Fastly, Inc. (FSLY) Earnings Call Transcript & Summary

And specifically within security, too, I think it's worth noting bot management has been just a really strong product for us. I mean we developed it in-house and have really gone against one of our largest competitors, as a viable solution. It's been around for quite a while now. We feel like we're very successful against them. We've also rolled out another feature on top of that called ContentGuard, which, as Rich was talking about earlier, is really very useful to help mitigate scraping of websites, but also setting up monetization streams around that. We talked a little bit about this on our last call with a publisher that co-innovated with us on that, and it turned out to be a very successful product for them, and we're seeing a lot of interest in that product from other companies. So feeling really good about the bot management side as well, which I think you're going to hear more about in the next year or so from Fastly.

I think on the earnings call, we talked about like bot management and DDoS growing triple digits on a year-over-year growth basis. Our WAF continues to be like our strongest security product and it continues to -- continue to like grow that business pretty nicely, too. So I think overall, we're feeling really good about the full product suite, like those 3 products, in particular, are driving a lot of security growth.

How is the team feeling about -- obviously, we're up to 5 security products at this point. How is the team feeling about the potential expansion of that product family, be it across the application security stack? Or do we start talking about network?

Yes. I think for us, I mean, we are definitely focused on the edge, like we want to like stay on the edge from a security perspective. I would say that the security product team is constantly looking at other kind of like what's next, right? And they want to make the existing 5 continue to be better. And so there's a lot of investments on making the 5 better, but then there's also like other ideas that they have. Obviously, they don't want to talk about -- like preannounce anything, but like the team has worked hard at work. So Kelly Shortridge, who is our Chief Product Officer, ran security product under Kip, when Kip was Chief Product Officer. So Kip became promoted into the CEO role and then his replacement was Kelly, who ran security products and is now the Chief Product Officer. Both Kip and Kelly will be at our Investor Day next week. And so it's a good opportunity to hear directly from Kelly around the product portfolio that she thinks about and how she's seeing the product evolution.

Makes sense. You guys put up a really good net retention rate here, really, in the first 2 quarters of the year. Can you just walk us through what are the primary levers or really what's causing the uptick? Is it traffic? Is it just a better pricing environment? Is it cross-sell? Help us on that.

Yes. I mean what you're calling out is absolutely right. We produced 117% NRR last quarter. I think that's a 4-year high for us. So like we've done extremely well in getting that up. I think when Scott came in, in 2024, one of his big focus areas was like how do you really continue to make sure that existing customers are very happy with our product and loving the product. And then I think Kip broadening that portfolio really helped out a lot as well. And so I think if you had to dissect the NRR, it's actually hard to break out, is it upsell of the existing like delivery services products, because we see a lot of upsell. I mean network services growing 17% year-on-year, right, which tends -- it actually happens to match the 117% NRR that we have. But I would say that, that alone is not enough to like drive that 117%. It's really the cross-sell opportunities. And I think that security now is almost -- it's like a $44 million a quarter revenue business. If you add in like our other compute, that's almost like a $50 million quarter business, a $200 million run rate, and just having that additional product SKUs really helps with the cross-selling. And so I would say it's a combination of the 2, like really the customer focus. Artur Bergman, who's our founder, is still actively involved in the business. He's actively involved with existing customers. Some of these customers have been with us a long time, and he's still very involved. And I just think that maniacal focus on customers has kind of been what's helpful in driving that 117%.

So speaking of your compute business, Compute@Edge is more of a CPU serverless-based architecture. Why is this the right approach to take? I get asked a lot, why doesn't Fastly get into the sort of GPU inferencing game? Help us on that.

Yes. I mean we do get asked the same question quite a lot. Two major things to point out here. One of our peers, our competitors has been approached by some LLMs to build out a large scalable capacity with their sort of central compute approach. We see that as a scenario that we wouldn't find a lot of value prop in what we do at Fastly. We believe our intellectual property is centered around the software that we run on this unified scalable network. And it's basically built pretty much with off-the-shelf hardware, for lack of better words, but it's completely software programmable. And it can accommodate many other needs for our customers. One thing we see happening is the way it's positioned is around a multi-cloud approach with open models around any sort of adventure a customer wants to take. We have the sandbox, if you will, that can cater to all these different possibilities. One of the issues when you go the route that Jim was describing is you're basically building sort of a capacity slug for a specific purpose-built customer applications. Sometimes you give up that opportunity to be positioned in this unique spot that we're at the edge to be multi-cloud, be more open model and you are very limited with how that scales. For us, we see the value prop around that software is being that. And we will see probably over time, more customers turn to us to run their workloads around agentic orchestration layers because we have that unique positioning. The other side of that, too, is I think from, basically, just a financial perspective, it's very costly. The return on invested capital remains to be seen. We would obviously probably look at opportunities like that. But for the time being right now, we don't see there being a good return on the investment versus also just the value prop we bring with our technology stack.

So how has Fastly been packaging Compute@Edge at this point for AI developers? And what is the margin profile of this business relative to the rest?

