저자: 미상 · 외국인
요약 보기
ARK Invest가 2024년 8월 13일 하루에 팔란티어 약 4만3,874주(약 790만 달러)를 매도하고 Cerebras Systems 약 10만6,941주(약 2,800만 달러)를 매수한 사실을 소개. 영상은 AI 기술 스택(응용-소프트웨어-컴퓨팅 인프라) 구조를 설명하며 두 거래의 의미를 해석하지만 ARK의 의도는 알 수 없다고 명시. 매수 배경으로 역발상, 가치 재평가, 리스크-리워드 등 가능한 동기를 나열하고, 칩 시장 경쟁·고객 집중·AI 지출 둔화 등 위험을 제시. 결론적으로 특정 종목 추천 없이 지속적 ARK 거래 패턴 관찰과 사업 지표 확인을 권고.
다듬은 글
Hello everyone and welcome to The Explainer. Today we will unravel an extremely interesting mystery that lies at the heart of the artificial intelligence boom. If you follow the AI sector at all, you know that Palantir has become a real darling of the software industry . I mean, its stock is soaring as investors are betting big on enterprise data platforms.
But recently we noticed one very unusual rotation. Ark Invest's Katie Wood, who, as you know, is famous for her aggressive innovation bets, has actually started reducing her stake in this successful AI giant. So, the main question is : what exactly are they buying and selling? We'll break down specific deals, analyze the AI technology stack, and find out what this massive capital shift suggests for the future of AI investment.
$28 million, $20 thousand. This is the exact amount of capital that was suddenly poured into a surprisingly different, much smaller AI hardware company on August 13, 2024. Think about it for a second. This wasn't just a minor portfolio adjustment or a cautious hedge. Absolutely not. It was a large-scale, decisive statement.
Investing tens of millions of dollars in a single day deserves our attention, especially when those funds are being taken out of a recognized market favorite and funneled into a company that has just been severely punished by the market. Okay, let's dig into this, because the contrast here is simply striking.
On that particular Tuesday, ARK Invest announced the sale of 43,874 shares of Palantir . Based on the closing price that day, this position was worth approximately $7.9 million . But at the same time, they purchased 106,941 shares of Cerebras Systems, which totaled the same $28 million we just talked about.
So, we see a very clear divergence. Dumping millions from shares of a popular AI software developer while aggressively accumulating shares of a chip manufacturer. Now, before we go any further, actually, forget about that. It is extremely important to remember that no explanatory note is attached to daily trades.
We don't know ARK's exact intentions, and we're definitely not going to make false assumptions here. Instead, we will look at the structural realities of these two businesses to understand the real signal. Chapter one: software vs. infrastructure, AI technology stack. And this perfectly illustrates exactly what we need to pay attention to first.
To understand why an investor might make such a sudden rotation, we must shift our focus to the fundamental structure of the AI industry itself. The AI technology stack basically operates according to a strict three-tier hierarchy. At the very top, we have the first step—applications with artificial intelligence.
These are tools that you and I, the end users, interact with directly. Immediately below them is the second step , AI software, which includes operating systems and data management platforms. But at the heart of all of this, literally holding the entire ecosystem together, is the third step, computing, that is, the fundamental hardware.
Look, AI models require enormous computing power. Computing resources are required to train them . To run inferences on them, calculations are also required . As AI workloads become exponentially more complex, the amount of computing required is growing rapidly, creating a huge infrastructure market, and this is where this deal becomes so interesting.
Let's use a short metaphor to illustrate this distinction. Imagine the entire AI economy as a huge, rapidly expanding city . Palantir operates exclusively at the software level. They are up there, building profitable businesses, high-class services , shiny skyscrapers inside this city. And their economy is highly scalable, right?
Because software can be copied without the need to produce physical goods. Cerebras, on the other hand, is a hardware layer. They work "in the mud": pouring concrete, laying power lines , and building heavy physical infrastructure under the city. And this requires enormous research, exhausting engineering work, and significant capital investments.
This is a completely different economic reality. Let's move on to the second section: "The Cerebras Adventure" and investing against trends. What's really interesting about this slide is the timing. Betting on this underground infrastructure layer is not for the faint of heart, especially when you look at the recent history of the company that the Arc Foundation decided to support.
Cerebras has just experienced a very painful share sell-off due to financial results. The market had genuine concerns about the company's prospects. So, Arc was not bought during the euphoric rise of meme stocks. They bought when the broader market was nervous and taking a step back. This is literally the definition of a strategy “from the opposite”.
