Excellent. Hello everyone, and thank you for joining us for our Growth Stock Conference and today's session with Akamai. My name is Jonathan Ho, and I'm the Cybersecurity Analyst for William Blair and Company. Our speaker today is Ed McGowan, who is the Chief Financial Officer of the company and has been with Akamai, and is with Akamai. 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. Ed, could I maybe have you start us off with a little bit of a level-set discussion on the Akamai story for those who are less familiar with the company? Maybe we can dive into a little bit of Q&A after that to go through the business.
Sure. Thank you, Jonathan, for having us today. Really appreciate it. Good to see everyone out here. So how many of you are familiar with Akamai? Show of hands. Okay. A few new folks to the name. Akamai was founded 25 years ago, designed to solve what was called the World Wide Wake problem, which essentially is the, you know, the internet was not designed for performance. Lots of choke points in the internet. The original start, reason for the company's being was to essentially solve the performance problem on the internet. How we did that was distributing servers all around the world close to where the end users are.
If you're requesting to watch a movie or to have a banking experience or a shopping cart experience, the concept was to deliver that as close to where the user is as possible, even though the origin may be thousands of miles away. It also enables companies to scale. Think of big, you know, live streaming events. The NCAA tournament was probably the first time we saw, you know, the massive demand for users coming online and wanting to watch television on the internet. It wouldn't have been possible if it wasn't for us because we could deliver that experience close to where the user is outside of the choke points on the internet. That was our first evolution of the business. We call that content delivery. We were originally formed as a, what's called a CDN or a content delivery network.
That business is, today, about a third of our revenue. We have evolved quite a bit from those early findings. I think the key to the CDN business is finding ways to monetize the platform outside of just serving content. We have been very successful in doing that. A lot of you may know us as a cybersecurity company. The second evolution of the company was leveraging the platform that we had and all the data that we see. We see every internet user multiple times a day. We know where all the bad actors are. We have some unique capabilities by being so distributed to solve some very challenging security problems on the internet. We have today about a $2 billion security business. It is growing about 10% a year.
Our primary solutions are about, say, a billion eight of our $2 billion of revenue comes from what's called web security. That is basically protecting websites and web applications from all sorts of different security challenges. For example, denial of service attacks is one way that bad actors will try to extort money from our customers by flooding the origin with a tremendous amount of traffic using infected devices. I'm sure you guys have seen emails, phishing attacks, texts, et cetera. Oftentimes users will click on those unknowingly and become part of what's called a bot army. A bot army is something that can be turned on and all the devices will go attack one particular site and try to take it down.
We've developed a number of solutions, including capabilities to help protect the websites from defacement or from people trying to steal information. It's called a web application firewall. We've also developed some technology to help our customers understand what bots are coming to their sites. A lot of the traffic on the internet today isn't actually human. It's machines. It's things like price scraping bots or search bots that are trying to index the internet, malicious bots that are trying to do things like credential stuffing where they may have a bunch of stolen usernames and passwords and they go to a website and just keep continually trying to put those in to try to see if they can get a hit.
We've developed technology that enables our customers to understand what those bots are, block them if they're bad, do different things. Maybe it's a price scraping bot. You wanna give it disinformation. That business has grown tremendously over the last 10 years. We've essentially evolved the company from which 10 years ago was about 85% CDN to now today about 52% of our revenue comes from security. That was a big and growing business for us. We've also evolved from just protecting websites and web applications to now working with the CISOs on protecting their employees and the applications that they run behind the firewall. We have some capabilities to replace a VPN. We also have capabilities to do micro-segmentation, which essentially will segment out your network and then enable you to get great visibility of what's going on in your network.
If something does get in, you can stop it from spreading. You can, you know, find things that get through all of your other types of protections that you may have. You might have endpoint security, email security, et cetera. Sometimes things get in. This enables you to be able to stop that before it becomes a big problem. That part of our business is growing extremely fast. It's growing at about 30% a year right now, 35%, excuse me. A great part of the business that's expanding. We recently acquired a company that gives us the ability to protect APIs. API right now is one of the most exploited areas of weakness for our customers in terms of protection. Most modern applications, almost every modern application, both web and non-web, use API.
We acquired a company that was a market leader in that field, and we expect that to be an extremely large business for us over time. We are very excited about our security business, especially with some of our newer security offerings. The last thing that I want to talk about and where the other portion of our business comes from is called compute. We have always been in and around compute in terms of leveraging our platform to do what we called edge computing. Our customers would come to us and want to run sort of lighter weight applications, say, for example, a waiting room application. You might go to a website, let's say they're selling tickets. Sale goes on at 10:00. Everybody hits the website at the same time.
