Fastly, we are an edge cloud platform company. When I say edge cloud platform, think of us differently than the kind of central cloud where the hyperscalers play. You know, the hyperscalers like AWS, GCP. They're massive data centers in remote locations. They're not in metro, close to the user. Part of the edge cloud, we basically have more data centers, smaller, but those data centers are in metro areas. The value that we provide is basically being close to the user, so there's a performance gain for the end users, and then a cost savings because we're saving the, our customers' egress traffic going back to the central cloud to do things.
If you imagine, like a big Fortnite game download, you're only copying one game download to our data centers, and it's going and getting distributed to all the users in these metro areas. There's, like, massive traffic savings. That's the edge cloud platform we play in. Edge cloud was historically built on a content delivery network, you know, which is caching and storage of, like, big files, getting closer to users. You know, version two was kind of doing security at the edge, which is really kind of about preventing bad traffic and then helping, you know, good traffic go through the network and just routing it and being more efficient and security being a big player.
I think that, you know, version three is I think what a lot of people are just talking about and excited about is, like, compute at the edge. Which is, like, how do you do computational exercises, how do you do Inference at the edge so that, you know, performance gets better. I think that, you know, this is kind of we're seeing the third flavor of kind of the edge cloud platform.
All right. Great. That's very helpful. Talk to us about some of your biggest takeaways from the quarter. Specifically what's different, you know, operationally, you know, with you guys now versus kind of what we've seen last, you know, last few years with the company.
Sure. We reported our Q4 results. Our Q4 results, we had our fourth quarter of re-accelerating revenues. Revenues in 23% range. The re-acceleration wasn't in one particular business line. Network services accelerated to 19% year-on-year. That's our kind of CDN business, our traffic business. Our security business accelerated. Our security business grew 32% year-on-year. We have an other business line that also accelerated, and that grew 78% year-on-year. Other consists of our compute business as well as our observability business. All three business lines really seen a lot of strength. In addition to kind of the top line growth, we had very good performance from a gross margin and operating margin perspective.
Gross margins, we hit record gross margins of 64%, the highest we've had in company history. For all of 2025, we were profitable and free cash flow positive, and we were free cash flow positive all 4 quarters. I think from an execution perspective, it was just a combination of the changes that we've made over the last 18-24 months as a company that really allowed us to not only re-accelerate growth, but also do it in a very profitable and gross margin and operating margin accretive way. Some of the changes that we've made over the last 18-24 months, you know, we had a new go-to-market president who joined in mid 2024. You know, he spent 2025 really kind of revamping our go-to-market, you know, execution engine.
You know, if you look at the 15% year-over-year growth we delivered in 2025, we did that with sales and marketing expense declining. You know, we're producing 15% more revenues, but sales and marketing expense declined on a dollar basis for the year. On top of that, you know, like we broadened our product portfolio. We went from one security product 24 months ago to five security products today, which also kind of really helped the go-to-market engine and execution. You know, Kip is our new CEO who joined, he joined over two years ago, about two years ago, but he was named CEO seven months ago, so he was the prior Chief Product Officer. We had some changes around the Executive team that really helped.
I think the combination of go-to-market changes plus kind of executive changes, plus the broadened product portfolio has really helped kind of, you know, give us the results that we had, and that you saw in Q4.
All right. Great. Yeah, you talked a little bit about some of the share gains you've got from routing. You know, you mentioned earlier performance and resiliency, that sort of thing. How much of that in, you know, we saw in Q4 and in the guide was kinda share shift versus sorta baseline traffic growth that you're seeing out there?
Yeah. I would say that, if you look at the network services business, I think we did, you know, see improvements in traffic growth in general. Traffic growth, we saw a noticeable pickup in Q4 traffic overall. I think what really helped it is it wasn't just traffic growth, but also price erosion improvements. I mentioned on the earnings call we had a mid-single-digit price erosion in the quarter. I think the combination of kind of, like, more volume plus kind of, like, price improvements over time have just really helped the story and really helped the business. I would say that, like, you know, our traffic share gains isn't just in the network services side where, like, we're taking and growing faster than the market.
