Fastly, Inc. ($FSLY)

Welcome to today's session with Fastly. My name is Jonathan Ho, and I'm the Cybersecurity Analyst for William Blair. Our speaker today is Rich Wong, who's the CFO of the company. Before we begin, I'm required to inform you that a complete list of research disclosures or conflicts of interest is available at our website at www.williamblair.com. As a reminder, the breakout session will be held in this room following the presentation. And so yes, with that, I'll hand it over to Rich to do a brief overview of the company, and then we'll follow up with a brief fireside chat. Thank you.

Great. Thank you for coming. I really appreciate it. I think we're at the last presentation for the day. So really appreciate you sitting through this. As mentioned, Rich Wong, the Chief Financial Officer. I've been with Fastly since August of last year. Very excited to be here. I had spent my career in tech and so had been a 2-time CFO before this and then had done some time across like Wall Street and investment banking and a number of other companies. Before we open, I do have kind of the forward-looking statements disclosure. I'm not going to read this here, but I just want to make sure that it's available and you guys take the time to read it at your leisure, if you want to fall asleep. Before I begin, I thought maybe I'd start with a high level about Fastly and what we do. So Fastly, we are an edge cloud platform company. We serve from the edge. Our mission is to really make the Internet a better place where all experiences for users are fast, safe and engaging. I just think that like the edge plays a very unique spot in terms of delivering Internet traffic across the world, and we play this -- in this role at the edge. Some of the stats about us, we have last 12-month revenue of $653 million, with the most recent quarter growing revenues at 20% year-on-year. And this revenue growth and this revenue number is across 3 different business product lines that we disclose. We have a delivery services revenue, we have security revenue, and then we also call other, which is a combination of our compute plus observability. We play in a very large TAM. And so we talk about a $22 billion TAM in 2026 for those 3 business lines. We also have over 600-plus customers that we define as large customers. Large customers is defined as customers with over $100,000 annualized revenue in the quarter. And then we have a pretty global employee base of 1,100 employees around the world and then a last 12-month NRR of 113%. We're very proud of the last box on the bottom. This is kind of some of the customer stats around what we do for our customers. We do it with a 95% plus CSAT score, processing over $5 trillion average daily requested -- request served. And then we -- if you look at performance from a business perspective, we outperform our competitors with a 32% faster time -- average time to first bite than other CDN providers. We play in, as I mentioned, the $22 billion market. We believe it's a large and growing addressable market. And we have a unified platform. So whether we're serving delivery traffic or we're doing security on our network or we're doing compute on our network, it's one network that we support. And we think that having that unified platform makes us better from a technology perspective. It makes our customers have a much better customer experience. And then we're seeing some pretty good tailwinds with Internet traffic and especially with AI trends that are happening in the industry. I think that the combination of continued traffic growth, more moving like live streaming, streaming events as well as more transactions processed on the Internet as well as kind of the AI tailwinds that we're seeing out there provides a unique opportunity for the edge. We are mission-critical to a lot of our customers. We are very vital for what they do on a day-to-day basis, whether it's their monetization strategy, whether it's their consumer experience or enterprise experience, we are mission-critical to our customers. And then I think what's really important is that we have multiple levers for continued growth in our business. And then if you look at us versus our competitors, we are a clear market leader with a differentiated technology. We win where performance matters. Our customers choose us because of performance requirements. And so definitely something that we're very proud of. I think I mentioned on the kind of tailwinds that are supporting us. I think we started first with the first box here where there is more traffic growing on the Internet today, greater coverage, a lot of like service affordability out there, 4K TV dominance. And so you start seeing a lot of more access and more pixels kind of going across the Internet. And then you combine it with like what we see around like peak events, downloads, right? Like you see like big gaming platforms doing downloads that are big massive files, a lot more live events. And then more and more the market and traffic is transitioning from like what we've seen in the past to more like AI-generated content, AI queries, et cetera. And so we think this growth plays in our favor. We think that 2026 -- 2025 traffic growth was in the order of 15.4% year-over-year and we reached almost like 33 exabytes per day. So definitely good tailwinds on the traffic growth and the reasons why traffic is growing on the Internet. For those who are not as familiar with the story, this will help kind of position with the edge cloud. I mentioned that we are