Rather than showing a bad experience, meaning the site either crashes or you show some kind of an error message, you can develop a nice waiting room where it can tell you where you are in the queue, et cetera. We could do A/B testing, show an ad to one person, show a different ad to another, see how they're performing. Usually these applications were run using JavaScript or WebAssembly. We've recently moved into full stack computing, competing against the big hyperscalers. We have full stack computing capabilities that came through an acquisition we did about three years ago, a company called Linode, which was essentially focused on small, medium business. We acquired that company. We've added about 20 core locations. We've enabled our network now to run, you can run containers at any one of our 4,000 locations.
We're very differentiated from the hyperscalers in terms of being more distributed. We're a lot cheaper and it performs much better. That business is growing extremely fast for us. It's gonna grow about 15% this year. The sort of businesses, the legacy business is about half of the revenue today. We were exiting some of that, but we're really focused on what we call our compute infrastructure services. That business is growing about 40-45%, a couple $100 million . Very excited about that. If I'm up here 10 years from now talking about the business, I would expect that would be the biggest part of our business. With that, I'll go back to Jonathan for Q& A.
Excellent. Excellent. Thanks for the overview, Ed. You know, a lot has changed with Akamai over the past few years with the Linode acquisition, a critical element to help you move from largely a CDN and security-focused provider to now include, core cloud computing. Can you put it all together for us and maybe help us understand the grand vision of how all of these parts fit together?
Yeah. So, you know, one of the most attractive things about Akamai is our business model, right? So we've recurring revenue, customers buy multiple products from us. As a matter of fact, 70% of our customers today are both CDN and security customers. In fact, we've got about a third of our customers buying more than one security product. We're very successful in terms of leveraging that platform and building new capabilities on top of it. Our services all run on the same platform. For example, I might be serving a video to Jonathan's house and blocking an attack at his neighbor's house, right? The same server can do both things. We can apply the web application firewall rules at the same time I'm delivering a banking transaction or shopping transaction. There's an awful lot of leverage in the model.
Now, in terms of the compute business, we've enabled our network to now be able to run code in any location. You now can run things like we talk about Apple Private Relay. They came to us several years ago with a problem of wanting to offer a solution where you can essentially surf privately, if you will, but you didn't want to introduce latency into the equation. If that slowed you down, customers wouldn't adopt it, adopt it. They tried doing it with a hyperscaler. It didn't work. Came to us. We now run that in hundreds of locations and enable this to work so that the end user doesn't realize that there's, you know, a lot of compute going on behind the scenes.
Fantastic. Fantastic. You know, maybe similarly from a financial perspective, can you help us understand sort of the constituent components of the business and what does that margin profile sort of look like over time? What do the growth rates look like for those businesses?
Yeah. So there's an enormous amount of operating leverage in the business itself. If I think about the Salesforce, my Salesforce sells all products to all customers. We do have some small overlay sales. We do leverage a channel. About a third of our business comes through the channel. I expect that number to increase over time. Security, for example, most of our Guardicore sales, almost all of them, all of our sales for micro-segmentation go through the channel. A lot of our API sales will go through the channel. So we should get some significant leverage there. We also get leverage in engineering. The same team that builds the CDN platform deploys the servers. Also, you know, those servers run security, but also our compute locations are built by the same team. So I get some significant leverage there.
Also leverage with our engineering teams. In terms of the costs, people are our biggest costs. Labor is by far the biggest cost. The next cost is the cost to run the network. One of the secrets to Akamai is we started off in the early days really embracing a partnership with the big carrier networks. Our business is very critical to have good relationships with the networks. As a result, we deploy our machines in their networks, oftentimes in many locations. By doing so, we help the performance for their end users and also save them an enormous amount of money so that they do not have to build out infrastructure to go get content that the customers are requesting. A lot of that lives locally on our network.
As a result, we get free bandwidth in some cases, free co-location, and free power. About 80% of our bandwidth today is free. We're able to get enormous profits from our CDN business despite the fact that it is a pretty competitive business. In terms of the gross margins, we run cash gross margins in the low seventies. Our operating margins are around 30%. As I talked about with the operating leverage, with the growth of security and compute, we believe that our operating margin will improve over time. We're still in an investment mode now where we're investing. We're a little bit off that 30%, which last quarter we were 30%. We got it to just slightly under that this quarter. We believe that we can, as those products begin to grow, our margins can expand over time.
Excellent. Excellent. I don't think we can have any discussion at this conference without talking about AI. I think, you know, first of all, you know, help us understand, you know, how does the AI opportunity fit within Akamai and what do you need to do, you know, potentially from an investment standpoint to help, you know, I guess realize this opportunity and how do you monetize it?
