저자: JimCramer · 외국인
요약 보기
AI 에이전트 트래픽 급증으로 CDN 기업(Cloudflare, Akamai, Fastly)이 수혜. Cloudflare는 고평가라 조정 대기 권고, Fastly도 비싸 신중, Akamai는 PER 16배 이하로 매력적이라 매수 추천. 엔비디아는 PER 14~15배로 성장 대비 저평가, GPU 수명이 길어 감가상각 우려는 과도하며 인프라 수요가 공급을 오래 초과할 것으로 봄.
다듬은 글
and Fastly. Cloudflare shares are up 121% from their February low. Akamai is up 51% from its lows last November. And Fastly experienced a real " roller coaster", but finished very confidently. Now it's up 159% per year. These guys work in a pretty simple industry. They build networks of servers around the world and help manage digital traffic to make everything run faster and more reliably.
But this is not a new business. I consistently recommended Akamai back in 2005 when the show started. And it showed excellent results until the financial crisis hit. Recently, they have focused more on cybersecurity and data collection, which is one of the reasons I continue to recommend Cloudflare.
In general, the thesis about content delivery networks is that these companies thrive when internet traffic increases, and the emergence of AI agents has significantly increased this traffic. I think this is the main reason for the success of these promotions. That's why they're "on the horse" now. But each of these stories is very different.
Let's start with Cloudflare. NET, which is the largest among content delivery networks. Its shares have risen by over 2,200% since going public in 2019. We've invited CEO Matthew Prince on the show many times, and he's been very apt in predicting what the advent of AI will mean for the rest of the internet.
We last spoke with him in June. He said that traffic from AI agents has already exceeded traffic from humans. And how accurately he predicted it. This is the essence of agents. If you're comparing prices on something, you might only check a few sites, and an AI agent will visit thousands of pages to find the best deal.
So, there's a lot more web traffic now, and someone like Cloudflare needs to help us deal with that . how such fantastic numbers with steadily growing revenue over the last year or so. Next up—Akamai. Late last year, they announced a new initiative, the Akamai Inference Cloud, which is essentially a network of many smaller data centers spread around the world that help AI models process information.
Matthew believes we are moving from a period of training AI models to a world where we actually use these models at scale, and that requires a lot more infrastructure. In early August, we had Akamai CEO Dr. Tom Layton on the show for the first time in over five years . I'll let him explain himself what his company is trying to do.
We are unique in that we can support AI infrastructure needs from core to edge . So regardless of your task, agent, or model you're working with— if training is needed, we do it in the core, but the real day-to-day use—is inference when queries are sent to the model . It's like a team work that is better done alongside a robot or user to make it fast, scalable, and efficient.
Since that interview, I've been thinking a lot about this Akamai plan. First, it's quite logical: you need distributed data centers, especially with the development of physical AI, because it requires high bandwidth and proximity of the infrastructure to the user. This reduces latency and increases speed.
Also, lately we've seen large centralized AI data centers become politically toxic , and it seems to me that Akamai's idea of building smaller centers might be easier to push, don't you think? This is an investment year for Akamai: they are spending to acquire and build infrastructure for their inference cloud, which will significantly impact their bottom line.
But management says there is huge customer interest and expects business growth in the fourth quarter and a real boom in 2027. By the way, last Thursday Akamai announced a major contract with Anthropic for $11.6 billion over seven years. Obviously, Anthropic needs additional capacity and is willing to pay for it.
That works out to $1.66 billion per year—quite impressive, considering it's more than a third of Akamai's expected annual revenue. Initially, the stock rose in response, as one would expect, but now it has lost all of its gains and more. To me, this looks like a buying opportunity. Finally, let's talk a little about Fastly.
This is a company I haven't paid much attention to lately. I remember Fastly as a short-lived favorite during the pandemic. The stock skyrocketed at the time , largely because it was a content delivery network for TikTok, which was then rapidly gaining popularity. But then Fastly lost most of its business to TikTok, and the stock plummeted.
For years, it looked like a losing stock that had lost its momentum. They only started making a profit last year. The stock has been showing incredible growth recently— up 159% year-over-year thanks to impressive financial performance. If that wasn't enough to get Wall Street's attention, Fastly reminded us of itself again last week.
First, the company held an investor day last Tuesday, where management explained how Fastly is transforming from a regular content delivery network (CDN) into a much more interesting player in edge cloud computing, fueled by the rapid growth of web traffic from all those AI agents. Second, it happened right when everyone was starting to realize the power of the Meta news platform, which really fueled interest in the stock.
Ultimately, the development of generative AI turned all three of these content delivery networks into big winners. So, did you miss your chance, or can the stock continue to rise? A difficult question. I love Cloudflare, but it's selling for 279 times this year's earnings. On the other hand, the company's shares have always been expensive, and if you let that scare you away, you'll miss out on huge profits.
