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

Okay. Welcome back, everybody. I think we are now live. It says live. All right. Great. So I'm Brad Zelnick, once again, with the software team here at Crédit Suisse. And for this session, we are delighted to be joined by the folks from Fastly. With us today, we've got Joshua Bixby, CEO; and Adriel Lares, CFO of the company. And thank you, guys, really appreciate you making the time.

It's a pleasure to be here. Thanks, Brad.

Awesome. Maybe just for a quick intro, Joshua. Many listening in today likely know Fastly, but for those that are less familiar, can you spend just a couple of minutes talking about the company, what Fastly does, why it's such an important part of our lives, especially today?

Sure. It's a pleasure to sort of speak about what we do. It's been an exciting few months for us. So if you extract out to a 30,000-foot view, when I wake up in the morning and get on my phone and I click on The New York Times, I'm speaking to a Fastly server. So we have servers all around the world. We perform some really important functions for the most innovative companies in the world. We allow their sites to be fast, we allow them to be secure, and we allow them to scale. So when all of us go to The New York Times or all of us go shopping or all of us are listening to music, it's at those moments where all of those interactions are happening with our servers and we are able to deliver not just the content, not just the images and let's say, the style sheet, but actually, the entire experience is delivered from our edges because we allow the programmers from our customers to put their code at our edge so those experiences can be personalized and they can be delivered exactly the way our customers want. So in some ways, we're like the modern web server.

That's very helpful context. And maybe as we think about Fastly the company and Fastly, FSLY, the stock, many investors -- it's human nature to apply pattern matching, to look for things so we can take something we know and reference something that we've seen in the past. And I think there's some debate around what bucket does Fastly fall into. Is it another content delivery network, which I think does some of the same things and has been around for a while with other companies out there? Is it a developer-friendly edge computing platform? How should investors think about your differentiation specifically as it relates to developer appeal, which is -- you talked about programmers in your original opening remarks?

Indeed. So yes, I mean, we are built by builders for builders or by developers for developers. And so if you think about what we provide to our users, it is a platform upon which they can build. So we think of this as developer's edge, an edge cloud, a platform upon which our customers can put their code in order to serve their users close to them and securely. So CDN, traditionally, is sort of the legacy term that would apply to the delivery of static objects. And that's still very important. We want the objects on websites to be close to us. We want them to be accessible. We want them to be able to scale so when millions of requests come in, in a second, we can just automatically deliver that. So that's an important foundation. And we are in the CDN business. But the reason that we are a product for innovators is because it's not just CDN. You need to be able to have the objects. You need to be able to have your code and then you need to be able to have them secured. And it's the 3 of those together which create this modern category. If you think of this category historically, it's been a category that's been very sort of enterprise-heavy. So think the opposite of the cloud, right, is a developer you can't sign up. You can't get a moment of inspiration to write code. You need to get an army of professional services people. You need to have a steak dinner and go through this entire procurement process. Fastly is like the AWS for the edge cloud market. We allow developers when they have inspiration to sign up. All of our documentation is live. Our pricing is transparent. They can just get it, use it and start immediately. And that's very attractive to the modern buyer because the modern buyer today, a lot of that power is centralized with these developers.

Got it. That's very helpful. And maybe just getting into the architecture a little bit, Joshua. And I know we don't have Artur, your founder, here, who is deep technical, but I know you can speak to this certainly better than I can. But as we think more broadly about your differentiation, it's not just through software, but it's how you've actually architected the network, right, which isn't as distributed as traditional CDNs. Why is that? What are the trade-offs? What are the pros and cons? What are the use cases for which Fastly is maybe more suited and perhaps the ones where maybe it's less suited?

