All right, good morning. Thanks everybody for being here. My name's Frank Louthan, and I'm the Senior Analyst here covering Telecom, CDNs, Cable, a whole bunch of other stuff. I'm very glad to have Akamai back with us again. We've got Ed McGowan, CFO of Akamai, here for the presentation. So, Ed, why don't we start out just kind of level set a little bit. Tell us about Akamai, what you do, how you fit into the space and you know, in the business you're in. And kind of start there, and then we'll jump into some questions.
Sounds good. Can everybody hear me? Nice to see everyone. Thanks, Frank, for having me here. So yeah, Akamai's been around for 25 years. We pioneered the CDN business, back in 1999. We've changed a lot over time. We now have a $2 billion security business and about $500 million compute business. So we have three different businesses: CDN, which is our legacy business, compute and security. And you know, the company's very, very profitable. We're approaching $4 billion in sales, got about 30% operating margins, very, very attractive free cash flow. And that's pretty much, at a high level, who we are.
All right, great. So we'll talk about the delivery business, and we'll get into some of the others. But, you know, tell us a bit about your network that's kind of a base of a lot of what you do. Get a lot of questions from investors about, you know, newer networks built on SDN technology. What are some of the advantages to your network, and how long, and what would it cost to replicate that, if that is even possible?
Yeah, good question. It's funny. There's. I've been hearing this for years, actually. Even going back to when we first started, Akamai's a legacy network. And, you know, the term SDN, we were doing software-defined networking before it was even a term. So, you know, our. If you think about our platform, we do multiple things on our platform. So we have machines that are in, you know, 4,000 locations around the world, probably the most distributed platform, or one of the most distributed platforms in the world. And, those machines do multiple things. They deliver content for our customers. They serve video. They serve uncacheable shopping content or banking transactions. They do security processing. So, for example, I might be serving a web page to you, and at the same time, I'm applying web application firewall rules.
So I'm doing security processing at the same time, from the same machine, by the way. I think there's a misnomer that we have a whole sub-series of networks. Now, we will. We'll talk about our network in different subsets, but it's all the software can run all of our products. There's one exception, which is our Prolexic network, which was a denial of service at the IP layer, which essentially, think about that as scrubbing centers that are separate. And then, obviously, Linode is a separate company that we bought a couple of years ago. But I think it's people misunderstand and think that we have, you know, each one of our products runs on a separate network, separate servers in all of these locations. That's just not true.
Now, in terms of your question around how difficult is it to replicate, I think if you just look at even just CDN, for example, we've had a lot of competitors over the years, not make it. And I think the biggest challenge is economics. I'd say the economics is the hardest thing for someone to replicate. Today, we have 70%-80% of our traffic is free. And it's because of two things. One, we have relationships with a lot of these ISPs where, by having our content servers in their networks, we save them a tremendous amount of money by having the content stored locally. A lot of the traffic that we do serve is cacheable traffic. So not having to go back all that traffic or build peering links saves them a lot of money. So we get favorable economics in return.
And also, several years ago, we spent a lot of money to build out a backbone. So we've got a sunk cost in our backbone. So we don't have a lot of transit costs or anything like that. So I'd say economics is the hardest thing. And then just building those relationships, getting as distributed as we are, is very difficult, where a lot of folks that build distributed platforms today might be in a dozen locations in the really congested part of the internet, in the core data centers, which is okay for some amount of traffic, but it does at some point, it becomes challenging. It's more costly. And also, you have performance challenges, especially on days where you've got a lot of congestion on the internet.
All right, great. That's helpful. So let's talk the newest part of your business is compute platform. Let's talk about that a little bit. It's pretty well integrated in your business. How's that changed your go-to-market strategy and, and, you know, how you're approaching customers, and, and what products do you lead with, with compute?
