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

Yes. I mean we do look at kind of flow-through on gross margin and operating margin. I think our last 12-month gross margin flow-through has been like 96%. I mean, so for every dollar that we have like generated, we've kind of flown through 96% back to gross profit. So you can see a lot of like operating leverage as we've kind of grown the business and expanded and done cross-sells into other areas. When you think about gross margin by business line, it gets a little bit harder and trickier because really like Fastly has one network, right? We have one network that supports all the products. And so whether a customer is using delivery, whether a customer is buying security product or even they're doing compute, they're using the same 166 points of presence that we have around the world. And because of that -- and they're using different resources, right? So delivery may be using some bandwidth and some potential like I/O. And then on security side, you may be using some -- a different kind of part of the server. And then when you're using compute, you might be using CPU capacity. And so -- but it's all the same servers, all the same network gear and supporting all 3 businesses. And so as a result, we don't do gross margins by business. I would say that having all the businesses together is highly complementary. So like if you use the compute as an example, our constraint was never kind of the CPU side on our business. It was more the I/O side. And so when you launch a compute business, you're using kind of stranded resources that were not being utilized effectively, right? And so when now you start selling compute into the business, now you're using a different side of the servers that were not being used, but it's all free, right? Because the servers were there, the CPU capacity was there. And so what you're seeing is you're seeing our mix shift changing and you see security and compute a bigger portion of revenues, a lot of that is like incrementally like higher margins just because like they were just stranded reserves that were not being utilized effectively. We're very much -- some of our peers actually have to support multiple networks. And so I think that for them, it's a little bit different where they may be able to break out -- but for us, we're 100% on the same network. All the servers run all the businesses, whether it's bot management or DDoS, it's still the same servers running and powering our delivery business.

I wanted to spend some time back on the flagship business, the bread and butter, the network services, 3/4 of the business. Zooming out, the Internet is growing, traffic is growing. machine and AI and bot traffic as a proportion of the Internet is growing. I think you have your own internal research that's also come out saying we've had a 6x increase in at least the traffic you're seeing flowing through your network. So back to the basics on the network security business, you just put up 17%. You had a very strong back half and first half of this year in that classic core side of the business. Can you walk through some puts and takes and assumptions because the 2 variables here are Q, traffic growth, P, pricing, which I think you've been very transparent about. So help us walk through how some of those -- or those 2 particular variables have played out in the first half of the year and how you're thinking about them certainly for the back half of the year? And then as a general matter, I mean, Q is really up.

Yes.

Yes. So I think just to talk about pricing since it's a topic I think we get asked about a lot. It has been a very relatively stable market for the last, say, 3 to 4 quarters for us, probably 3 to be precise. And if you go back and look at this industry, the content delivery network industry, there were a couple of point players that unfortunately went out of business about 2 years ago, and they introduced what we would characterize as like an irrational pricing environment. So there were some rough moments that the whole industry went through. That's been totally absorbed into everyone's models that's still remaining for well over a year now. And so we've had this environment where pricing has been relatively stable. When we say stable, it's still declining sort of mid- to high single digits year-over-year, but we also get commensurate gigabit traffic on top of that. So it goes through sort of a Moore's Law dynamic where we're seeing more traffic as a result as the environment is growing, as Fatima pointed out. In terms of the Q side of it, the quantity, I think where we've been very successful, and Rich was alluding to this earlier, is we have many more opportunities and on-ramps with our largest customers in cross-selling and upselling different pieces of our overall platform, specifically in security. So we are seeing dynamics where we may close a new piece of security with an existing customer. And then when we go back to do a renewal, we're at the table, they will give us more traffic as a result of that longer term. So they'll commit to that. And so that's sort of feeding a lot of the growth that we're seeing. I think in the market overall, there has been a shift towards higher performance delivery for both static websites as well as where we do very well in live entertainment and streaming type media. And we've continued to be probably a share gainer across most of those markets. And certainly, if there's any large live sporting event, there's a good chance that Fastly is behind the scenes powering that.

And Vern, just to double-click on that, thinking about the composition or, call it, the complexion of that traffic, how do you expect that? I mean, based on the data that you're seeing in your network, how is the complexion of that traffic changing between live events, between corporate traffic, between episodic events, which I suppose would be more live event-centric and AI traffic and bot traffic and machine-generated traffic?

