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

May 7, 2026

NASDAQ US Information Technology IT Services earnings 63 min

What were the key takeaways from Akamai Technologies, Inc.'s May 7, 2026 earnings call?

In Q1 2026, Akamai Technologies, Inc. reported total revenue of $1.074 billion, reflecting a 6% year-over-year increase. The company announced a landmark 7-year $1.8 billion deal, the largest in its history, which is expected to significantly boost future revenue growth, with management projecting double-digit annual growth by 2027. Non-GAAP earnings per share for the quarter were $1.61, down 5% year-over-year, as the company continues to invest in cloud infrastructure to support its expanding AI-driven services.

What topics did Akamai Technologies, Inc. cover?

  • Landmark Customer Deal: Akamai signed a historic 7-year $1.8 billion commitment for cloud infrastructure services, marking the largest customer deal in the company's history. This deal is expected to validate Akamai's value proposition in the AI space and significantly enhance future revenue streams.
  • Strong Security Revenue Growth: Akamai's security revenue grew 11% year-over-year, driven by demand for its web application firewall and API security solutions. Management noted that 'CISOs are feeling more of a sense of urgency' due to the increasing sophistication of AI-driven attacks, leading to heightened demand for Akamai's security offerings.
  • Cloud Infrastructure Services Performance: Cloud Infrastructure Services (CIS) revenue reached $95 million, up 40% year-over-year, indicating strong demand across various sectors. Management highlighted a rapidly building pipeline for AI-specific use cases, which is expected to drive further growth.
  • Guidance for Future Growth: Akamai anticipates significant acceleration in overall revenue growth heading into 2027, projecting double-digit annual growth. The guidance reflects confidence in the company's cloud infrastructure capabilities and the impact of recent large contracts.
  • Investment in Infrastructure: To support the growing demand for cloud services, Akamai plans to invest approximately $800 million in capital expenditures over the next 12 months. This investment is aimed at enhancing the company's infrastructure to meet customer needs effectively.

What were Akamai Technologies, Inc.'s May 7, 2026 results?

  • Total Revenue: $1.074 billion (+6% YoY)
  • CIS Revenue: $95 million (+40% YoY)
  • Non-GAAP EPS: $1.61 (-5% YoY)
  • Operating Margin: 26% (in line with expectations)
  • CapEx: $206 million (19% of revenue)
  • Full Year EPS Guidance: $6.40 to $7.15 (null)

Akamai's strong Q1 results and the landmark $1.8 billion customer deal position the company well for future growth, particularly in cloud infrastructure and security services. However, the decline in delivery revenue raises questions about the sustainability of growth in that segment. Investors should monitor the execution of the cloud strategy and the impact of ongoing investments on profitability as catalysts for future performance.

Earnings Call Speaker Segments

Mark Stoutenberg

executive
#1

Hello, everybody, and welcome to Akamai's earnings call for the first quarter of 2026 — we appreciate you being here this afternoon. Today you'll hear from our Chief Executive Officer, Tom Leighton, along with our Chief Financial Officer, Ed McGowan. Before we go further, I want to flag that remarks made on this call contain forward-looking statements, including guidance on revenue and earnings; those statements rest on expectations and assumptions as they stand now, which carry particular risks and uncertainties and involve multiple factors that might cause actual outcomes to diverge materially from what is stated or suggested. These elements include, but aren't restricted to, consequences stemming from broad economic conditions, bringing any acquired business into the fold, international political events, plus additional risk elements laid out in our SEC submissions. Remarks made during this session capture the company's perspective as of May 7, 2026, and we assume no duty to revise any forward-looking statements. Please note, we'll be discussing specific non-GAAP financial measures on this call. You can find a thorough reconciliation between GAAP and non-GAAP figures within the Investor Relations area of akamai.com, listed under financials. Having said that, I'll pass things over now to Dr. Tom Leighton, our CEO.

