this very high-performance, durable, efficient cloud storage platform. To get Overdrive out the door, what we were able to do is from a software perspective, we were able to optimize the utilization of the platform. If you're a customer coming to us and saying, "I would like to use your platform for backup or archive," the utilization of that is pretty low. We don't have to, you don't need a lot of performance. You just need the amount of storage space.
If you're a customer that comes to us and says, "I want to use you for my surveillance application," that's a customer that needs to upload a lot of data, but not download the data because with surveillance, it's mostly inbound, footage that sometimes, but not very often, gets looked at. For a customer that is selling their large AI data set, that's a lot of getting the data out and getting it out really fast. What we were able to do with the software side for Overdrive was basically optimize, enabling customers to utilize the performance of the platform more aggressively than others who didn't need it as much. That's on the software platform side. On the hardware side, the nice thing is the software for the most part takes care of all the performance. It's all hard drive-based storage, so it's very cost-efficient.
We did make some optimizations because we're seeing more performance demands from our customers in general. We made some additional enhancements, things like some CPU upgrades across the platforms, some networking upgrades, but things where, unlike potentially others who would've had to spend, you know, tens of millions of dollars to upgrade the core infrastructure, the core infrastructure all works. The software was able to optimize that. We were able to make a thin layer of CapEx improvements to raise the performance level even further.
Like your R&D team would probably kill me for like the oversimplification, but because you have this platform layer, right?
Mm-hmm.
You can see the different use cases, you leverage software to optimize your pipe to ensure that you could deliver this level of throughput required for Overdrive.
Pipe and storage. I think one of the really key things is the storage platform itself has to enable this incredibly high level of performance. The pipes are a piece of it, but if you can't get the data off of the storage systems, there's no data to send through the pipes. It's really optimizing the storage utilization to enable you to get the data and then send it through the pipes.
From a, so from a customer use case standpoint, and I just want to jump for a second to like net expansion, net retention, that you guys have demonstrated. Backup, archive, storage, like they all seem relatively close. Like they're all cousins here.
Mm-hmm.
Can you walk us through like what is the journey from a typical customer? How long are the commitments that they're signing with you and what that upgrade or expansion cycle looks like?
A lot of customers, we have a couple different go-to-market channels, right? A lot of customers just show up, self-serve. They come to the website, they enter an email, they enter a credit card, and they just go, right? Most of those customers are small, but some of them have grown to be $10,000, $50,000, $100,000, on a, you know, credit card type basis, self-serve motion. We've overlaid sales because we've been moving up market. Heavily direct sales focused approach on that. We also do channel, and we also do an OEM style powered by approach, and that has been moving up market for it. You know, those sales cycles are three-ish months typically. They are typically signing a one-year contract. Sometimes they're multi-year. We have 90% gross customer retention.
An average customer stays with us like a decade. And then we have net revenue retention, you know, on this quarter, we reported B2 net revenue retention of 117%. Our customers get on the platform, they stay on the platform for a really long time, and they expand on the platform when they're on it. One thing we announced in this Q1 was that we signed our largest total contract value customer ever, right? Multi-million dollar contract, over a multi-year period. That customer was on AWS. They decided to switch from AWS and got onto our platform last year and signed a six-figure deal and moved some of their data over. It was a six-figure deal, one-year commit.
One year later, after becoming fully comfortable with the platform, they signed a multi-year deal, moved all of their data over, and upsized the size of the deal. That's kind of the journey, I guess, is coming, trying, starting, and then growing over time.
Largest TCP deal ever. They started with you guys last year with a six-figure. What was that bake-off most recently for you guys versus AWS? Again, that they've seen the value of the platform here. What was that sales cycle to then get that convert and get that large TCP deal? Are we talking again, was it ever since they started with you? Was it multiple months? Like.
The original, I mean, we had our, you know, our first contact with them was years ago, just kind of a, "Hey, you know, this is us, this is Backblaze, and, you know, we'd be interested in talking." At that point, they were all in on AWS and thought that that was just fine and that was where they were going to stay. At some point before they signed last year, they started actually engaging with us because they felt like their AWS bill was getting too expensive. The customer service that they were getting and the engagement they were getting, they didn't love. And they started looking at choices. They did some cursory considerations of other things, but they honed in on Backblaze as the likely choice for them.