Yeah, great question. So we're making several changes this year to go-to-market. If you think about last year, we spent a lot of time doing two things. One, we spent a lot of money building out another 13 core data centers. We have about 25 core data centers. This year, we'll be building out 100 distributed sites, which are much smaller, not as much CapEx. So there's a lot of build. And there was also a lot of innovation that was going on in compliance and getting the platform that we bought from Linode, which is really focused mostly on small, medium business, although it did have some enterprise customers, getting it ready to take on enterprise workloads. So, connecting it to our backbone, building up more capacity, adding more compliance functionality, and things like that to be compliant with things like PCI, etc.
So now that we've done that, we have from a go-to-market perspective, we now have changed our compensation plans that all of our reps have to sell compute. If they don't sell compute, they don't hit accelerators. They don't get to Titan's Club . That's our President's Club. And I can tell you from running the field I used to run sales before I came into CFO that will have a dramatic impact on the amount of deals that we see. Now, in terms of what products do we lead with, it really depends. I was on a sales call the other day. We were going through a renewal.
And I have this expression, "It's so easy, a CFO can sell it." So I said, "I want to go see if I can get a compute sales cycle here." So we're going through with a big customer, a very large customer, probably spends $4 million-$5 million a year with us. And they're going through cost cutting. And I said, "Okay, well, we can negotiate our deal. Maybe you save 10%-15%, whatever. It's not going to make a dent. How much do you spend on egress? Just I know you use your application is hosted at one of the major clouds." I said, "I bet you spend more on egress than you spend on your entire CDN business." "Yeah, that's a good point. Well, we can eliminate that cost for you.
By the way, we can also help you with performance. If you're in, say, the advertising space, you could run your advertising decisioning in 100 locations closer to where the user is. "It's more cost-effective, and it's a lot lower latency." Sure enough, we have a sales cycle now open with that customer. So it's very easy for us to take what we're doing for customers today and open up an opportunity. Think about media, for example. We do a lot of work with social media companies. When you're doing things like live transcoding or uplink and encoding, doing that close to where the user is and getting that content loaded cost-effectively and monetizable as fast as you can makes a lot of sense, all right?
That's something that we're already talking to them about performance of their application, scaling the delivery, going into how they actually prepare the content and put it up on the network, how they store it, very natural for our sales reps to sell. So easy a CFO could even sell it.
That's great. So one of the questions we get from investors is, you know, how do you compete with some of the larger compute platforms? You know, maybe walk us through what's a typical type of product and things like that, and how you compete with them and, you know, what your advantage is versus the larger ones.
Yeah, so I mean, if you think about the world is moving more towards a multi-cloud architecture. And there it's not like we it's a zero-sum binary game where we win and someone else loses. There's opportunity for, you know, a lot of cloud providers to win. I mean, just look at the market today. It's dominated by three players, but massive market growing very, very rapidly. And there is a we, we actually sort of got pushed into this market by our customers and by some really large customers that came to us and said, "We think that your platform you have a unique advantage over what the current hyperscalers provide today." There's some interesting use cases on our on our website. One in particular is with Apple, where they were doing a Private Relay.
If you think about that to work, you can't have everything hauled back to a dozen or two dozen locations. You have millions of users all over the world. You need to run that application out on a distributed platform. We have several other customers that were asking us very similar things for, say, video or advertising. And so we realized that there is a niche here for us to potentially create a pretty interesting business. Also, we found ourselves that the fastest growing line on our P&L was our cloud spend. You might say, "Well, wait a minute. Why would you have cloud spend?" Well, we're obviously being a security company. We see massive amounts of data. And when you have a lot of the data just think of almost like how you think about training AI versus inference engines.
You wouldn't have all that data out on the edge. What you want to do is take all that data into a central location, process that data, come up with rule sets, you know, make your, your software a lot smarter than when you're out on the edge, you're doing a lot lighter weight compute. So we were using hyperscalers for some of our data collection and processing, and it was costing us a fortune. So we decided to move all of that internal. And we became sort of customer one, our first $100 million customers ourselves. And what we realized is that we could do it a lot more cost-effectively. One of the big pain points we see with customers is we talked earlier about egress. And what egress is, if you store data in a central cloud and you access that data, the data moves around a lot.
