저자: Kevin · 외국인
요약 보기
AI 자체보다 광고 수혜 기업(특히 Meta)이 유망하다는 주장. Meta는 DAU 4%·광고 노출 19%·광고 단가 12% 증가, 영업이익률 41%로 가격결정력이 강하다고 봄. 구조조정과 컴퓨팅 자원 판매로 이익 개선 가능. 과거 780 고점 대비 하락해 600달러 이하는 매수 기회로 제시. 연말 2배, 4년 내 4~5배 가능성 언급. 레버리지·부채 리스크도 일부 언급.
다듬은 글
Hello everyone, this is Kevin, speaking to you from London. We need to talk about the potential for AI stocks and why I think they can more than double, and possibly even more. I think go much further over the next four. So, that's worth a drink. So, to substantiate this point, you should know what really got me thinking about this the last few days is an institutional study I read about Intuit.
I was supposed to be reading about Intuit, but I got sidetracked by another promotion and opportunity that came up elsewhere. So let me read this to you. Intuit does a review, like QuickBooks, Mailchimp, right? They run Credit Karma as part of this. Intuit is conducting a review of its $1 billion U.S.
media account. So, the media account is essentially how Intuit is promoting its AI services in what looks like one of the biggest tenders of the year. Reviews , as they say, cover their entire business. QuickBooks, TurboTax, Mailchimp, Credit Karma, and so on. And essentially, they're going to try to figure out how to: "Hey, how do we market and promote our AI services, which are now integrated into Mailchimp and other products, to companies and individual small businesses so that they use our products not only more, but more effectively?" And right after that, I got off the plane and started seeing ads in the London Heathrow terminal.
This is a certain company . It's called Cohere. This is a Canadian multinational company. They already have about a quarter of a billion dollars in revenue, which is actually pretty impressive for the 450 employees they have. And right after the Cohere ad came out , I saw this ad and thought, "Wait, I thought you guys were doing great with your income ." Is that really the case if they don't use the Starlink strategy?
But I was a bit surprised . I'm like, "What? Seriously? "Claude has to advertise himself." So then I see these Claude ads, and it's basically just banner after banner: AI ad, AI ad. And I think, "Hmm, that's really interesting." You see, Starlink did the same thing before the API, before the IPO . So maybe this is just a strategy, I think, before the SpaceX IPO.
Elon Musk has reduced Starlink prices by 50%. He spent a lot of money on advertising, then included in his IPO prospectus how much the number of Starlink users had grown. I am not criticizing this strategy. I think this is exactly how companies act before an IPO. I mention this to say: such advertising could be a way to " pump up" user growth figures before an IPO , or, given the context of the conversation, it's part of a strategy.
It is quite possible that the next direction for large investments will be not so much investments in AI companies themselves, but investments in the world's largest advertisers. One of my favorites here —as much as I hate to admit it—is Meta, which is currently very profitable with a PEG ratio below 1.3.
Frankly, the company could potentially double its capitalization by the end of the year if the software market recovers, although there are no guarantees yet. But its assessment fully allows for a doubling by the end of the year, and in the four- year perspective, a growth of four to five times at a fairly rapid pace.
Interestingly, according to institutional research, advertising agencies are growing faster than Goldman Sachs analysts predicted. Their models predicted some growth rates, but in reality there is a sharp jump in advertising spending, especially from large AI companies. So, for me, it looks like this: I'm already convinced that Claude, Gemini, Grok are essentially consumer goods , and the most important thing to do is attract users.
I think that's what Grok, Meta, and others are fighting for—for users, but Meta already has a huge number of them . Elon Musk constantly points out that X, i.e. Twitter, has about 600 million monthly active users . That's great, but Meta has almost six times more users on their apps every day. Elon uses a monthly metric for X, although I personally noticed that my activity in X dropped dramatically.
You can follow me there where I rarely post , although I don't think they promote my content at all. Personally, I think that when you go into X, you see more and more AI garbage every day. This is pretty bad. Well, never mind. Elon cites a figure of 600 million monthly active users. Um, let me make sure I said everything correctly.
It was 600 million. If I said 600 billion, it was a mistake. 600 million. Meta has 3.56 billion of them. So that's about six times more. And again, they use a daily activity metric. But what is interesting is the large number of Meta. You start to wonder how many people in the world actually use Instagram and Facebook, where most of Meta's advertising money comes from .
But they only grew by about 4%. Although the number of active daily users grew by only 4% year- over-year, their ad impressions increased by 19%. And the average price for advertising increased by 12% year-on-year. So, despite user growth being 4%, companies are paying three times as much money compared to audience growth because they see an improved AI ROI that they are willing to pay for.
