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

February 13, 2024

NASDAQ US Information Technology IT Services earnings 76 min

Earnings Call Speaker Segments

Tom Barth

executive
#1

Appreciate it, operator. Hello, everyone, and thanks for being with us on Akamai's call covering the fourth quarter of 2023. Joining us today are Tom Leighton, who serves as Akamai's Chief Executive Officer, and Ed McGowan, Akamai's Chief Financial Officer. I want to point out that what we say today contains forward-looking statements, and those include guidance on revenue and earnings. Such forward-looking statements carry risks and uncertainties and depend on a number of factors that might make actual results differ materially from what those statements express or imply. Those factors take in any effect of macroeconomic trends, the integration of any acquisitions, and any effect of geopolitical developments. More details about these factors can be found in Akamai's filings with the SEC, which include our annual report on Form 10-K and quarterly reports on Form 10-Q. The forward-looking statements made on this call reflect the company's view as of February 13, 2024. Except where the law requires it, Akamai disclaims any obligation to update these statements so they reflect new information, future events or circumstances. As a reminder, some non-GAAP financial metrics will come up during today's call. A detailed reconciliation of GAAP and non-GAAP metrics is available under the financial portion of the Investor Relations section of akamai.com. Before handing the call to Tom, I want everyone to know that I have transitioned to Head of Akamai's investor roles to Mark Stoutenberg. A lot of you have met Mark over the past year, and I'm confident you will find him extremely helpful about all things Akamai. I'll be moving into another role internally, and I want to thank Tom personally for bringing me into the Akamai culture 10 years ago; it's been a privilege to work with so many wonderful people, both here at Akamai, and of course, all of you externally. I wish you all happiness and good fortune. And now I'd like to turn the call over to Tom.

Tom Leighton

executive
#2

Thank you, Tom; I truly value the remarkable contributions you've made over your decade at Akamai. It has honestly been a joy to work with you, and I'd like to join you in extending a warm welcome to Mark as he steps into your position. Turning to our Q4 results, I'm pleased to report that ACMI closed out 2023 on a very strong note. Fourth-quarter revenue rose to $995 million, and non-GAAP operating margin was 30%. Non-GAAP earnings per share for Q4 came to $1.69, up 23% year-over-year and up 22% in constant currency. For the full year, revenue was $3.81 billion, with non-GAAP operating margin of 30%. Full-year non-GAAP earnings per share were $6.20, an increase of 15% year-over-year and 16% in constant currency. Security revenue in Q4 grew 17% year-over-year and 18% in constant currency, representing nearly half of our total revenue for the quarter. This security growth was driven in part by continued strong demand for our market-leading Guardicore Segmentation Solution, as more and more enterprises came to Akamai for defense against malware and ransomware. Customers who purchased Segmentation in Q4 included one of India's largest global tech companies, as well as a leading payment systems firm based in Hong Kong that handles 15 million transactions daily. We've also seen significant interest in our new API Security Solution, launched in August. Early adopters of this offering include three major financial institutions, several top retail brands throughout Europe and North America, and a notable industrial automation company in Europe. Customers are seeing the value of this new capability, with some already paying over $0.5 million per year for the service. And 2 of the world's best known software companies recently signed on and are already spending about $0.25 million per month on cloud computing with Akamai. Akamai's API Security Solution also drew praise from industry analysts in Q4. KuppingerCole named Akamai an overall industry leader in its API security and management leadership compass report. And Gartner affirmed our strong and expanded API capabilities in its market guide for cloud, web application and API protection. Now shifting to cloud computing, I'm happy to say we accomplished what we set out to do last year regarding infrastructure deployment, product development, kickstarting our partner ecosystem, onboarding the first mission critical apps from some major enterprise customers, and achieving substantial cost savings as we moved our own applications from hyperscalers to the Akamai Connected Cloud. After launching Akamai Connected Cloud last year, we rolled out 14 new core computing regions worldwide, bringing the total to 25. We further enhanced our cloud compute offering by doubling the capacity of our object storage solution and introducing premium instances for large commercial workloads, designed to deliver consistent performance along with predictable resource and cost allocation. Additionally, our Akamai Global Load Balancer is now live. This integrated offering is designed to steer traffic requests to the data center that works best, cutting latency and making sure there's no single point of failure. Over the last twelve months, we also improved how we go to market via our cloud computing partner program. Thanks to the program's collaborative setup, Akamai has a unique method of engaging customers in a consultative way—gaining a deep understanding of their requirements and pain points, then providing a complete solution that combines the partner's technology with Akamai's distributed compute platform. A number of leading SaaS and PaaS providers, as well as cloud data and processing platforms, are already working with us as partners. We're thrilled with our progress and eager to add more new partnerships throughout 2024. Momentum is growing too among some of our largest customers, who are shifting mission-critical applications onto our cloud platform. Consider one of the most widely recognized social media platforms anywhere: it spent about $5 million with us on computing services in the prior year, and in the current year it's already on track to more than double that amount. And 2 of the world's best known software companies recently signed on and are already spending about $0.25 million per month on cloud computing with Akamai. In Q4, we also welcomed a major global media measurement and analytics company, and we displaced a hyperscaler when Blu TV, Turkey's biggest VOD streaming company, chose us. As that streaming customer described it, our platform felt easy to use, automated and intuitive, cloud agnostic—which made a smooth multi-cloud migration possible—and affordable because of extremely low egress cost, all backed by a trusted and reliable partner. In addition to surpassing our $500 million full-year cloud computing revenue goal last year, we realized significant cost savings by moving several of our own applications off hyperscalers and onto Akamai Connected Cloud. Our bot manager and enterprise application access solutions were among the first to make that shift. More than 1,000 customers depend on these products together, and they generate over $300 million in annual revenue for Akamai. Still, all of that is merely a beginning. Today we're excited to reveal the next stage of our multi-year plan to transform the cloud marketplace—extending cloud computing to the edge by embedding cloud computing capabilities inside Akamai's massively distributed edge network. Codenamed Gecko—short for "Generalized Edge Compute"—Akamai's new initiative combines the computing strength of our cloud platform with the proximity and efficiency of the edge, positioning workloads closer to users than any competing cloud provider can. Traditional cloud providers support virtual machines and containers across a relatively small group of core data centers. Gecko is designed to extend that capability out to our edge pops, bringing full stack computing power to hundreds of locations that used to be difficult to reach. When we roll out cloud computing across Akamai's global edge network, we'll be positioned to take advantage of the operational tools, processes, and observability we already have—enabling developers to build across the full spectrum of compute and giving them a uniform experience extending from centralized cloud right out to the distributed edge. Nobody else in the market offers this today. In the latest phase we're announcing now, Akamai plans to bring virtual machine–supported compute to around 100 cities by the end of this year. We've already stood up new regions on the Gecko architecture in 4 nations, as well as in towns where no major hyperscaler presence exists. The list of those initial locations appears in today's press release. Following that, the next thing we'll add is container support. After that, we aim to create automated workload orchestration, which will make it easier for developers to build applications that stretch across hundreds of distributed sites. Alongside a set of our enterprise customers—the ones eager to give their own users better experiences by moving workloads closer to users, devices, and data sources—we've been conducting preliminary Gecko trials. As they evaluate Gecko for tasks such as AI inferencing, deep learning for recommendation engines, data analytics, multiplayer gaming, quicker banking transactions, e-commerce personalization, and various media workflow applications including transcoding, their early feedback has been extremely favorable. In short, I'm tremendously enthusiastic about what lies ahead for Gecko as we extend full stack compute to the edge. Turning to content delivery, I'm pleased to report that Akamai remains the unmistakable market leader, providing the scale and performance that the world's top brands need while we help them deliver reliable, secure, and nearly flawless online experiences. Delivery is still a core profit driver for us—one we draw on to build new offerings that fuel Akamai's future growth. It's also a significant engine behind our security and cloud computing businesses, because we gain the competitive and cost advantages of selling delivery, security, and compute together as one bundled platform. As we've said on prior calls, we're selective about less profitable delivery opportunities. That discipline will continue, and in some cases intensify, through 2024. Let me be clear: we still aim to be the best at delivery for our customers, and we believe this disciplined approach will benefit our business by allowing us to steer a larger portion of investment toward security and cloud computing, which now represent nearly two thirds of Akamai's revenue and are growing rapidly. And 2 of the world's best known software companies recently signed on and are already spending about $0.25 million per month on cloud computing with Akamai. To wrap up, we're pleased to have delivered what we promised in 2023. Our cloud computing efforts are advancing as we envisioned. Our expanded security portfolio is helping us deepen customer relationships. And we continue to invest in Akamai's future growth while at the same time enhancing our profitability. I'll now pass the call to Ed for more detail on our results and our Q1 and 2024 outlook. Ed?