Yes. I mean I think from that perspective, too, and it should be elaborated more on the last point there, is we have a lot of toolkits around how to develop on our compute layer. We also, as I said, adhere to more open standards, MCP server, things like Really Simple Licensing. These are technologies or standards, if you will, that are incorporated already in our stack, and we're open to that. Our competitors tend to take some more proprietary flavors around some of these things. But that's the way to entice developers to come to our platform, is that we continue to be sort of cloud neutral, model neutral and open for this development opportunity. That's the ethos that we're sort of built on, and we continue to drive forward with.

And then the second part of your question was around like the margin profile on compute. I would say that because we're leveraging the same existing network, the same existing infrastructure, we have the ability to be able to use excess stranded CPU capacity, right? So when you're selling delivery services, our delivery services was historically more like I/O bound. And when we work with our customers and we're selling compute and those compute resources end up using CPUs, the CPUs are basically there sitting idle, and we're able to like really leverage and more effectively use the assets that we have. And so as a result, what you see is like in the last 12-month basis, like our gross margin flow-through was 96%, meaning for every incremental dollar of revenue that we generated in the last 12 months, 96% of it went back to gross profit. And so you can kind of see the interplay of like as we kind of diversify and broaden our security portfolio, and we're broadening our compute portfolio without having to do incremental, like building a whole new network. We're using the same exact network, that gross margin ends up like just accruing the benefits of using stranded capacity that was there.

Got it. We're just under 5 minutes left. Any questions from the audience? Otherwise, I got plenty to talk Vernon and Rich about. No. Well, I know Vernon is really excited for GTA 6 coming out. And so I guess, how are you guys thinking about this upcoming release? What is Fastly's exposure to gaming? And how do you guys view the online gaming opportunity even just longer term?

Yes. It's actually interesting because you watch and hear what investors or even analysts are saying about like GTA, it will be like record-breaking downloads. And we do have -- and we do partner with customers on the gaming side, like gaming downloads is a portion of our revenues. We had a really big Q4 beat last year. And with that big Q4 beat, we had attributed some of it to like record gaming downloads that we saw in Q4 of last year. What we do when we enter a quarter, like when we guided our kind of full year at the last earnings piece, we knew some was episodic, some was durable. We kind of factored in the durable beat and we raised the guidance for Q3 and Q4. And then what we do is we layer in some of the episodic events that we think will happen in the second half of the year. And so the episodic -- the unknown episodic events are going to be the GTA 6 launch. And then I think the other big unknown that could be like very different is going to be the midterm elections. Elections happen every 2 years roughly, and the midterm elections are not as big as the presidential elections, but I think there's a lot of interest that tends to happen. A lot of the big news providers end up using our delivery services business because we have like instant purge. We have capabilities that our competitors don't have, where like they need high configurability. They need high control over like articles that get refreshed. We have this ability where like our customers can just push a button and be like -- purge everything that's out there and replace it with this updated article, and then all the users end up seeing the same article. And so like these big midterm elections end up helping us out a lot. So those 2 bigger unknowns, we look at historical norms and kind of factor in what we think that traffic pattern will be like. And we remain prudent, right? Like you're never going to want to build a revenue forecast that's at the peak that everyone is expecting it to be. And so when we guide, we guide with -- like close to the pin, with some upside potential. We wouldn't want to factor all that in. The other 2 episodic events happen every year, which is going to be holiday shopping and NFL, right? And so those are 2 other kind of episodic events that tend to happen in kind of Q4 time frame. And so we factor those in. Those are a little bit more known to us because those 2 happen every single year. I think the 2 kind of bigger question marks are going to be GTA plus midterm elections.

Well, I tell Vernon every year, he can join the Bills Mafia. It's okay. [indiscernible] Cleveland Browns fan. Just with a minute to go, obviously, Analyst Day next week, and any kind of thoughts around -- or puts and takes around fiscal '27? I know you're not going to guide here necessarily, but -- and also how we should think about infrastructure spending going forward given what's going on in the environment?

So Investor Day is going to be next Tuesday in New York City, Nasdaq MarketSite. I encourage you guys all to go because you're going to get to see Artur Bergman, who's our CTO and Founder; Kelly Shortridge, who's our Chief Product Officer. You'll hear from Scott about the transformation that he's done and where he's focusing his attention. I think it's exciting. We won't -- we won't do guidance on 2027, but I think what we will do is give some more visibility in terms of like a longer-term model that we think would be appropriate. I think it's going to be very -- like a good opportunity to get in front of investors and talk about where we see the next 3 years kind of playing out and what kind of flow-throughs that we expect. And I think I also talked about the idea of like multiproduct adoption. I think that's pretty exciting for me to talk about as well. So I encourage you all to go if you have time, New York Nasdaq MarketSite.

Perfect. Well, that's a perfect spot to end it. Thank you guys for joining us and look forward to hearing from you guys next week.

All right. Perfect.

Thanks, guys.

Thank you. Keep up whatever you're doing.