We're also growing faster than the market in security. We're also at the same time, doing share gains with some of the security sales that we're seeing. Then, of course, you know, our other business, we're also, you know, winning business there. I think overall we've been executing on all business lines. It's a combination of upsell with existing businesses as well as the way we've changed our comp plans to really encourage cross-selling.
So with the price erosion, you know, improving a little bit, how much of that is more of a mix shift of what you're seeing, versus or is it kind of across the board and we're seeing improvements in maybe some of the legacy traffic, you know, pricing as well? How should we think about what's driving that improved price decline?
Good question. I think it's a, it's a number of factors. I would say that the way the industry has evolved around the edge cloud platform companies. There was this, like, buildup of capacity in 2022, 2023 timeframe, and what we saw is, exits by some of these players that has actually made it much healthier from a pricing perspective, right? As companies like Edgio went out of business, they were doing irrational pricing that really kind of ruined pricing environment for everybody. What you're seeing now is a combination of a few things, right? One is that you have rational players. Us, Akamai, Cloudflare, we're in it. We're here. We're rational players. You know, we're going to make sure that when we sell, we're selling based on, the value and performance that we have.
you know, I think that you know, for us, you know, we're all focused on making sure that, you know, as public companies, you know, we deliver, you know, margins and profitability for our investors. I think combine that with kind of, you know, since they're exiting, there's a rational supply of, you know, in the market. Rational supply means rational players in the space. I do think that traffic in general from the edge is getting more complicated, right? I think that if your traditional CDN, people still think of CDN as a commoditized business. If you think about CDN and the traffic that's happening on the network today, it's a much more complicated traffic.
When you couple that with security, you couple that with, like, what's happening in AI. I think that, you know, is there a barrier to entry potentially of new competitors coming into the space? You know, new players can pop up. Given the complexity, I think that, you know, it creates somewhat more of a barrier to entry around irrational kind of behaviors in the market where there's excess supply again. I do think that, you know, that also helps the pricing environment when y ou have, like, a slig ht more barriers to entry, but also a more conversations with your customers around performance selling.
All right, great. One of the things I thought was really interesting out of the quarter was start looking at the guide and, the revenue commitments that you're now getting from some of your customers. Maybe walk us through, you know, what's kinda giving you that increased level in the guide and those commitments and what is sort of, you know, how much of your revenue is now are you now able to get those types of commitments and so forth?
Yeah. For those of you who, you know, read the earnings report, we reported record RPO. RPO, the Remaining Performance Obligations grew 55% year-on-year. We also broke ou t the current portion of RPO, and the current portion of RPO grew 37% year-on-year. We're definitely getting greater commitments from our customers around, like, contractually committed dollars of revenues that we see in the future. If you take that current RPO and divide that by, like, the midpoint of our guide of $710, like, we're about 35% committed. You know, I think 35% committed for the year is definitely better than where we ended the prior year, right? If you look at the prior year current RPO divided by the revenues of 2025, we're about 30% committed.
That gives us more confidence with kind of, like, the revenues just because we have greater commitments. It's not 100%, and I don't think we'll ever be at 100%, but I think the closer we get that to, like, 70% or 80%, we'll start feeling much better about the guide that we give. I feel, to me, you know, we've triangulated and looked at our customers and the contracts we have in place, and, based on that, you know, bottoms- up build by customer, we felt really good about the guide that we had, if you look at the way we build our model plus the commitments that we're getting.
All right, great. It's a nice shift. It's like you guys said you can differentiate some between some of the agentic traffic and browser requests.
Mm-hmm.
Maybe break that down, k ind of what percentage from both and how you're seeing that those types of traffics in the network.
Yeah. What conference wouldn't be complete without the AI question? Yeah, we do see an increasing amount of agentic AI traffic on our network. We haven't disclosed to the street what that is exactly yet, but we do-
Do it today. Come on.