Yes. I mean, frankly, we are faster all around the world than the legacy provider. So I would say in all situations, you want speed. We also are in a situation where we can scale in these locations. So if you think about the way that it was done, in the 1990s, the Internet was very different. And so it made a lot of sense to put hundreds of thousands of servers across thousands of locations. The Internet has completely changed over the last 2 decades. And we're now in a situation where with many fewer locations, we can be faster than competitors. So you're right, it's about the software, but it's also about the way that we've architected it. And the reason, this is hard for people to think about, right, which is, if you're in a location, it makes sense that if you're sitting right next to me, you'd be faster, but that's not the case. And there's a number of reasons for that. The first is, the Internet doesn't work that way anymore. So if you're in a certain location, let's say, somewhere like Santa Fe, you're on an AT&T cell phone. You're going to come out on the Internet in Dallas. And so if I'm sitting in Santa Fe with 3 servers in a small server rack somewhere, that actually doesn't help me because I've got to go my phone to Dallas back to Santa Fe. So actually, what you need is you need to be really big in Dallas. There's about 16, 18 locations all around the U.S., for example, where you need incredible density in order to actually deliver the modern Internet. So it's almost like a highway system, which changed very much how small towns and even the United States used to interact with travelers, like now you're on the highway and those highways come out at very designated points. So that's the first thing. But the second thing that's really important is at those locations because the Internet has become this vast encyclopedia of information, you can't just have 3 or 4 old servers. You need massive density of storage. And so you need to be in a location where you can put that massive density. You have to be in the right location, which as I said, in the United States, like 16 to 18 locations. You need massive density of storage. And then you need the compute capability to work on that storage. So all 3 of those things are necessary. You can't get that when you sort of randomly throw hundreds of thousands of servers around the Internet. And that is why, as I say, we are -- if you look at the sort of third-party rankings of performance and other things, we showcase really, really well in that. And as you pointed out, that's about the software, but it's equally about the architecture.

Got it. That's extremely helpful. And maybe if we could turn to what's been really exciting, Compute@Edge. And since we've come to know Fastly and Artur, your founder, he always said that delivering content has been the way to build towards the goal of becoming a more general computing platform, which recently culminated in the limited general availability of Compute@Edge, your serverless platform. Can you talk about the interest you're seeing here, some of the early use cases? And I've asked you in the past to size the opportunity. But maybe without putting numbers to it, what convinces you that it's really that massive?

Yes. I think what people forget about Fastly is our first foray into this market almost 10 years ago was with a compute platform. So we have been a compute platform our entire history. When Artur founded the business, his focus was, in order to deliver the modern experience, I need to deliver compute at the edge. So it started with that vision. Now what we saw was wide adoption across our enterprise customer base. The vast majority of our customers, if not all of them, outside of some markets like video delivery, are using edge compute today. So we've got 9 years of history watching how compute gets used. One of the things that we heard from our customers was, we wanted to do more. We want to do it in the language of our choosing. Because in the first incarnation of this, we did ask our customers to use a language that we -- an open-source language, but a configuration language. And there's really good reason for that from a security, a scale, a performance perspective. Now as the technology has evolved, we've seen the opportunity to allow the customers to work in languages of their choice and do more and do all of that while maintaining the scale, the performance and the security. Because if we went back and said, oh, you can just work in any language, but we've got to do 10x the number of servers or it's going to be slower or we have to rate limit you. The number of requests that we serve to the largest vendors in the world, we just can't go back to them with all these limitations that we see in other products. So a ton of experience. That experience is like a flywheel. As people write code and they share that code, the entire platform gets better. And that's exactly what we're seeing with this new version. So I think of it as the next sort of incarnation of this, not the first one. And what this is allowing organizations to do is to do more. And one of the reasons that initially in the first incarnation of this, we didn't allow people to do more is because security is really important here. You're on a multi-tenant platform. We need to make sure that whatever you do, we can keep safe. And so there's a number of really important innovations in this next generation of technology. The first is around security, which is we actually isolate each request in its own encapsulated area so that it can't go out and do damage to anyone else. That's really important. A lot of other vendors will mix all those requests together, and that creates a real challenge. We also are able to bring these up in microseconds. And that's really important because we're sitting in the way of the most important traffic in the Internet. So if it takes 50 milliseconds or 100 milliseconds, that time matters. And then obviously, we need to be able to scale in the same way without taking away the really significant architectural advantage. So what we have seen over the last year as we've been bringing this to market, and as you mentioned, we announced the production readiness of the product, is just some incredibly creative use cases. I'd point out a few areas. One is around personalization. We're seeing customers, let's say, in an e-commerce experience, who have always done this game of, I want to make it fast, I want to make it personalized and they stretch you between these. It's almost a zero-sum game there. And we're allowing -- with this new Compute@Edge, we're allowing organizations to not make those choices. They can be performant and personalized because that personalization can happen at the edge. Critical use cases, and I think really exciting ones. It's a big area of innovation. Data and compliance is another big area of innovation. We're also seeing a lot on the machine learning and inferencing from models. Those are areas that have really come up, but we're seeing wide adoption and we're seeing a tremendous amount of excitement.

