Akamai Technologies, Inc. (AKAM) Earnings Call Transcript & Summary

And I'm really excited to be hosting Akamai Technologies, CEO and Founder, Tom Leighton, thank you so much for being here.

Thank you.

Well, I think we have a lot of ground to cover, so I'm going to jump right into it. But I think a good place to start would be just to kind of take stock of the year in terms of Akamai's business performance, specifically around you all taking some very strategic and necessary steps to revitalize certain growth areas and refocus the business to higher growth areas. So just from that perspective, if you can talk to us about key achievements over the course of the year and more specifically around some of the financial and business targets and goals that you also shared about 6 months ago.

Yes, we've had a good year, a good year for execution, getting ahead of plan and where we started the year, raised full year guidance on top and bottom line on the last call. Really excited about our cloud infrastructure services. That ended last year with an ARR of about $0.25 billion, and we're on target to deliver 40% to 45% growth there this year. And that has just a really bright future. It's a huge market. Akamai is differentiated in our distributed approach to get the business logic closer to end users, which gives better performance and also a lower cost given the extensive nature of our platform. So very excited to see that. Security performing well, especially with our API security and Guardicore segmentation solutions. We ended last year combined, they were about $0.25 billion, and we're on track this year in ARR to grow that 30% to 35%, which is really exciting to see. We just released a new product, AI for -- firewall for AI, very early days, but a lot of customer interest there. Everybody is deploying all sorts of agents and AI apps, and they need specialized security. AI is a great tool, but it's a huge vulnerability. It's something that has access to all of your data, is speaking to the public. It's not deterministic like a normal application. You never really know what it's going to do and already been some pretty bad headlines for companies that got burned there. So very excited about that. And delivery is stabilizing, which I think is a relief to a lot of folks to see that. Last year was not a good year, but doing much better at this point. And I think the prospects are good going forward there. So overall, we've been very pleased with the developments this year and the execution.

Because you brought up delivery, I think I'd love to peel back the onion on that a little bit. As you said, the last couple of years, there's been quite a high degree of variability in that franchise for you. And of late, there has been a much greater degree of stability. And so the recovery we've seen in that business, how sustainable is that recovery? And just to kind of bring it down to more tangible drivers of what's going to ensure that the business continues to stay more stable rather than have maybe more of the erratic growth patterns that we saw more recently?

Yes. I think the delivery marketplace was impacted by a bunch of companies that were selling below cost. And we have the best cost basis in the business. We're the biggest and best by a good margin, but it's not helpful when there's competitors out there that are doing anything to get some revenue. And...

On $1 for $0.90.

Exactly. And doing it at scale. Now 4 of them went broke for the obvious reasons. There's still a couple of players doing that, but a lot less than there were before. So the market is still very competitive, but not nearly like it was before. And traffic levels are picking back up as well. So that's a good combination. And I'm optimistic that going forward, we're -- it'd be a much better situation.

I know Ed and I have talked about this offline rather. But at the risk of being oversimplistic, the delivery business has just been a product of pricing and volume, right? So you sort of alluded to pricing being far less irrational. I'm not going to say pricing is rational. It's far less irrational. So I think that is easy to appreciate. On the volume side, you did speak to the traffic volumes picking up a little bit. So what would you attribute that recovery in traffic, especially in the absence of new content releases and OTT launches? I know there's a hit-driven component to some of the gaming side. But would love to get a little bit more granular on the V side of the P times V equation for delivery.

Yes, great point. And traffic growth is not like it was back in a time where there were a lot of new OTT providers or early days of COVID where everybody is locked up inside. So it's not at that level and doesn't need to be for us to have a good business there. The improvements in traffic we're seeing is really pretty much across the board, no one sector you'd point to, which is also good to see. So I think we're at probably more of the new normal at this point. And there are potential drivers that could take it up further. If we start doing a lot of video associated with Gen AI, how the web changes, how we interact with it, text doesn't drive a lot of traffic. But anything with video, if we start wearing headsets a lot of the time, that drives a lot of traffic. So there are potentials for more. And of course, we always like to have better quality on our videos and better quality means a lot more traffic. And so that's helpful.

You brought up AI, and it begs the question, could that potentially be a very strong force in creating an inflection point in growth for Internet traffic, especially for consumer applications. I think there's somewhere -- 80% to 85% of generative AI traffic is geared towards consumer applications, which kind of interestingly dovetails with the delivery business, and then we can naturally talk about the compute business. But just curious what you think the further democratization and mainstreaming of AI and consumer-facing generative AI applications is going to do to traffic volumes over time.