F. Leighton

executive
#2

Mark, thank you. It's a pleasure to share that the year has begun strongly for Akamai. Within only a handful of months, we've hit significant milestones in our cloud computing approach, which represents a decisive moment in how our business grows and evolves. For a long time, Akamai has been recognized for running the biggest distributed platform on earth for delivery and security offerings, delivered on a worldwide scale, with a name built on reliability, quality, and trust. Today, we're drawing on that worldwide footprint plus our many years of know-how serving the largest enterprises on the planet, so that we can turn into a provider whose infrastructure is essential to the AI-driven economy. Back at GTC in March, we introduced the first implementation of NVIDIA's AI grid anywhere at global scale, and we shared plans to roll out thousands of NVIDIA RTX Pro 6000 GPUs. By bringing NVIDIA AI infrastructure into Akamai's enormous distributed platform, and by using intelligent workload orchestration throughout our network, our aim is to shift the AI market away from standalone AI factories and toward one unified distributed grid built for AI inference. When we push AI inference out to the edge and pair it with our extensive deployment of CPUs for delivery, security, and Functions-as-a-Service, customers gain the ability to run sophisticated models within milliseconds of their end users; that brings the responsiveness of local compute together with the scale of the global web, boosting performance while cutting latency and cost. Anyone who was at GCC heard NVIDIA describe Akamai as a crucial participant in the industry's AI infrastructure ecosystem, and the market has responded very favorably as our capabilities expand quickly across a broad range of enterprises. Today, we're thrilled to share yet another significant milestone for our cloud computing approach and for Akamai's evolution. We have signed a landmark commitment worth $1.8 billion over 7 years for our cloud infrastructure services, from a leading frontier model company. It stands as the biggest customer deal ever signed at Akamai, and it follows the $200 million CIS deal announced in February with a major U.S. tech company that likewise sits at the forefront of the AI revolution. These AI leaders picked Akamai because their AI workloads require the scale, performance, and reliability that our cloud platform delivers. Plenty of other enterprises have picked Akamai for comparable reasons. To illustrate, since the year began, a top cloud and digital infrastructure provider in Asia selected our VPUs to serve their low-latency live streaming media offering. A U.S. AI company selected our GPU platform to drive their voice-first solution for optimizing business operations. An AI-powered video intelligence platform in India selected our GPU platform to scale video analytics and computer vision workloads for retailers. A U.S. consumer AI platform selected Akamai cloud to run and scale live personalized agents. An AI commerce company in India selected our distributed inference platform to drive their ad personalization engine. And 2 premier global retail brands selected our distributed data capabilities to boost the performance and resilience of their online retail applications. Yet all of this is merely the start. Our prospect pipeline is big and growing quickly, with companies turning to Akamai for cloud solutions, and a number of them have extremely substantial requirements. To meet this robust and rising demand for our cloud infrastructure services, we intend to keep expanding our physical infrastructure along with our cloud sales and support organizations. And, as Ed will discuss shortly, we now foresee a meaningful step-up in our total revenue growth as we move into 2027 and afterward. Now on to security. I'm happy to share that Q1 was likewise a solid quarter for our security portfolio, with revenue up 11% year-over-year as reported and up 9% in constant currency. Growth in security was driven by robust demand for our market-leading web app firewall, API security and Guardicore Segmentation solutions. Our WAF especially is drawing rising interest from customers who want to put the newest defenses in place against vulnerabilities that the ever-stronger frontier models and AI-powered attacks could expose. Frontier models are reshaping vulnerability management, and we take pride in being among the industry's must-have security providers, working alongside the Frontier model companies to help make sure AI-enhanced defenses get deployed safely and quickly. Thanks to our early access to their vulnerability detection programs, we're bringing our expertise to bear in helping keep major enterprises and critical infrastructure protected. Naturally — and this matters to grasp — attackers will likewise leverage more sophisticated AI technology to devise even more powerful methods of causing harm. That means major enterprises are going to require Akamai security solutions even more than they did previously. Take, for instance, the many legacy systems and billions of deployed devices that cannot be patched. They will grow far more exposed as AI advances, and they will require our security solutions to remain safe. For the devices and systems that can be patched, the process of patching still takes time, frequently days or weeks, and they will need our protection in the meantime. We have witnessed this situation before, when zero-day attacks appeared. And given the advances in AI, we should expect zero-day attacks to show up far more often. There's also a mounting difficulty around scale, since AI lets attackers seize control of more devices and assemble enormous bot armies, and we are now seeing attacks at unprecedented volumes. Only a few weeks ago, we shut down a series of app layer attacks carrying millions of malicious requests per second originating from millions of widely distributed IPs. Akamai is able to defend against attacks like these thanks to our widely distributed platform. Our WAF operates in 4,300 locations across 700 cities, intercepting attack traffic exactly where it enters the Internet and long before it can gather onto the target. Possessing an excellent WAF with the necessary defenses for the newest attacks is clearly essential, yet that by itself won't be sufficient in the coming age of AI. WAFs must be deployed across a vast distributed platform. And that requirement gives Akamai a unique advantage relative to the competition. To wrap up, we think demand for Akamai's security offerings has never been greater, since bad actors are leveraging AI breakthroughs. Why? Our enormous platform scale lets us soak up attacks, we have unmatched visibility into live attack traffic, we're deeply connected to the early warning community so we can deliver current defenses against the newest zero-day threats, and we have a big, highly seasoned human security operations group armed with cutting-edge AI tools that sharpen visibility and cut response times. On top of that, our product lineup is inventive, fast-moving, and AI-powered, helping stop intrusions and reduce harm when intrusions happen. Among customers who chose Akamai during Q1 for API protection of that sort were a top telecom group based in Africa, a prominent investment management firm in South America, a premier investment bank in the Middle East, and a leading global fintech business in the U.S. Those who broadened or newly adopted our Guardicore Segmentation offering in Q1 included South Korea's foremost telecom carrier and media company, a major European banking group, and a top U.S. health care firm. A number of the big renewals we closed in Q1 came with added security services too. Take one of America's top retailers: once we shielded them from unwanted bots through the holiday shopping period, they ramped up their usage under a contract valued at $24 million. We concluded an expansion deal valued at $80 million spanning 2 years with a video game company that ranks among the world's biggest; we also concluded an expansion deal valued at over $20 million with a consumer electronics firm operating globally out of Korea. Meanwhile, a professional services company counted among the largest worldwide broadened its adoption of our ZTNA offering to safeguard remote access at scale as it shifts critical applications toward a 0 trust model. Recognition from the big analyst houses keeps coming our way for how well our security solutions perform. Case in point: during the prior quarter, we earned a 99% recommendation score as customers' choice in Gartner's Peer Insights report covering micro segmentation; and in the past month, we stood alone as the sole provider named Customers' Choice in Gartner's Peer Insights report covering API protection. To finish, we're delighted that our growth strategy has gained traction and is producing transformative possibilities for our company; we hold the view that Akamai occupies a singular position to power and profit from an AI-driven economy as it develops. By delivering robust compute right at the edge, close to data and users, Akamai is powering and protecting the coming wave of agentic AI; quarter after quarter, the enormous chance before us grows clearer, and we're investing aggressively to seize it and let Akamai accomplish for cloud and AI what we already achieved in security and CDN, driving substantial growth for our business down the road. At this point, I'll hand things off to Ed, who will share additional detail on how we performed and what we expect for the second quarter and the full year. Ed?