This, in essence, is the definition of affordability. We noticed that Meta's costs, for example, increased by 35% year over year, which usually makes you think, "Wow, that's a lot." But they actually increased their operating margin to 41 %, despite a 35 % increase in overall expenses. They are increasing margins even as costs rise, indicating their pricing power is increasing .
They have high pricing power, and that's very good. I know that many people don't like Zuckerberg, and there's a lot of talk about "social media is harmful to children," that we need to " protect children from social media," and there are also antitrust risks and questions about the responsibility of tech giants for the mental health of users.
These are real risks, and we don't want to downplay them. But what's interesting is that they now assess your likelihood of buying something on Instagram or Facebook, and if you're a potentially valuable customer, they're pushing your algorithm into a more resource-intensive architecture to offer you products more efficiently because they know they can make more money from you.
So now AI determines how likely you are to spend money, and depending on that probability, they invest more resources into the algorithm for your conversion. And structurally it works. This is kind of genius. So it's like AI is deciding when to spend more on AI to sell you something, which is a little creepy in a way.
You know, the company had kind of petitions within the company itself where users or employees of the company didn't want their data to be removed because Meta was looking at and probably implementing the practice of tracking your mouse movements and, you know, everything you do on your computer so they could train new AI models to essentially replace you, which , you know, would basically lead to this massive rise of socialism.
While socialism does lead to more consumerism in some cases , where else would you find things to buy if not on social media? So this is where this really strange cycle of doom arises. But, be that as it may , they are so actively using artificial intelligence at Meta that by the end of the first quarter they increased the number of staff by 1% year-on-year.
But then they decided: oh, we're actually going to downsize our company by 10%. They are laying off about 8,000 workers. Not only are they laying off 8,000 employees, they are also reassigning 7,000 employees to AI-related initiatives . And I'm assuming that if some of these initiatives don't work, let's say you put 7,000 employees on 10 initiatives.
If eight out of those 10 initiatives don't work, well, then they'll just fire eight out of 10 of those reassigned people. So, whatever it is, you know, 5600, right? Workers simply end up being laid off. So, like you're on a plateau, which sounds harsh, because the company is actually making more money, and that's crazy because, you know, revenue growth is up, their revenue growth is up 33% across the board, which is just fantastic.
And their total net income has grown significantly beyond that. So, remember, with a 4% increase in user numbers, somehow their revenue increased by 33%. You have to think about the scale of how much more money they are making from advertising, and how that relates to the commercialization of artificial intelligence.
All of these AI companies will need customers . But, be that as it may , their net income increased by 60%. Now, part of this is due to the benefits of probably depreciating the AI equipment, which then gives you more tax benefits. So, if you actually look at the pre- tax income, it's not that attractive.
It is always good to pay attention to this nuance. You only grew about 18.3%. Uh, and if you actually look at the operating profit , uh, right now you have, let's see, 1755, that's about plus 30%. So, not all of this was reflected in net profit. Well, and part of that is because they spend a ton of money on research and development.
They spent 45.6% more year-over-year on R&D. And they did n't actually spend much more on marketing. This tells us a few things that I find very interesting. They spent only 5.4% more money on their own marketing. So who is Meta targeting? Well, Meta is orienting, for example—that's one of the reasons why I like to travel, —you know, I book a car with some rental company and I get a WhatsApp message from an AI, I can chat with that AI , and of course, ask for live support when I need it.
But bots have really gotten pretty good, and Meta owns that platform, right? WhatsApp. So they sell these AI bots and infrastructure to businesses to make it easier to interact with customers. And all over the world, although less so in the United States. Globally, WhatsApp is very popular, I think less so in the US, but I think there are quite a few users there too .
I just don't use it in the US. I only use it when I'm abroad. Eh, but I find it very interesting, because even though the company has more of these products and services, they only increase their own advertising by 5%. But at the same time, their income increased by 33%. Which to me again shows the power of pricing.
People go into this business because they make money on Instagram or Facebook. The vast majority of their advertising revenue comes from Instagram and Facebook, not Threads or anything else. So, yes, they potentially want to monetize computing power. Uh, which Mark Zuckerberg recently explained some of his, uh, views on this.
His idea is that there are companies, and SpaceX was an example of what inspired him to do that. He says there are companies like SpaceX that are making deals with Anthropic or other firms that are looking to collaborate with SpaceX because they need that computing power. They want to provide themselves with the capabilities to, you know, quickly accelerate the development of their LLMs and get ahead of the competition.
And Mark is essentially saying, hey, you know, we have the computational architecture, the AI model architecture, and the talented AI people. We have a kind of triad in the AI business. And all this is financed by huge advertising revenues. So you have this triad: AI and huge advertising revenues, helping each other.