I think it helps. And the question is how much. If we move into a new world of the web where it becomes more voice than click, that's good because voice drives a lot more traffic than reading the text. If it gets to a world where there's video interaction that what you're seeing back is a video presentation of some kind, that's huge. That's a really wonderful world to be in for content delivery. We do the delivery for a lot of the entities that would be involved in that, including one of the largest Gen AI search engines. So I think there's good potential there. It's early days. Still today, it's basically text.

Just to kind of put the delivery conversation to bed, the business is declining less. What is it going to take for the business to actually get to a path where it's not in an x growth mode?

It's a little bit better on the pricing stabilization and decent traffic levels. And if something does turn and generates a lot more traffic, that's a huge win.

And large customers and how they've been behaving vis-a-vis in-sourcing some of the delivery capabilities, how far down the path are we that, hey, the largest customers for whom it makes ROI sense to do delivery on their own, have we run the course on that? What are the risks of more customers incrementally shifting more volumes to a DIY approach and more of a multi-CDN vendor management approach?

Yes, pretty much -- I think all the big guys that would do it have done it. I think all the big guys do have some kind of multi-vendor. In fact, some of them may be decreasing the number of vendors, especially with 4 of the providers gone now. So I think -- yes, that's a factor, but I don't see it really changing at this point. Now you ask, okay, who does it make sense for? I don't really think it makes sense for any of them. I think it's better. We provide a better capability at a lower price point, but some of them just think it's important for them to have that capability.

Tom, you and the management team have worked really hard to take Akamai from a 60% delivery business to a 60% compute and security business, which has a higher horsepower and torque in terms of growth. So with that in mind, let's talk a little bit about security. The broader strategy there and the insights you can offer us at a portfolio and SKU level because underneath the hood, you've got multiple enterprise-grade security capabilities, but they have wildly different growth profiles. And so just from a broader portfolio strategy perspective, how are you thinking about security and how we should internalize some of the product level growth?

Yes. We've made a lot of progress in security. It's now the majority, over half of our revenue is just security. We're one of the largest security vendors in the world. Not many security companies are over $2 billion in revenue like Akamai, getting good growth. And you're right, there's a dynamic that some of our products, which are market-leading, have a lot of penetration now. And so they're a little bit more mature and they're going to grow a little slower going forward. And some of our products are newer, very big markets growing really rapidly. API security, a lot of potential for continued growth, segmentation, leading products to stop ransomware. So a lot of growth there. And I think you'll see us over time continue to develop new capabilities, think about firewall for AI that should be very important over time. And there'll maybe other acquisitions that we do. We've been very pleased with the acquisitions of Noname and Guardicore. And if we can find the right next company like that, that's reasonably priced, which is hard in this market, yes, we would do that and increase the amount that we can provide our customers and strengthen the security platform that we sell and the growth.

Is there a white space in the security portfolio that would behoove you to be more assertive and aggressive on M&A? I appreciate that you are going to be very particular about the valuation context and all those things. But any specific areas of interest or opportunity, again, within the security portfolio that would be more accretive to the overall platform?

Generally, our strategy is to look for capabilities that are close enough to what we do that it makes sense to package as a platform that we know the buyer, that we can be successful in selling it and really growing it. So durable long-term growth. And so those are the areas that we look at for security.

How are you positioning yourself both from a brand recognition perspective, but also from a go-to-market perspective within the cybersecurity proper arena? There are [ mass ] forces in that space in and of its own right. So how are you ensuring you have the brand recognition? And then maybe the companion question to that is how much of your penetration of security capabilities within your customer base is predominantly an add-on for your very large delivery customers that essentially are glomming on to security as an adjacent functionality on top of their delivery?

Yes. For our major verticals in security, you say, led by financial vertical, they all know who we are. You go to any bank, financial institution, CISO, we're one of their largest vendors, 2 or 3 sometimes largest for security. And so we have pretty good brand recognition there. Now as we grow the portfolio, we would increase marketing for certain capabilities, but the buyers know us today in security.

And from a KPI perspective, what are you tracking vis-a-vis demand generation vis-a-vis there are certain episodic dynamics around, hey, ransomware attacks are through the roof or zero-day attack volumes are through the roof. And I guess this particularly pertains or would pertain to the DDoS business, right? Are there any particular KPIs or macro drivers that you feel the security business is maybe more sensitive to both on the up and down? And how do you mitigate some of that in the context of having a more durable and higher growth in the security business?

