Here we go. All right, let's get started. Welcome, everyone. Thank you all for being here at the UBS Tech Conference. I'm Roger Boyd. I cover cybersecurity infrastructure software. Very happy to have the management team from Fastly up here with me, CEO Todd Nightingale and CFO Ron Kisling. So thank you both for being here.
Thank you. Thank you.
Awesome. I think there's a fair amount of people familiar with the story, but Todd, maybe we could just start with the journey Fastly's been on. Hard to summarize a decade of growth. You've only been a part of it for a little bit, but, how would you characterize the company's evolution from really kind of a specific performance use case to a much broader edge platform?
Sure. I've only been a part of it for a year, but I've been an admirer for years before that. You know, the content delivery space was really invented by Akamai more than 20 years ago, and I think it started to stagnate a little bit before Fastly entered just over 10 years ago. And I think the reason is the reason that Fastly entered, and the reason there was a real need for that, was this focus on performance and user experience. And that is the initial niche that Fastly really delivered on, was performance, user experience-focused content delivery. And because of that, businesses that were deeply focused on that user experience flocked to the platform, starting with publishers, then media and social media, and then high tech. And that was great.
But, while that fueled the growth of the company for the first few years, what became more and more a part of the differentiation wasn't just the performance, but the dynamic nature of the solution. And that dynamic nature allowed people to build real personalization into the edge and be able to deliver personalized experiences without compromising performance. That's why we say at Fastly, we deliver the best possible user experience that is fast, safe, and engaging, and that engaging piece is really personalization. As the years grew, this idea of truly dynamic CDN really expanded to edge compute, fully bespoke workloads being able to run at the edge, and in Fastly architecture, run using the language that the developer chooses to develop in, and that's a very powerful combination, CDN plus security.
The market has also continued to evolve, where folks, the big, important part of the market is really less interested in picking just a point solution for a CDN or a point solution for edge compute, but really looking for an edge partner, an edge cloud platform. That would be a solution that includes content delivery, edge security, edge compute, and some type of analytics or edge observability solution. That is the transformation that we're seeing right now, and that's really where the focus of our investment is, being and building the best edge cloud platform on the market, and, you know, in the fullness of time, leading that space.
Yeah, makes sense. Continuing on with that, we've seen a lot of changes in the competitive landscape, some consolidation, some mergers, some smaller competitors exiting the business, specifically on the delivery side. Can you just give some context on consolidation, why we're seeing it? Are you surprised we're seeing it this early, and, what, what it means for Fastly?
Yeah. A big part of the reason I joined a year ago is, you know, what I believe is that we're gonna see the same market transformation in the edge cloud that we saw in the central cloud. And it's this, this transformation that a point solution is just not gonna be a viable solution, just the CDN play, just an edge security play, just observability. It is a platform. It's a complete platform at the edge that's gonna matter, and I think that's what we're seeing. We saw StackPath and Lumen both shut down operations, sell their contracts to Akamai this past quarter or two. And I think it's a direct result of this phenomenon, you know, being a content delivery-only solution is just not viable in the market today, and it is a tough, tough business to be in.
The other side to that is. Well, we even saw it on the compute side with Akamai purchasing Linode as well. The other piece here is, I think it is healthy for the market, this type of consolidation, because what we have done is removed some of the lowest price options. And so it'll help us sort of maybe build a little cushion around the pricing for the next year or so, maybe two years. And that's, that's healthy for the market. I feel like that's a good tailwind for all of us. But the real trend we're seeing is a focus on an edge cloud platform. That's the real market that will matter. I think in a couple of years, we won't even be talking about edge security market and TAM or CDN market and TAM.
There will just be the Edge Cloud, just the same way there's only an IaaS market in central.
Yeah. And just to talk about your own competitiveness in the industry, there's been a pretty marked focus on operational efficiency. You outlined a lot of it at the Investor Day in terms of POP performance and efficiency. Can you just outline what's changed over the last year that's made you kind of better able to compete in this market?