But he actually says: "We are ready to sell capacity if we are offered profitable deals, because we can make a lot of money from it ." I think we are at a turning point, as Meta's performance has been declining for quite some time. It was n't a very attractive stock to own. I note that I invest in Meta myself, and the shares have fallen in price.
They have been showing a downward trend from the 780 mark since around August 2025. Yes, this has been going on for some time. And whenever the price drops below $600, it looks like a good buying opportunity. Now it's about 670, 669 to be exact. But the point is, Zach says in Q2, " Look, we had great year-over-year revenue growth in Q1 .
We barely increased our acquisition costs , but we're making a ton of money. We're getting more efficient every day at how we convert users into customers. We're digging deeper into their viewing history, which gives us a better understanding of who they are and what they like, and that also increases conversion." They clearly stated that they doubled the length of user interaction sequences to train deeper interest modeling, and this improved content delivery.
In short, they make money by making even more money. And in the second quarter, Mark Zuckerberg moves to the model: "Okay, let's start making real money." Let's cut costs where our initiatives are not succeeding. Let's redistribute resources where possible. Let's make these high-margin deals for computing power while we're being offered that kind of money.
And why not? And keep in mind: Meta has a lot of off-balance sheet leasing obligations, which I do n't really like because it hurts the balance sheet . Not what I like . They have enough cash and marketable securities, and if you add in accounts receivable , it's about $17 billion . I have written down: 23 billion in cash, 57 billion in securities, and 17 billion in accounts receivable .
If I add all of this up, I get almost a hundred billion in liquid assets in the first quarter. And I compare it to their accounts. They have about $48 billion in accounts receivable and $79 billion in other long-term debt. So, we have enough funds for current expenses, but not if we add all the other long-term debt here.
So you know, the balance sheet is getting a little tighter, especially when you consider some of the off-balance sheet risks . But we know about these deals with Blue Owl or the data centers that Meta has invested in, which is very profitable for the company. So, you know, Zuckerberg is kind of hedging his bets the old-fashioned way .
And I think the second quarter is the tipping point where he says, " Look, I'm going to bet on AI, but now I'm going to hedge on some of the costs." We will reduce stock payments and reduce salary costs for laid-off employees. Of course, we will stop spending money on salaries and stock compensation.
We will pay more money and give more shares to those we want to retain. And when someone offers us an unheard-of amount for computing power, we will simply sell it. This is not necessarily a tipping point for the AI market unless capital spending starts to decline, and there are no signs of that yet.
You know, this company, for example, in the first quarter still spent $19 billion on capital expenditures , even though it had cash flow of $32 billion. So in the first quarter they still had $13 billion in free cash flow, even though many other companies, even some of the "Mag 7," are burning through almost all of their free cash flow, whether it's Microsoft, Amazon, or Google.
You can analyze these companies separately, for example, Microsoft. Meta keeps some stock, actually quite solid. I mean, even last year in the first quarter they spent $13 billion on capital expenditures with a free cash flow of about $9 billion. So, they still have enough cash flow to pay their bills.
The business is growing incredibly. In my opinion, as they become more efficient, the second quarter will probably be the bottom for Meta, assuming there is no recession, of course. (This isn't personalized financial advice either, but the company is clearly cleaning up, uh , improving staff efficiency, computing power, increasing margins where possible.) They've tripled the number of Meta glasses users.
I'm a really big fan of Meta glasses, especially when traveling with my family. A huge fan of theirs. Of course, you know, this is not sponsored content . I'm just saying that there are moments that I capture on them that I would n't otherwise be able to capture because of the time it takes to get my phone out or, uh, the awkwardness of holding something when you're trying to do something.
For example, let's say you're taking a baking class. It's hard to bake when your kids have their hands in flour and still have to hold your phone, when you can just put on your glasses and capture those moments, which is also a little weird because that's essentially what fuels AI, right? You're helping Meta learn visual AI better than any other company because they literally get a view through the user's eyes, and I wonder how much it might cost to, uh, sell that data to, say, the robot manufacturer Optimus sometime in the future.
But I am very optimistic. Uh, I have to say, uh, the price target I have for this company is over $1,377 by the end of the year . I don't know if we'll achieve this by the end of the year. I admit that the software and Magnetic Seven really took a back seat, but I, you know, have a lot of confidence in this title, so I want to be very frank and share some thoughts.
And I think that the commercialization of AI is happening, and advertising companies, companies that sell advertising, have inventory to sell —even Netflix can be included in that group, and Google with YouTube and so on—they're going to make a lot of money advertising the very AI products that they're also competing with, which is both incredible and amazing.
Anyway, thank you very much for watching. Think about subscribing to the channel , and, uh, wow, we didn't even do any sponsorships in this video or, well, I guess we do n't do any advertising or any pitch. This is madness. Well, let's not spoil everything. OK. See you later.
Comments 0