Akamai CEO：AI基础设施投资将推动Q4及明年增长

Don't have to rehash all of the numbers from yesterday, but I do just wanna get to this idea. I mean, the revenue growth is there. The profit growth isn't, but that seems to be somewhat deliberate as you're investing a lot back into this business. Why? Yeah. The profit growth follows the revenue growth. So we're investing a lot now to build out the infrastructure, and then that's followed by the revenue and the profit growth next year.

These big deals we're doing all in have margin operating margins from the low to mid twenties to low thirties, very consistent with the company profile, drive a lot of profit growth starting next year and into the future. Well, what you said on the call, I think, did no acceleration in this quarter, q three, but a hockey stick Yeah. For q four? Yeah. Really? Well, you know, cloud infrastructure services grew 39%, so not exactly bad.

No. No. And we think for the year, gonna be 50. So a big you know, we'll start seeing the revenue from these deals hit in q four. And then next year, we think it accelerates further enough to drive the whole company into double digit growth. Some of the concern and some of the reasons why you've seen, the shares down, you suspended the buyback to help build out this buyout. And based on the numbers I saw, you're guiding CapEx at basically about 40 plus percent of revenue, which is high for a company of your size.

Is it not? Well, yeah. And we're making these investments to drive all that revenue and profit growth starting q four but into next year. And remember that this year, we've already bought back a lot of shares, more to offset the equity dilution plus additional. And over the last fifteen years, we've probably net reduced our share count by 15 to 20%. Well, I'm curious about that. I mean, obviously, you Akamai, a pioneer in sort of this idea of, content delivery networks.

I mean, you know, we always kinda say, you know, you're you're kind of the the hidden, sort of, juice behind behind we're entering this new phase with AI, and we talked about the deal at the top here, which I think focuses a little bit more on AI robotics. What are you trying to build now? Is this I I I mean, I know the CFO has talked about this being sort of a once in a generational opportunity. Are you making a big pivot here?

Yeah. We're gonna do the same thing Yeah. For AI that we did for content delivery and then cybersecurity. And think about robotics. You got a a fleet of robots or devices, might be cars. First, they got all sorts of sensors. They gotta look at what's going on around them Mhmm. And feed that back in, and and that's a lot of video. And if you're doing this at scale, you need to do that, you know, more distributed fashion just like, you know, delivering a sporting event.

You can't do that from a big data center. You gotta do that in a distributed fashion. And then you need the AI to process that video and issue instructions to the robot or the device. And if it's a car or if it's a robot interacting with a human, you want that to be fast. You don't wanna have latency. And so doing it in a distributed fashion like Akamai is uniquely capable of doing is great for performance and for cost.

Well, forgive me for for, being a little ignorant of this, but are you doing something that the the hyperscalers that are building all these data centers that they can't do in terms of fulfilling that? Yes? Exactly. I mean, that they're very good at training the big models. Mhmm. You know, we are very good at training the medium sized models and doing the inference, actually executing on it. In fact, all the hyperscalers use our cloud infrastructure services for the kinds of applications where they wanna have the logic, the compute logic be close to the user for performance.

Well, talk about sort of this acceleration there. I mean, I mentioned the hockey stick and all that, but give me the sense here of the demand that you're seeing, the capacity that you have or you think you'll have to meet that demand. Is that in sync? Yes. Of course. Yeah. We wouldn't take on a deal that we didn't have the capacity to fulfill it. And it is a hockey stick because this year, we've signed up already 2,800,000,000 in multiyear commits for the cloud business, really, you know, just taking off.

Well, I mean, is there a concern about customer risk? I mean, when you say commits, I mean, mean, these companies can't pull back. And for some reason, they feel like, okay. We're spending too much and, you know, maybe we do that. But you're confident that the pace of spending that we're seeing, whether it's Akamai or to the broader space, is going to continue? These are committed contracts. I mean, the only way out of that is the company goes broke.

Mhmm. I don't think that's likely, in these scenarios. And even if that in that crazy case of it happened, what we're buying, the assets, still have tremendous use. Mhmm. You know? So the in the case of GPUs, we are still selling at capacity, the GPUs we had six years ago, at a higher price than we sold them for six years ago. So so when you hear people criticize this AI build out and try to make parallels to the telecom boom and bust, what do you say to that?

Well, I think the concern there really is more around these gigantic data centers and training the gigantic models with the latest hardware. Mhmm. That is not what we're doing. We are doing modest size build outs for the enterprise use of AI. That's not going away. That's gonna keep growing. What about your cost? I mean, we we've talked a lot on the show, you know, price of memory chips, price of everything, really, to build out, this AI infrastructure is through the roof.

How are you managing that? Well, in some cases, we've raised prices. And when we take on these long term deals at the large commits, that's all factored in. So we've already locked down what our costs are in the contracts of certain cases. If there's a variable cost, that is passed through so that we're protected. When you're assessing the customers that you're taking on and more importantly, the concentration of customers, we've seen with other companies, you know, a huge chunk of their revenue is coming from two or three players.

Right. Does is that a concern for you? Do you think that the breadth of customers that you will have will be there? Oh, yeah. We're, you know, very distributed. We have a small handful of customers over 1% of revenue. Okay. Collectively, all the big ones, less than 20% of revenue. Mhmm. And and those are the name brands that I I don't think people worry about having financial problems. My last question, this is less Akamai specifically and just more about what you've seen over the years in your career.

I mean, what are you excited about most when it comes to just everything going on right now with technology and the way it's being integrated into, our society? It's an exciting time, in good and in bad ways. And for us in two dimensions, you know, driven by AI, because it's changing everything. And a fast moving landscape, we're we're in a great position to help enable that. The downside is AI is used to for cyber attacks.

The bot armies that we're seeing, probably 10 times larger now than they were a year ago because the adversary is using AI to take over devices. The penetration rates for enterprise is much higher, gonna get worse, you know, as you have more AI in the hands of the attackers. They find more zero day vulnerabilities. Now, again, that's a tailwind for our business cause we protect pretty much all the world's major enterprises.

We see that zero day first Yeah. Put rules in the defenses to protect our enterprise customers. Tom.