an edge cloud platform company. You see us playing in the middle between the central cloud and the end users. So think of the central cloud as where the hyperscalers play, right? They have big massive data centers. Those big massive data centers are in remote locations where power is cheap and space is cheap. And -- but they're not close to the end users. And so for people to hit the Internet and hit those hyperscaler data centers, they leverage technology that us edge cloud providers provide. So we started first with edge -- the content delivery network where we cache big files on the edge. And then more and more, you see security moving to the edge where we're the first points of entry to the traffic, and we can block and provide more intelligence to that traffic, right? And so security plays a big role in that. And then more and more compute will happen. So if you look at the central cloud and edge cloud, we're kind of synergistic, like we partner with the central cloud players, the big hyperscalers because those are far remote locations and the customers need a better user experience. We also help lower cost for our customers, right? Because those customers save on egress traffic that goes to the central cloud. If the central clouds are far away, there's a lot of bandwidth costs that end up happening. And if we're caching the content or we're providing more security and blocking that traffic to even entering the central cloud, we're saving our customers lots of money. So the edge cloud plays this unique role that helps improve the user experience from an end user experience and then it also kind of saves the customer a lot of money from egress traffic. If you think about what's better done at the edge, now that you see how the edge plays with the central cloud, the first use case of the edge cloud was the content delivery network, right, content and application delivery. That was kind of the first use case that was where a lot of time was spent. Years later, they've realized that, hey, all the traffic is going through the edge cloud. If we can layer in intelligence, and provide security at the edge, that's an even better value proposition. And so more and more companies have moved from just purely content delivery into security. And then I think that with agentic and where things are moving towards, compute will also be done at the edge. And so I do think that it's still very nascent and very early, but it is a good tailwind and another better use of the edge cloud. And then because you're managing traffic as it comes into the Internet, there's a lot of observability tools that can be used as well. So I think that from a use case perspective, these are the 4 big use cases of the edge. This is how we divide up our business. And then when we disclose our revenues, we disclose it as delivery services, security and then we put compute and observability as other. The reason why these are really strong use cases for the edge is because speed matters a lot. Customers do not have the patience to wait to load pages, and they don't want to make transactions or make purchases where speed is slow. I think what else matters is that in this world of like real live news updates, think like New York Times and election nights, right? You need very fast like purging technology, which allows more personalized dynamic content being displayed. So these content delivery networks like us, you need really good technology because the content is very dynamic. In the world of social media, in the world of personalized content across the web, you're going on a shopping site, you're going to get recommendations on what else to buy. And so everyone needs personalized experiences. Having an edge provider who can provide configurability is really important. And then I think that the reason the edge is being used is because of resiliency and mission control. When you have a central cloud that houses all your data center information back there, like if you're under a DDoS attack or you're under like a different event, having multiple POPs at the edge and blocking traffic and providing more resiliency because you're distributing your content across multiple POPs is a much better experience. And so people choose the edge for resiliency and mission control. And then I talked about enforcement and security, like if you can enforce security at the edge, even better, right? It's a much better experience, you're blocking information or you're letting in the right players. And then I also talked about lower cost. Being at the edge allows a lot of the customers that we have save them from egress traffic. Our footprint today, we have 167 POPs throughout the world, delivering 578 terabits per second. If you compare us versus our competitors, what you will notice is that we have fewer POPs, but much more powerful POPs. And the powerful POPs enables us to have better performance. And it's -- our POPs are very software configurable and very programmable using off-the-shelf products. And so it is a very much more capital efficient and better way to run the network versus our competitors. We were formed 15 years ago based on a very modern architecture. I think we were founded by a Chief Technology Officer, Artur Bergman, came as the CTO of Wikia. He has always been very focused on the customer experience and on being technology first. As a result of that, we are recognized by analysts and investors as -- our customers as one of the better products out there, right? And so we are constantly listed as a leader in the spaces we play in. We've won Customers' Choice Award 7 years in a row. And so across a different number of categories, we tend to win awards because we do have a higher performance and better product out there. So we are recognized by industry analysts and experts. When you think about like the -- the revenues that we have, it cuts across the 3 kind of circles that you see here, where network services is 73% of our revenue. That's the content delivery side of the business. That's growing 11% year-over-year. Our security has been kind of the bright spot, 22% of our total revenues. That's growing 47% year-on-year. That's also the fifth consecutive quarter of accelerating revenues in our business. 