저자: Kevin · 외국인
요약 보기
AI 토큰 가격 하락과 에이전트 확산으로 사이버 위험이 커져 사이버보안 수요가 구조적으로 증가한다는 주장. 크라우드스트라이크 매출 성장(연간 반복매출 약 34%, 장기적 50~70% 가능성)과 골드만 리포트를 근거로 저평가를 주장. 동시에 본인이 보유 중이며 강의 할인 쿠폰 마감 등 구매 유도 발언 포함.
다듬은 글
The next frontier of AI stocks that will be huge winners, truly huge winners, may already be right in front of us. And I personally not only own a stake in the company we're about to talk about, but I truly believe that Wall Street's valuations for this company right now are a joke compared to what they should be .
This means that on paper, the company looks overvalued to Wall Street. In my opinion, given where we are headed with artificial intelligence, the company looks undervalued. Now, to understand this, we must first understand why. Artificial intelligence, as the Wall Street Journal, The Information, and others have already suggested, is becoming cheaper to implement.
This is great. This means we can use it more often, but it also creates a problem. Take a look at some of the latest models, for example. Alibaba's Qwen costs only $0.41 per million initial tokens. The cheapest American option or model here is the Muse Spark. This is a Meta development for $0.69 per million initial tokens.
And then there's Ox Alpha, just released by the same people who created GLM-5, and it's priced at around $0.50 per million initial tokens. Compare this for a moment with the cost of advanced models, the so-called top models. They cost $25 for Claude Opus, $50 for Fable 5, and the Kimi K3 Moonshot, which has been much talked about due to the fact that they essentially distilled Claude Opus to achieve the performance it has .
$15 per million initial tokens. All of these costs are coming down rapidly, in part due to the open-sourced releases we're seeing not only from companies like Meta, but also, in particular, from Chinese companies, especially DeepSeek, as DeepSeek tries to raise more funding. And what these companies are doing is allowing them to raise more money through this graph right here.
If you look at the AI adoption according to Ramp Capital, which is associated with venture capital and therefore, in my opinion, is a little biased in favor of AI, but that's okay. They indicate that AI adoption is growing along this orange line. Bloomberg then superimposed the rate of price growth on this orange line , essentially the pricing of APIs at OpenAI and Anthropic.
And what you start to see are literally perfect S-curves. OpenAI doesn't really have that long of an initial phase, so it looks a little less like a perfect S-curve, but you can clearly see the token prices peaking. They actually lowered prices twice: on July 31 and August 20. You can see Anthropic, it's more like a clear S- curve.
You see pricing leveling out. This is partly because open- weight models make access to AI cheaper , allowing agents to be launched and achieve 90–95% of the performance of state-of-the-art models. Of course, when I talk about open-scale models, it sometimes causes confusion. These are the prices you get when you launch on AWS or any other platform .
You can run these models on your own hardware, and then the only marginal costs would be the engineering labor and the energy required for the computations on your CPU and GPU stack. You can do a lot of damage with these models. So what will be the next frontier in the development of artificial intelligence?
CrowdStrike explains this quite aptly, and I think they created a good image for it. They suggest that we should fear an “ arms race” from swarms of agent bots, where AI stimulates cyberattacks from countries such as Iran, Russia, or China. We have already seen Iran hack our medical companies, there have also been attacks from China and Russia.
We've been through all this before. oop emerges: what more attacks like the one that happened to...who is there? Boston Scientific was just hacked. Thousands of workers were sent home because the networks were down due to a hack. So what's going on? AI is making cyberattacks cheaper and more frequent, forcing more investment in cybersecurity to protect.
But as AI becomes cheaper, the number of attacks is increasing again, further stimulating spending on defense technologies. Maybe I'm too optimistic about cybersecurity, I want to clarify that. But I believe cybersecurity is the next investment frontier for AI, even if the growth rates of cloud providers, hardware manufacturers, or Nvidia start to slow down.
When it happens—or maybe it doesn't— it won't matter much. In my opinion, cybersecurity is just beginning its journey to a tipping point. I think this is the next milestone. Stocks aren't particularly cheap right now, if you look at Wall Street prices. Here we need to approach it with our own vision: what do we expect from the growth of these companies?
And we're going to look at some of these arguments in detail by examining their reports in a moment. I want to remind you that the J Holo coupon expires tomorrow. We bought CrowdStrike and some other cybersecurity companies before this crazy CrowdStrike takeoff, and we're in a nice profit now. I mean, the stock is up 19% today.