Edward McGowan

executive
#3

Thanks, Tom. To pick up where Tom left off and before I begin, I'd like to share directly how thrilled I am about the new customer deal worth $1.8 billion that we announced today; it's a strong testament to what Akamai offers in the AI era and shows plainly the size of operations we're capable of. To seize this momentum fully and back the faster growth we're expecting, our spending will run a bit ahead of revenue, and you'll notice that in the revised capex and operating margin guidance I'll walk through when I get to that section; in our view, these outlays in the CIS portfolio matter greatly so that we're positioned with the right foundation for the substantial demand coming our way. In addition, thanks to the $1.8 billion win we announced today, plus the 4-year, $200 million CIS agreement unveiled last quarter within our quickly expanding pipeline, our expectation now is that total company annual top line revenue growth will hit double digits come 2027; we're eager to provide further specifics over the next few quarters. Plainly, this marks a remarkably thrilling moment for Akamai. So with that, let's turn to the Q1 results. Our first quarter performance was solid, with total revenue reaching $1.074 billion, an increase of 6% versus a year ago as reported and 4% in constant currency. Cloud Infrastructure Services, or CIS, revenue began the year strongly at $95 million, climbing 40% year-over-year as reported and 39% in constant currency. As Tom pointed out, CIS wins are coming from a broad range of industries, geographies and use cases. What's even more heartening is that our pipeline for AI-specific use cases is expanding quickly. We kept up very robust momentum in security as well, with revenue of $590 million, rising 11% year-over-year as reported and 9% in constant currency. First quarter strength remained fueled by our rapidly growing API security and Guardicore Segmentation offerings, together with solid growth from Web Application Firewall, our biggest product. Turning to delivery and other cloud applications. Revenue came in at $389 million, a decline of 7% year-over-year as reported and 8% in constant currency. These figures matched expectations, reflecting the wraparound effect of the Edgio transaction in 2025. We anticipate that this impact on the rate of decline will ease as the rest of the year unfolds. International revenue totaled $530 million, an increase of 9% year-over-year, or 5% in constant currency, accounting for 49% of total revenue in Q1. Currency movements gave revenue a boost of $2 million sequentially and a favorable $19 million year-over-year. Now on to profitability. During Q1, non-GAAP net income came to $239 million, or $1.61 in earnings per diluted share, a drop of 5% year-over-year as reported and in constant currency; these figures reflect our larger colocation investments, greater depreciation and higher head count expenses, all connected to our strategic investment in cloud infrastructure services in the first quarter. For the first quarter, our non-GAAP operating margin came in at 26%, which matched what we had anticipated; looking ahead, we anticipate that margin will stay within that same band through the rest of this year, since we're stepping up spending to seize the exciting growth opportunities in front of us. CapEx in Q1 totaled $206 million, equivalent to 19% of revenue. That first-quarter CapEx came in a bit under our guidance, largely because of timing as well as favorable pricing. To be specific, certain expenditures moved out of Q1 and into Q2, and we also gained from component costs that came in lower than we had expected. Turning to cash and how we allocate capital. In the first quarter, we used roughly $206 million to repurchase about 2 million shares. We closed the first quarter with about $975 million left under our existing repurchase authorization. Our approach to capital allocation hasn't changed: keep repurchasing shares over time to offset dilution from employee equity programs, and stay opportunistic on both M&A and share repurchases. As of March 31, our cash, cash equivalents and marketable securities totaled approximately $1.7 billion. Before I move on to guidance for Q2 and for the full year 2026, I'd like to address a few housekeeping points. First, CapEx for Q2 should rise considerably, since we begin taking delivery of the NVIDIA GPUs that we mentioned on last quarter's earnings call, and we also make up for some CapEx that slipped from Q1 into Q2. Second, operating expenses should increase in the second quarter, mainly because of ongoing investments in go-to-market plus the effect of our annual employee merit cycle, which took effect on April 1. Third, we expect revenue from the $1.8 billion customer win to begin ramping in Q4, and we expect to generate roughly $20 million to $25 million of revenue during the fourth quarter. Lastly, on CapEx related to this win. We expect to spend a total of approximately $800 million to $825 million across the next 12 months in support of this customer. Of that total, we expect to deploy roughly $700 million during the second half of 2026, with the rest landing in the first half of 2027. Now on to guidance; for the second quarter, we project revenue between $1.075 billion and $1.1 billion, which is up 3% to 5% as reported and in constant currency versus Q2 2025. At current spot rates, foreign exchange movements should have no material effect on Q2 revenue relative to Q1 levels, and a positive $2 million effect year-over-year. At those revenue levels, we expect cash gross margins of approximately 70% to 71%. Gross margin is being affected by the substantial rise in colocation as we speed up growth in our CIS business. Non-GAAP operating expenses for Q2 are projected at $346 million to $357 million. For the second quarter, our EBITDA margin should land around 38% to 39%. Non-GAAP depreciation expense is projected at $144 million to $146 million. We're looking at a non-GAAP operating margin of roughly 25% to 26%, and given the revenue and spending picture I just walked through, Q2 non-GAAP EPS should come in between $1.45 and $1.65. That EPS figure bakes in taxes of $47 million to $54 million, which works out from an estimated quarterly non-GAAP tax rate of about 18.5%, and it also assumes a fully diluted share count of roughly 146 million shares. Now on to CapEx. Because of what I mentioned earlier, our second quarter spending should be about $433 million to $453 million. That's roughly 40% to 41% of total revenue. As we look toward the whole of 2026, revenue should reach $4.445 billion to $4.55 billion, representing growth of 6% to 8% as reported and 5% to 8% in constant currency. On cloud infrastructure services, we're lifting our guidance to at least 50% year-over-year growth in constant currency. Momentum in DAS should keep building across the back half of 2026, primarily from the scaling of our AI opportunities plus the effect of the 2 very large transactions we announced in Q4 and today. Beyond that, security revenue growth should remain in the high single digits on a constant currency basis in 2026. And for delivery and other cloud apps, we still foresee a mid-single digits year-over-year decline on a constant currency basis. Based on current spot rates, our guidance assumes foreign exchange will boost revenue in 2026 by a positive $20 million on a year-over-year basis. Switching to operating margin. For 2026, we estimate a non-GAAP operating margin of about 26%, measured at today's FX rates. Now to CapEx. At this point, we expect full year capital expenditures to be roughly 40% to 42% of total revenue, and that includes the $700 million impact from the $1.8 billion contract we referenced earlier. Before moving on, I'd like to add some more color regarding our CapEx outlook. As Tom pointed out, the demand we're seeing for CIS, our GPU deployments included, is exceptional. Our present GPU pipeline is well above both existing and projected inventory, which means we could place more GPU orders in the second half of the year to satisfy that demand. Our current annual CapEx guide does not include this; should we place another GPU order before year-end, we'll update CapEx guidance on a later earnings call. Now to EPS. For full year 2026, non-GAAP earnings per diluted share should fall in the range of $6.40 to $7.15. This EPS guidance incorporates the impact from the very large win; the non-GAAP earnings guidance rests on a non-GAAP effective tax rate of about 18.5% and a fully diluted share count of roughly 147 million shares. With that, I'll close things out; and Tom and I are glad to take your questions. Operator?