You know, I think the efficiency, the operational efficiency, it allows us to generate a better financial result and to focus more of our OpEx spend on driving growth. In order to do that, you have to drive real operational rigor across the business, both above and below the line, both in your capital outlay as well as your OpEx outlay. You know, to Ron's credit, over the last year, 18 months, I think we've done a really good job transforming the business to run more responsibly, to be grittier in our spend, and to be more focused on the outcomes that matter. Building a budget process that allows our departments to really get every value of every dollar and not run a one-size-fits-all across all the functions.
Building out a demand planning function that gives us the ability to predict our business pretty well. I hope you see that in our guide and our results. But also, and maybe even more importantly, allows us to build out our infrastructure in an efficient way. Instead of having to build out everywhere, a little bit haphazardly in the hopes that traffic will eventually land there, we can build out to a demand plan that's being meticulously projected, and that really has changed the efficiency with which we're able to build out our infrastructure. And then how we look at our spend across the teams. There's a growing pain in, I think, every business that's running the race to a billion dollars in revenue, and you have to get over some of those growing pains.
Every team doesn't get to choose their own SaaS solution for every single program. Every single bit of outsourcing that had to be done in a time-sensitive way, two or three or four years ago, doesn't have to continue to be outsourced forever. There's operational efficiencies and operational rigor that has to be built in the business, and we've done quite a bit of that.
Yeah.
In the last year. It shows up in the bottom line, and I think there's still probably a decent amount left.
Yeah. We're gonna come back to that, Ron, don't worry. But on security, I think a few years ago, there was this perception that Fastly was kind of the best-in-class performance product, maybe not so much in security. And I think since then, it's been a pretty good story. You talked about 15% of the business coming from security today, growing 30%+. What's been changing there? There's been a lot of focus on how do you better integrate the web firewall into the data plane. And how would you grade Fastly's ability to compete across the different areas of application security going forward?
Sure. Ron would want me to tell you that the numbers disclosed are for the Signal Sciences not the full security business.
Correct.
I got you. Yeah. Well, I think the first big step here was the acquisition of Signal Sciences a few years ago. That gave us the, you know, highest efficacy, lowest false positive, next-gen WAF solution. And it was purchased, I think, at the right time, because we are right now in this transformation of people moving from a traditional WAF and frequently an appliance-based web application firewall to a next-gen WAF, usually, deployed at the edge. That is a great place to be, with the best solution to be in the middle of that technology transition. Just like we saw it in the IT firewall, the next-gen firewall transition, and that Palo Alto story, I think we're seeing the same thing here.
That, I think, really built the crown jewel of our security solution, and it was high performing. We decided to take a tack of sticking with our platform strategy, being the best cloud provider. And what that meant was taking that intellectual property and deploying it in our infrastructure, building one cohesive edge that runs edge compute and content delivery and security on the exact same machines deployed all around the world, thousands and thousands of machines. And that's exactly what we've done over the last year, since I joined, and now we have next-gen WAF, that Signal Sciences intellectual property deployed in our edge. We've always run a highly sophisticated DDoS prevention, distributed, DoS prevention system, but we've added managed services around that in order to build a more complete solution. Some of our competitors have that type of managed service.
Having it helps people transition over to Fastly because they can get the exact same offering, better, better visibility around it. And those two, DDoS prevention and next-gen WAF, super important, but there is a third leg to the security stool, which is bot mitigation. And Fastly has been lacking in that. A year and a half ago, we launched a solution there through partnership with a startup called HUMAN, and it kind of lacked the platform strategy, that full unified experience on a single platform, although it's a great test. And so we've pivoted away from that, and we've organically built a bot mitigation solution, leveraging the data that we have at the edge, built into that edge, and it's super exciting. It's in beta right now. We have a ton of customers running the beta.
It'll be in limited availability, start closing deals before the end of the year, and GA in Q1. And that'll really complete, I think, the security portfolio today that we really need, which is super exciting. And, you know, moving forward, we're already launching features in API security. We've done a ton of groundwork on more data sovereignty solutions. Our security proxy product, while not applicable to every single customer, has had amazing traction with the browser folks. We just closed Mozilla, our third win in that space. And I think there's a ton of early research that people are looking at around anti-piracy, around more sophisticated edge security products. So I think there's a ton, a ton of road to innovate here, and actually an opportunity for the security business to be as large as our content business.