Yes, showing the value is really important in security. Now our most motivated buyer is somebody who just got hacked or even better if it's the nextdoor neighbor from somebody who just got hacked. And so you do see episodic buying patterns sometimes. What was it 6 to 9 months ago with KillNet, they went after hospitals and organizations that just had never really thought about it. And then, well, that, in a sense, drove a lot of business our way because we have the leading solutions to stop the denial service attacks, hadn't really done business with those companies before because they didn't think they needed a solution. Now they know they do. And that can drive an uptick in the business. Ransomware, yes, that's why you're seeing such spectacular growth with our segmentation because a lot more industries realize, oh my goodness, they need a solution, and we have the best one. So in peace time, if you will, in a sector with a company, it's important to keep showing the value that what we're providing to them and making sure they understand what's going to happen to them and the cost if they don't have defenses.

I want to shift gears to the compute business. This entire franchise got a nice turbo boost and shot in the arm with the acquisition of Linode, and you've really built organically upon that acquisition with more CapEx and infrastructure-oriented investments or really building that out. But there have been very, very big changes in this segment and in this offering, both from a product and capability standpoint, but also from a messaging standpoint and then naturally also from a go-to-market perspective. So can we -- we've come a long way, but I think a recap would be very helpful, especially in the context of you shedding and rationalizing some low ROI areas within that portfolio. So I would love to have you kind of spend a little bit of time talking about the evolution and -- the ongoing evolution of the compute franchise.

Yes. It's -- that's a really important question. The Linode acquisition was transformational for us. And we put a huge amount of investment into it, not just on the capital increase in size or increase in footprint. We're now in 36 cities, a whole new architecture, much more scalable, much more reliable. We can get various accreditations for it, more functionality. We're building out an ecosystem of partners. We have pretty much full stack media workflow to compete with the hyperscalers there. So a lot of investment and growth. And that's captured in our cloud infrastructure services, which we call revenue, which is stuff that really comes from Linode and our edge computing, our JavaScript Function as a Service. And we're just now going live with our managed container service. We can take customer containers and support it in any of our edge regions. And we got the first few customers now that we're running containers in 150 cities, and we can scale that to 750. In fact, one of the first big users is a hyperscaler because we can get their containers in more cities than they can. And so that is our future in compute. We had legacy solutions that we developed for customers that did compute in our platform, but not in the way you'd think of as a hyperscaler doing it. And some of those, yes, we're phasing some of those out because they don't make sense for us to invest in going forward. And probably next year, maybe when we talk about compute for the year, we're really just focused on cloud infrastructure services. It's where all our investment is going and where all the growth is. And I think a very exciting trajectory.

You talked about a 40% to 45% ARR growth aspiration for CIS in particular, this year. So what does that actually assume? And what gives you the confidence that this is the right ZIP code of output that you can achieve? And then what are you seeing in your customer base from a behavioral workload migration to Akamai's footprint standpoint that's giving you this confidence that it's going to grow at this hyper growth rate, which is orders of magnitude above company level?

Yes. So well, for this year, it's contracts because at this point, pretty much the revenue we're going to be getting in Q4, we have a very good solid understanding of. Maybe something moves a month or 2 crossing into next year, but basically, that is pretty well understood. Longer term, as we think about it, we think we have a compelling value proposition in an enormous market. We can provide better performance and a bunch of the apps need that. A bunch of the apps out there that customers want their compute logic close to users. So the user gets a better experience. AI is a tailwind there for apps where you want to talk to the thing, chatbots, inference engines, buying decisions get made based on that latency is really important in that context. And we can do it at a lower price point because we have the world's most distributed and I think, efficient platform. And so particularly for applications where there's a lot of data moving around, very expensive to use the hyperscalers, and we can provide a lower price point. And in today's world, if you can cut your cloud bill in half, that's a big deal.

Hard dollar savings. So bridging us back to your 20% CAGR aspirations over the next 5 years for the compute franchise, you're going to exit at ideally 45% or higher this year. Decelerating to 20%, what does that imply?

Yes. So that's the thing. The compute number when we talked about it, there's 2 pieces. Cloud infrastructure services, which is doing fabulous and our legacy compute applications, some of which we're deprecating or handing over to partners who then migrate to our cloud platform. And so that legacy part, think of it as flattish, you got the part that's really booming. The booming part is going to cross over and be the majority early next year. And so that's how you get this is 40% and then the whole thing being, say, 20% over the longer term. Probably we're going to change -- maybe it will change how we report next year that says, look, let's just focus on the part where we're investing in and that's growing. And that's the bigger number.