Roger Boyd

analyst
#4

A question, and congratulations on that landmark deal. If possible, Tom, could you give us the broad strokes regarding the sort of competitive field involved in securing that deal? Are you going head-to-head against our hyperscalers, or is it neo clouds? And whatever you can share about the type of use case — is it inference? Is it agenetic workloads? Also — regarding your compute-enabled POPs, in what way is this customer making use of the Akamai network overall?

F. Leighton

executive
#5

Right. Regarding this particular deal, I'm unable to share further specifics. Broadly speaking, though, yes, when it comes to our cloud infrastructure services, we do find ourselves competing against both the hyperscalers and the neo clouds. That's the main source of competition; they choose Akamai owing to our demonstrated capacity for managing and scaling intricate distributed systems, our capacity to secure the required data center space and locations worldwide in order to interconnect that with the globe's largest and highest-performing delivery network along with leading security solutions; on latency and scalability, we deliver the best. With a presence spanning 4,300 locations throughout 700 cities and 130 countries, we likely work with a greater number of data center companies than anyone else. So yes, the competition we face is considerable. There's competitiveness in every deal. Yet we possess distinctive capabilities as well, which I believe explains why our pipeline is so robust and why we're securing some extremely large deals.

Roger Boyd

analyst
#6

Great; and now regarding security, I'd like to hear you break down what you're observing on the demand side. A strong outcome in Q1. What are you noticing when it comes to conversion rates and sales cycles? Is there more urgency coming from organizations considering how to reduce the blast radius and protect themselves against an AI driven threat environment.

F. Leighton

executive
#7

Yeah. I doubt I've ever witnessed CISOs this worked up or carrying such a feeling of urgency as they do right now. During the past several weeks, a couple of months, I've gotten to sit down with plenty of CISOs from the largest companies globally; often, the CEOs and senior executives too, and they're deeply worried about what unfolds once attackers obtain advanced AI with the newest AI frontier models, which it appears they will. That will expose far more vulnerabilities. We'll witness the equivalent of far more 0 days, and at this moment they are genuinely scrambling in many instances to ensure every one of their applications, their agents, their APIs is protected by Akamai. And as you might expect, most of the world's major banks depend on us for security, and they see a fairly large wave of new attacks heading in their direction. So this is — I can't think of a similar period where there's so much anxiety about what will occur with security and simultaneously so much recognition of what Akamai delivers through our security platform.

Patrick Colville

analyst
#8

This next question is for Dr. Tom. When I look at Akamai, what has made it valuable for over 30 years is that distributed setup — 70 cities, 130 countries. With this huge deal, does that use case need to be highly distributed, or should we view it as running out of just a handful of data centers, like under 10?

F. Leighton

executive
#9

I can't discuss that recent deal. That said, when you consider what Akamai brings to the table, you've zeroed in on something essential — our distributed architecture really has no equal. I mentioned several use cases during the prepared remarks. And yes, those depend heavily on our distributed platform, because you want agents and applications and business logic situated near users and near the data, which delivers low latency and stability — especially for anything involving video processing or video generation, which even calls for a great deal of scale. Akamai stands alone in that regard. So I'd say without question, what we can provide is extremely compelling.

Patrick Colville

analyst
#10

Yeah, and, I suppose, a congestion follow-up, if I may. Ed, you brought up this somewhat painful idea that the CapEx guidance could need to go higher.

Patrick Colville

analyst
#11

Appreciate it. This follow-up, I think, is directed at Ed. Ed, so you handed us the CapEx guidance — yet after that you made this sort of understated remark that CapEx might have to rise even more. Could you walk us through the finer details of why CapEx could go up partway through the year? And I suppose what the implications of that would be.

Edward McGowan

executive
#12

Right. Thanks for asking, Patrick. What I said was that for the platform we have an extremely, extremely robust pipeline. And right now we're only beginning to bring the majority of those chips online, and that pipeline is really big. It's bigger than the inventory we hold. So clearly, we want to pursue that pipeline, begin closing all those deals, turning it into contracts, and so on. And the reason I hedged somewhat is, first, obviously we won't fulfill that entire pipeline. But second, obtaining the chip takes some time. So even if we placed an order, it could slide into the following year. So what I'd like to do is give it one more quarter and, if we actually are able to put in an order and get it before the year ends, we will definitely do so and tell you all. To me that's a very bullish thing to say. And once more, I simply didn't want to show up and shock you with another, whatever the figure is, a couple hundred million or whatever that order turns out to be, without at least providing a bit of context for it.

John DiFucci

analyst
#13

I'll start with a question for Ed, and then I have a quick follow-up for Tom. I appreciate all the CapEx detail. But when I look at the CapEx tied to this mega deal — and I think this is where Patrick was heading — this spans a long period, correct, 7 years? Are you on the hook, for instance, right now we're seeing memory costs running higher than we'd have expected maybe a year ago. When you locked in this deal, did you also lock in the supply? Or does that leave you exposed if something like that were to occur, say, 2 years from now, with prices climbing again?