So if you want to join our fundamental analysis, I invite you to join. For example, we reviewed CrowdStrike back in June. When we analyzed CrowdStrike in June, here are some notes we took. We saw that total revenue grew by 25.6%, and gross profit grew even faster. You could say they have very good margin growth.
Operating expenses increased by only 14.8%, and they have a significant price advantage. Good cash flow, good balance. So how did things change yesterday after the earnings report was released? Why did CrowdStrike suddenly soar 19% in a day, pulling up the IGV we were betting on as part of a software recovery in Q3-4?
IGV has a significant share of software. And this ETF , consisting of 102 stocks, is up 7% today. For an ETF with 102 stocks, a 7% increase in one day is pretty impressive. This makes people wonder: is this really a tipping point? Maybe, who knows, maybe it's just speculative trading. Maybe everything will go back soon.
But I think there's more going on here than meets the eye. In particular, Wall Street expects annual recurring revenue for companies like CrowdStrike to grow at a certain level. But in reality it is growing at a completely different pace. This expansion is already happening. For example, CrowdStrike beat expectations for net new annual revenue by 16.4%.
This means their total annual recurring revenue is now growing at 34% per year. And Wall Street still predicts earnings per share growth of just under 20% per year for this company. And the irony is that this "cloud moment" that occurred at the end of the first and beginning of the second quarter is just the beginning of a rapid increase in cybersecurity spending.
Mostly because I think people see a defining moment, like a "Mephisto moment" or something like that, and they say, "Oh, we should probably do something about that." And then there is a delay between this realization and actual action. So I think the second quarter doesn't yet give a full picture of what this turning point in growth will look like.
I think their annual recurring revenue growth rates could eventually exceed 50, 60, or maybe even 70%. Some of their Falcon Shield products are growing in about the same way now—they're doubling year-over-year. So, let's take a moment to look at what Goldman says, and then move on to some recent reports.
Here Goldman notes: “ Importantly, we believe that AI security—and the text breaks off—will disproportionately benefit CrowdStrike due to the strength of their technology stack , lightweight agent, threat graph, human feedback loops, and willingness to buy next- generation assets, essentially buying up other companies.” That's what's happening too.
Rich companies eventually buy good companies with quality products, right? What's interesting about CrowdStrike is that with their lightweight agent, they can place CrowdStrike in the Microsoft Store and AWS Store, so businesses can easily find the Falcon cybersecurity product where they already operate— great for small and medium-sized businesses.
And of course, if you're a large contractor, they'll send their front-end deployment engineers, like Palantir does. You know, Palantir is actually one of the integrators for Boston Scientific. Palantir is playing around with cybersecurity with its Apollo program, but, you know, how far are we there?
You know, the question is open. And I'm not pessimistic about Palantir. I want this to be clear. I actually think Palantir has a much better valuation right now, at least even according to Wall Street forecasts, than this company. Okay, this company is expensive now, and you see for yourself what the expectations have to be to say, "Oh, yeah, okay, this is really justified." But it's also worth noting that CrowdStrike is partnering with another company that I think is currently going through the process of hitting rock bottom, and that's Cerebrus.
Part of this is because they see the power of the subtle latency in artificial intelligence that can be achieved with SRAM chips. They see this potential and are partnering with Cerebrus to enable AI data detection and response to cyber threats, responding to them much faster than, say, an attacker or enemy could ever react .
So, if you have faster hardware, you can solve a cyber problem faster than a cybercriminal with, say, an open model running on a slower product. There is also some potential between Nvidia's Groq and Cerebras platforms, which compete in the SRAM space. But this only indirectly relates to the discussion of cybersecurity.
Goldman Sachs states the following. They believe they see an annual recurring revenue of 51%. They believe the tipping point in security spending planning is happening sooner than expected. I must say that this confirms my prejudices to some extent. Again , we send out alerts about cybersecurity companies.
We buy shares of cybersecurity companies, we work on the platform. This is included in the Meet Kevin membership. Many people are signing up because our coupon expires tomorrow. Another reason why many people sign up is for lifetime access; The opportunity to secure lifetime access to the Meet Reinvest and Homes AI product disappears after the expiration date tomorrow.
So a lot of people are emailing us, "Hey, can we combine these two products together?" Current course participants write to us requesting a code for a package offer. You can also do this by writing to [email protected]. Write: "Hey, I'm a course participant. I want to add Homes AI." You can do it. But this is actually a move to an annual recurring revenue or subscription model, which is CrowdStrike's model.