No. We're not gonna be doing that today, unfortunately. By the way, we're not a seat licensed-based company. We are a consumption-based company, which may also be an implication for an AI answer there down the road. There's a joke in there somewhere. What we're seeing is a sort of what we would try to characterize as rising tide lifts all boats. For those of you that aren't familiar with what we're talking about, when you go into one of these LLM search tools like a ChatGPT or a Perplexity AI or what have you can see the tool scraping other websites. When that's happening, that's basically an agent going out and searching on your behalf. The traditional model was a human would go in and click through to scour for these answers.
A couple of interesting things are happening because of that. Obviously, it's a force multiplier with the traffic that's happ ening on the web. We're benefiting from that in a sort of an indirect way. Our customers are out there at these websites seeing more traffic on their websites. We're sort of like a consumption-based toll that they pay on that, so we benefit from that. What's also happening is you're now having a harder time as a content enabler on a website distinguishing between what's an agent or a bot versus what's a human. There are a lot of tools out there.
We already have a security tool called Bot Mitigation, which has its roots in, if any of you know the sneaker drop phenomena of, like, a new sneaker comes out and basically a robot, an agent goes in an d buys all the sneakers, or Ticketmaster drops a new Taylor Swift concert. You're competing with robots. There's a lot of money paid to filter out those requests versus actual humans, and not have to go through the torturous CAPTCHA things that we all have to deal with. That's a wide open field now for a lot of opportunity in how that traffic gets managed. There's also a lot of new licensing models that ourselves and our competitors are wrapped around.
We have, we have one standard that we've developed, called RSL for Real Simple Licensing that helps users distinguish and put tools in their place to figure out which is gonna be like benevolent agent traffic that's gonna scour the website versus agents they want in, where you might have the LLM actually or someone else actually paying for the access to those assets, because when you're not in there, you're not as a human seeing the banner ads and paying for that. There are some i nteresting models that are evolving around that, and we feel like we're really well positioned as that unfolds, especially around our security products, which have a high degree of AI in them right now.
All right. With that, you touched on some of these, but, you know, where do you expect to monetize AI first? Is it going to be in delivery traffic, this AI bot mitigation or compute and inference sort of things at the edge? How should we think about that?
Yes. I think to rank them, I would say definitely more near term, it's the sort of rising tide. I mean, the traffic increases are happening out there. I'd say more medium term over the next maybe two years, you'll probably see us have a much broader role in that security mechanism, trying to distinguish between, again, like what's, you know, human versus agent, and then also benevolent agent versus malevolent. There are a lot of new developments happening in that space around the security engines that we have out there. That some of that does go hand in hand, like sort of glove in hand, if you will, with our DDoS or WAF products.
Certainly around API protection, which is basically the interfaces into people's websites and their programs, there is a lot of discussion around how to manage that. We're looking at implementing basically AI within that to help our users manage around agentic inbounds and see how we can monetize around that and how they can make money as well as us.
All right, great. For the inference and, you know, computed edge, you know, where are you on maturity there, and what do you need to invest in? What kind of hardware, what other types of things do you need to invest to make that work?
The computed edge platform for us is effectively like a compute fabric. I think Richard already kind of explained, like this is an emerging area. It's been talked about for probably a good decade now. The idea is we would take a workload out of the central cloud and put it on the edge and process that faster. A great example that all of you use every day, well, not every day, but probably every week, is e-commerce. If you're on a big website and you're putting something in a shopping cart, unfortunately, you might get a recommendation at the last second. There's a high probability that's run through a computed edge workload. It would probably be us or one of our competitors.
Also, when you're shopping online, there might be other processing around the transaction itself that takes place at the edge. If you click buy and you were, say, in Boise, Idaho, and you had to go to a server all the way in, you know, New Jersey and back, you'd start to get pretty upset, you know, going into the central cloud at wherever that process is handled. You probably are accessing a server that might be, say, like in Denver by ourselves or one of our competitors that gets back to you much faster. That's what we're talking about here. There's a whole, which I can't do justice in like 30 seconds on, but there's a whole industry around Time to First Paint, like abandoned shop carts. Any time there's latency around that, customers get upset.