They needed to make sure that they felt fully comfortable because it's, they are effectively counting on Backblaze for the existence of their business, right? It's, you know, this is not a backup or an archive. This is the fundamental way their business runs. They have an application, their customers use the application. If we don't work, they don't work. Their company stops working. They had to feel totally comfortable that we were going to be durable, performant, available, as well as cost-efficient with great customer support. Once they got comfortable with that, they signed this expanded deal.
Awesome. Awesome. You had noted on the go-to-market earlier, let's tackle that for a second, because I know you guys are going through this transformation. Where are we with the direct team as far as hiring of reps, ramping of reps? Anything there would be beneficial.
We started that go-to-market transformation middle of last year. We hired our new Chief Revenue Officer, Jason Wakem. He's brought on a number of people. He brought on a VP of Sales, most recently a VP of Customer Success. He has ramped and instilled rigor in the sales process. Through Q1, we were hiring the rest of the planned rollout of reps. We are done. We've hired all the people that we intended to hire, which is not to say we'll never hire sales reps again, but we are done with the current plan. It takes about six months for a rep to ramp. We are about halfway through that journey with the sales team. Some of them are fully ramped because they were on before.
Some of them just started on, you know, in March, so they are newer to it, but kind of on blend, probably about halfway through.
You had cited, in addition to the direct piece, also like channel and OEM, right? Direct versus channel, how should we think about revenue contribution? Also, just on the partnership front, like I think about Pure Notable. That just stuck with me as far as the powered by Backblaze partner. Can we talk about what you guys are doing on that partnership front as well?
Yeah. We have self-serve and we have the sales side. The sales side direct is still the biggest contributor of that. Channel was a little earlier in the journey, although it's, you know, it is a growing part. The OEM part is also interesting because we only launched that in Q1 of last year. We talked about how, in the last year, we announced an edge compute partner, a transcoding partner, an MSP partner, all of whom have built and integrated Backblaze into their offerings so that customers can directly get storage from them. We also announced that we had a very large deal in Q4 that was a powered by deal, it was a seven-figure deal. We've signed a number of others.
That's an area where, I think it is not a large number of customer partners that sign up for that, but those tend to be larger deals in size. In terms of the partnerships, Pure Notable is a GPU infrastructure provider. We've partnered with them where they are using our power by, they've built Backblaze into their GPU platform so that when a customer shows up and wants to use their GPU platform, they almost de facto have to have storage somewhere. We are the default storage provider for their customers. When you spin up their GPU cloud, it offers you Backblaze as the place for you to keep all your data.
We are also on our side, we have customers that come to us, they want to use data with us and they want to send it to a GPU provider. We partner with, you know, various providers, but Pure Notable is one of the places that they can send to. They provide liquid cooled, highly efficient GPUs for customers.
You guys also have the partnerships that you're talking to in the media and entertainment space too. I think about like Iconic, Sweet Studios or Studio Sweet. Yeah.
Yep.
Can we talk about the media and entertainment use case as well and how you guys are moving into that space through partnerships as well?
Media entertainment is not just Hollywood. Every company at this point has media, right? Pretty much every medium to large company has a group whose focus is producing commercials, producing customer support materials, producing employee training, producing the trade show materials, all of that content. The flows of how they do that have somewhat of a similar approach, right? You generally create or shoot the content. You have a media asset management system through which you organize all the photos and videos and other content. You do editing on it, and then you distribute it in some way. You possibly distribute it out through the website or you distribute it out through a streaming platform, et cetera. That is kind of the flow. Backblaze supports all parts of that media workflow.
As customers are shooting or creating the video, it can land on Backblaze as a destination for the storage. Media asset management systems like Iconic use that as the base, for all of the organization of all the assets. Sweet Studios is a partner that enables you to work remotely with your data, as almost a virtual file system. The underlying storage of that is Backblaze. If you want to distribute that data out, you serve it out and stream it out from Backblaze. We are the underlying platform for the media workflow that you would use.
Awesome. Okay. If I just jump for a second, I know macro has been top of mind for a lot of folks. Go back to the most recent earnings call. You guys had cited some extended or elongated decision making from like a handful of customers. That was a little bit of like a boots on the ground view. Can you just further qualify what you guys are seeing in self-serve? I guess the number of customers that are seeing that pause and if they're opting in any way for different commitments or how they're using Backblaze differently today versus just four months ago?