We are very bullish on the future of edge, and I think we see it much the way that you do. Maybe -- and you touched on this a little bit, but because it's such a significant opportunity, you're seeing a lot of other serverless platforms emerge, whether it's AWS' Lambda@Edge or Cloudflare Workers, for example. Is there anything that you can hone in on that speaks to the differentiation of your serverless platform versus some of the others that are out there? And frankly, the competition, like is there competition or is it more, I'm going to use the serverless capability of the platform that I am already using for some other reason?

Yes. So I think what we're seeing in the enterprise customer base, remember, we're dealing with the largest customers in the world and those that aspire to be, is that you start looking at the capabilities that products should have. So I can't speak to looking through that lens of a small and medium enterprise. You think of the largest enterprises. They're looking for scale. They're looking for the security and they're looking for the performance. It's actually the similar criteria they've always looked at us through. And through those 3 lenses, we provide really important capability versus one class competitor, which is more on the SMB side. So I'm going to limit, you can only do 100 of these a second or I'm going to mix your request with everyone else's request all together. And if there's a security vulnerability, well, that's not really my fault. So there's one sort of viewpoint which is enterprise-class security, scale and performance. And I think that really separates the product from a certain class of competitor. Now on the other side, when you have vendors that are at scale, you then look at the capabilities of edge computing along with all of the delivery and security capabilities. So now what organizations are looking for is they want an integrated, holistic view. And actually, many of our customers use us to actually go in between other central clouds. So we are a point at which they are saying, okay, I'm going to give some traffic to Microsoft, some to Google, some to Amazon. And what our customers come to us all the time and tell us is, if I am just in on -- with one of those vendors, I lose a lot of pricing, scale and I really get locked in. And so one of the things we're hearing from our customers when you look at another class of competitor is actually, I don't want to be all-in on all of my edge compute because that's where I'm going to centralize all my security and login information, that's where I'm going to centralize my load balancing. If you look back historically at a classic data center, you never had Cisco go Cisco from the edge router all the way down to the database server. There was good reason for that. It was because it makes a lot of sense to have choice. And there are points in the network where you decide, I'm going to centralize, but that will provide me choice at another layer. So I was talking to a CTO recently of a big gaming company. He said, listen, for every $1,000 I used to spend in my data center, $950 went to Dell and HP and Super Micro in this commoditized server layer. But I spent a lot on [ diversifying against ] Cisco, and I did that really intentionally in the sense that, that was a very expensive dollar per CPU. But in total, that gave me choice. And I think what we're seeing right now is that with these central clouds, that choice is becoming more and more important. We're moving into a -- very much a multi-cloud world. And that's one area where we see edge compute be a really important linchpin of a multi-cloud strategy, and people are really pushing against using one cloud, one central cloud all the way through the stack, just like they did historically in the data center structure.

That's great context, Joshua. And it's helpful to understand the edge opportunity from multiple perspectives and how you see Fastly uniquely fitting into that world. Maybe just a question about how investors should think about the adoption curve of edge computing and your ability to compete. I mean, you've spoken about your ability to compete, but is there anything that you can compare the adoption curve to and how you all think about it from a resourcing, staffing and buildout perspective? What's your expectation 3 years from now? Is that an S-shaped curve or how should we think about it?