Ed McGowan

executive
#3

Thanks, Tom. I also want to say how much I appreciate Tom Barth — his ten years leading investor relations have been truly outstanding. Moving on to the numbers: fiscal 2023 was a year I feel great about, delivering non-GAAP EPS of $6.20 and marking another period of double-digit earnings growth for shareholders. In the time ahead I'll walk through what happened in Q4, share some perspective on 2024 — covering a handful of items that should give investors a clearer picture of what's shaping our coming results — and wrap up with guidance for both Q1 and the full year 2024, starting on the revenue side. For Q4, revenue reached $995 million, up 7% from the prior year on a reported basis and also in cost and currency terms. During Q4, both our compute and security segments delivered ongoing robust expansion. Revenue for the compute division reached $135 million, representing a 20% increase compared with the same period last year, both as reported and on a constant currency basis. Feedback regarding our cloud compute products continues to be highly encouraging, and we're quite enthusiastic about the initial momentum coming from enterprise clients. Now looking at security, that segment brought in $471 million, which is an 18% rise versus the prior year period and 17% higher in constant currency terms. Growth within our security offerings remains fueled by notable gains in the Guardicore Segmentation Solution, alongside our market-leading web app firewall, denial of service, and bot management products. Furthermore, as Tom pointed out, the early progress of our newly launched API security solution has been quite encouraging. In Q4, we closed deals with 17 API security clients, and 4 of those carry yearly contract values above $500,000. Delivery revenue came in at $389 million, which includes roughly $20 million tied to the contracts we picked up recently from StackPath and Lumen. International revenue reached $479 million, an 8% increase compared with the same period last year, or 6% growth in constant currency, and that accounted for 48% of total Q4 revenue. And finally, currency movements hurt revenue by $4 million quarter over quarter, while giving us a $6 million lift year over year. Now on to profitability. For the fourth quarter, we posted robust non-GAAP net income totaling $263 million, which works out to $1.69 per diluted share. That represents a 23% increase compared with the same period last year, or 22% when measured in constant currency, and it came in $0.07 above the top end of the range we had guided to. The primary drivers behind these better-than-anticipated EPS figures were the ongoing gains from the cost-reduction efforts we've described before, along with roughly $6 million in transition services—TSA—expenses that landed below what we had anticipated on the StackPath and Lumen contracts, because our services team moved those customers over to our platform far more quickly than we had projected. Capital expenditures for Q4 reached $143 million, which is slightly under 15% of revenue. We felt quite good about how we kept concentrating on reducing how capital-intensive our delivery business is. Thanks to that work—combined with really robust profitability—we produced exceptionally strong free cash flow in Q4. Turning to how we allocate capital: in the fourth quarter we used about $55 million to repurchase roughly 500,000 shares. For the full year, approximately $654 million was directed toward buying back around 8 million shares, and as 2023 ended, about $500 million remained accessible through our current repurchase program. Going forward, our plan is to continue buying back stock so that dilution from employee equity plans is gradually offset, while we also remain prepared to move opportunistically when it comes to M&A as well as share repurchases. Ahead of discussing our outlook, I want to highlight a couple of points that ought to help investors see the business more clearly. The first concerns the revenue we earn from delivery. In the first six months of 2024, seven of our ten biggest CDN customers have contracts that come up for renewal. As we've noted before, this type of renewal typically leads to a drop in revenue at first, and then we tend to watch revenue rise once more as traffic expands over time. We have already factored the expected impact of those renewals into our guidance for Q1 and for the full year 2024. Beyond that, per Tom's remarks, our plan is to continue refining the delivery side of the business. That means changing the way we charge specific high-volume traffic customers for what they use on our network, with profitability always front of mind. To illustrate: for delivery destinations that cost us more, we plan to introduce a premium charge; we expect to keep improving the balance between peak and day-to-day traffic; and when it comes to API traffic versus download traffic, we plan to negotiate separate pricing. By choosing to drop certain traffic that earns us less, we'll arrive at a pricing approach that's both better balanced and more profitable — which, in our view, is the correct strategy for this company. Also, a 15% worldwide minimum corporate tax is something OECD nations continue to move toward putting into law. To be specific, Switzerland's Federal Council announced in December 2023 that these rules would apply there beginning in 2024. As a result, we expect our non-GAAP effective tax rate to climb about 1.5 to 2 percentage points, landing near 18.5% to 19%. We figure this higher tax rate will cut Q1 non-GAAP EPS by roughly $0.02 to $0.03 per diluted share, and reduce full year non-GAAP EPS by about $0.12 to $0.15 per diluted share. Our Q1 and full year 2024 guidance already reflects the effect of this tax rate change. Also, we're anticipating a big increase in free cash flow for 2024 compared with 2023. The primary driver is a substantial drop in capital expenditures next year, along with our continued emphasis on profitability. Don't forget that the full expense of constructing the Gecko compute sites that were announced earlier today is factored into our Q1 and full year 2024 capital expenditure guidance. And fourth, I'd like to call your attention to the typical seasonal patterns that affect both revenue and earnings throughout the year. On the revenue side, our strongest quarter tends to be the fourth quarter. On the profitability side, during Q1 we take on significantly greater payroll tax expenses tied to the Social Security tax reset for staff members whose contributions hit the cap in 2023, along with costs from stock vesting under employee equity plans, which are usually weighted more heavily toward the first quarter. Another point to keep in mind is that our company-wide annual merit-based salary raises take effect in Q3, so operating costs in Q3 typically come in above Q2 levels. Lastly, looking at 2024, we expect greater turbulence in foreign exchange markets, caused by the hard-to-predict timing and scale of Federal Reserve policy shifts and how those shifts affect interest rates. Given that, projecting how FX will move during the back half of the year is a very difficult task. So, for the entire year, our plan is to give annual revenue growth, security and compute revenue growth, non-GAAP EPS growth, non-GAAP operating margin, and CapEx solely on a constant currency basis, using 1231 to 2023 exchange rates. For the upcoming quarter, though, we'll offer guidance on both an as reported and a constant currency basis. As a reminder, roughly $1.2 billion in yearly revenue comes to us in foreign currency, with the Euro, Yen, and Great British Pound representing our biggest non-U.S. dollar revenue sources; on top of that, our costs in non-U.S.