Edward McGowan

executive
#14

Yeah, thanks for that — it's a good question. I was lucky enough to be heavily involved with the team on both ends of this deal. So, yep, we've gotten the supply chain lined up; we expect that everything we need to provide this service across the 7 years will arrive over the next 12 months, and of course, you noticed how the CapEx was split out, with most of it landing this year. So we expect a big chunk of it to come in. Now, slippage and delays can always happen. That said, our contracts include mechanisms to handle it if, for instance, 6 months down the line, prices were to rise. So that's something we factored in. We have that piece handled. When you look at this deal from the revenue angle, the way to understand it is that we're putting in a fixed amount of capacity, and there isn't any usage component attached. It's simply a usage — pardon me, a committed arrangement spanning 7 years. That means once all of that capacity is fully ramped, we'll begin recognizing the revenue on a full-year basis. As I mentioned, I'm expecting a small contribution this year. Then in the following year, we'll see a partial year as the rest of what needs to be deployed comes online. After that point, it continues on through the remaining 6-plus years.

John DiFucci

analyst
#15

All right. So in a way that ends up appearing — despite the fact that what we're talking about is consumption, it will sort of come across as a subscription. Would you say that's right?

Edward McGowan

executive
#16

Precisely. Yeah, that is precisely how I view it.

John DiFucci

analyst
#17

Fantastic. All right, that's great; and Dr. Tom, video streaming makes up a piece of what your delivery segment does. Also, back in March, we observed that open AI confirmed they had discontinued their AI video generation platform, Sora. I'm just wondering, do you anticipate that will have any impact on your delivery or compute business projection?

F. Leighton

executive
#18

Nope. We work alongside OpenAI regarding security vulnerabilities, assisting in locating them and safeguarding our customers against the related attacks. However, open AI isn't a customer of Akamai, and never has been. So indeed, zero effect on us whatsoever.

Aidan Daniels

analyst
#19

Aidan Daniels here, filling in for Jackson Ader. Given this sizable agreement, when you're distributing capacity down the road, how should we consider what it means for whatever on-demand GPU capacity you can keep offering later on? In other words, how do you strike a balance between what's already locked in through this deal and preserving room for newer, incremental demand that comes along?

F. Leighton

executive
#20

No; access to our platform is available either on demand per token or per VM hour. Beyond that, we also handle large tranche deals. So at this stage, it isn't truly a question of choosing one over the other. Should we require additional GPUs — and as Ed mentioned, that could very well happen — we would buy more.

Aidan Daniels

analyst
#21

Great. Then I have a single brief follow-up question. I realize there's a limit to what you can share regarding the deal. But how should we approach thinking about the split between CPU and the GPU inference cloud moving forward? Is there some kind of framework we could use when considering this deal?

F. Leighton

executive
#22

Regarding this deal, we're not able to comment. That said, broadly speaking, when it comes to inference and AI, you truly need both. And one aspect of the value we deliver is our ability to supply whichever computational resource fits the workload best, which could be CPU, could be GPU, since being as efficient as possible is the goal. Plus, you want it located as near to the user as you can manage, so performance ends up optimal. So it's a combination, and the blend of CPU versus GPU required varies from one application to the next.

Fatima Boolani

analyst
#23

Let me start with something broader on strategy. Rather than relying mostly on renting out capacity as it's needed, you've chosen to lean toward dedicated capacity to meet demand and address some of the supply tightness out there. I'd like to understand the reasoning more deeply, mainly because the equity rates and the market rates for what could otherwise be almost purely a rental or GPU-as-a-Service operation are considerably more appealing. So I just wanted to hear the rationale, the thinking, and how you weighed the choice to tilt the network and the platform toward bigger customers, extended commitments, and greater dedicated capacity? And I have one more question after that, if you don't mind.

F. Leighton

executive
#24

Actually, we pursue both. And in many respects, the larger, bigger deals with long-term commitments are the more appealing ones. You get the commitment; and yes, on the big deals the pricing would come in lower, but we also back the on-demand model, where you can purchase by the token or by the hour and the pricing runs a bit higher, though there can be greater costs tied to that, such as bringing the customer on board if you have a rep working the account. Still, both are appealing and we back both. So it isn't a case of us picking one over the other.

Edward McGowan

executive
#25

The single point I'd toss in at this moment, simply to chime in briefly. It's the customers who are genuinely pushing this. When I look at what's in our pipeline, many of our customers are after dedicated capacity — a set number of GPUs, that kind of thing — since the marketplace has a scarcity problem. So instead of paying on a consumption basis, they're able to secure somewhat better pricing and reserve that capacity for themselves. So honestly, it's a market-driven phenomenon more than anything else.

Fatima Boolani

analyst
#26

Thanks for that. And Ed, while I've got you here, you gave us a pretty clear signal that if the pipeline keeps expanding and evolving the way it is — and evolving in a positive direction the way you're watching — you'd be more than willing to keep putting down CapEx and getting lots brought online and lit up. But I'm curious about where the money and capital would come from to pay for those investments. Do you think you can handle that organically just by operating the business? Or should we be thinking that some other form of capital might get drawn on as you scale up with a heavier CapEx profile for a few of these bigger customers when demand hits?

Edward McGowan

executive
#27

Yeah. So up to this point, we've had no trouble at all funding these build-outs out of our own capital. Clearly we're a very profitable outfit that generates plenty of cash. Sure, during the years when we're pouring a lot into investment, cash flow will come in a little lower. But once the deployment is actually done, these assets throw off incredible free cash flow. That's one of the appealing aspects. We — in terms of cash and equivalents, there's $1.7 billion sitting on the books right now; on top of that we hold a $1 billion line of credit we could draw on if necessary. And beyond that, naturally, our credit is excellent and we'd have no difficulty raising money in the capital markets should the need arise. As of now, we haven't made any announcements on that front. And if we keep landing big deals and require capital, we'll definitely pursue that. But to this point, we've managed by relying on our own funds.