It's accessed frequently. It can cost you a fortune, very cheap to store, very hot expensive to access. We serve hundreds of terabits per second of traffic across our network today. We've got one of the largest backbones in the world. So our cost to add that traffic is virtually nothing. So that's a competitive advantage for us to go in and take some of that business and say, "Look, you can put that on us. That being accessed frequently, it's not going to cost you anything." So there's a cost advantage. There's a, a scale advantage, a performance advantage. And, you know, being the most distributed cloud, we think, has some pretty interesting applications for us.
All right, great. So on the call, you talked about this new Gecko product that you have. You know, how are you building that out? And what sort of solutions does that offer or going to offer when it gets, you know, fully pushed out that you don't already have?
Yeah, so, Gecko or Generalized Edge Compute I think I said it right. But anyway, is going to be. Think of that as being in today 100 locations, probably in some of our larger CDN locations where we already have space, power, and, you know, favorable economics. We've got our CDN infrastructure already there. And there, you're building out several racks of equipment with the Linode technology so that if someone wanted to run full-stack applications, they can run it in over 100 locations today, leveraging us. And that could be for things like advertising, for encoding, transcoding, for banking transactions that have low latency requirements. It could be just you want to have something running in a location where one of the hyperscalers don't offer that location. It might be in a small country somewhere. And we happen to have a deployment there.
And you'd like to have it localized. Maybe there's a data localization issue where you need to keep that data local and we're the only option in that particular country. AI inferencing could be run in a situation like that where you use your training models in the hyperscaler, but you're running a much lighter weight AI decisioning engine out where things are being processed, maybe where a video's being served or an ad's being served, etc. So a lot of different use cases for that. And actually, that was some of the early push from our customers. We sort of did this on a custom basis where we built out and ran custom code in hundreds of locations across our network. So we've seen demand for this, early, early on. And I think there's a lot more interesting applications in the future.
All right, so now that we're not going to disappoint anyone searching on the term AI in the transcript, let's maybe talk a little bit about, you know, the breadth of your network in an AI world and how you can, you know, benefit from that and your customers can.
Sure. Yeah, so we think of AI in three ways. One is obviously the, you know, for as a company, the productivity gains you can get from it. Today, we're using AI in our security products. So we've been doing that for years. In terms of the use cases as a customer, so could customers use us? We do support GPU today. I don't envision us being a source where people will be doing large training models. It just doesn't make sense. We're not going to go make that investment to go build out for, you know, massive training models. But for inference engines, you can do that. You can actually do that. And we're doing it today using CPU and really sophisticated algorithms. You could run GPU out there. And we can certainly do that. Our platform supports it.
If a customer wants us to do that, we could do it. But, I'd say from a customer perspective, that's probably the most interesting area for us is on the inference side.
Okay, great. So put this into some numbers. So what, you know, walk us through kind of what it takes to achieve the high end of your 2024 guidance and, and what might cause you to be at the lower end of that range.
Yeah, so there's a lot of things going on. I'll start with the delivery business. So we had probably, I think people would say, somewhat disappointing guidance around delivery. We saw traffic growing fairly nicely going into Q4. Q4 is a little bit weaker than we normally see in a Q4. So I started to see internet traffic slow a little bit. We have seven of our largest customers renewing. And in the delivery business, you know, we talk about our largest customers being, you know, greater than 1%. Our delivery business is less than half. So you can imagine those customers are fairly large delivery customers. So they have an outsized impact. Those are all renewing in the first half of the year. So the outcome of those renewals, if they're better than expected, that can help us get towards the top end.
If we see better traffic growth than what we're seeing now, that obviously, you know, has a big impact on the top line, but also on the bottom line, probably more so on the bottom line. We're very heavy, you know, very levered towards quick growth and traffic usually yields great benefits on the bottom line. Security, security, we talked about 14%-16% for this year is our expectation. We don't plan on any headline-grabbing events to drive upside in revenue. We've seen it over the last few years. We've seen even things like our legacy Prolexic business will get a boost sometimes from things like the Killnet attacks that hit the healthcare industry a couple of years ago. So there's always an opportunity for that to happen. Compute, you know, we it's a newer business for us. We've got obviously a big pipeline.