Yeah. So there's a couple different ways. I won't, I won't get into the internal use of AI in terms of, you know, cost savings and scale and stuff like that. I'll, I'll focus more on what we're doing with customers. So we use AI today with our security products. It's, you know, something that we've been using for a while. Obviously determining, you know, how to make security decisions is, is critical with, with you leveraging AI. So that's something that we've been doing and we'll continue to do. We just announced a product, what we call AI security for basically large language models. And if you think about a customer that's using a large language model, there's a couple of potential problems. One is it's a significant cost to run one of those.
If you were ever hit with a denial of service attack, you could potentially have millions of requests and run up your bill, and that could be a problem. You also worry about potentially leaking sensitive information or personally identifiable information. Being able to understand the request that's coming in, that it's a legitimate question, it's a legitimate user, and then the answer that comes back out is something that you would think of as not leaking sensitive information, but also a lot of these models do what's called hallucinate, where they might give you a crazy answer and it could do some serious brand damage. Basically, we enable the customer to put in essentially a firewall effectively to make sure that the customer, the questions coming in are legitimate questions.
They're, you're not getting hit with, you know, denial of service attacks. Also, the answers going out are not leaking out sensitive information. That's just an example of innovation and creating a new product line for us. In terms of compute, if you think about one of the value propositions we have around low latency and being massively distributed, when you run some of these lighter weight inference models, being close to the user and being able to not have to proxy traffic back to a central server and introduce latency is critical. Things like voice to text or ad decisioning. You might have an ad decisioning model that's running very close to where the user is, and you can deliver that experience cost-effectively as well as with a higher performance.
We think that with AI, AI creates an enormous opportunity for us, from both product innovation, also with security. Security becomes a lot more challenging. The more sophisticated these AI models get with some of the deep fakes and things like that, just creates a much more challenging environment for our customers. It just gives us a lot of runway in terms of, you know, new security offerings and things like that.
That makes a ton of sense. And we're rapidly heading into a brave new world. I think there's a ton of opportunity for companies to help secure that. Maybe digging a little bit deeper into the security market, you know, micro-segmentation and API security have been two of the biggest growth drivers for your security business. Can you talk a little bit about these opportunities, why they've been growing, and you know, maybe the role you could play in a zero trust world?
Yeah. I'll start with segmentation. Traditionally, segmentation has been one of the more challenging things for people to do. Typically it was done either physically segmenting your environment or using machines and a lot of equipment. It was very challenging and very costly to do it. We have an agent approach where we basically run agents on your machines or your cloud instances, et cetera. It's very easy to deploy. Going back to your AI question, we actually developed a large language model to help our customers to do the deployment and do the investigation in terms of, you know, segmenting out their environment.
So it's, if you think about from a security perspective, it's probably one of the smartest things you can do as a CISO, because no matter how good your perimeter security is, something always gets in. Especially now, go back to the AI point, with AI, the fakes are getting much better. The likelihood that even though your employees are smart and this whole zero trust thing is don't trust any device, don't trust any person, is about somebody is inevitably gonna click on a bad link and stuff is going to get in. What segmentation does is it limits the damage. It can identify malware in the environment and you can shut it down before it becomes a headline-grabbing problem. A lot of times people just weren't doing it because it was very difficult to do.
We've made, we've taken the difficulty out of it. We've been able to scale. We've got massive scale with some of our customers. Today that's a couple, you know, bought the company about $30 million. We've grown just to well over $100 million. We expect that to be a very significant contributor to revenue over time.
Excellent. Maybe combining the last two topics together, this notion of AI and security together, you know, when I look at the Akamai platform, you have both the edge, so the ability to deliver performance, you have security, you have also now the ability to link back to the core compute as well. When we look at sort of the evolution of the space where we're gonna see agentic AI come out, MCP, we're going to see all these evolutions happening at a rapid pace. You know, how does Akamai maybe serve as a potential platform for that to take place?
Yeah. So, you know, one area that's started to get a lot of traction is, you know, use of GPU. So today the platform handles GPU. That's a potential area for us to expand and grow. The advantages there is being so massively distributed, we can run GPU pretty much anywhere. We also think you could use CPU too, in some cases with some of these inference engines that don't require as much memory and compute. It's not as compute intensive. That's another advantage to us. You know, there's enormous opportunity there. Also, just as inferencing becomes a bigger part of day-to-day life, those models are critical in terms of latency. You can't introduce latency into any experience. Otherwise users will get frustrated and not, you know, leverage the application.
I think that could be a potential game changer for us. Being as distributed as we are, I think we have a significant advantage there.
Got it. Got it. Maybe switching to the core cloud opportunity, you know, we do get a lot of questions in terms of, you know, what are the use cases that customers are using Akamai the most for? And how often do you sort of have the ability to win against hyperscalers?