Arti Vula

analyst
#28

This is Arti Vula with JPMorgan, sitting in for Mark Murphy. It's really good to see the traction you're building with those big agreements involving firms at the leading edge of AI technology. You closed a sizable one last quarter, and now another this quarter that makes the prior one look small. So at a broad level, could you help us see it from where you sit — it feels like out of nowhere, you're landing several of these sizable contracts. Has this been developing in the pipeline for some time? Or have things moved somewhat quicker? What has shifted such that so much of this business has arrived at your door, seemingly quite fast from where we're watching. Then, a brief follow-up: as you commit financial and operational resources toward CIS and those sizable agreements, does that alter the way you view the remaining business segments?

F. Leighton

executive
#29

Honestly, this was the plan from the start; so — and we're quite happy to be carrying it out. The aim was to be rolling out a distributed inference platform, a distributed compute platform, that enterprises truly across the board, including numerous major customers, would want. And naturally, Akamai's customer base includes plenty of the biggest enterprises worldwide. And as we've discussed previously, their spending on compute runs 10x or more compared with what they put toward our legacy offerings, delivery and security. So this is precisely what we told you we'd be doing. And at this point we're producing those outcomes. The platform has reached a stage where that's possible. And I believe you'll witness more of it moving ahead.

Sanjit Singh

analyst
#30

Nice work on the major steel and company milestone. About that, maybe this is a silly thing to ask, but regarding that $1.8 billion contract, does it lean more toward a public cloud opportunity? Since I realize a portion of the public cloud business also includes a GPU element? Or was this — was it specifically for Akamai Inference cloud? That was my first question, and then I had a follow-up.

F. Leighton

executive
#31

Right, we truly can't say anything further regarding this specific deal. But clearly, there are plenty of companies where we've signed contracts that we did discuss across the spectrum. And those deals relate to our inference cloud and our cloud capabilities for our GPUs and our CPUs. And it truly comes down to our ability to bring the right hardware for the particular application and have it situated where you get the best benefit for the use of that application.

Sanjit Singh

analyst
#32

Okay, that's reasonable, Tom. My second question relates to the delivery side of things. A number of folks out there are speculating about whether there could be a fresh growth driver for CDN and delivery, given a future where millions — maybe billions — of agents are out there invoking tools, carrying out tasks, running web searches. Has your team gone back internally and reconsidered its thinking on the long-term growth potential for delivery? Or do you still view it mainly as something to mine for profit and gross profit dollars that then fuel the stronger chances in security and compute?

F. Leighton

executive
#33

Yeah, good question. If you consider how agents are multiplying and what lies ahead, the main engine of growth will be the compute platform — the cloud platform underpinning all of it — and we are positioned very well for that. Then there's a major security concern, since AI and agents create an entirely new attack surface; you need not just your web application firewall and API security, but also dedicated protection for AI. So we enjoy a genuine tailwind from STG, our security technology group, on that front. Beyond that, with agents, you must figure out what an agent actually is, who is operating it, and what its intent is, whether you're delivering or safeguarding an app, a site, or another agent. The answer you provide is highly customized to whatever the customer wants done when an agent of this type shows up and engages with you. That's why we've built plenty of capabilities in this area, and they generally sit inside our security offerings. Now, on delivery: sure, some traffic that humans used to create will instead come from agents. Fine. That won't cause a dramatic shift in how many bits you're pushing out. Things begin to shift once agents handle video or produce video — imagine visiting a commerce site and wanting to see how you'd look in a sweater you're considering, and a video is generated of you wearing it; that boosts the site's return and drives substantial traffic. We're only in the earliest phase of witnessing such use cases, and they're being tested right now, which could mean additional traffic for delivery. Still, the largest effect on us lands in the cloud business, and after that in security. Delivery matters greatly, works very synergistically with our overall platform strategy, throws off significant cash for us, and we're directing much of that cash toward expanding the cloud business.

Mike Cikos

analyst
#34

Nice work on the solid quarter and landing that customer. I just want to confirm I've got the mechanics of this arrangement straight. So you committed to 7 years at $1.8 billion. Should we expect the entire $1.8 billion to appear in RPO? Or does that figure also cover potential renewals? Is it entirely take or pay? Just trying to be certain we grasp how the deal works.

Edward McGowan

executive
#35

Yes, of course; I mentioned this briefly before, and someone followed up about capacity — dedicated capacity as opposed to paying by the hour. This leans toward the dedicated capacity side. So once the capacity is stood up, we'll recognize the revenue ratably across the contract term. Like I said, we'll book some revenue this year and a partial amount next year — not a full year next year, since we're still building and bringing the capacity online. Regarding how RPO gets accounted for, we'll capture the bulk of it in the following quarter. And once everything is delivered, it will ultimately all land in RPO. There are just a few unusual mechanics tied to the first 12 months and how we're handling, how we're taking receipt of the goods, and we discussed a pricing mechanism to address whether prices rise or fall, that kind of thing. So there's a bit of subtlety involved; but after we have this fully operational, you'll find it reflected in our RPO. Some portion will come next quarter, and from there it will grow.

Mike Cikos

analyst
#36

Thanks for walking through how that works. And then for Dr. Leighton, just so I'm sure I've got it straight too, and it's really good to hear that your biggest security offering, the one with WAF, is growing more strongly, which I honestly didn't anticipate. Could you dig into that once more in terms of what's behind it? Is it truly this elevated threat landscape we're sitting in? Or is there another factor at play?