Yeah. Yeah, makes sense. You hit upon two topics I wanted to dive into, but API security, it feels like the CISOs we talk to and partners that we talk to, it feels like that's a growing attack vector, and hence an area where we're gonna see some demand. I'd be curious to get your perspective. There's some startups in the space, there's some others that are approaching it. What makes Fastly and other web app firewall vendors the right vendors to kind of bring in that solution?
Yeah, I think we're gonna see API security move to the edge, just like we saw in DDoS, just like we saw in the WAF. And the reason is, it's best delivered there. Your API security should run with your API proxy. By pushing that to the edge, you gain an enormous performance benefit. There's an architectural benefit to running those things in the same place. API proxy, which allows for API result caching at the edge, API security, which is able to leverage all of the security signals that are coming off that edge platform. I believe API security is gonna move to the edge for technology architectural reasons, just like we saw, you know, DDoS and WAF.
Yeah.
I'm super, super excited for this market, and I think, again, builds out this reality in the market, that you must be a platform player to win, and, and that's what we intend to be, the best edge cloud platform.
Yeah. And then on browser relay, you've seen a number of pretty important wins there. I think three of the top four browsers worldwide, Mozilla this past quarter. Can you just explain what you're doing for these companies? There's a couple of different levels to it. And then if you think about the other delivery web app firewall vendors, the key web application security vendors that are in that market, what makes Fastly different? I think you're the sole vendor for some of these-
Yeah
... solutions.
Yeah, it's an exciting space. And I mean, to be honest, I was a little worried that there's, like, not that many buyers, like how many browsers are there? But we're starting to see the beginnings of interest from other folks. And the reason is, exactly what you asked about, like what, what does this do? It essentially provides an additional layer of security, to browsers by delivering anonymity to them. In order to sort of protect the, browser or the cloud, from divulging data to, any sort of cloud service, there has to be a third party, who's basically understands half of the connection, and it's that, proxy, security proxy that we deliver. And for that reason, in fact, it can't be brought in-house. So like, we announced this architecture around, Oblivious HTTP.
A terrible name for a product, by the way. And that's the name that the Google architecture uses. Well, the point of Oblivious is, it allows, the Google side of the connection to not be aware of all of the personally identifying information on the other side, and adds a layer of security. In order to do that, it requires a third party. It can't be brought in-house. The whole point is, we only understand half of the connection. They only understand half the connection. None of us can put together all of the data from an architectural, but fundamental point of view, and that is why that security proxy is so important. It splits the connection so that no one is holding all of the personally identifying information at any given time. From a browser point of view, it's a no-brainer.
We're starting to see some interest on the ad placement side of the house, and that's a much bigger market with many more buyers. And that gets me excited, because that is a repeatable motion with hundreds of buyers, and they're gonna care. The reason I think we've been so successful in this space is we are the performance leader, pretty much unequivocally, and we have a very developer-led motion here. This technology is standards compliant, it's open, it's a very transparent development model, and we co-develop it with our customers in the space, which has been very well received. Yeah.
Maybe zooming back out to the Edge Compute environment, how do you think about the evolution of that market? I think it's an interesting story where the initial use cases kind of stemmed out of delivery and CDN, and you're seeing customers kind of transition that to more of like an Edge Compute-type product. But how do you think that evolves from here? I think there's all sorts of estimations about how big Edge Compute could get.
Yeah.
How do you see that market evolving?
Yeah. I think it comes down to personalization, as the first big use case here. There are other use cases, but the most repeatable use case, and where I get excited about this, is where, you know, not just one or two, like, high-tech customers are building a very bespoke use case. We see a ton of that, which is awesome, and I'm very excited about it. But when I see a website or an app building personalization use case, where they're doing content recommendation or product recommendation, where they're doing, some type of direct, conversion of a shopping cart conversion or, order processing, and they're using edge compute, because by taking just a couple of workloads, they can move them to the edge and rapidly, radically change how fast those things happen. They can change the user experience.