You know, we were just talking about that, that we almost wonder whether we should have even not said that on the call, because it is certainly, because of the macro, a lot of people kind of focused in on that. It was a, I guess a small comment, not a big comment on what we are seeing. On the self-serve side, we have thousands of customers showing up every quarter. They have continued to show up, do not seem to be affected by the macro. On the existing customer base, we had 117% net revenue retention. So net of customers who left, the existing customers added 17%. Again, do not see any impact on that front. On the new customers on the direct sales front, many of them, again, did not make any commentary, did not see any issue, but some of them were slowing down and pausing decision making.
What's interesting is because we are so low cost and because people need their data, in some ways, both the scenario of growth in the economy and a scenario of a recession in the economy can actually be good for us. If it's growth and people are investing and they want to do more, that's great. We can help them with that. In a recessionary environment, customers are looking, where can I get more out of my current spend? We are a great way to do that. We cut your cost by 80%. It's a great way of doing that. The scenario of, I just don't know what the world looks like and I'm going to not do anything isn't great for anybody, right? We saw some customers kind of freeze and wait.
We were just having the conversation that already it's only been, you know, a matter of weeks. Already those customers seem to be getting past that. I think that, I think we're in that there's not a lot to that, as far, but we did see some of them. We wanted to be cautious and share that we were seeing some of it. One of the things you, you know, you talk about the earnings call, you know, as you saw, we beat the top line of our revenue guidance, as well as beating the top line of our EBITDA guidance. I think a lot of that is because the go-to-market transformation that we're doing is outweighing any macro impact, right? We are executing better than we were in the past. The AI use cases are providing a good tailwind for the company.
Between those two things, you know, we continue to be confident in, in where we're going.
Yeah. Just like one of the things I was trying to think through as well is self-serve has been untouched, right? I appreciate the transparency of the comment on the handful of customers, but at the same time, I don't know, I'm trying to explain it away, but I would have thought the self-serve channel is much more macro sensitive. If you guys were going to see something anywhere, it would probably be hitting you there and the fact that you're not would indicate that this might just be just a handful of customers, like what you guys are describing here.
Yeah. And we haven't seen anything on the self-serve. I think if you think about it, the thing about what we do is it's not for, for most companies, it's just not optional, right? Like, you know, like this company that I was talking about, you know, the one that signed their, you know, the largest deal that we've ever signed, they don't have a choice to not keep the data. They have a choice to use us or Amazon at, you know, five times the price, but they don't have a choice to not do it. They'd shut the company down. The use cases that we have, a lot of them are like that. The companies are relying on us for their core business. Even in a scenario where that's not that critical, like a backup, right?
It's still hard for a company to say, you know what, we're just not going to keep any backups of our systems and we're just going to hope that nothing happens, right? I mean, that's also generally not considered an acceptable practice.
Awesome. To your point earlier, as far as the offset, when you think about the full year guide, we were talking about go-to-market, but maybe you could take a second just to talk about the bookings or even the AI customer traction you're seeing, whether it's customer count increase or data volume coming from those AI customers.
We saw a 66% increase in the number of AI customers. We saw a 25 times increase in the amount of data from those customers. The AI thing is definitely real, and it's driving real meaningful growth for Backblaze. Our largest customer now is an AI company. The primary driver from a use case perspective of growth in Q1 was driven by AI. It is meaningful. We're being, I guess, pretty constrained on what we're calling AI. When we call an AI customer, what we are saying is it's a company who either explicitly has the dot AI domain name, so they are, you know, focusing on that as their thing, or it's a use case that we know is specifically, it is what they are doing as AI.
I think over time, all of this is going to get blended away because everybody's going to be doing AI in everything. It's going to be hard to say, well, this is an AI company because everybody's going to be an AI company. We're being very specific right now about like these are literally companies doing AI as the core of what they're doing. It's still very material.
All right. I want to be true to my word here. Happy to keep going on my side. Any questions out there in the audience? Yeah.
What could the growth rate be for B2 beyond this year? You kind of talked about the acceleration sequentially over the course of 2025. I don't know, I guess whatever it is, it'll be against tougher costs from this year.
Yeah. In terms of what will the growth rate be beyond 2025? What we have published is an outlook for quarter over quarter acceleration of growth this year. We accelerated growth in organic growth in Q4 over Q3, Q1 over Q4. We've shared an outlook that Q2 will be faster, Q3 will be faster, Q4 will be faster, and Q4 will exit at least at 30%. We have not provided 2026 guidance yet. What I will say is it's a $100 billion market. We obviously own a very, very small fraction of that. We don't believe there's any reason why growth shouldn't be higher than 30%, right? There's nothing kind of inherently limiting to it. You know, we want to get to that stage first and prove that out and then, you know, kind of set our new targets.