18 to 24 months ago, we had 1 security product, and now you see the 5 security products we have today. So that's really evolved and definitely helped see continued acceleration across our security products. And then other is a little bit more nascent. I mentioned compute and observability. That's about 5% of our business, and that's growing the fastest at 67% year-over-year. That continues to also grow on a year-over-year basis. And so we're definitely, for me, continue to invest in our core business, which is network services, but we have growth drivers across security. And then we have kind of areas that we're playing with because in the future, we do think that compute will play a bigger part. I mentioned some of our growth drivers, but being a platform play is definitely very helpful and nice, like I think starting with content delivery, but then layering in having 5 security products has helped us gain momentum across our business where we are delivering 20% year-over-year growth again. We have accelerated our R&D momentum. I mentioned 18 to 24 months ago, having 1 security product with Al. We now have kind of the full suite of 5 edge security products, launching DDoS, bot management, API security and client site protection. You see across our businesses, we're growing faster than the market in all 3 of our businesses. And you see that because we are doing competitive takeouts with our customers. And so we are winning where performance matters, and we have a better go-to-market sales execution motion than we've had in the past. And then I think with the other area that we're investing is international markets. I think that if you look at what we've been doing, especially in 2026, we've invested a lot in our APAC region. We have a new leader who's in Singapore. Prior to her joining, we were supporting Asia from our London office, right? And so you can imagine that experience being not as where it should be. And so now having more feet on the ground in the Asia Pacific region, I think that is a definite opportunity to see that. We're investing in POPs internationally. And so in that slide around our POPs, you've seen some purple circles where we're investing in, for example, Latin America this year because World Cup is going on, right? And so like we are continuing to invest internationally. I went through some of the financial highlights. Here, you see the same numbers kind of repeated. I would say that if you focus on the bottom box because the first 3 we've talked about, our record gross margin last quarter was 65%. And so we had record margins in gross margins there. If you look at what we've done on a last 12-month basis, we've actually had incremental flow-through on a gross margin perspective of 89%. So every dollar of incremental revenue that we've generated, we've flown 89% through to gross margins. We're working really hard to increase customer commitments. Yes, delivery services is a consumption-based business, but it doesn't mean that we can't have RPO. And so you'll see our RPO growing 63% year-on-year, last quarter at $369 million. And then I think the last one is we had $42 million of last 12-month cash flow positive. If you look at the last 5 quarters, we've really turned a corner. We've had now 5 consecutive quarters of free cash flow generation. Very proud of that. If you look at full year 2025, we have had record gross margins and operating income. So going to this page here, you'll see our gross margins at 61%, really picking up. And then I mentioned last quarter in Q1, we had 65.1% gross margins. And so definitely very proud of the momentum, not only on the revenue growth that we're seeing, but also on the gross margin flow-through and the gross margin contribution that we are having. And then if you look at the OpEx leverage over time, we're definitely doing that by making sure that we are judicious. We're investing in R&D, we talked about R&D velocity, but yet we were able to still reduce costs and still deliver continued product innovation around security products. We were able to deliver incremental sales with lower sales and marketing expense in 2025. And so very proud of the operating expense leverage we had. And as a result of this, what you're seeing is the blue and dark blue boxes moving positive, right? We've had a full year in 2025 of operating profits, and we had a full year of free cash flow positivity. And if you look at like the quarterly box, it's 5 consecutive quarters of free cash flow generation, and we have continued kind of operating income momentum across the board. So very proud of what we've done as a company and really like excited about what's to come going forward, especially in the opportunities around traffic, traffic growth, the agentic opportunity, the compute opportunity and the security opportunity in front of us. So with that...