One of the other elements that I wanted to discuss with you is the interplay or the necessary interplay between CIS and delivery in that, by and large, a lot of the visibility that you have in CIS momentum and workload migration to your edge is coming from your delivery customers, right? So to ask the question more bluntly, are you seeing a little bit of a shifting of the deck chairs in terms of what would have been delivery revenue as being maybe captured in CIS revenue? How should we disabuse investors of the notion that this would have been maybe categorized as delivery revenue and maybe delivery is ending up being a loss leading mechanism to drive more CIS workload growth. How should we disabuse us of that notion that, that is not the correct interpretation?

Well, that's not what's happening. We would do that. For example, some of the hyperscalers sometimes will say, "Hey, buy our compute, delivery is free." Sometimes they'll do that. And that's because compute is worth 10x as much. Now we would do the same thing. There's been a couple of times we've offered to do that. It hasn't happened because the compute is 10x the delivery, that's a slam dunk and probably higher margin. So that's not a factor today in our compute revenue. If we get a new customer in compute, we will pick up delivery, and that will be allocated to delivery. It turns out that the sweet spot, center of mass in our new compute revenue is big media, and they are big Akamai customers generally for delivery, but they still have the delivery portion separate from the compute portion.

Okay. In terms of driving greater leverage and, again, a return on investment and return on invested capital on your infrastructure footprint, all of the efforts around scaling the compute business would downstream have very favorable leverage impacts, right, because the value of that workload is significantly greater than all things being equal, a delivery workload being run at the edge, right? And I think you've been very candid about that. So just from a compute-related ROI perspective, what does that algorithm look like versus if you were only just back in the day just running delivery, right? What does that look like?

Yes. So very, very roughly $1 of compute CapEx when we are selling our Infrastructure as a Service or a partner service is $1 a year of revenue. And we're amortizing now over a 6-year period. So very attractive economics. Now some of the investment initially we made was for our internal uses. And there, we don't bill ourselves, but it does greatly reduce what we were paying the hyperscaler.

And this was the program that you undertook where it was roughly $100 million of hyperscaler spend that you effectively in-source to drink your own champagne, as they say.

Yes. And in fact, today, if we hadn't done that, it would be a lot more than $100 million for saving.

So the 1:1 ratio is the right way to think about it.

Yes, approximately. Maybe storage, a little bit less, but ballpark, $1 of CapEx is $1 a year of revenue, very, very roughly.

And I know a lot of investors who have been familiar with you have realized that you do have a very capital-intensive business, especially when traffic volumes for delivery purposes just going through the roof, supporting new launches and things like that. As you pivot to more of these higher-margin workload and use cases, right, how does that change the structural capital intensity of the business going forward, i.e., do you go from becoming a high teens, 20% plus CapEx player to low teens? And how long does that take to transpire?

Yes, great question. So first, with CapEx, most but not all is the deployed network. We have software capitalization, other kinds of things that -- infrastructure that would go into capitalization. Delivery used to be pretty intensive, but we've cut it in about half. The delivery business used to be about 9% of revenue, and that's down now 4%. And of course, we've changed our strategy there. There's a little bit less traffic growth, which helps. We're always working to make our platform more efficient in terms of its delivery capabilities, but we're not going after the huge spiky crowd kinds of things where you got to do a lot of build-out. We're not taking the super low-price business, which is a lot of build-out, but not enough revenue. And so the capital intensity is cut in half. Security, not so capitally intensive, so not an issue. Compute is capitally intensive as we're growing. And the faster we grow, the more it shows up in CapEx, certainly as a percentage of revenue. And that's -- it's a good thing. We want to grow fast. But if you're growing really fast, you're going to add a lot of revenue next year, you got to build out the CapEx to support that revenue. Now over the long haul, that's a very attractive proposition on the P&L. So today, yes, you're right, we're in the high teens when you put all those things together. It depends on the mix going forward and how fast compute grows. I want to have a problem where I come back and say, wow, compute is doing even better. And so yes, we're going to put more CapEx in to fuel that. I think investors will be very happy.

What would be the catalyst for that? Is it you moving into more on the inferencing side at the edge? Is that something that you're kind of seeing green shoots in this compute or CIS business in particular today? And would that be the telltale sign of, okay, you're seeing -- you're going to invest ahead of that sort of cresting?

Yes, inferencing AI in general is a strong tailwind. Yes, so that -- it remains to be seen how large, but looks very promising for us. And as we look to the future, more desire to have the workflows be closer to users for performance reasons. And we're in a great position to do that. We're at the best at being able to do that.