Ed McGowan

executive
#4

Amounts in dollars are generally much smaller than what we bring in as revenue, and they're mostly denominated in the Indian rupee, the Israeli shekel, and the Polish Zloty. Let me turn to guidance now. When we set our 2024 guidance, we assumed there would be no meaningful shifts, whether favorable or unfavorable, in the macroeconomic environment as it stands today. For Q1 of 2024, our projection is revenue between $980 million and $1 billion, which works out to an increase of 7% to 9% as reported, or 8% to 10% in constant currency, measured against Q1 2023. Based on spot rates right now, currency movements should add a positive $2 million to Q1 revenue relative to Q4 levels, while creating a negative $4 million effect on a year-over-year basis. With revenue at those levels, we're expecting cash gross margin of roughly 73%, both as reported and in constant currency. We project Q1 non-GAAP operating expenses will land between $305 million and $310 million. For Q1, we anticipate EBITDA margins of about 42% to 43%, as reported and in constant currency. Non-GAAP depreciation expense should come in between $127 million and $129 million. For Q1, we're looking at non-GAAP operating margin of roughly 29% to 30% as reported and in constant currency. Given the revenue and spending picture I've just described, we expect Q1 non-GAAP EPS between $1.59 and $1.64, which is an increase of 14% to 18% as reported and 16% to 19% in constant currency. This EPS guidance factors in taxes of $56 million to $58 million, using an estimated quarterly non-GAAP tax rate of about 18.5% to 19%. It also assumes a fully diluted share count of roughly 155 million shares. Turning to CapEx, we anticipate spending somewhere around $146 million to $154 million in the first quarter, excluding equity compensation and capitalized interest; that works out to roughly 15% of total revenue. Looking at the full year 2024, we're expecting revenue growth of 6% to 8% in constant currency; for security revenue, we expect growth of approximately 14% to 16% in constant currency. Compute revenue growth should be around 20% in constant currency; and we estimate non-GAAP operating margin of roughly 30% in constant currency. Full year CapEx should come to about 15% of total constant currency revenue, which again factors in the Gecko compute buildup; here's roughly how we expect our CapEx to break down. About 3% of revenue goes to our delivery and security business, about 4% of revenue goes to compute, about 7% of revenue goes to capitalized software, and the rest covers IT and facility related spend. We expect non-GAAP earnings per diluted share growth of 7% to 11% in constant currency; and as we said before, this non-GAAP earnings guidance rests on a non-GAAP effective tax rate of approximately 18.5% to 19%, along with a fully diluted share count of approximately 155 million shares. To wrap up, we're quite happy with how strongly 2023 finished; we remain enthusiastic about our growth opportunities and about improving profitability throughout the business; at this point Tom and I would be glad to take your questions. Operator?

Fatima Boolani

analyst
#5

Thanks for taking my questions. Ed, I'd like to dig into the delivery side, how the segment performed, the guidance, and a few of those modeling items you walked us through. Could you break out for us the organic traffic patterns you observed on the platform, separated from the traffic tied to the contracts you acquired via StackPath and Lumen? Then, as far as you're able, speak to whatever timeline applies to how you'll take in some of that extra acquired traffic from those contracts. And the final part here is the guidance, which, based on our math, implies delivery declining 6% in 2024 off your full year outlook. Could you address what share of that traffic you'd consider lower quality that's being peeled off — is that perhaps the reason we aren't seeing a sharper bounce in the delivery franchise? Then I have one quick follow-up. Thanks.

Ed McGowan

executive
#6

All right, sure. Let me see if I can get to each of those. I'll begin with how Q4 performed; I'll split it into a few buckets. In Q4, seasonality typically works in our favor, and we did get some of that benefit. On the retail customer side, bursting came in at roughly half the level we had a year ago. My sense is that's mostly tied to the zero overage. As that offering has gained traction, bursting has shown up less. Looking back several years, it's down more than 50%, 22 versus 21, and again, 23 versus 22. Macro factors might play a part, though it's hard to say. We aren't really the ideal gauge of how much folks consume — it's more about their surfing. Turning to gaming, gaming had a stronger year. A year ago, gaming was fairly soft. This year, it was solid. Gaming usually carries fairly low-priced delivery, so revenue upside is limited. Video rose quarter-over-quarter, though not by as much as a year earlier. So overall, I'd call it a decent fourth quarter for delivery. The softest area was probably more around high tech. Think of that as new connected devices — maybe connected TV, or a new, say, iPad or something along those lines, printer drivers, that kind of thing. That came in much weaker than a year ago. Last year brought a pretty nice bump in that category; so that essentially captures the dynamics for delivery in Q4; and I looked at that without the acquisition traffic included. Your second question, if I noted it right, concerned absorbing traffic from the acquisitions. We've mostly completed that. The services group executed an outstanding job moving those customers over. We managed it in roughly one-third of the timeline we had anticipated needing. A handful of accounts are still lagging behind, yet the costs tied to transition services are not material. That explains why we didn't highlight it. But by and large, that work is finished. Now, regarding this year, what's behind a 6% decline if that's the calculation you're applying to delivery? Much of it relates to renewals. Sadly, when it comes to these bigger accounts, as we've mentioned previously, only around eight of them represent more than 1% of revenue. And regrettably, seven of those are up for renewal during the first half of the year. That certainly creates some effect on the delivery business. Most of the revenue generated by those accounts originates from delivery; we've observed this before. It feels as though every 2 years, we run into this situation. Despite our efforts to sign staggered contracts, what frequently occurs is that a customer signs a 2 or 3 year commitment with a revenue target. They end up spending it over a shorter span. And we simply end up with a cluster of renewals at once. Then there was the other matter concerning shedding of traffic. Let me get a bit more granular about what we're doing. Last year, if you recall, we discussed being somewhat stricter regarding peak to average ratios. We intend to keep doing that. We performed fairly well. A few customers still need adjusting. This typically occurs when customers split their traffic and have plenty of day-to-day traffic, and they might divide that across 3 or 4 CDNs. Then they'll have peak events, whether that's a large video event or a large download event. And we receive the disproportionate share of that peak; so you want to ensure you're getting paid for it. The method for doing that is either you capture a greater percentage of the day-to-day or you simply cap the peak; so we'll keep doing that. And at times that might mean we lose a bit of traffic, but that's fine since you can see the outcome in CapEx. The new initiative we're pursuing now is that we've collaborated with product and IT to allow us to build at a very granular level on destinations. Previously, we would typically build around large geos, say for instance, a large geos, say for instance, the U.S. or Europe or Asia. Now we can go one level deeper. And so we can target some of the regions where our costs to deliver are somewhat higher. It's unclear. We're assuming that customers have choice. They may choose to take that traffic elsewhere. Some of our rivals don't place as much emphasis on profit as we do. Should that prove true, we're fine with it. As we attempted to take everything into account, those two elements are the main reasons we aren't observing stronger delivery results for 2024. I believe I've addressed every point you raised.