The machine-generated traffic without question, is where we're seeing probably the most growth. I mean, again, it's still very early innings and it's still very small, but relates to our platform itself, that's having an outsized impact. We have a product called ContentGuard that we rolled out, which is specifically designed to help, and a great example of this is a publisher where they have agents and bots coming in and scraping content off of their website being delivered through a Frontier LLM to an end user, and that publisher does not get compensated for that. So they now have full control to see who's accessing their content. They can set up licensing arrangements around that and actually set up a monetization stream around it. So it's been a very successful product. We have a lot of excitement around that one, and we recently rolled that out in the last 12 months. But we're seeing more use cases around those types of applications in the security realm and don't want to take any sizzle from Investor Day, but stay tuned. There are some interesting things there. And then also on the delivery side, yes, the requests are definitely going up machine to machine-wise. And we do see that in some of the programs that we're involved in. I mean if you look at where we're positioned, this could range from us being sort of a middle orchestration layer for inference and agentic AI and communicating with the frontier LLMs. And I think where we're at is a very unique position because we are one of the few cloud-neutral systems that are out there or architectures that are out there, whereas a lot of these other ones that are sort of quasi peers of ours, I won't name names, but they're building massive GPU farms and things like that. They're actually building more closer to centralized clusters around these activities. And so if you're someone that wants to put in a multi-cloud strategy around that and you want to have the flexibility over time, you'll wind up probably engaging with an architecture similar to ours. So that's sort of the positioning. So we think there's a lot of growth there potentially. Admittedly, we're not seeing it like today in terms of its proportion of the overall traffic, but it is picking up a lot behind the scenes, as you said, 6x for that traffic level.

Vern or Rich, so what I'm gathering from you is actually counterintuitively, even though your traffic levels are going parabolic on the network, from a monetization standpoint, from a P&L geography standpoint, you're actually seeing that sooner in the security business, right? So if current trends persist in the way they are, we should, as investors, see that manifestation within your security business. Is that a fair distillation?

Yes. I think like the nearest-term impact because of AI would probably be security and some compute. I would say that when we think about like AI and the way it's impacting our businesses, specifically even delivery, when we talk about traffic, you typically think about traffic as measured by gigabits delivered. AI traffic is very small bits of data, right? And so it's not like streaming where you're doing high-definition bandwidth transmission. And then I think that it's -- when we talk about like growing 6x on a traffic basis, that's from a request perspective, which is very, very tiny bits, but the number of pings that happens on the Internet is quite high. And so when you think about our products and the way we price, our security and our compute products are priced based on a request per second basis, whereas a lot of our delivery business is still on a per gig basis. And so it's very -- when you add up all those little bits from machine traffic, it's not still the same as a transmission. And so I think delivery will take time to kind of change over time. But I think right now, based on the way things are priced.

On the pricing side of the equation for the delivery business, I know I'm sort of jumping back and forth, but I think it's an important point to hash out. You talked about just more rational behavior in the marketplace, right? I'm wondering how much of that pricing rationality is maybe transcending into pricing power because you have had some of your peers actually raise pricing, right? And so has that been an advantageous opportunity for you to command more market share from a unit perspective? Or is the strategy that, hey, the market can bear a higher price, so there is an opportunity for you to realize better price on the average delivery contracts versus the historical cadence of general degradation in that market. Any thoughts there?

Yes. I mean I would say that from a pricing perspective, our competitors and us, we're very rational players now. We're going to like talk about the value that we create for our customers, those an -- a lot of those little ones have gone out of business, and so that pricing rationality is not there anymore. I do think that the edge cloud is growing in significance with like security and compute. And because of that, we're adding more value. And so a lot of conversations go into value and like how do we support our customers and how do we price our products across the board. And so we actually less and less think about like pricing on a delivery services side, it's more like, hey, what is your problem that you're trying to solve? And what are the product suites that we can bring to the table to do that. And yes, I think that given the environment we're in today, I think there's definitely more of a partnership approach to pricing as opposed to like a, hey, we are going to like be the lowest cost provider. Like it's much more about that partnership and the value creation, which I think is a much better position to be in. I think customers are more and more wanting to partner with us. They're wanting to make commitments. You see that in our RPO growth, and you're seeing that in our CRPO growth where RPO grew 38% year-over-year. Our current RPO is growing 44% year-over-year. And so I just think that partnership overall is definitely there. I can't speak specifically about like what our competitors are doing with price increases. I think they announced it, whether that's actually happening or not, like that's for them to discuss. But it's the approach that we've taken at Fastly is much more about the partnership with our customers, the problems they're trying to solve and like how do we price our products portfolio and the suite for them.

Rich, just staying on the pricing topic, but maybe taking a different angle to it. You've been able to deliver a lot of value and value-based selling to your installed base, right? And talking about, okay, let's not apportion a conversation around delivery services. It's not a portion of conversation around security. So generally speaking, has the incremental momentum in security come from better and more assiduously farming the installed base, right? And then relatedly, you do have about 1/3 of your revenue coming from your top 10 customers, right? So a little bit of a double-edged sword in that, okay, there is concentration here, but these are very immediate important customers. So how do you balance some of that ongoing wallet capture penetration opportunity with the fact that, hey, you maybe do need to continue to diversify the rest of the business. So it's not as weighty of a performance from the top 10. So how do you kind of juggle those very -- it's a tenuous dynamic. How do you juggle that?