I want to shift gears to kind of operations and then more specifically go-to-market. So we talked about the strategy shift at the top end or from a revenue perspective, but you're making mirroring adjustments on the go-to-market side and no pain, no gain, right? So there has been change in the way the sales organization has been structured and incentivized. So I think it would be worthwhile to have you shed light on what has changed? How far down the path are we? What things have you seen that give you encouragement? And all of us want results yesterday, right? But what's that time frame where you feel that the changes are going to be seasoned and you can gain your stride again on the go-to-market side after these changes?

Yes. So there's a lot more focus on hunting because our new products that are growing fast appeal to a broader set of verticals than we had. So a lot of opportunity in hunting. There's increased investment in specialists to support the sales of new products. So today, that's segmentation, it's API security, a few folks around AI security and compute. Now in another year or so, we may not need as many of the specialists on API security. I probably fold into the regular field. Maybe there's new products in security that those folks would be taking over. So -- and overall, increased investment in go-to-market because we have the products and the greenfield to go after. We've made improvements in rep efficiency. Contract lengths are going up in a material way by design. So we're making good progress now. We're starting to see some of the benefit. And this will be work that we continue certainly through the end of this year and probably into next year as well. And then we start, I think, seeing even more benefits in next year.

And historically, I think some of the changes you've made has been more reconstitution or realignment of existing resources, but I want to be unambiguous and clear, you are actively increasing the level of investment in terms of headcount added to the sales organization, more quota-carrying reps, et cetera. Is that the right interpretation?

Yes.

Tom, I'm going to take a step back and ask you a very big picture question. You've had a seat in terms of watching pretty massive technology and economic cycles, just having founded the company and being at the bleeding edge of the explosive Web 2.0 era and where we're at. I would love for you to give your perspective on how much of what we're seeing with AI and generative AI today rings alarm bells, but also harkens back to some of the things you saw when the Internet was on the precipice of just absolute explosion.

Yes. It is possible that we're going to see a major transformation in the web and how we use it. That will be a pretty fundamental transition that instead of the model of clicking on something or interacting with an app, and we'll be doing it in language, even better for us if it's video. There's already new protocols for the models to talk to each other and share data with Agentic AI that you would ask your device to set up something for you. It would go out and deal with all the entities out there to do it. It would be smart enough to know how to do that, what you want, and it would produce the result. So that's really a different kind of world than today when you want to do something or buy something or reserve something, you're doing a lot of that yourself and clicking or tapping on your device. That world may fundamentally change, which is a pretty cool concept. On the -- and that will create a huge demand for the compute around all that because it is compute-intensive. It does create massive new security issues, which happened with the original web. Everybody raced to adopt it, got great benefits from it and just massive security loopholes and challenges that we're still suffering from today. The security challenges are going to get bigger when we do that. The AI is fundamentally different. With the big enterprise codes and app today, they got good developers, it gets coded. You know what it's going to do. AI does not work that way at all. It's much more efficient to create the application, but you watch what it does, and it seems like it's working, but you have no promise.

False positive score.

Yes. You have no idea really that hallucinates. On top of that, it has access to all your sensitive data and it's talking to the outside world. And already, you're seeing some of the exploits there. That's not a good thing. And you have no real control on the insides. Once some data, maybe malicious data gets in, there's no way to get it out. There's nowhere it's stored. It's embedded into the billions of variables in the model. You can't get it out. So it's a totally different underlying world. We're going to get a lot more efficient all across the board, but security is going to be a huge, huge challenge.

Fair enough. I think my last question for you is you've had the opportunity to see a lot of investors today and possibly even yesterday. What is the single most prominent and most salient misunderstanding or misperception that you're finding investors are running into with the Akamai story?

Yes, it's a great question. In fact, we did an investor survey recently because -- the one common theme is, wow, you're undervalued. We are. And I bought stock...

I was going to say, you're putting your money where your mouth is.

Yes. Our Board Chairman bought stock. I don't know if he's ever done that before. The company bought back a lot of stock. And maybe the concerns were we had -- delivery was tough last year. That is stabilizing. Security, we have a fabulous security business. The majority of our revenue, good growth, market-leading products and compute, I guess the question people have is, well, is that real? Is that really going to -- yes, it's real. And so we're looking at ways -- some of the advice we got from investors in the survey was how we position ourselves and explain it, because you think about, wow, our cloud infrastructure services, $0.25 billion, which is not totally trivial last year, growing 40%, and we got a strong trajectory from there. So we just, I think, got to communicate that better and execute and show people, yes, we're doing what we said.

Stuff. We're going to be rooting and watching you from the sidelines.

Great. Well, thank you.

Thank you very much. I appreciate your time.

Thank you.