Fatima Boolani

analyst
#7

You certainly did. That was really thorough. Thanks. And here's a simple one, just regarding gross margins — you provided the guidance for cash gross margins for the first quarter; but speaking qualitatively, could you walk us through a few of the positives and negatives on a full-year basis as we consider how the business mix is shifting, your emphasis on pricing discipline, and also a U.S. versus international mix — any positives and negatives around the dynamics we ought to keep in mind for COGS and cash gross margins across the full year. That's all from me. Thanks.

Ed McGowan

executive
#8

Sure. Happy to. When it comes to the cost of goods sold, I'll begin with the favorable things happening there. Clearly, you brought up the mix. So as security business becomes a larger portion, that clearly benefits our gross margin line. And the second item is shifting our third-party cloud costs over to our own platform. We've done a great job on that so far this year. Tom mentioned it a bit in his prepared remarks. And there's more ahead of us, and we have a roadmap to achieve that. So that will generate some meaningful savings. The negative side, or what's sort of compressing them or preventing the margins from expanding as much, I should say — we could be somewhat higher — is that with the build of the Gecko sites, we'll take on some additional co-location costs. So as we begin building out those facilities, you don't have revenue to go against it. But also, as I mentioned last year, when we built out some of the bigger sites, under the accounting rules, when you make long-term Colo commitments or any kind of commitment for long-term leases, you have to straight line any commitment. So we wind up with somewhat of a disconnect between what we're paying in cash and what we're actually expensing; so we're able to offset that with the savings we get from the third-party cloud in the mix, but those are the underlying dynamics.

Fatima Boolani

analyst
#9

Thanks a lot.

Frank Louthan

analyst
#10

All right, thank you. In what ways does Gecko differ from the Compute work you're doing right now? Is it a fresh bundling of certain offerings, or is it a bit more distinct than that? Then, for my second one, regarding the CDN operation you bought in the prior year, how much do you anticipate that shrinking? I believe the aim was to move it toward some additional services, but what are you building into your assumptions on that?

Tom Leighton

executive
#11

Sure, I'll handle the first one. No, Gecko isn't a bundle. It's a brand-new ability, and by year's end we'll deliver full-stack compute across 100 cities, and eventually hundreds of cities; essentially, we're bringing full-stack compute — the sort of services available from Linode or a hyperscaler — into our edge pops; as things stand, we have 4,000 edge pops running function as a service. Depending on user demand, we'll launch JavaScript apps in milliseconds across 4,000 locations spanning more than 700 cities, yet that isn't full-stack compute. What we'll do now is bring virtual machines and containers into our Edge platform and support them there, which lets our customers achieve far superior performance and scalability, plus reduced cost thanks to the financial upside our edge platform delivers. So I think it turns into a highly attractive cloud computing option that simply isn't available on the market right now; it's not a bundling matter whatsoever. It's a new ability, and naturally, once we've got VM and container support, our aim is to make it operate the same way function as a service does, so we'll be launching containers and VMs on demand at the moment and place they're required, and nothing like that exists in today's market. Ed, do you want to field the CDN one, the acquisition?

Ed McGowan

executive
#12

Absolutely, glad to do that. Hi, Frank. Yeah, I'd say — just to remind everyone — when those businesses were bought, it was actually selected customers that came with them. What that means is we went through things carefully and decided not to bring some customers aboard; say, for instance, if a party was in breach of our acceptable use policy, very small accounts, and then — hard as it may be to believe — a few whose pricing we simply had no interest in taking on. So a sort of selection process had already been carried out by us. As you may remember, guidance I provided during the prior quarter was roughly $18 million to $20 million for this quarter, and the top end of that range was reached, which makes me glad. So pretty much everything we'd been hoping for has been migrated over. A handful of customers did churn, yet on the whole, everything has been gotten out of that acquisition — or those acquisitions, rather. Two of them were what we'd been hoping for.

Frank Louthan

analyst
#13

All right, excellent. Thanks so much.

James Fish

analyst
#14

Hi, everyone. To start, I just want to talk about Tom Barth. You've been a real professional and I'm grateful for all your assistance through the years, and I simply wanted to add my voice to what Tom Leighton and Ed expressed. I truly value the support over the years. Now I'd like to shift to security. Security came in somewhat below what I believe all of us expected for Q4. I understand there are certain factors beneath that are benefiting the business, but did any deals get delayed, and could that be feeding into your confidence regarding mid-teens growth for 2024? Give us some help on that confidence for continued mid-teens growth, and really, how is selling going outside the existing customer base, and what about penetration of those new security packages?

Ed McGowan

executive
#15

Hi, Jim; Ed here. Yes, honestly I felt quite good about how we performed and didn't observe many deals slipping out. We had no big license deals the way we did in Q3, so that could be — that always distorts some of the numbers, so none of that happened for us; but no deals truly slipped from a security standpoint. Very happy with what we're seeing at Guardicore, which keeps growing tremendously. That has been outstanding, and a great many customers are being brought into new verticals and similar areas. The channel has been performing incredibly well there. We discussed at length in our prepared remarks, API security. It's rare that I'm caught off guard, but the ARPUs there have caught me off guard. I've been extremely happy with that. That has been really, really encouraging to witness; and the strength in — we discussed the bundles that we worked on heavily during the last call; that has kept going quite well, and we're observing very robust growth in our WAF and our fraud products as well, bot management and our fraud protector. So really strength everywhere. Nothing thus far from macroeconomic difficulties; that can always shift. You never know what might occur, but nothing that we encountered in Q4. Regarding the projections for this year, we feel quite confident. We've typically been fairly conservative with how we approach security growth. We don't build in any significant kind of attacks where occasionally we'll see demand spike or anything similar. So we feel quite good about how security is progressing.