F. Leighton

executive
#37

Yeah. As you're aware, there have been genuine breakthroughs in AI, and it's becoming far more adept at uncovering weaknesses and assisting the attacker in seizing control of devices and breaking into enterprises. And our protections are required today like never before. There are billions of devices out there that simply can't be patched. And these days the adversary is able to discover paths into those devices and take them over. Consequently, we're witnessing attacks on a scale far beyond anything previously seen. Quite literally, attacks at the application layer originating from millions of distributed IPs, with millions of strikes hitting a target every second. And defending against that is impossible with merely a WAF sitting in a data center, or anything similar. You require the WAF platform that we offer in order to intercept all of that traffic and handle it, since you have to distinguish the malicious traffic from the legitimate traffic, and the volume of malicious traffic right now is enormous. So our platform, that physical infrastructure, is more essential today than it ever has been for our security services. And our customers understand this. And there's a heightened sense of urgency at present, because they realize the attacks are growing more sophisticated thanks to AI, and bigger in scale since the bad actors can take over all of those devices and launch their attacks from far more places. And that's the reason we're seeing something like our web app firewall all of a sudden in much greater demand; AI does assist on the defensive side, yet it doesn't fix that issue. And so all things considered, this is an extremely difficult moment for CISOs, and that's why they're coming to us to ensure that everything they've got is safeguarded by Akamai.

Mike Cikos

analyst
#38

Hearing about that halo effect is really encouraging. And once more, congratulations on landing such a significant customer.

Frank Louthan

analyst
#39

Yeah. I just wanted to ask a follow-up regarding the $1.8 billion and the way it gets booked. Does every bit of that show up as revenue? Might some portion be treated as upfront CapEx that's been paid? Or anything along those lines? Then a second follow-up from me: at present, how many locations has the Inference Cloud been built out to? And what does the plan look like?

Edward McGowan

executive
#40

Absolutely, I'll handle the first portion, and Tom, you can field the second. Yes, it's entirely revenue. Nothing gets offset against CapEx or anything of that sort, so the whole thing will be revenue.

F. Leighton

executive
#41

Absolutely. On the latter portion of your question — with Inference Cloud, every one of our 4,300 sites is covered. Functions-as-a-Service operates in a serverless fashion across each of those 4,300 sites. Our managed container offering is live in considerably more than 100 cities, and it could in principle be deployed across all 700 cities, though today it's active in well above 100. Full IAS capabilities exist in several dozen cities, and a couple of dozen among them come equipped with the new 6,000 GPUs. Naturally, the aim is for everything to be orchestrated such that any application or agent requiring execution lands on the resource that's most computationally efficient. Should it be possible to handle it on an edge server using the CPU already there — wonderful, quick, extremely inexpensive. If a container in the same city running on a CPU can do the job — excellent. If a cluster of GPUs at one of those couple of dozen sites is what's needed — fine. Once more, the point is to place it on the most efficient resource, near the user, and already prepared to operate. You don't want to have to spin it up in response to a request; and that's what our orchestration layer is designed to make possible. And this is how it fits in with the vision from NVIDIA with the AI grid, you think of AI like you would an electrical grid. And that's what Akamai is building.

Will Power

analyst
#42

Great. Congratulations from me as well on this enormous deal. Maybe just a few questions. First, to clarify — Ed, when you say more GPUs are needed, is that extra supply required so we can deliver on the new deal? Or is it tied more to what's in the build pipeline? And — I realize we don't know exactly when those GPUs could become available — but is there a concrete number for what we're talking about in terms of total cost? Then I have another question.

Edward McGowan

executive
#43

Yes, sure. So basically, if you go through the prepared remarks, we covered how all the CapEx required is included in the guidance to meet the $1.8 billion. That point stands apart from what I said about buying more GPUs. That purchase really came down to our pipeline performance and how fast we could move on it. And once more, you never know whether they'll actually arrive on schedule. We'll share more details later. It all hinges on the demand we're observing. Our pipeline looks extremely robust, with some really big opportunities — certain customers are looking to begin with a few hundred GPUs, while others want to kick off with 1,000 or more. Honestly, the possibilities are spread everywhere you look, and that list keeps expanding daily, which is fantastic. We'll give you a proper estimate down the road. The prior figure came in at roughly $250 million in CapEx. As for how large I expect this to get, I can't offer any guidance yet — though ideally what I'm conveying is that it's going to be a massive figure, since the demand on our side is substantial.

Will Power

analyst
#44

All right, yes. Is there a way to give us a sense of how you're viewing the effects on gross margin and operating margin? I realize 2027 seems like it's still not a full year. When you look ahead to 2028, how should we think about the way this shapes the overall financial model compared with — perhaps compared with where things stand right now?

Edward McGowan

executive
#45

Sure. Let me address this at a fairly high level. Consider some — particularly a few of those bigger deals, the type where somebody approaches us and says, for instance, I'd like 1,000 GPUs, as is the situation with this major customer, I'd like a set amount of capacity stretched across a lengthy stretch of time. Depreciation across that stretch is the largest cost factor in that situation. The expenses feeding into your cash gross margin are considerably smaller, correct? Those are your co-location expenses, possibly some bandwidth or networking expenses and similar items; and at times, there are certain people expenses. Yet broadly, these tend to scale quite nicely. So the expectation over time would be that your cash gross margin has room to get better. Of course, there's some give and take involved in bringing colo online ahead of anticipated demand, and you must bring it online before anything else and so forth; so it's going to take a while for this to work itself out. Even so, you ought to see your cash gross margin widen somewhat. Your EBITDA margin widen somewhat; and after that, when it comes to operating margin, the mix truly is what matters. We're prepared to take on a number of these considerably bigger deals, possibly at margins that fall below the 30% operating margin. Looking at the — without question, the GPU by the service rented GPU, far above the company operating margin. You're going to achieve substantial scale throughout the OpEx as we bring on certainly bigger customers. So our focus across the coming year or two is going to be squarely on seizing this growth. So right now we won't be at a point of margin expansion; but eventually, that will come about on its own and your free cash flow margins will get better and so forth; so that's roughly how we're approaching it, though we're genuinely thrilled about pursuing this growth opportunity. We will keep investing in order to capture it.