We've got some pretty big, exciting opportunities in that pipeline. The timing of those, the size of those obviously can help drive upside as well. You know, obviously, when you put guidance together, you have to be very balanced in your approach and looking at all the different things. But those would be the main things that would drive us to the upside.
All right, great. So on the distribution side, we always get the question about self-provisioning. And some of your larger customers kind of do it themselves. You know, how do you remain relevant in a media and cloud world where you've got a lot of those customers that are building some of that themselves? And how should we think about your, you know, how you can continue to be relevant there?
Yeah, so if you go back to, like, the 2014, 2015 time frame, the DIY phase, if you will, or craze, was something that sort of took the whole industry by surprise. So that slowed growth pretty significantly in the delivery business back then. We've seen that cool off quite a bit. And as a matter of fact, there's some large customers who were heading down.
Found it hard to do.
Well, it's hard to do. It's, it's expensive. That's the other thing. We've seen some large customers head down that path. Obviously, the ones that have their big platforms, the Facebooks of the world, they're not going to tear them out. But there's some others that we're talking about going in this direction of now reverse course. It's, it is expensive. Labor costs never go down. They only go up. You can drive bandwidth costs down to a certain point. And I think what if you see if you look at high-volume traffic today in the industry, it's a lot cheaper to go with a third-party CDN. And there's not as much of a strategic value, I would say, at this point. You know, maybe early on, when telcos had a little bit more of a pricing power in the market, that sort of has gone away.
So I don't really see that being a threat. You know, obviously, it's with some of the big larger, you know, larger, largest platforms in the world, they can take on more themselves. But we find actually them giving us more because we're more cost-effective.
Yeah, so there's been an expectation maybe that delivery business would flatten out. Or what? What's sort of your expectation on the trajectory of that business?
Good question. I, you know, we've sort of talked years ago, maybe two years ago now, three years ago, whatever it was, that we thought this business would be flattened down 4%. We've been surprised a little bit to the downside where we didn't get the traffic growth that we expected. You know, traffic on the internet has usually grown about 30% year-over-year. We aren't seeing that anymore. During the pandemic, you saw it double. The year after the pandemic, I guess, say, 2020 to 2021, very strong traffic growth. Then all of a sudden, it really moderated and was, is much below trendline. So that's something that, you know, if I were to look back at my models, I would have expected us to have higher traffic. So that was probably the biggest thing.
Then I'd say just the big renewals, if you say, well, how would you get to -4? It's when you have large customers renewing. And you assume that every couple of years, you're going to have that happen. So those are the sort of two factors. So it's really the lack of traffic growth is what's really causing us to have lower than, you know, sort of expected downside in the delivery business. So as soon as we see traffic increase, you've got like I said, you're very levered to having a good outcome. We have the capacity. And we have the infrastructure that a lot of that will just drop to the bottom line. But I don't expect that business to be, you know, flat or growing, certainly this year and probably not next year.
It's, you know, it's hard to tell at this point. I'm not going to go put a projection out there of what I think's going to happen next year. But, you know, we'd have to see a pretty big change in the just internet traffic in general for us to get on a path where it's, sort of flat to growing again.
All right, and so and how do you see that in overseas markets versus the U.S.? You know, what kind of drives that non-U.S. business as far as applications or security? What is it? And how, how should we think about those two parts of your business?
Yeah, it's interesting. You know, obviously, from the delivery business, the most of the content is still U.S. content. The biggest gaming companies, the biggest, you know, production of content is still U.S. So I, I, you know, in terms of the international business, if you think about it from a market share perspective, we tend to get more market share from those customers delivering outside the U.S. In terms of security, I'd say it's pretty much an even playing field. The attackers don't pick one country or another. They're attacking everywhere. So we're seeing good growth pretty much across the globe. It's interesting. We are seeing pretty good pickup of business in Asia with our compute business, so early adopters in Asia, which is pretty interesting. I wouldn't have expected that.