Tom Leighton

executive
#16

Perhaps to drill one level further into what Fatima asked about earlier regarding the CDN side, and also something Tom mentioned during his prepared remarks — is there a way for us to grasp how much of the CDN base you intend to, well, I don't want to phrase it as handing over at no charge, but providing at no cost, in order to capture that compute revenue? Or could you help us grasp the interplay between what we ought to anticipate regarding CDN versus compute in terms of dollars migrating? Thanks, everyone. Right. My view is that no percentage can be incorporated at this juncture. We've had certain conversations with a few very large media companies under which we'd provide discounted or complimentary delivery in exchange for a substantial share of the compute business; taken as a whole, that would be an excellent trade on our side — considerably more profitable and considerably more revenue, since ultimately, large media companies are going to spend 10x on compute compared with what they'll spend on delivery and even security. This is a practice we observe among the hyperscalers. Occasionally they'll provide the delivery at no charge in exchange for winning the compute business, given that's where the overwhelming bulk of the revenue sits, and it's very profitable; we'll keep you updated as things progress if that begins to make a material difference.

Keith Weiss

analyst
#17

Great. Thanks, everyone, for fielding this. This one's directed at Ed. All through 2020, we spent a lot of time discussing cost-cutting programs. There was plenty of talk about shifting workloads onto the internal cloud, which produces those savings. Honestly, to be direct about it, the absence of margin growth in the sky is somewhat disappointing. Is anything preventing that, or could there be some added spending, maybe tied to Gecko, that we're doing in its place? Or answering a set of questions along the lines of, why isn't there a bigger margin? Considering every one of those efficiency-type gains you folks have been implementing over the last 12 months?

Tom Leighton

executive
#18

Yes, I'll handle this one, Tom. You cut out a bit there, Keith, but I believe I caught the gist of what you asked. So yes, I felt we did a terrific job completing some acquisitions during the past few years, putting money into a compute business, spending a tremendous amount to construct our compute facilities, adding plenty of functionality, expanding our security business, making purchases there too; and we returned to a 30% margin in the prior year and are carrying that forward this year. We've consistently described that as a fairly healthy level at which to operate the company. We're putting money in because we perceive a chance to grow, and there's always a trade-off. You can't cut your way to greatness. Maybe we could trim a couple more points, but then what would we be giving up? I believe we've stayed fairly disciplined and balanced in how we've approached investing for growth while coming back to margin. We have some quite thrilling areas. We're experiencing strong growth in API security. It remains early days there. The product will keep improving. We've observed solid ARPUs there, so I'm quite enthusiastic about our efforts there. We've put investments into go-to-market and Guardicore, and that definitely has delivered results. And the investments in Computer Stein have produced some early returns. So once more, it's a balanced approach. We're here for the long-term, and I think we don't want to shoot ourselves in the foot and fail to pursue some of these large opportunities sitting in front of us.

Keith Weiss

analyst
#19

Got it. That's clear.

Tom Leighton

executive
#20

Thank you.

Mark Murphy

analyst
#21

Thanks a lot. Tom, you've clearly performed quite well in the compute segment. During your prepared remarks, you brought up the onboarding of apps that are critical for submission. Could you provide some insight into what sort of pipeline you're seeing for those essential applications arriving in 2024 and up to now? How effectively is your infrastructure managing the heavy demands of those bigger workloads when it comes to stability, reliability, and uptime? Then I have one quick follow-up.

Tom Leighton

executive
#22

Sure. Bringing compute customers aboard is a major priority for us in 2024; the foundational infrastructure is ready. Naturally, we're currently expanding Gecko. Yet even using what we already have across 25 cities, our aim is to bring in lots more apps vital to operations from large corporations. To give some instances, consider social media and live transcoding. Two enormous firms now rely on that through our platform — one handles live sports broadcasting, while the other deals with live content generated by users. With another client, we host their e-commerce sites in a manner that delivers superior performance since it's nearer to the end-user and costs less. AI inferencing serves ad targeting and content personalization. Once more, you need that done extremely quickly. There's simply no time to backhaul everything to one centralized spot. You need presence in numerous locations globally for improved performance. And once more, we can deliver it more cheaply. We even have a major bank — among the biggest worldwide — currently employing our edge compute to register credit cards, specifically their users' credit cards, with Apple Pay, since Apple Pay mandates completing that registration within 60 milliseconds. The sole method for them to achieve that speed is by operating on Akamai Connected Cloud; thus, numerous distinct applications are already live on the platform, with proofs of concept underway; so our priority for this year, and I believe the pipeline looks strong, involves adopting far more apps essential to operations for large corporations. Naturally, we're the foremost major player; we already have massive applications operating on the platform quite effectively; and we deliver it through a multi-cloud approach. As I mentioned before, we've now integrated global load balancing along with failover functionality, ensuring a dependable service that performs at a high level. So I'm genuinely thrilled about what lies ahead this year.

Mark Murphy

analyst
#23

Yes. It's really nice to hear how this connects to inferencing and Apple Pay too. So I appreciate those details. Ed, I'd like to ask you something: you're giving fiscal year guidance on a constant currency basis, and naturally we all get that predicting the actual swings in the spot market is going to be extremely hard, but if we were looking at current FX levels, do you figure it would push that 6% to 8% revenue growth figure up or down? If we tried to convert it into reported U.S. dollars in our models as of right now today?

Ed McGowan

executive
#24

Yes, that's a fair question. I think today's CPI report showed you nicely how fast things are able to shift. At first the market believed there would be plenty of rate cuts, and now suddenly that appears unlikely to occur, and clearly currencies and interest rates move together very tightly. So if you take a look, I provided the 1231 figure, so clearly the dollar has become somewhat stronger. I gave you those figures regarding total non-U.S. dollars, so you can sort of run some calculations; it would still fall within that range. Clearly, it would be somewhat of a headwind simply because the dollar has strengthened since 1231, but I believe you can take our Q1 guidance and kind of work that in and consider the usual seasonality that exists and arrive at an answer. But it would still fall within that range, though.

Mark Murphy

analyst
#25

Just to make sure I've got this right — if we convert it into USD, does it remain within that 68% band, or do you mean it's still roughly mid-single-digit, though possibly about a point less?

Ed McGowan

executive
#26

No, well, what I'm saying is that using today's spot rates, basic arithmetic indicates yes. Suppose you simply took the guide's midpoint, correct? If that's what I apply, along with how the dollar has performed, it stays inside that band. Now, you might wonder whether it could come in higher. Clearly, back at 1231 the dollar sat at 101; by November it had reached 106, so it keeps swinging around. Naturally, should it shift, you can run the numbers — five or six points lower, and you might potentially land beyond that band. So once more, the thing is I'd just wind up handing you all an enormous range, which wouldn't serve any purpose. What I'd prefer instead is to simply provide the tools so you can work it out on your own and examine — truly grasp — what the underlying core business looks like. Does that work? Understanding that strikes me as far more essential.