James Fish

analyst
#46

Okay, based on what you've previously said about power, where the big site uses — I believe you mentioned this, Ed — 5 to 10 megawatts and the smaller ones use only a portion of that, that puts you past 300 megawatts. So at this point, you don't bring in enough revenue online for this; so what portion of that power serves non-compute services? And is that the reason you need to add, from what I can gather, about another 40 megawatts just for this one deal? It seems sort of enormous, and you folks ought to be able to handle this if all that power goes to compute. So could you just take us through where you stand regarding megawatts and what the plan looks like by the end of '27?

Edward McGowan

executive
#47

Yeah. So I didn't fully track all of that, but let me begin by stating that your math regarding what delivering this specific deal would require isn't correct. I'll just stop there; it's much lower than that; regarding our capacity, we have — the big — if you consider what consumes the megawatts of power we possess, the CDN and security business accounts for a small portion. Think of that as being kilowatts in certain cases, perhaps a megawatt or 2 in some of the larger CDN deployments; so running the CDN business doesn't demand a ton of massive power. As for the compute business, it's far larger, particularly when you have customers wanting, for instance, a few thousand GPUs at one particular location, or they're spread across, say, 20, 30 locations, and they've got plenty of CPUs. So you do tend to see far more power demand there. And what I've mentioned is our typical deployment for some of our bigger locations, we discussed having 40 core compute locations. And those fell in the 5 to 10 megawatts range, expandable to, say, 20 to 30, and again, it truly depends. And we can go somewhat bigger than that, but there's ample opportunity for us to secure additional colo. We expect to light up far more going forward here. So if the worry is that we lack access to sufficient power or can't obtain colo, that isn't a worry for us whatsoever at this moment. We — as Tom mentioned earlier, we've got fantastic relationships, and we're a highly appealing client for some of these data center providers. We've got outstanding credit. We're not a do-it-yourselfer like a hyperscaler. We've got far better credit than, say, some of the neo clouds, and we do take substantial chunks of colo in certain cases and actually assist some of our colo partners in building out. So I'm not worried about that whatsoever. And the power dynamics differ for each one of the different products. GPUs, you know, they draw more electricity than CPUs do. That means there's arithmetic involved, and I've already laid out that arithmetic. On the CPU side the numbers come out a lot smaller, since the power draw isn't as heavy. Beyond that, the kind of gear you happen to be running also plays a fairly big role in how much power gets used. Some — we keep seeing plenty of fascinating hardware vendors roll out products that are amazingly power-efficient. So honestly, giving you a precise answer to that is tough, but just understand that we have zero real worry about securing sufficient power to run these things, and the power needs differ quite a bit. Still, we build that consideration into every deal we sign and make sure we won't accept anything that doesn't make money.

James Fish

analyst
#48

Understood. That tracks; and only to build on the security part, usually you provide us an API and Zero Trust as opposed to sort of the core. I suppose, what was the direction of that? And then what was the direction of compute during the quarter between — with enterprise.

Edward McGowan

executive
#49

Yeah. So on security, API and Guardicore weren't broken out by us, we did state it accounted for most of what's fueling growth; and I'd simply say the growth rates resemble those from the prior quarter. Just keep in mind, the prior quarter included a decent amount of license revenue. So once you strip that out, it's essentially comparable, expanding at about the same rates when you remove the effect of the license revenue. So still extremely, extremely robust growth rates there. And then regarding compute, you brought up enterprise compute. That isn't broken out by us in that manner, really, the way to think about enterprise compute is CIS, which gets broken out on its own, and we do give what we previously called our application services, which sits within the third bucket, the delivery and app services. So that figure is broken out for you; so 40% growth was for CIS year-over-year, and we anticipate that will accelerate.

Jonathan Ho

analyst
#50

I'd like to add my own congratulations too. I've just got one question. Considering the kinds of mega customers you're onboarding onto your platform, does that create a bigger chance to upsell them after they've joined, or are there possibly extra services? Or might they return to the well should they keep expanding their growth further?

F. Leighton

executive
#51

Yes. As you're aware, demand for AI is climbing quickly; we're truly at an early stage there. And yes, I'd anticipate there's lots of space to expand the current base and, naturally, bring in additional customers at that scale.

Jeff Van Rhee

analyst
#52

That question's pulling me into the mix. Two things, quickly. First, Tom, maybe there's been a lot of national backlash against AI data centers and the electricity usage tied to them. As you move into transactions this big, how do you plan to avoid getting caught in the crossfire of that kind of community-level resistance to the wider AI compute landscape, so to speak.

F. Leighton

executive
#53

Right, I doubt our profile in the mainstream media resembles that of the massive hyperscalers at all. So I really don't see that as a concern on our end. So yes, that's not something keeping us up at night at this point. Perhaps it's even a nice dilemma to face once we've grown far, far bigger than where we stand now.

Jeff Van Rhee

analyst
#54

All right. Then, moving to the second point — regarding security, with what was said about AI turning into a tailwind in that area, do you figure this year probably marks the bottom for the growth rate, meaning we ought to expect perhaps a pickup once we reach '27 and afterward?

Edward McGowan

executive
#55

Time will tell. Look — we provided guidance covering the year. Clearly we're quite happy with the results from Q1. There's — we genuinely like what's showing up, particularly in API security, — very early stage still, penetration is low. Guardicor keeps expanding at a very steady clip. So yes, we'll monitor how things develop, and we'll bring you updates along the way.

Operator

operator
#56

That brings the Q&A portion to an end. Today's conference call is officially over. We appreciate your participation in today's presentation. You're welcome to disconnect 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.