I'd say we've got sort of an advantage over our traditional customers in that we invested in that infrastructure in terms of our people and the infrastructure you need to run your business outside the U.S. well ahead of folks. It's difficult. It's expensive. It's hard to do. We've been able to do it. It's been the fastest growing part of our business for many years. You know, I think there's still a lot of opportunity there.
All right, on the security side, so how when you win with security, who are you competing with? And what are customers giving? You know, why are you winning over the competition, in that business?
Yeah, so I think about our security in sort of two different businesses. There's the web security, so protecting websites, web applications. And there, we're, you know, the market leader when it comes to things like web application firewall . And there, you're really competing against sort of the traditional way of doing things, you know, doing something with a box and having a box in a data center that's providing your firewall. That business, we do see some competition. They generally try to compete on price. In the security market, competing on price isn't usually a successful endeavor. It's really about capabilities, expertise. And switching actually brings some risk to the people that make a decision to switch.
If you get attacked after you've made a switch and somebody exploits a vulnerability, say, from a competitor that doesn't have the type of protections, you can open yourself up to potentially some very unfavorable outcomes, you know, either you lose your job or you get sued or whatever. So we don't see a lot of that on our end, within the web security business, web app firewall, which is the biggest. There's bot managemen t, which is sort of a new space that we sort of pioneered, which is understanding that a lot of the traffic on the internet today isn't human. It's machine. So understanding what machine is interacting with your website, what do you do with it? How do you should you block it? Should you, you know, give it wrong information if it's a price-scraping bot, that sort of stuff?
There's smaller competitors out there that have tried to catch up to us. But we're sort of a market leader in that space. API security is a newer security area. There's a couple of startups that are in that space. That's a huge opportunity where if you think about modern applications today, everything uses APIs. There's a tremendous amount of attacks that are going on right now with OpenAPI gateways. And there's not a lot of good solutions. So I think we could be a market leader there. On the enterprise security side, where our strength is, microsegmentation and access. So think of replacing a VPN with our remote access product. And with segmentation, there's one startup that we've sort of leapfrogged since we bought Guardicore.
But there, you're sort of – what's interesting is with segmentation, people generally think about segmentation as something that's very difficult to do. We've actually been able to do it pretty easily with our software agents. And think about that as having a firewall basically on every application and being able to segment your network and be able to identify threats as they get in. And there's an interesting story about one of our competitors that talked about vendor fatigue and vendors getting upset about spending all this money, but stuff still gets in. Well, that's a perfect opportunity to use segmentation because what segmentation will do is identify something that gets in and block it and stop it from spreading. So no matter how good your endpoint security is, your filtering, whatever, stuff still gets in.
And this is a great way for you to be able to identify things that are in, block it, stop the spread, and limit the damage.
All right, so what have we seen with pricing in security? I think a major pure-play security company just lowered some pricing. How do you see that holding up in your business and the outlook for the year for security pricing?
Yeah, we haven't seen pricing pressure in security. We've seen some competitors try to compete on price. You know, sometimes you've got good relationships with your customers. They'll show you some desperate attempts of somebody saying, "Hey, I'll do it for, you know, half the price or a third of the price." And generally, that's a warning sign. If you're in security, you generally don't want to answer that call. But we haven't seen the pricing pressure there. Is it possible over time? Sure. It's technology. So over time, you know, it could creep into the business. But we have not seen it.
Okay, great. Why don't we see if we got any questions from the audience? And I might have a couple morning. All right, great. So all right, Ed, so why don't we talk about, you know, how should we think about uses of cash here? And, and how do we think about, you know, buybacks, debt repayment, that sort of thing?
Yeah, so, we've been producing a tremendous amount of cash. Our free cash flow is going to be up pretty significantly this year as our CapEx is down. We did a major build for compute last year. We've got our policy is to, from a buyback perspective, offset dilution from our equity plans. But if you look over the last five to 10 years, we've reduced our share count by about 1%, a little over 1% a year. You know, that's not our stated objective. But we are opportunistic with our buyback program. And we, you know, traditionally have been, you know, retiring about 1% or so of our outstanding shares. You know, the best use of cash for us is through M&A. You know, that's the preferred use of cash if we can do it. Obviously, we're reinvesting back in the business.