Mark Murphy

analyst
#27

Got it. Thanks a lot; I'd also like to express my appreciation to Tom Barth for many wonderful exchanges through the years.

Madeline Brooks

analyst
#28

Hello, everyone. I really appreciate you letting me ask a question this evening. I'll ask just one about security. Besides Guardicore, I'd like to hear about the remaining Zero Trust portfolio for this year — what trends have you noticed, and are you perhaps sensing more competitive pressure given how much the market has grown? Then I have a follow-up.

Tom Leighton

executive
#29

Sure. When it comes to Web App Firewall, we've held the top spot in that market for a decade, ever since we launched that space with Web App Firewall offered as a service. Ten years in, competition does show up. Even so, we remain the leader by a solid margin. It's a healthy growth area for us; we've layered on plenty of additional capabilities, bot management and, more recently, account protector, client side protection, which lets shoppers at commerce sites remain secure when they visit the site. You're going to need that nowadays for compliance. There's a brand protector too. That means spotting phishing sites and preventing them from taking, user data. Naturally, we've offered, denial of service protection for a very long time. We hold the leading market position there; then on the enterprise side, naturally, there's Guardicore, which as we mentioned is performing quite well; and API security genuinely excites me. My view is that over the longer term, it will grow into a market as large and as crucial as Web App Firewall has. Our aim there is to take the market leader position; and already within the go-to-market motion, Web App Firewall and API security reinforce each other strongly; we created a really simple method for doing a proof-of-concept for customers of our Web App Firewall. That's producing substantial early traction for us; in addition, we've connected with many of the load balancers and other firewalls on the market so we can bring aboard new customers who aren't, using my CDN or Web App Firewall. So I'd say, multiple areas within security are delivering great results for us.

Madeline Brooks

analyst
#30

Thank you very much, Tom. Then I've got a single fast question regarding compute as well. When we look at the results reported up to now, particularly among hyperscalers such as AWS, Microsoft, and Meta, we've somewhat picked up on this idea of an optimization turning point when it comes to cloud computing — which suggests that perhaps during this year we'll witness somewhat greater spending on new workloads. I'd like to know whether you've picked up on any of those patterns among your own clients who are considering compute for the very first time, or perhaps whether you're noticing a bigger appetite for compute during the year ahead compared with 2023. Many thanks.

Tom Leighton

executive
#31

Right — compute represents a massive market, and it's expanding quickly, with fresh applications constantly being built — not merely ones moving from a data center over to the cloud, but completely new applications as well. That's the area where we're experiencing considerable momentum. Additionally, there are situations involving lift and shift away from a data center or away from a hyperscaler; yet it remains a massive market and an excellent space for us to do business in. And even the ones who are optimizing — I suppose that isn't particularly favorable for the hyperscalers, but we're involved in that movement. It benefits us greatly, since we're able to assist customers in cutting cloud spend; and our early adopters of Akamai Connected Cloud have given us really positive feedback that they're achieving substantial savings; so the push toward optimization works in Akamai's favor.

Rishi Jaluria

analyst
#32

Fantastic. I really appreciate you fielding my questions. Let me join my peers in expressing gratitude to Tom Barth; collaborating with you over these 10 years has been terrific, and I'm eager to see what you do next. I'd like to focus, perhaps returning to Gecko. First off, could you share some detail on edge inferencing and how those applications actually appear? We tend to hear plenty of theoretical discussion about it, yet when you're engaging with customers and having those discussions, how does it play out in practice? What makes Gecko stand out for this? Then on the financial side, though I realize it's still early, within your guidance for the year are you factoring in an actual contribution from Gecko on the compute line? Or might that be something that, once it gains momentum, delivers additional upside past what you've modeled? Thank you.

Tom Leighton

executive
#33

Sure. I'll begin with edge inferencing; a few of the illustrations I mentioned are precisely what occurs for commerce sites as they determine in real time which content will actually be delivered to the visitor arriving at the site. Ad targeting, which ad do they receive? Anything tied to personalization. Regarding security, an enormous amount of inferencing gets applied to examine real-time data. Take, for instance, our very own bot management solution. Is the entity arriving at the site a bot, or is it a human? And even when it's a human possessing valid credentials, is it the correct human? You rely on AI and inferencing for that, and it has to happen extremely quickly. Sending it back to the centralized data center isn't feasible, since you're dealing with a huge volume of users that must be handled in real time, particularly during something like a live event. That's why being at the edge is important, since you gain scalability, you can manage it locally, you achieve excellent performance, and you can make it real time. Akamai's distinctive value proposition through Gecko is that we'll now be capable of supporting this, not across a few cities, but across 100 cities by the close of this year. So whatever you can place in a VM, a virtual machine, which covers most things, you'll be able to run that across 100 cities. And eventually across hundreds of cities, since we can deploy this within general Akamai Edge pods. After that comes containers, which accounts for essentially everything else you'd do in cloud computing. And then being able to bring it all up automatically. For compute, this represents an entirely novel idea that I consider quite potent, and there's significant intersection between pursuing that and inference engines, in which you attempt to perform something smart grounded in whichever end user or end entity happens to be engaging with the app; regarding Gecko specifically, we are currently only at the beginning phases of rolling it out. Nine cities are covered by us, and the 10th additional city should come online in roughly another month. We'll have reached 100 cities in total delivering compute by the time this year concludes. Thus, not much revenue tied to Gecko has been included in the guidance for the present year. The bulk of that would arrive in the following year. Consequently, guidance for revenue this year rests upon the initial 25 core compute regions that had been established by the close of last year; naturally, we'll roll this out at maximum speed and to the degree customers are willing to embrace it. Ideally, circumstances will arise in which we choose to expand further, adding more compute capacity, given how enormous the demand for compute is at Akamai. Ed, is there any additional perspective you'd like to offer concerning the guidance?

Ed McGowan

executive
#34

No, I believe you got that right, Tom. We aren't truly expecting anything. What I mean is, one thing we're doing during this year involves altering the compensation plan for our reps, so every one of them has to sell compute this year. So you might see situations—reps can get quite inventive. In sales, I picked up that behavior follows comp. So through pushing on this and requiring it of every rep, we ought to see far more use cases, far more opportunities, and so on; so there's always a possibility we might get caught off guard by how creative our field is in delivering opportunities to us, but nothing material regarding Gecko went into our assumptions.

Rishi Jaluria

analyst
#35

Alright, that's fantastic. Many thanks, everyone.

Michael Elias

analyst
#36

Good. I appreciate you fielding our questions. I have a couple, if that's okay. First, regarding Gecko, I assume the points of presence you already operate are currently handling security and delivery workloads. So looking at it from an architecture standpoint, could you walk us through what it means to extend the compute platform into those locations? Does that simply involve extra co-location rollouts, plus on the CapEx side, network hardware and machines? Any details you can share about how that expansion would actually work mechanically? Then secondly, Ed, during last year you mentioned enterprise sales cycles getting longer. I'm just wondering what you're observing when it comes to how your customer base is behaving on purchases. Anything worth flagging? Thanks.