Let's go into the fireside chat. Rich, thank you for giving us that overview in quite a bit of depth. One thing I think that investors really want to understand is just the strength that you've seen within the security business. What's maybe driving that? And what capabilities have you added to allow Fastly to achieve that stronger security growth?

Yes. I would say that security growth would probably be a few factors, right? One is that we talked -- I talked about the product velocity, right? Once you start going from one single security product, which is the web application firewall and going to launch the full suite at 5, right, and getting to parity with your competitors, I think that R&D velocity opens up doors. When you only have one product, you get shut out of RFPs because customers will say, like can I scale with you? And if you only have one product, it's hard to scale with you. So having the 5 products was very key because across like the edge security, now we are at parity with our peers. I think that's number one. I would say number two is that we really revamped our go-to-market execution, right? Scott Lovett had joined mid-2024. And prior to Scott, we didn't really have a security selling motion. Scott came from Akamai, but he also spent time at Imperva, where he was doing security sales. So he brought in sales leaders who knew how to sell security. Selling security is a very different kind of like motion than selling content delivery. And so having the know-how and then changing comp plans to be able to like do cross-sell opportunities and kickers is really key. So I think that if you couple product velocity with go-to-market changes, that's kind of the 2 biggest drivers. I would say the other big driver is like the proliferation of like the complexity of traffic over the Internet, right? The complexity is happening, more DDoS attacks, more bot management, like more bots scraping the traffic. And then -- so I think just being -- having the right products at the right time with the right kind of like products is definitely a good tailwind in our favor. So I would say 2 executional one and then a market dynamic has been helpful.

Excellent. Excellent. I mean, in terms of AI, I mean, this has clearly been a top of mind discussion area. Where does Fastly potentially benefit from Agentic AI traffic? And where does that maybe show up in the P&L? Like which business units do you typically see that in?

Yes. I mean I'd start first at the highest level, which is like how does agentic impact traffic flows across the Internet across the edge. I would say agentic use cases are still early, but I do think that it's going to increase traffic on the Internet in general. It won't increase on a gigabyte basis because AI traffic is going to probably be small bits compared to like streaming or video. But on a request per second basis, on a like number of times it's getting pinging the Internet, it's going to be massive. The growth is going to be so much faster. And I think that, that's going to be a big. Now how does that impact our revenue lines? Starting first with security, that's where we're seeing the majority of the impact from AI right now. With the proliferation of AI, like you're seeing more DDoS attacks, more bot management. Like if you're a customer and you're creating content, so let's say you're in New York Times, if AI traffic is scraping your content, like you're losing eyeballs, which means less ad revenues or less subscription revenues. So your monetization model is kind of at risk. So there's definitely a need for like bot management, like what are good bots? What are bad bots? Fastly, help me block the bad bots who are scraping my content, but not paying me for it. So with Fastly, we've adopted what we call -- it's the RSL, the real simple licensing standard, and we're actually helping drive a lot of that, too. And so I think we're seeing the initial kind of on the security side, definitely more of the like AI traffic and how it kind of provides tailwinds. I do think that compute is very early days. I think that right now, a lot of the AI is more inference. Inference is going back to the central cloud. Like the hyperscalers are building big massive GPU data centers. And a lot of that is actually more about information gathering and information gathering does not require as much workload from the edge. But as you move to agentic, you get -- go away from information gathering and you go towards execution. And when you go through execution, there's going to be orchestration that is required, right? Like you're going to need to have like perhaps 10 agents or 5 agents working on your behalf. These agents will sit in different hyperscalers in different locations. And orchestrating that work has to be done in a location. And I think the edge is perfectly situated to really orchestrate that work across agentic, right? It's still early days, like how many companies really have agentic apps out there, right? But I do think that when you start saying, "hey, tell me what to do in New York for 2 days" and you're doing information gathering, that's not orchestration. That's just information gathering, that's the language learning models, the frontier models. But when you get to like, "hey, work with a United agent to book my ticket", "I work with an Uber agent to pick me up", like there's going to be agents working on your behalf and doing execution. Well, that orchestration needs to happen somewhere. And I believe that it should happen at the edge. I think the edge is kind of the right place to kind of do it because you need to be efficient about how traffic is kind of moving around the Internet. And then I think on the last piece, we have network services, like how does agentic impact our network services revenue. I think that right now, like a lot of us in the space, we price on a per gig basis. And I do think that there's opportunities to think through how we think about pricing and how we think about like the impacts there. Is it more on a request per second basis? Is it -- how do we kind of do that? I think overall, network traffic is going to increase because of agentic, but how do we take advantage of the opportunity? And I think that's still kind of -- we're not there yet.