But we're free cash flow positive. So it's really a combination of M&A and buybacks in terms of, you know, use of capital outside of, you know, investing in the business.
On M&A, what have you seen in the M&A market? It's been an interesting observation here at this conference. We have the most corporate attendees we've ever had. And as an analyst that's done this more years, and I won't admit, you know, this I think it's M&A. There's been just a lack of it, a lot less of it. And a lot of talk to some of my peers. We're not getting the calls about a scheduling conflict a couple weeks out and that sort of thing. So what do you see in the M&A market? Have interest rates helped or hurt that? Do you see opportunities in M&A more or less? How should we characterize it?
Yeah, well, I mean, obviously, from a big M&A perspective, we've got an administration that's anti-M&A. So we're not seeing, you know, the big deals anymore. But it's interesting. You go back a few years ago with interest rates near zero, every company that was private was thought they were worth a fortune, right? You know, everything's in the terminal year. And you get crazy valuations. You would think that those would have corrected. And you've got a lot of disciplined buyers like ourselves. We can't go out and afford to pay, you know, 100 times revenue for a company or anything like that. You would think that that would start to correct. What we're seeing is a lot of companies back in 2001 raised a lot of capital. I'm sorry, 2021 raised a lot of capital. That's starting to dry up.
We're hearing from a lot of companies, "Hey, we're either looking at a down round or potentially exploring M&A." It's starting to see a little bit more activity. You still have stubbornly high valuations, especially in security. I think you're going to see folks like us and other companies in the space that have cash that are going to be a lot more disciplined and patient.
Okay, great. And any other questions from the audience? No? All right. Sorry. So you know I'm going to ask a dividend question. You know, where are we on that? You know, what are the right parameters for you to consider a dividend? And you know, how far away are we from maybe making that decision? You've obviously got a lot of cash. You used it well. You at least hey, you're doing the buybacks. But you know, what's that sort of the next step? How should we think about that?
Yeah, I mean, I think there's, obviously, the markets where in security and compute in particular, just massive opportunities. And, I, I think we see a lot of growth opportunities ahead of us. And the best use of cash being M&A to accelerate our opportunities in those spaces. It's something that we talk about with the board. And we explore. I think there's always been a stigma around dividends that in tech, if you're paying a dividend, you'd you have nothing better to do. It's starting to see that change a little bit. You see some big tech companies starting to announce dividends. So it's something we'll think about. Obviously, we do have the profile to be able to pay it. We certainly produce a lot of free cash flow. And we could certainly do something like that. But it's, it's once you initiate it, you can't take it away.
People want you to grow it. So it is something you have to to really think about and think about your shareholders and what, what do they want? They want a lot of our shareholders want to see us grow. And you don't want to paint yourself in a corner and not have the capital to be able to grow. So it's, again, something we think about. We certainly have the profile to do it and maybe someday. But right now, we're not. It's not in our plans.
Okay, and on the preferences on M&A, is it more security? Is it more, doing the compute? You did take in a couple of orphaned delivery guys last year. But maybe there's more opportunities there. But what are your thoughts on the longer term, you know, growth of the business?
Yeah, I'd say it's kind of more the same, what we've done. You know, opportunistically, if something presents itself and it makes sense from a scale perspective, we always look at those. Those you can't count on those. If they come up, they come up. You evaluate them. You do it if it makes sense. On the tech tuck-in side, you know, we'll, you'll probably see us do a bunch of those. We've done those over the years. That's acquiring talent, sort of accelerating your roadmap, maybe getting some technology going to bolt into something. And we've had a pretty good track record with that. Adjacencies and security is an area that we've done. And we've been pretty successful, especially with Guardicore. So we'll look to do, you know, stuff like that. You know, I don't see anything transformative.
You know, I don't think we are in a position where we need to do something transformative. So I think it would be kind of more of the same of what you've seen over the years of what we would be doing.
Okay, great. All right, Ed, thank you very much. We got a breakout session after this if anyone wants to continue the conversation.
Thank you very much, Frank.
Thanks. I appreciate it.