Tom Leighton

executive
#37

Okay. I'll handle the first question; when it comes to Gecko, generally those are existing Akamai Edge locations. More specifically, they're usually the bigger ones where a substantial amount of equipment is already in place; connectivity to our backbone already exists. What we would do is bring in more servers; for compute purposes, that means a more powerful server, along with extra COLO space for those machines. Yet everything else in terms of infrastructure is typically already present. It's already hooked up, and delivery plus security are already running at those sites. So deploying Gecko this way turns out to be quite an efficient approach for us. And Ed, would you like to address the second question?

Ed McGowan

executive
#38

Right, yeah. Look, bringing in fresh accounts is never easy when conditions are like this — that's just how it goes. The security area is likely the single place where that isn't true; with the SEC reporting requirements now in play, plus the disclosure I added about CISOs potentially facing criminal charges over breaches and similar situations, that space tends to behave differently. Cybersecurity keeps taking up more of the Audit Committee's attention. That's a budget line that, first, rarely gets trimmed and, second, usually keeps growing. Still, yes, landing new accounts is tough. I do believe an environment like this actually works in our favor — this connects to what we were discussing in those earlier questions about getting cloud spend optimized. If you've watched what we've been doing, we've put an enormous sum of money into it. So that, I think, can be another advantage for us right now; though without question, getting new customers is somewhat harder, yet we're still performing quite well. Sure, things could shift. For us, though, it hasn't become a big issue so far.

Michael Elias

analyst
#39

Great. Thank you.

Ray McDonough

analyst
#40

Great, thanks for squeezing me in. Tom, perhaps a follow-up to something asked earlier. When we consider Gecko, you said your edge sites currently lack full stack compute. Yet how much effort goes into converging what's already present at those edge locations with the work you've done on building out node capabilities? Should we anticipate one shared software stack spanning both edge and centralized sites? And if that's the case, do you plan to have it ready by year-end?

Tom Leighton

executive
#41

Yes. Excellent question. What we're doing right now, as I noted in response to the previous question, is putting more hardware into existing edge regions, generally the bigger ones; the network is already there; delivery security is already situated there; so it's a matter of some additional color and servers. And yes, the aim is to have everything running on a single common software stack. At first, we have the Linode stack migrating into these edge pops for Gecko; but once we obtain support for virtual machines and containers, then afterward, we intend to add the software stack we possess for delivery, for security, and for function as a service that automatically, for instance, spins up JavaScript apps and milliseconds based on end-user demand, we want all of that operating on containers and VMs. That way you don't need to plan in advance regarding how many VMs you'd like in each of these hundreds of cities. It simply occurs according to end user demand. You automatically get new ones spun up, load balancing, fail over, truly a very compelling concept. And that isn't available in the cloud marketplace today. That is the vision — I think you really captured it when you spoke about the common software stack because only Akamai has that full edge platform today, that software stack around delivery and security that will now be including compute.

Ray McDonough

analyst
#42

Thanks for that context. And as a quick follow-up, when we consider how the Linode footprint grew last year, could you help us grasp, to the extent you're able, what portion of the capacity already constructed served internal needs — supporting those third-party cloud cost savings — versus capacity that's live today and producing revenue? And I realize we've covered this before, but any insight into what customer utilization we should anticipate that may be baked into your guidance as we progress toward the end of the year? That would be useful.

Ed McGowan

executive
#43

Yes, absolutely. So most of the increase in the compute guidance will come from enterprise compute. So the capacity we constructed over the past year. So if you sort of revisit and examine the numbers in terms of — we've sort of indicated, broadly, that $1 of CapEx equals $1 of revenue. You can sort of examine our current compute build-out activity alongside what we did the prior year. We mentioned we had approximately $100 million, all in, for our internal use; so that leaves a fairly substantial amount remaining for customer demand; now clearly, the way buyers operate these days, they select a location, and so on. So it won't be precisely a dollar for dollar at this moment, but it's a general rule of thumb; so I'd say most of what we constructed is for customer usage.

Ray McDonough

analyst
#44

Excellent. I'm grateful you let me squeeze in. Thanks a lot.

Tim Horan

analyst
#45

Thank you, folks. To build on what Ray asked. My assumption is that the aim is to arrive at a solitary platform, so that clients can reach the entire set of offerings with relative ease through, I suppose, a single entry point. What's your timeline for reaching that point? Also, regarding the Gecko offering, my impression is that it's fully serverless. Is it additionally a platform for development? Thanks.

Tom Leighton

executive
#46

Correct. Regarding a single platform, meaning the ability to perform all tasks in unison with identical software, such that our Edge software operates alongside the Linode software to launch VMs and containers, that won't happen before 2025. What we're doing initially is merging the infrastructure, and naturally, clients are able to purchase the offerings together as a bundle. In numerous instances, we already provide unified reporting. However, when it comes to fully automated provisioning and genuine serverless capability for VMs and containers, expect that in 2025. Now, let me think; also — what else did you ask me?

Tim Horan

analyst
#47

So Gecko, that fresh offering, is mostly a serverless thing, from what I gather. And is the full set of developer tooling available so they can host their entire applications on this platform?

Tom Leighton

executive
#48

Yes, that's right. It really hinges on what your definition of serverless is. To start with Gecko, you'd run things the way you do with Linode. You pick how many containers and VMs you'd like across the different cities. It's quite friendly for developers and behaves the same as Linode, so anyone who knows that environment would find it works in — well, by the close of the year, 100 cities for your virtual machines. Now, if your definition of serverless means — and nothing like that is available right now in the market for VMs and containers, where they simply spin up on their own the way we already do for Function as a Service and JavaScript — that arrives in 2025.

Tim Horan

analyst
#49

If you wouldn't mind my asking, this really sounds similar to something Cloudflare does, yet according to you, it isn't available right now. Tom, could you perhaps explain a bit about how yours differs, and what exactly you're up to?

Tom Leighton

executive
#50

Sure, excellent question. VMs and containers get zero backing from them, to say nothing of serverless or other options — support simply isn't there. Full stack cloud computing isn't something they offer.

Tim Horan

analyst
#51

Understood. Many thanks.

Alex Henderson

analyst
#52

Excellent. So it looks fairly evident that Guardicore plays an essential role in the expansion of your security business. And clearly it's affecting the total growth rate. I was hoping you might provide some indication of how the security product lines appear when Guardicore is left out, in terms of how fast they're growing. Even a rough estimate of that growth would be extremely useful. And then secondly, I was hoping you could discuss a bit your — you brought up inferencing, but I believe it arrived somewhat like a secondary consideration rather than the main emphasis. Could you address your engagement with AI inferencing at the edge and how much that demands either the 2025 type of structure or what requirements you have in that area and whether you're deploying GPUs at the edge to enable that?