Inherently, you know, we've been moving up market. We're about halfway through the, you know, the sales ramp of the team that we have, not to speak of teams that we will add. We just hired a VP of Customer Success who's revamping that part of the business. You know, we're very proud of the 117% net revenue retention. Again, I think that there's an opportunity for that, you know, in 2026 and beyond to even be higher. We have 100,000 customers on B2. When we've looked at those, you know, kind of sporadically, there's so many customers in there who, you know, they're paying us $1,000, $3,000, $5,000, but they are the IT director at a billion dollar revenue company. There is much more that we can do with them than what we're doing.
Revamping that customer success to drive NRR, working on, scaling the direct sales, leveraging the OEM, leveraging the channel. I think there's a lot of opportunity there.
All right. Tariffs are like a, a touch and go subject in day to day. But have you guys thought about potential impact of tariffs if you're looking to, I don't know, optimize or retool your supply chain, what that might have as far as impact on gross margins from where we sit today?
We looked at it. Luckily, there wasn't really any impact, partially because the set of things that we are buying and from the places that we are buying aren't currently, as we, you know, depending, you know, ask, depending on the day, but aren't affected. We did do a modeling exercise to say if there were 10% tariffs applied on all of our imports, from all the places that they come from, that would phase in over the course of six years and max out at 1.5% impact on gross margin six years from now. If there were tariffs and if they were at a 10% level and if they were on all of our equipment, it still would be pretty minor.
Obviously, if, you know, those tariffs end up being 30% or 50% or something, you know, that would be more major. We'd have to consider what to do about that. At the current kind of levels, nothing at the 10% level is nothing meaningful.
In the meantime, you guys are still executing to improve the mortgage structure on your side. Like I think about the non-GAAP gross margins in Q1 was up two points year on year, just because of some of the restructuring activity and optimizations that you guys had in the past and how you're talking about how in Q2 there should be anywhere from like a 200-300 basis point improvement to GAAP gross margins because of the extended useful life for the underlying hardware.
Yeah.
It's not like you guys are just sitting still on that front either, which is great.
No, I mean, I, you know, we are an infrastructure as a service company. We really do have equipment that we have to buy, and that really does cost money. At the same time, we do see a continual path to optimization. In Q1, gross margin increased because we were able to effectively get more utilization out of the system. In Q2, you know, like what we've talked about is that the margin increases because the analysis we did on the useful life of the equipment is actually showing up to be longer than what we were amortizing for. We continue to look for ways through the software and through the optimization of how we run the platform to continue driving more efficiency out of it. I think that there is definitely space for us to continue to do that.
The other thing I'll just mention also is we've also introduced new features and functionality. We've launched Live Read, we've launched Enterprise Control, we've launched Event Notifications. These are things which have a potentially different margin profile. So there's also not only the cost side of it, but also the revenue side where we have an opportunity to increase margin.
On that gross margin point, I know we were talking about this a little bit earlier, but can you just help us think about how Backblaze balances that underlying utilization coming from your customers versus the continued data center rollout to ensure that you're appropriately, well, growing, moving into new territories, et cetera, and best servicing your customers? Like how do you walk that fine line between utilization and the DC rollout?
It's very important, because it is, it's part of the core of, you know, how we run as a company, right? Is driving efficiency in our ability to leverage the hard assets, right? And so part of that is the platform is good at utilizing the capacity of the CapEx that we have. Part of that is being thoughtful about where we invest. With new data center locations, what we aim to do is find anchor tenants with whom we do a new facility. We launched Canada just a couple of months ago, and we did that with a partner who was an anchor tenant. They committed that if we put our data center in place, they will sign up for a certain commit level with them. We work together. We said we will do that.
We also do our, have a shared go-to-market to go after the opportunities there. That's the mental model that we're using as we aim to deploy additional locations. It is not a build it and hope they come, but find partners or customers to work together to justify building out a data center and then leverage that as the starting off point.
The last point on that is if I think about the push that you kind of feel on a global basis as far as data sovereignty, is that in any way pulling you into new geographies faster than what you had initially anticipated? Or is that a secular driver when you think about the market backdrop that you're executing against?
It is a secular backdrop. It is a benefit. I would say it is a smaller benefit than AI. Part of the reason we launched a Canadian data center is because Canada has certain laws and regulations around that. We saw some customer demand around having a region in that location. Just on the scale of the tailwind, we see AI as the bigger tailwind.
All right. That's awesome.