Yeah. So it's, you know, first of all, we're more cost-effective. We do see a lot of customers come to us in sort of the early days where they have applications that are very chatty or, or they're accessed very frequently. You might store your app, you know, compute, use your compute at AWS and use AWS as your origin or any one of the hyperscalers. There are some hidden charges in there, including what's called egress fees. That means every time you access the data and you pull it out, you're paying a fee. We have customers in the CDN business that might use a hyperscaler as an origin, and we might get 98% offload, meaning 9.8 out of 10 requests can be served off the CDN, and we never have to go back to the origin. You think, wow, that's phenomenal.
In some cases, our customers will say that 2% miss is more than my CDN bill. It is really expensive when you put your data in to get it out. We have customers coming to us. We do not charge for egress fees because we have connected our few locations with our backbone. We serve hundreds of terabits per second a day. It does not cost us anything to serve the traffic. There is not a lot of traffic that typically goes with these applications, but it is very expensive. That is one area. Low latency is another one. We actually had a customer who was trying to comply with Apple Pay, which had a 60 millisecond latency requirement, could not get that done in a hyperscaler. They came to us and we were able to comply with that.
We are seeing a lot of, I'm impressed with what I'm seeing across the board. I had originally assumed we would be very strong in media, and we are doing pretty well there, but we've got customers in manufacturing and retail, banking across the board. We've done a good job penetrating a lot more verticals. We're cheaper than the hyperscalers. We do not compete with our customers, and we perform better. That is really the main value proposition.
Excellent. Excellent. And then when it comes to having both cloud and edge, can you help us understand, we're having both sides of the equation, you know, potentially, along with security and now these AI capabilities, you know, putting it all together, you know, what are the strengths here? And then, and then how does that become beneficial to that customer?
Yeah, I'd say performance and reliability are probably the two biggest strengths. You've got a lot of redundancy built into the, to the network and the platform. Performance is a big one, for us. Cost is another. We're, the leverage that we have across all of our different products, we do pass a lot of that savings onto our customers. So it is a very attractive financial proposition for our customers.
Got it. Got it. We can't unfortunately not talk about the CDN business. Just to maybe hit on, you know, some of the pricing pressure and slower growth that have impacted the business over the past few years. You've also reallocated resources away from the business to drive higher profitability. Is there light at the tunnel now? Is there, in terms of the business stabilizing, you know, particularly given the exit of some of the smaller players in the space?
Yeah. The CDN business, if I were to draw, you know, I've been at the company for 25 years, and if I were to draw two lines, one would be up into the right and the other one would be down to the right. Down to the right is unit pricing. Up into the right is traffic growth. Really the business is that simple. It's all about your unit pricing and your traffic growth. Historically the internet was growing 30% plus. For probably the first 15 years of our existence, our traffic was doubling every year. We got to the point after the pandemic where we saw massive growth during the pandemic, but traffic started to moderate quite a bit. We saw, you know, there wasn't a lot of new streaming services. Gaming was kind of weak. Traffic growth moderated more than we had expected.
Therefore we saw a decline in the business. Now pricing started to moderate some, but unit prices still keep coming down. We have seen a lot of our CDN competitors go out of business. They generally compete on price and there's been about four or five that have exited the market. We've been a benefactor of that by buying some of those contracts from these players that have exited the market. You do have a little bit less competition, but we are starting to see traffic pick up again, which is good. We far exceeded expectations last quarter on CDN, still declining, but we're getting down into the sort of mid to low single- digits in terms of declines. We think that that business ultimately is a kind of flattish to down a couple of points. It's really a strategic asset for us.
It produces a lot of cash. As Jonathan mentioned, we were able to move about 1,000 people out of the CDN business into compute. Some of the engineers and the folks that build the platform. We are able to get some, you know, better profitability out of that business. As I talked about, we have run multiple services on our machines. Having the economic advantage you get by carrying a lot of traffic, the data that you get powers the businesses for security. It is very strategic. Also, if you are gonna be in the compute business, all of the hyperscalers have some type of CDN, right? Think about it, you are now taking applications from on-premise to putting them in a cloud, you are introducing latency.
Therefore you need some way of trying to deliver a performant application to your end users. Having a CDN as part of a compute business is mission critical.
Yeah. Just on that last point around your CDN coverage, I mean, aren't many of the hyperscalers also Akamai customers?
They are. Yeah. As a matter of fact, we, the nice thing about our business is we do not have any 10% customers. We have, I think there are eight customers that are 1% of revenue or greater. Of those, the hyperscalers are in that bucket. Not all of them, but some of them are very big customers of ours. Yep.
Fantastic. Fantastic. Why don't we open it up to the audience for questions? We have a couple of minutes left. Anyone have any questions for Ed? Don't be shy. Okay. I won't keep everybody from lunch. Thank you so much.