So I would say that our product suite comes from both sides, right? Like with the fuller security suite, we're able to do very effective cross-selling with existing customers. But we also can land with new customers just with our security product. A lot of times, they come in having a security problem, and we can actually -- it could be an inroad in terms of how we sell to customers. And so I would say having that fuller product breadth allows us to kind of really do both, cross-sell existing customers plus land new logos. I would say that when we think about like the products and the suite we have, I mean, I think that it's -- our top 10 customers are adopting the multiproduct. And so you'll see the strength of that business. I think our top 10 kind of -- it was 37% of revenues in the most recent quarter. Like does that concern me? I would say these are actually very good customers with really good margin profile, as you can see with like the 65.8% gross margin, record gross margins, even with that higher concentration. What makes me feel really good about these customers is it's not -- the growth isn't coming just from delivery services where they can just shift and move businesses. Like if you look 3 years ago, 5 years ago, like that's risky, right? Because when you're only selling one delivery services product. But the reality is these customers are now buying security. They're buying our suite. They're doing compute. That makes them a lot stickier. And so we actually have internal studies that have shown that when customers buy multiple products, like they just become much more sticky. So am I concerned about the 37% -- not really because they're really good customer contracts. They're much stickier than they have been historically. I do think that when you look at the outside the top 10, outside of the top 10, it's still growing double digits on a year-over-year basis. I think the most recent was 12% year-over-year growth. We need to do a better job with that. I acknowledge that. But I do think that like we are not going to -- the best way to really diversify is to really grow that non-top 10. And even outside of that, I think we had 624 large customers. So just continue to grow that portion of the business as well. But I wouldn't do it at the sake of cutting back the support that we give to the top 10. Our top 10 is a very good set of customers.

Rich, I know you disclosed net retention rate at the company level. I'm wondering if there is a meaningful distinction or delta between the net retention rates of your top 10 customers versus the rest of the base. Even any directional contouring would be helpful here, just so most of us can appreciate that the top 10 are indeed all in on the Fastly platform, and there is a lot of attractive LTV kind of latent in there. Anything you can sort of -- any color you can give us around that?

Yes, I'll give you 2 data points on that. One is that -- so our NRR is 117% in the most recent quarter. I think that's like a 3- or 4-year high for the company. And so we've definitely been very proud of that like NRR kind of increase. I would say that given that top 10 is now 37%, you can kind of imply that like NRR for the top 10 is actually higher than the 117%. And so we feel really good about that because it just means they're just embracing and adopting more products at a bigger scale, right? And I think it's a really kind of good data point to have. I think the other data point is that if you look outside the top 10 or even outside the top 50, it's still very healthy NRR, right? And we're not like at an NRR where it's like dropping to like 105% or even 100%. So I think that like our NRR even among the different cohorts are very strong and very good, and I feel really good about where we're at.

On the -- so we talked a ton about network services. We talked a ton about security. In the other bucket, again, a lot of option value in that bucket. A lot of your peers and the industry at large, we are going through a generational capital expenditure cycle, right? So what behooves you to invest or not invest in scaling out your CapEx and infrastructure footprint? You've been more judicious. But what's holding you back? Maybe what do folks in the investor community not understand about your network topology or architecture that precludes you from having to take similarly aggressive steps as some of your direct peers, some of your orthogonal peers who are, again, undertaking a generational CapEx investment because ultimately, you're paying the price on component price inflation, which I think you've managed around. So just generally commentary on are you -- do you want to get into the race? Is it too late? Does it matter? What are your thoughts there and then why?

Yes. So our most recent earnings quarter, we talked about 10% to 12% of revenues invested back in CapEx. It actually is an increase. I think it's almost like a 40%, 45% year-over-year increase on CapEx spend. So we are making the right investments we need to do. I think when we think about it, like the beauty -- and we have a much more network efficient architecture, right? Like we're running one network supporting all of our products. That enables us to have a lower percentage in CapEx as a percentage of revenues, which is very helpful. So when you compare us with us versus our peers, even at our size and scale, we're making the investments we need to support the revenues that we expect, and we're doing that in a much more efficient way. I would say that like orthogonally, the peers are doing a lot of GPUs, which like when we look at it, it doesn't make sense for us to put GPUs in, like we can service our compute business with existing CPU. And so I just think that like for us, we're going to be continuing to invest in CPUs and our business. And when we see the opportunity for GPUs, we may consider it. But for now, like I just think that we can be much more efficient the way we are.

Last question for you.

Sure.

If you had your magic wand, to wave away any one glaring investor misunderstanding or misconception, what would that be? And I know you'll have Analyst Day to expand on it, but...

So I can give my one wish. I just think that this business on being on the edge cloud is much broader than just a CDN business. I just think that there's a lot of misconceptions around this being a CDN commodity business, like there's a lot of intelligence in the edge now, right? And this intelligence is making the product suite that we have much more important around security, bot management, DDoS. It's not about just pushing bits of traffic across the Internet and speeding it up. It's actually a lot more secure dynamic, more e-commerce traffic is happening, more hospitality, like just there's so much happening on the Internet today. And I just think that people still have this mindset on like commodity CDN business, and that's not what we do.

I appreciate that. I look forward to getting some more details in a few weeks.

All right.

Thank you very much...

Thank you guys a lot who can join us.

Thank you.