Why that relates to computing is you need some of that computing on the edge to make this more streamlined and efficient. When it comes to AI, there's gonna be this concept of Inference at the edge, where you will put those AI, those compute workloads to use on your behalf to have agentic traffic going out and doing things. Again, it's almost like, now my analogy probably won't come together perfectly, but the idea of having this workload closer to you that you're orchestrating as opposed to having to go back and forth to the central cloud. As the end user, you're gonna be upset 'cause it's very slow, and it kind of already is today, by the way, with a lot of these. They can probably be sped up by partnering more with ourselves and our competitors.
Also on their side of it, they're paying a lot of money 'cause every time you have to go back up into the central cloud to Google or Amazon or AWS, you're gonna pay egress fees, and those are very costly to go back and forth on. We eliminate, and our peers help customers eliminate that by locating these assets on the edge and having faster retrieval.
That just o utlined my whole coverage base and why it's very, very important here, so it's a beautiful thing. All right. One of the things you've emphasized last couple quarters has been cross-selling across your platforms. Maybe talk to us about that, and then how should we think about, you know, top 10 versus non-top 10 cohorts and how those are growing? Is it, you know, evenly distributed? How should we think about it?
On the cross-selling, I think that, you know, I mentioned that Scott Lovett, who joined mid-2024, the things that he's done to, like, really fix the go-to-market engine was, you know, not only kind of revamp the way we do segmentation and how do we go after different verticals, but he really emphasized a lot of on, like, how do we sell not just content delivery and network services, but how do we get into security? Scott, you know, by way of background, he came from Akamai, but he also had spent some time at Imperva, and he had experience not just selling content delivery, but also selling kinda security overall. I think that was a really helpful kind of skill set that he brought in. He's upleveled his leadership team and brought in people who know how to sell security.
He also kind of revamped the 2025 commissions plan, right, and introduced, cross-selling kickers. The idea is that, you know, you're gonna have bookings targets, but we'll give you kickers if you can do cross-selling. Those efforts really kind of revamped and trained the way the sales teams went to market, which is really helpful. That alone doesn't help if you don't have the product portfolio to go with it. I had mentioned that Kip came in as Chief Product Officer and really broadened our product, you know, efforts, our product roadmap around security, going from one product to five. I think when you have one product, it's great. You can get an RFP.
A lot of times you get kicked out of the RFP because in the RFP, they may say, "Hey, you only have one security product. We can't really grow with you." Now by having the kind of be at parity with our competitors and having five security products, that actually keeps us in the running for security. I think if you combine those two, that really kind of increases the cross-sell motion that happens at the company and really starts, you know, seeing the kind of re-acceleration that you saw in security, where, you know, we had our fourth sequential quarter of, you know, security acceleration, and it grew 32% this quarter. Last quarter, it grew 30%. Not only are we re-accelerating it, but we're also doing it with a big customer win.
In Q3, we talked about a top 10 customer who historically had only bought Network services and, you know, had entered into security. Even if you exclude that one customer, you know, we'd still saw in the re-acceleration of security on that. I think the cross-sell motion definitely helps. Frank, I think the second part of your question was around how do we think about top 10 versus non-top-10 customer. Our top 10 customer concentration did increase. We had 32% last quarter, we went to 34%. The question is like, do I worry as a CFO that, you know, it went to 34%? I would say absolutely not, because if you look at it, even we had greater sales in the top 10, and that contributed to more, we had record gross margins, we had record profitability in the quarter as well.
These top 10 customers are very profitable to us. They're, you know, very accretive to us, we should definitely continue to focus on selling as much as we can to these top 10 customers. I would be very worried if the non-top 10 did not grow fast. Our non-top 10 continued to accelerate. Our non-top 10 grew 20% year-over-year, and that was an acceleration from the prior quarter where it grew 17% year-on-year. Not only was top 10 growing, but non-top 10 was also growing. As a CFO, like, I don't worry that the top 10 had grown in Q4. I, you know, I think we'll probably still be in the kind of low to mid 30% range from a top 10 customer concentration perspective.