However, if you want to buy Cloudflare now, I recommend waiting for a significant market pullback. Given the situation in Iran and the actions of the Federal Reserve, I'm sure someone will give you that opportunity. Fastly has a more reasonable valuation, trading at over 50 times this year's earnings, but that's still pretty expensive and the stock isn't usually wildly valuable.
Honestly, this is the company I have the least confidence in. So again , I strongly advise you to wait for a lower entry point. But Akamai, I think, is worth buying right now because they are cheap. They're selling for less than 16 times this year's earnings forecast, and you essentially get the latest partnership with Anthropic for free, which I thought was awesome.
Sure, they're having a bad year because of all those expenses I mentioned, but things should change for the better soon. Here is the main conclusion. A side effect of the AI revolution is that it is already creating an insane amount of web traffic, and someone needs to help us keep the internet running smoothly.
This is where Cloudflare, Fastly, and Akamai come to the rescue. Although at this point, Akamai is the only company that is cheap enough that I can recommend it with complete confidence. I want to take a few calls and start with Linda from Pennsylvania. Linda! Hello, Jim. May the Lord bless you and Lisa.
I feel like you are part of my family, and when you are on vacation, I love all the extra research you provide me. I hope Norman and Bowl are doing well. Jim, I'm calling about an IPO that I've been trading since inception, I made money on it, but it's down 26% since April. I know there are debts and yesterday you had C-C-C-On, Michael and Trader.
So, I'm wondering if I should cash it in and maybe wait for something better, or stay? Well, first of all, tell Linda that I thank you for those kind words, and thank you for what you said about Lisa. You know, she's the one who doesn't let me... you know, she says, " You need to get some rest, Jim." " You really should do this." So that's great advice.
I really like Michael and Trader. The stock has still almost doubled, you know, it's doubled from its IPO price . I have tremendous faith in him. Stocks really lose a lot of money. So the question is, will you have enough stamina? If I were you, I would say, “You know what? I want a small stake in CoreCivic, a small one because I don't like their balance sheet , but that's part of the business." I really like Intrepid though.
He is a straightforward guy. Rob from Florida, Rob. Congratulations, Jim. Hello, Rob. Jim, what are your thoughts on Axon? Thank you. Okay, I like Rick Smith. Anyone who has been watching the show for a couple of decades knows that I think Rick Smith is great. I really think Axon has real competition, they have a good product , but I have to say: when there is competition, it means pressure on gross margins, and with that pressure, the stock cannot be sold at a price that exceeds 54 times earnings.
So the answer is: I'm taking a break from Axon right now. I feel great about Corey, but I was a little too negative. I think Corey Weaver is the one Jensen wants. He considers him the best builder. I'm not going to go into that. Or at least at Traders Point, Jensen says it's the best builder. Since AI is generating all this web traffic, I think these content delivery companies are worth paying attention to.
I like this idea. If you look at Nvidia today and compare it to some of the biggest names in tech, the valuation argument becomes very interesting. Nvidia trades at around a 14-15 multiple to earnings, while Apple is closer to 35. Now add in bout 17% year-over-year. And the problem is not a lack of demand.
for AI. So you , deeply integrated into the AI ecosystem, and yet its market valuation, according to some investors, is surprisingly low. For some analysts, this makes Nvidia a much clearer opportunity than slower-growing tech companies . But there's another part of this Nvidia story that doesn't get enough attention.
How long does ver start buying it. The previous generation depreciates, and within a few years, old equipment supposedly becomes worthless. But this assumption may be too simplistic. One cloud infrastructure operator recently provided an important example with Nvidia A100 GPUs. The was re-contracted for additional workloads every AI workload requires the latest GPUs.
The latest equipment can older accelerators can still be useful for physical AI computing, . So when a newer generation comes along , the old equipment doesn't necessarily disappear. It can simply move on to other tasks. This means that the useful life of AI infrastructure may be much longer than some investors initially expected.
And if a GPU purchased years ago continues to generate revenue for several let's take a look at the deployment of Nvidia's latest infrastructure. The first Blackwell early deployments include NVL 72 configurations. But getting a GPU is only one part of the process. To operate an AI cluster, you need GPUs, racks, networks, data center capacity, electricity, and software.
Everything has to work harmoniously . You can't deliver 95% of the infrastructure and call it done. The entire system has to work, and this iy transform individual pieces of equipment into useful computing power for AI companies and enterprises. Now demand significantly exceeds supply . Some vendors say that virtually every GPU .
And that's how unusual this AI hype has become. Of course, the market reacts to shortages over tim produced and more cloud capacity is being deployed. Eventually, supply and demand converge. The question is not whether this imbalance will last forever. This is not true. The real question is how long demand can remain higher than supply.
And NVIDIA's long-term forecasts suggest that this imbalance could remain significant for years to come.
Comments 0