Tom Leighton

executive
#53

Ed, would you like to handle the first one, and I'll take the second?

Ed McGowan

executive
#54

Absolutely. Let me take the first question. Since this is a year-end call, I'll share these figures at a broad level, though I won't do this each quarter. Looking at what we previously referred to — what we now call the Appen API security bucket, our biggest bucket, which covers bot management, our fraud offerings, our web app firewall, and our new API security product — that grew over 20% in Q4. So that's been amazing. As for Guardicore on its own, when I adjust for the one-time software we had in last year's Q4, it's expanding by roughly 6%. Meanwhile, infrastructure and services are seeing growth below 10%.

Alex Henderson

analyst
#55

I want to make sure I understand — does that figure represent growth for the entire year, or is it the rate for Q4?

Ed McGowan

executive
#56

That number is for the fourth quarter. For the full year, I don't have it on hand.

Alex Henderson

analyst
#57

That's enough. I simply wanted to know what it was.

Tom Leighton

executive
#58

All right. Yes, regarding the question about inferencing and AI and so on, yes, we are building full stack compute so we get excellent performance at a cheaper price point and make it available across hundreds of cities. And AI inferencing is one of the many activities you'd perform with that, and it isn't something we tacked on later. As a matter of fact, AI has been part of our products for, well, 10 years; bot management, for instance, operates on Akamai connected cloud; it's among the many workloads running there; so it isn't an afterthought. Right now there's a tremendous amount of hype surrounding AI. And much of that, I believe, is warranted. And I believe far more compute will be consumed as a result of AI. And among our customers who use Akamai connected cloud, it's a powerful use case. Having said that, AI isn't everything. In fact, our largest customers do media workflow, they do live transcoding. And AI isn't involved in that. So I'd say AI matters as a use case, one among several; specifically now, you asked about 2024 versus 2025, it's already being done on our platforms. You don't have to wait for the year to end unless you want it in 100 cities, since that arrives at year's end. No reason to hold off until 2025 when the instances get spun up automatically rather than by design in advance, which is how compute operates today; so you mentioned GPUs. Akamai has GPUs deployed, and we're deploying additional ones. We've used them before for graphics. And in the future, likely for Gen AI uses; at present we aren't really deploying them in the edge pops. And that isn't — simply because there's no need; it doesn't make financial sense. And in the edge pops, you'll handle the inferencing; and for inferencing, GPUs are an option, but CPUs are also in use. And at present, CPUs give us superior ROI. So I suppose there's considerable misunderstanding there too; now GPUs matter greatly for training, particularly large language models, and that will happen in the core, and we aren't supporting that as a key use case at present; theoretically we could, right, we possess all the technology needed, but that isn't where our focus lies when it comes to achieving the best ROI for our platform. And moreover, with these models, the bulk of the work, the bulk of the compute happens while you're running them for inferencing. You perform the training, and then you spend — just so it learns, you get it ready, and then you operate it. And operation is where the vast majority of cycles go, and CPUs can handle that. Also, quite often—like with the examples I brought up, whether it’s personalization, security, or data analytics—those things get handled at the edge, and there are solid reasons to do them there on CPU-based hardware.

Alex Henderson

analyst
#59

Great. Thanks for the complete answer.

Jonathan Ho

analyst
#60

Hi, good afternoon. I have just one question. How significant is the capability for global load balancing? And what might that indicate regarding your capacity to either bring in additional customers or generate revenue from that offering? Thank you.

Tom Leighton

executive
#61

Yes, that helps a great deal since it makes things far more scalable. With failover, reliability improves substantially. And I believe it's the fundamental capability that, of course, we've possessed for what seems like forever in delivering security, and now it's offered for compute; so I think that matters, and it substantially expands the market we're able to pursue for compute.

Jonathan Ho

analyst
#62

Thanks.

Tom Leighton

executive
#63

Operator, there's room for a final question.

Rudy Kessinger

analyst
#64

Hey, thanks for fitting me in, everyone. Ed, if I've got the math right — even taking out that $100 million of CapEx and calculating the prior year, which was meant for shifting internal workloads from last year into this one — it comes out to roughly $400 million in compute CapEx. And returning to that idea of $1 of CapEx equaling $1 of revenue capacity, $400 million of CapEx works out to about $200 million of compute growth for 2024 versus 2022. Do you think maybe you're overbuilding? Or what gives you confidence in the pipeline and the ramping usage to put so much into yet another round of build-out this year, when growth hasn't yet accelerated — correct? Your guidance calls for 20% growth next year, which is flat versus Q4.

Tom Leighton

executive
#65

Yes. So I'll just — let me tackle that first part. What I'd say is, when you examine the underlying pieces of what's growing, it's really that enterprise compute opportunity expanding extremely, extremely, extremely quickly — those figures, the percentages, it would be kind of silly to break them out, since they're starting from small numbers and growing by very large amounts. Beyond that, part of our strategy involves being competitive and having large core centers across many cities, and that calls for a bigger build out. So we've got a lot of capacity we need to sell. On top of that, we're seeing demand in particular cities, and you have to add more capacity where that demand appears. Then there are the Gecko sites we're building out — it isn't a major — well, it's a fair amount of capital. But I believe that's another major key differentiator for us; and as Tom said, we see a big opportunity in that area. So I realize plenty of people have been questioning whether we can take on large workloads, and so on. We obviously have a great deal of capacity out there. Like I mentioned earlier, we've changed our compensation plans so that our reps are now required to sell compute. That means we'll be seeing much more of it; we've accomplished an enormous amount on the platform side when it comes to adding functionality. We built out the platform, hooked it to our backbone, and we have plenty of new compute partners. The platform is ready to sell, so we're quite optimistic. And I think we're building in a fairly responsible way. As I said, our CapEx is relatively modest for this business at present. So I think we're in pretty good shape. And with that, that will wrap up today's call. I want to thank everybody for joining, and have a great evening.

Operator

operator
#66

That brings us to the end of this call. We appreciate everyone joining us for today's presentation, and you're free to log off at this time.

Read the full transcript via the API

You're viewing the first half of this call. Get the complete Akamai Technologies, Inc. transcript — plus 133,000+ transcripts from 7,000+ companies, speaker segments, AI summaries and full-text search — through the EarningsCalls.dev API.

Get the API View API docs →

This call discussed

For developers and AI pipelines

Programmatic access to Akamai Technologies, Inc. earnings transcripts and 133,000+ others is available through the EarningsCalls.dev REST API. Plans from $39.99/month — full transcripts, speaker segments, full-text search, and the recently-added /api/v1/transcripts/recent polling endpoint for ETL pipelines.

More tools for stock research

OptionsBell: an unusual options activity scanner that tracks options flow across all US stocks and emails you when large trades appear in Akamai Technologies, Inc. or any ticker on your watchlist.

StockMarketScan: a stock screener with alerts: 24 pre-built screens refreshed every 5 minutes, watchlists and email notifications. Both offer API and MCP access and fair monthly pricing.