Yes. I mean, I think we've seen 2 of these curves already, and this would be the third, which is we saw the adoption of a modern edge cloud. We're right in the middle of this incredible adoption of the secured edge and how that transforms. And sort of behind that is this next generation of compute. Well, because we work with developers and because we work the way they want to work, it's really more about them than it is us. And what we hear from our largest partners in this and customers is, we're going to try it, we're going to put it out, and we're going to see how it adopts. One of the beauties of this, those who have digitally transformed -- it's not a great word, as you know, but like this idea that I'm going to experiment, is that we can see it real-time. So we'll see a customer write lots of code. I mean, we see these great early indicators, see customers write lots of code. Then they go put it out in 5% of their traffic in a certain region. One of the beauties of this process is, if it goes to 10% of that region, we know something's working. If they take it out, we know something isn't. And so we see in real-time as this technology starts to expand into the sort of circle of the customer and expand out into more circles, more sites, more users, more traffic. So we are already seeing those curves, and those curves look exactly like you'd see standard adoption. I try it. I like it. I buy it. And if you look at our land-and-expand process that we've seen in every cohort that we've talked about, it won't be very different. You're going to see an experimentation. You're going to see an adoption. And you're going to see an expansion. And that's exactly what we should expect in this technology. Like basically, everything else that we bring out in the edge is going to follow that same strategy. I was talking to a CTO at a big travel business. And they're like, listen, in the old days, for every $100 I spent on my data center, there was just a way that I looked at the math. Let me tell you the way I look at the math in the cloud world. For every $100 I spend, they happened to be a Google customer.For every $100 I spend on Google, I spend $5 to $10 on security. I always have. I always will. I spend $5 to $10 on delivery. I always have. I always will. And the question is, how much am I going to spend on edge compute? And in their estimation, it was somewhere between $10 and $30. And that's because of the movement of workloads and of work and of importance out to the edge. So we haven't figured out -- those equations out. That's speculation, obviously, based on what one customer said. But I think we already know the $5 to $10 on security and delivery. I think the unknown, which we are exploring as we speak, is what happens when modern web serving technology becomes available to you on the edge. When all these microservices are stitched together in one location, and that's the edge, what does that look like? And we have some early indicators of what it looks like. It's very promising.

Very cool. Joshua, if I could switch topics a little bit. I want to talk about, you've disclosed what had been your largest customer, and we'll see where it goes forward. But as we think about the recent events, ByteDance was your largest customer, recently churned off or has been expected to churn off the platform given geopolitical issues. And I think some investors are surprised at how quickly they were able to do so. And I appreciate you've always said you shy away from taking commodity traffic, but obviously, you would satisfy your customers' needs when thinking more holistically. How should investors be comfortable that you haven't maybe become less discerning in the customers and use cases that you take on?

Yes. I mean, we have shied away from speaking specifically about that use case unless we have to. I think if you extract out and you sort of look at these geopolitical risks in general, what you see is this is a process where, as a vendor, you want to support your customers through this process. So we have said in this context and other context, we're going to stand by our customer. We are going to help them put themselves in a situation where they can survive no matter what the outcome is. And imagine, in these situations, the outcome could be that Fastly or other companies can't work with a certain customer. And so I hear this sort of sense of, well, timing and it went quickly. I think the one thing people should know is, we helped in this process get our customer in a situation where they can still exist, and that's important to us. And that means helping them in any way we can. So I guess I look at it through a slightly different lens, which is, these are very uncertain times in this regard. And survival is really important, and we are going to help our customers survive. I think the real telling case will be in the future when we see resolution, what that will look like in the future. But right now, I think I wouldn't read too much into that because this is a question of, at a certain point, and those points have been well-defined by a calendar, it's possible we can't work with this particular customer. So time is not -- you can't go slowly. That's not an option. And you're going to make a choice to defer other things. Maybe it's speed. Maybe it's performance. Maybe it's scale. Maybe it's something else in order to achieve these because they're such a binary outcome. So I think we need to let the future speak to that. I wouldn't jump to too many conclusions until we see resolution. We don't know when that will be.