I do think that, you know, I watched the non-top 10 just as much, and I think that we had a really good kind of outcome where the non-top 10 continued to grow.
All right, great. Talk to us about the typical profile of a, of a security- first customer. What's kinda the size? What are they buying? What, you know, what convinces them to jump into Fastly when they're, when they're coming from that direction?
I think, you know, for us, you know, when we do initial customer land, like, we're agnostic if it's gonna be a network services customer or it's a security customer. I would say that, you know, our WAF product, you know, we've had around for almost like, you know, since 2020. The WAF keeps on getting better and better. The WAF is the web application firewall. Customers who buy that, you know, just they know. You know, we win the Customer's Choice Award, I think six or seven years in a row, we've done extremely well on that product. I think for us, we've been agnostic around, like, do we go security first or do we go network services first? We are, you know, an edge cloud platform company, we're gonna go after both.
Our historical bread and butter has been on the network services side. This is an area that we're still investing in and we'll continue to focus on. In those customers who wanna go security first, we're totally happy to go security first and then potentially cross-sell network services on top of that. Now, the reality is 76% of our revenues come from network services, and so it's still, you know, a, you know, network services kind of motion. Security, you'll see security playing more and more. I don't know there's a specific trend on, like, is it a certain vertical or certain type of customer or not. We can have a big customer be security first, and we can have a longer tail customer be security first. I wouldn't necessarily call it any verticals, or spaces.
Yeah. I would just add to that too. If you look at the landscape of what we compete in, which is called the web application API protection security bucket, which is very different from, say, like, an Endpoint security solution. Again, I probably can't do that justice quickly in front of this audience, but in that specific market, we're very under-penetrated. We're probably less than 5% market share. Our nearest competitors that have the same setup are probably 2x-3x times that, depending on the company. There are very small point players that are in the space, only have the security without the delivery side, and they're getting more and more marginalized as the days roll on, and we're taking a lot of share in that area as a result.
All right, great. Maybe take one more question, and we'll get some questions from the audience. Your, you know, your tagline, many on-ramps to one destination, you know, talk about platform convergence and so forth. You know, talk to us about, you know, how you're packaging and pricing those bundles to get folks in. That's a question that comes up a lot with investors, but how you think... How should we think about that?
Yeah, I think for us, we've always been, you know, a product that's been higher performant. You know, we sell to a lot of technologists who really like our product. We typically will work with them on, like, what do they need, what problems are they trying to solve, and what are the right products that we have to help them solve it. I think that from a product and packaging perspective, we've been very kind of solutions-oriented with our customers, and I think that we need to kind of simplify our packaging and pricing a little bit, and rethink about how we go to market on that front.
I think that, historically, because we were founded by a CTO who, you know, loves is a technologist at heart, like, we've made our product very solutions-oriented, around, you know, solving the most technical and hardest problems. As a result, we get the best, you know, the highest the customers who want the best performance. We do have some work around, like, how do we think about packaging and pricing and simplifying that part of it. I think that's more to come there. I think that, you know, I think for me, that's a huge opportunity to kinda unlock in the future.
All right, great. Well, why don't we see if we have any questions from the audience, if anyone has anything? No. All right. All right. Well, if you have one, grab one before we end here. I got a couple more. All right. You issued $180 million of 0% converts. You paid, you know, repurchased some notes. Talk to us about capital priorities between growth CapEx, tuck-in M&A, other things. How should we think about the capital priorities?
Sure. Just as a little background, I mean, I joined the company seven months ago as the CFO, and I looked at kind of our balance sheet. One of the things I wanted to make sure that we do is clean up the balance sheet, you know, stretch, you know, look at the pending maturity that was coming up in March, actually this month. I just wanted to be proactive and refinance that and kind of, you know, issue a convert to kind of take that one out. I think from a capital perspective, you know, we are in a balance sheet perspective, we're in really good shape. I think that, you know, we have over $350 million in cash on our balance sheet.