They can either drive greater engagement or better shopping cart conversions. That's incredibly compelling for them, and those are the use cases that we're really starting to see take hold. I think this is exactly the pattern that we're gonna see. Use cases where moving workloads to the edge can derive a radical improvement in the user experience by lowering the latency. Those are the processes that are gonna move. I know some of my competitors are targeting, you know, moving organization's workloads over into their compute infrastructure. That's not really what the Edge Cloud is good at, not what we're built for. We're built for user experience, performance-driven, and we're gonna be able to differentiate our customers with our customers' experience by moving those workloads over.
We don't compete with AWS or GCP for all of those central workloads. We move a handful of edge workloads, of latency-sensitive workloads to the edge to drive user experience. How big is that? It's, you know, bigger than 1% or 2% of the central cloud workload base. It's probably less than 25%. The more important piece for us is it's far, far bigger than we need it to be to invest in.
Yeah.
It will certainly be a multi-billion-dollar market, and we're excited to be sort of at the beginning of it.
Yeah. There's been a lot of talk about the potential to do AI inference at the edge. Some of your competitors have been a little more vocal about it, but how do you, how do you see that playing out? And I, I think you'd argue that you already have some machine learning type use cases that are, are deployed today. What, what does that look like, and, and, and, and really, how do you see that playing out?
Yeah, I think it's an incredibly exciting space. And again, I think it's an important place where there's a partnership between the central cloud and the Edge Cloud. These models need to be trained in the central cloud. The reverse propagation through the system that trains the model really shouldn't be done at the edge because it's not latency sensitive. And we are gonna differentiate and add value in those user experience moments. Once you have a trained model, and you wanna run a customer, let's say, a customer support chatbot or product recommendation engine, media recommendation engine, running those services at the edge, that's a really powerful thing to do because it can radically improve the end user experience by lowering the latency.
It doesn't take the enormous amount of data and processing that the training does, and it, it gives our developers an opportunity to start using AI at the edge without having to, like, lock, stock, move enormous existing infrastructure over. And so I think it's, I think it's exciting. I think we're probably a couple of years out before it starts to really drive revenue impact, which is probably why I'm less vocal about it than others, but the hype is real.
Yeah.
We try.
Yeah. Good stuff. Ron, maybe we'll switch over to you. It's been a pretty remarkable story about OpEx control and financial discipline. Pretty, pretty impressive leverage this year. How do you-- Todd talked a little bit about some of the changes-
Yeah
... you made in terms of operational efficiency, but how do you think about that going forward? What levers do you have-
Yeah.
-on the pole?
Yeah, I think, you know, Todd spoke about a lot of the levers we've put in place. I think, you know, it starts with really good planning. What I would characterize as just, you know, financial rigor in terms of looking at expenses. And so we've built a planning process. It's comprehensive in terms of looking at how are we organized in terms of the various organizations? Do we have the right level of management, right span of control, so we have efficient organizations? Looking at the number of contractors and temporary resources. It's something that, you know, as you grow, you tend to bring in a lot of those. Really assessing those and whether we need employees, and we've made significant savings there.
And then particularly around SaaS and other applications, eliminating duplicate SaaS, establishing a purchasing function to negotiate with vendors, and get better pricing, all are part of the processes we put in place. And then lastly, the planning process is focused on driving alignment around what our goals are, so that the investments we're making in product are very tightly aligned with what the go-to-market resources are, and the operational support. And that's allowed us to drive, you know, significant, you know, leverage in our operating expenses.
I think this, combined with the progress we've made in gross margin from being able to project demand much better, a real focus on what are the cost drivers there, have driven a big improvement in not only our operating margins, you know, an 8% improvement in the operating losses versus last year, as well as significant improvement in free cash flow. You know, a year ago, mid-year, we're looking at -40. You know, that number is down to, you know, negative single digits this year, and we expect to be break even from a cash flow basis in 2024 as we continue to drive leverage around these areas.
Makes sense. Maybe a question for both of you. You, you've talked about the opportunity internationally, and the question that we get is just really around the margins there. How do you be efficient there? You've talked about the ability to leverage the channel. Can you just talk about how you're approaching that and as you think about the timeline of efficiency and profitability in some of these growing regions?