That makes a ton of sense. I mean, I think around the AI topic, the other major concern here is what's going to happen with Mythos in particular. And there seems to be a lot of talk about significant increases in known vulnerabilities that are going to be out there. Not all of those vulnerabilities can be patched. And so what role can your WAF products play in maybe that opportunity?

Yes. So I think that we are big believers on like even like AI and how we can use AI to identify vulnerabilities. I do think that we've been very good about making sure that we have internal technology to like identify correlation and traffic flows and figure out like, which ones are bad. But I think that there's opportunity to continue to leverage AI in that front. And so like Mythos would be a great example of like how do we leverage AI to really identify vulnerabilities ahead of time. But you have to kind of couple with like the know-how that we also have in seeing traffic flows. I think that because we sit at the edge and we see a lot of traffic before it even hits like other places. We're in a very good position to combine our know-how with kind of like the AI and learning models to really identify those vulnerabilities for our customers. And that's part of the reason why our security products continue to win awards, because we're kind of taking the best of both, like we're seeing the traffic upfront, and we're combining our know-how with kind of the tools that are out there.

Excellent. Excellent. Just one last one. So do you need to make more incremental investments in compute or infrastructure, particularly on the GPU capability side? And are there opportunities to partner with some of the larger frontier models that are out there?

Good question. I think we get this a lot. In terms of like how are we doing compute today, right? Compute is still a pretty nascent market for us. It's still -- I think it's like I showed you the compute -- our other number. We are doing a lot of that compute with our existing CPUs, right? So we have a bunch of servers that are sitting in our POPs. They have CPU capacity, and we're able to process it through the CPUs that we have. I think that we are continuing to co-innovate with our customers. So that co-innovation means like what do you need from a customer perspective? Like how do we solve that problem? And so we're working with them on like how do we solve that best. And we find that right now, we continue to solve their problems through existing compute resources. We are ready to do GPUs if our customers need it, but we're not in a position to like want to put GPUs in the edge if there's no need for it, right? Because once you start deploying it, one, they depreciate pretty quickly, especially with the iterations that are kind of launching. But if they sit idle, like that's a gross margin impact. And so I think that co-innovation with our customers is actually really key because that co-innovation allows us to have leading views on that. Now the second part of your question is like, are there opportunities to kind of partner with the frontier models? And I think absolutely, right? I think that we welcome and we work with the different players out there. And depending on the use cases that they see, we're always working with a lot of players out there across all different types, including the frontier models.

Excellent. Excellent. We can take one question from the audience, and then we'll break for the breakout session. Let's -- go ahead.

Yes. So the question in the room was that there was an 11% growth rate for delivery services revenue. As we see the orchestration layer happening on agentic AI, like where should we see that? Would we see it on the 11% pickup on delivery services? I think the orchestration layer ends up happening across a number of areas. I do think that like we should see it on the compute side, right? On the compute side, we're going to need to like be able to orchestrate and do computations around like, hey, this person is asking me these things. These are the 3 criteria. I need to work with like 10 different agents to go out there, but you need kind of like that compute to happen so they can send out instructions. Now I think that every compute needs to have some storage and some caching information. And so there might be some pickup in the delivery services as well. I do think that having multiple lines of businesses enables us to play together in that orchestration layer that needs to happen. But it's still early days, I don't want to get people too excited about it, but I do think that from agentic world, it's going to change the way the edge plays, and we need to be ready for that.

Perfect. Perfect. We're going to take a quick 5-minute recess and then we'll continue the discussion.