We don't have any near-term maturities that we have to worry about or think about. We are generating cash. We've generated cash for the last four quarters. we're kind of in a net cash position now, which is very good position to be in. I think that gives us kind of the ability to be much more proactive in how we think about how we spend and deploy capital in the future. We've historically not been very acquisitive company. You know, our biggest acquisition was in 2020 when we did Signal Sciences to get our WAF product and which got us our entry into security. We've done some tuck-in acquisitions to do acqui-hires. beyond that, we haven't been extremely aggressive on that front.
I think given the evolution of the company and what we've done in 2024 and 2025 to, like, really clean up execution and really, you know, fix, and think about how we operate and execute, I think we've earned the right to think to do that. I think that we also are a Technology-first company. Our bar is extremely high. And, you know, there's obviously things that we wanna build out and broaden our portfolio and make our products better. So we're gonna be open to it. I do think that the bar is high, given that we are a very technology-first. Our customers love us for our technology and how superior we are from the performance and reliability and scalability perspective.
With that, with M&A, you know, what kind of assets would be interesting to you? Is it more of a technology asset or a specific security product? What are sort of the criteria to get over that high bar that you mentioned?
You know, I think that we're actually doing some deep thoughts and analysis around the company around, like, you know, what are our markets that we play in? What are the gaps that we have to fill? What are the adjacencies, and how should we think about it build versus buy? You know, I think one of the things that we wanna do is we're gonna probably do an investor day in the second half of the year where we'll lay out our thoughts along our market. You know, thanks, I don't wanna give away any information that, you know, we're working on right now. The bar is gonna be high for acquisitions. You know, we are definitely open to it, right?
I do think that we're at a point where we are really happy and comfortable with the broader portfolio we have today, and we're just gonna continue to be opportunistically and look at it. Stay tuned for a second half of the year kind of investor day.
All right. That'll be great. The last couple minutes we have here, you know, been a lot of change here. Got new management team, a lot of things going, you know, move in the right direction. What are the maybe the top two or three misconceptions or one or two misconceptions about the company that you'd like to kind of address with investors?
I would say that the biggest one is that, you know, there's been a lot of investor like, you know, attention recently, but prior to that, like, there wasn't a lot of investor interest. I think that people had kinda thought about the market we play in being kind of content delivery network, and it's a commoditized space. I think that what you're seeing at the edge is very different now, right? Like, I think that, there's a lot more value being added on the edge when you start thinking about, like, the traffic flows and how much more complex it is and how security plays a role and how AI plays a role.
The need for a real good edge cloud platform company has become just much more massive, and I just think that the players with the right technology and the right kind of components that kinda fill not just content delivery, but security and also in the future, a real good agentic AI product will win. I do think that the evolution of that industry hopefully continues to get away from price erosion and how the traffic flows to more like. What's the value being created? How do we improve the performance for end users? How do we provide more security for the, you know, the content creators? How do we actually help them with their monetization engine?
you know, with these AI bots going out there and scraping content from The New York Times, like how do we help them, you know, monetize their engine in a way that, like, they're getting the rewards from the AI companies that are scraping them, not just, you know, going out there and allowing them to kind of be scraped and not get any value extraction from it.
All right.
I think that's like number one. I think that's like the biggest misconception. I think, you know, Fastly, you know, we are A lot of customers will say that we are high performance. We are a high performance, and I think the question is, like, how does a higher performance translate into the macro conditions that we're seeing in the edge, right? I personally, I joined seven months ago. I chose Fastly to join because I just felt like the complexity of traffic, the complexity of things happening at the edge, having a higher performing product positions us well to kinda be a market winner here. I personally chose to come to a company with better technology because I think that, you know, the edge is at the right place, and I think Fastly has the right technology.
That's great. Yep. The Q4 numbers kinda showed that was heading that direction. Well, this is great. Well, thank you very much for the time, folks. We have a breakout session afterwards if anyone's interested. We'll join you down there. Thank you guys very much.
Thank you. Should we turn this off? Thank you so much.