You see, when CrowdStrike makes $6.6 billion a year, they start the next year with $6.6 billion. Any growth they have, minus those who unsubscribed, is based on that starting point. And that actually gives a company like CrowdStrike justification for a significantly higher valuation than, say, Apple, right?
Apple currently trades at a PEG ratio of 3.7 with growth estimates of 10%. They can be quite low. To be fair, CrowdStrike trades at a PEG ratio of 8 on estimates of 20% growth. I believe these growth estimates are underestimated. If I take a 50% growth estimate, I'm trading closer to where Apple is trading, and I think the upside potential here is even greater.
We'll see. We will talk more about financial indicators. But the bottom line is that businesses with annual recurring revenue, in my opinion, deserve higher multipliers. Apple has something similar with its subscription services . Perhaps that's why Apple is currently trading at a slight premium to the level at which I believe the company should be valued .
So, if we really look at CrowdStrike's latest financials, we can see a little more after this Goldman stuff. So, the Goldman piece here, again, confirmation bias warning, I caution you on this part. We believe that a certain turning point is occurring in budgeting. I think so , so there is a certain risk.
Flexa's revenue doubled. This is the 2.29 billion. I think it will grow much faster than Wall Street thinks. And from Goldman, they point to EPS growth somewhere here in their financial statements. They are forecasting EPS growth for the next 2 years of at least 39% and 33%, based on some of the dips that have occurred before.
Considering that the company fluctuated between profitability and losses. We only had one profitable year. I think it was 2024. In 2025, you returned to negative figures, and before that, everything was negative. Maybe I'm confusing '24 and '25. It doesn't really matter. The bottom line is that you are only now really getting to profitability, and this is another reason why the company may look expensive.
And maybe it really is expensive. Maybe it's worth waiting for a better opportunity to enter, right? But look at the document here. Annual revenue is up 25% compared to last year . I think that's too little. I think it will grow faster. This is an invasion of agent problems. This is the agent invasion.
Agents are like people and they multiply the risks, you know, because of the access to data they have. Agents can act unpredictably, act collaboratively, and exceed their limitations, causing harm. The combination of agent-based AI with endpoints and cheap, open-source AI makes me very, very optimistic about cybersecurity.
Oh, yes. Sorry. Maybe I'm being too categorical here . But I think it's worth paying attention to these three factors. So , you have this three. First, it's a decrease in the value of tokens, let's call it that. This could be related to open weights or something else, right? Then you have agent systems, as well as more endpoints.
So an endpoint is a device, like a phone, a computer, a Linux server, a laptop, an iPad, whatever, right? Every endpoint that someone provides access to, whether a CrowdStrike agent or any other, is a risk. And most people don't have endpoint management software . This is a big alarm bell, and it will change.
There are other players in the endpoint management industry that are smaller in size. You know, this is already . So, how much is CrowdStrike worth now ? CrowdStrike is now worth $230 billion . There is another player in the endpoint management space whose shares are up 11% today and in which we are also investing.
They are worth only $7.8 billion. So, this is a little tip for you if you don't want to join us on the Meet Kevin courses yet. But this is another great endpoint manager . And so the whole of cyberspace gets a boost from this. But I think there's a trio at play here: lower token value leads to more agent usage , which leads to agent usage on more endpoints, and all of that correlates with growth—let's use a different color.
All of this correlates with the rise of cyber risks, right ? Increasing risk . All three of these factors increase the risk. It can be argued that more and more endpoints are now using token generators. So, we have an acceleration: more people are using endpoints, more people are using AgenTech, open model weights are getting cheaper , which we talked about at the beginning— all of this just increases cyber risks .
Of course? Class. So, now that we've put it all together, what do we think about it? What do we think? Well, let's go back to the table. So, we have a conversation about AgenTech swarms. We have information from their earnings report about the exponential growth in vulnerabilities being detected using Frontier AI.
And the problem is the delay between identifying a risk and addressing it. And that's where CrowdStrike has another great opportunity, because they get access to a lot of their own data, a bit like the old days of Tesla FSD, full autopilot, where everyone was like, "Oh yeah, everyone loves FSD. They have the data.
They collect all the data before anyone else. They're going to be the leader for a long time." To be fair , that's how it was. I love FSD. So the problem with data is—and that 's why CrowdStrike, you know, the big players keep winning—when you have data, people come to you because you have that data.
So people will come to you because you have data, and by having data, you can protect them from more threats. The more data you have, the more protection you can provide. So people come because you have more data. But with more people coming in, you increase the amount of data because you now have access to more companies.