Yeah. I feel like this is a great opportunity for us because there's so many attractive markets where we're still operating subscale. Not so much from a delivery point of view, but from a go-to-market point of view. We have large multinationals who are using Fastly delivery around the world in every major market, which is great. And so I have infrastructure already built in those markets. And in most of those markets, it's reached a point where it can be delivered highly profitably, etc . But the go-to-market engine is still pretty nascent in some major, you know, Western European markets, and certain parts of Asia and even some of the most attractive parts of Latin America.
We have a real opportunity to win more deals for domestic traffic in those regions and really leverage the infrastructure that's already been built out for big multinationals. We're investing in the international team specifically to make sure that we're doing that, that we're taking every opportunity to monetize the infrastructure that's out there by reaching for customers not in the U.S. There's another twist to this, which is, if we do find ourselves serving a ton of traffic in a particularly hard-to-reach region, not one of these, like, major Western European regions, but some other country, it can be a little bit of a hit to margins in the short term. We just saw this last quarter.
At the same time, it's an opportunity for us to take that traffic and use it to negotiate better rates and deploy the right infrastructure and become more margin efficient for our whole business in that market, and give us an opportunity to then go in and sell more domestic business at higher profitability. So there's not exactly a straight line when it comes to those hard-to-reach markets, but the really bigger prevailing trend here is there's a bunch of mature markets where we do have deployed base, and we're just starting to reach the local customer set, which is awesome.
Love that. Yeah. Ron, on capital allocation, you're—as you mentioned, you're verging on cash flow profitability. You have the convert out there. There's CapEx requirements. How do you think about the various uses of cash, and how do you think about M&A, given you've had some success-
Yeah.
with a fair amount of these tuck-in acquisitions?
Yeah. I mean, I think we've looked at capital allocation with kind of what I would like to think the same rigor that we've kind of looked at the operations of the business. You know, over the last, you know, year and a half, two years, we've, you know, opportunistically repurchased, a lot of our, convertible debt, bringing it down from just under $1 billion to about $390 million, at a, at a sort of a nominal savings of $110 million. The debt matures in 2026. As we look forward, I anticipate we'll do some form of refinance before it becomes current in March of 2025, but it's a much lower balance.
But broadly speaking, one of the things, if you look at the, the planning that we put in place and demand forecasting, has really allowed us to, you know, invest in our CapEx, you know, in line with demand, take advantage of how efficient our network is from a hardware perspective, and dramatically bring down our CapEx. If you go back to 2021, our CapEx spending was between 12% and 14% of revenue. This year it's gonna be 6%-8%, and we see that as kind of the, the medium-term outlook in terms of sustainable investment in CapEx. So that provides, you know, a nice tailwind to our cash generation.
Yeah. Good stuff. Maybe last question. Looking into next year, how are you thinking about the growth drivers? You've had some pretty strong customer metrics. Net retention rates been very strong. Sounds like customer engagements are generally getting much, much bigger. How do you think about the balance between new logos and upsells as you think about next year? General optimism across the board, or?
I'm pretty optimistic across the board. I think new customer acquisition is really our go-to-market focus, for sure. Not so much because it helps us drive the 2024 number, which it does, but more importantly, it helps us drive the 2025, 2026, 2027 numbers. If you look at our Investor Day data, you can see cohort analysis and how customers ramp over the first two to three years of their existence on the Fastly platform. It's how we compensate our sales team. We incentivize customer acquisition, for sure.
I feel really good about that because we've got the international piece we just spoke about, but we've also really started to see the benefits of the packaging changes we've made, of the product simplification changes we've made, starting to drive new customer acquisition in verticals, in new verticals for us, or really adjacent verticals. And that, starting to see that early success, makes me certainly optimistic there, which is awesome. On the cross-sell side of the house, on the customer expansion side of the house, I think it's gonna be cross-sell that'll be the biggest driver, especially content delivery customers cross-selling into security because of that unified platform. And, you know, we're getting closer and closer to a fully unified management plane, and we'll be delivering that in the first quarter or so, and that is just a phenomenally powerful solution.
You're a content delivery customer, you want to use next-gen WAF or get, you know, DDoS, managed service, a single click, that is compelling.
Platform usability.
Yeah, excited about that.
Awesome. I think we'll wrap it there. Thank you, gentlemen, both for being here. Everybody in the audience, thanks for coming.