So this is a crazy flywheel that's spinning here, and advanced AI, and cheaper access to it—all of that amplifies the effect. Net income, retained earnings, they're pretty clever about revealing that. In some quarters they mention it, in others they don't, but their net income retention ratio was around 112%.
Keep in mind that Palantir has over 150% revenue growth with a 3% customer base growth rate , which is very attractive for Palantir. Sorry for constantly mentioning Palantir. They have very, very... they have huge potential, understand? Very, very, very huge potential . Great potential. And sometimes, when comparing to other companies, I think it's important to know that it doesn't have to be CrowdStrike.
I'm not trying to convince you, like, " Hey, you should do exactly what I'm doing." Because I can buy stocks, they can go down and you can lose money. And, you know, hopefully they'll grow back. It's always a risk, right? We can also buy stocks and they can go up. I am not your financial advisor. My goal is to provide as much detail as possible.
And if we can make money together, that's great. But ultimately, the purchasing decision is yours, right? For example, we just received an email from a user, specifically regarding a software deal. I just replied, “Hey , I have what they got here. I am now in the black $37,000 thanks to Kevin on Path .
Well, that's very nice. Well, congratulations and thank you very much for sharing this. This is very cool. This is another topic that we have been analyzing for some time. We believe the bottom was around nine dollars. Shares have become very cheap. But in any case, remember, join us at mekevin.com.
You will get lifetime access to the courses in this perspective. And you'll be able to see a lot of this analysis very early on, thanks to this exchange of ideas. But what's really interesting is that they're collecting data from their endpoints and funneling it into their pricing policies. So , there is truly great value here, which cannot but be admired.
Then , if we look at it, it was a real boom for Claude at the end of the first quarter. There is a risk that part of the boom we are seeing in Claude usage is 400 % growth in Claude usage, 100% growth in client agent usage on endpoints in recent months. There is a potential risk that we have gone through a kind of boom cycle and it will eventually go into a downturn because of what we saw in the first quarter.
So, it's too early to say. If we look at the financials, the cash flow is quite good. We have about $1.8 billion in cash flow per year. This works out to less than 1% cash flow return. So , a bit expensive compared to their market cap. So who knows, maybe this is a peak, right? Still very optimistic about cybersecurity, but worth noting.
They are also just reaching profitability. If I look at their balance sheet, I have $6 billion in cash, $1 billion in checking accounts. That's cash and accounts receivable, $1 billion in accounts receivable, which means I have $5 billion in net free cash. If I subtract their long-term debt, I subtract about another $1 billion.
So, I really do have about $4 billion in free cash. Which is good. I mean, to be fair , for a company with a market capitalization of $200 billion, that's, you know, maybe not a lot, but the balance sheet is very good. So the balance looks good. Revenue, another 25.8% year-over-year revenue growth. Their subscription costs are actually decreasing, with 22.1% indicating a growing price advantage between revenue and subscriptions.
But when compared to last year, their subscription costs were 22.8%, which also means we get a bigger price advantage. Their gross profit margin is at Nvidia's level—75%, which is really good. Now, to be fair , they have a lot of general and administrative expenses, and they're just getting to profitability.
They are barely making a profit, so there is some risk that this will become a "star" for many investors on Wall Street. Of course. You can see that their R&D spending has now increased by 29.8%, so it's slightly higher than their current revenue. But their total general administrative and selling expenses, along with R&D, grew by only 17.5%.
So, to sum it all up, their price advantage is growing. They simply invest more in R& D. For example, sales costs alone have increased by 14%, but their revenue is almost twice as high—25.8%. That's pretty good. Maybe I'm too optimistic. So, just mentioning this. Well, the impact of data centers, they also have a lot of their own data center equipment.
Uh, you know, over a billion dollars worth of GPUs. It seems they rent data centers, but have a lot of their own equipment. Also, most of their income comes from the United States. About 65% of revenue comes from the United States. So that gives a little more understanding of where he's coming from.
No country, no other country has more than 10% influence on CrowdStrike's revenue. However, they have other regions, like Europe or others, that account for over 10%. OK. This gives us a full picture of how I see open-source models driving down costs, potentially leading to more cyberattacks, how endpoint models create more potential cybersecurity risks and a greater need for protection, how a company like CrowdStrike has an advantage, and while they are very expensive by Wall Street standards, I see them as undervalued because people don't properly consider how important it is to have cybersecurity in their business and the fact that they can deploy it on, for example, Microsoft or AWS.
Oh, great. The fire alarm went off . Crisis! The house is burning! I need to figure this out. In any case, thank you for being here. And, uh, see you after the fire. Okay, friends. .
Comments 0