Stocks are falling, ordinary investors are panicking, and a large fund comes in with $28 million. But remember: a falling stock doesn't always mean a good buy. Sometimes the market is absolutely right. So why take such a huge risk? We need to separate what we know for sure from what it could mean. Looking at the situation completely neutrally, we can identify four possible motives for this purchase.
First, Arc simply believes that the market overreacted to Cerebras' latest results. Second, they are betting that the future of specialized AI computing is much bigger than Wall Street thinks. Third: at the appropriate prices, Cerebras may simply have offered a better risk-reward ratio than Palantir.
Or fourthly—and this, friends, is important to remember—these two agreements are completely unrelated. Professional portfolios are constantly rebalanced, and one deal does not always directly fund another. Chapter Three: The Battle for AI Chips, a Fierce Market. If Arc is truly betting on specialized AI computing, they are entering an extremely competitive arena.
The semiconductor market is not a friendly place for beginners. Cerebras does not operate in a vacuum here. They are waging a brutal war against real giants. Nvidia is clearly the undisputed king in this field, with AMD a powerful and relentless competitor hot on their heels. Alongside them, there are dozens of other specialized processors.
Because of this reality, Cerebras must answer a truly existential question. Why does the market really need their architecture? It's never just about creating a technically powerful chip. To survive in the long term, Cerebras must attract large customers, guarantee production capacity , ensure a reliable supply chain, and create comprehensive software support so that developers can use the equipment.
So, the purchase by Arc is not proof that Cerebras will win this battle. This is just proof that Arc is willing to risk significant capital to find out. But let's change our train of thought for a moment and look at the other side of the coin. We talked a lot about why they might have bought Cerebras, but why sell a stock that is showing such incredible results?
Palantir is perhaps one of the most well-known public AI companies today. Why the hell would you give up on a fast- growing winner? Again, we must avoid jumping to sensational conclusions. There are completely standard, logical reasons for a fund to press the " sell" button. First, they could simply have booked profits after a huge rally.
Quite logical , isn't it? Second, they can simply manage their overall position size so that Palantir does not dominate their portfolio. Moving on to the third possibility: because the market has recently placed too much hope in Palantir, the valuation may have become too high. When stocks are rated as perfect, they must consistently outperform expectations, which is extremely difficult to do.
And finally, fourth: they can simply reallocate capital to areas where they see higher growth potential. Chapter Four: Reality Check , AI Risk Assessment. So the main takeaway is that this whole rotation signals a willingness to move from the relatively scalable world of AI programs to the intensive, heavy world of AI hardware infrastructure .
And if you're an investor looking at this area, it's time to soberly assess what this really means. Understanding the specific dangers that smaller hardware companies face is extremely important . The first is the cost estimate. Listen, even a great company can have horribly overvalued stock. The second is competition.
As we have already mentioned, the giants of the semiconductor industry act ruthlessly. Next is completing tasks. Building the physical infrastructure for AI requires a staggering amount of ongoing capital investment. Fourth, customer concentration. Smaller chipmakers are often heavily dependent on just a few large customers.
If even one leaves, revenue will take a huge hit. And fifth— technological changes. Hardware is evolving at an incredible speed. The revolutionary chip you create today may become completely obsolete tomorrow. But the main risk hanging over all of this is a complete slowdown in AI spending. Think for a minute about the entire ecosystem.
Chips, massive data centers, cooling systems, huge amounts of electricity required. If the industry as a whole ends up spending too much money on all this infrastructure relative to the actual software revenue, the source of capital could suddenly dry up. If the applications on top don't bring in enough money, infrastructure spending on the bottom will stall.
Chapter five: what to look at next, the big picture. Let's move on to how this creates a clear plan for your own research in the future. Because, frankly, watching the daily fluctuations in a stock price won't tell you the real story. We need to focus on specific business metrics. For Palantir , you should track their revenue growth, profit margins, the level of actual adoption of their AI platforms by enterprise customers, and whether their sky-high valuation remains justified.
But for Cerebras, the metrics are changing. Yes, you still track revenue, but you should definitely pay attention to gross margin, their physical manufacturing capacity— can they actually produce what they've invented at scale—and their cash flow. Because creating physical hardware burns cash incredibly quickly.
But if you want to know whether this particular rotation was a one-time adjustment or a large-scale strategic shift, here's the strongest signal to watch for . The following ARK operations . A single-day transaction of $28 million can sometimes be just noise in the portfolio. But if we start to see a pattern, for example, if ARK consistently increases its share week after week while systematically reducing its presence in Palantir, then we will have concrete evidence that one of the largest innovation funds is fundamentally reevaluating its faith in the entire AI ecosystem.
Ultimately, the story of AI investment is becoming much larger than just one software company or one chipmaker. At the heart of this entire boom is a huge fundamental need for computing power. So the question I want to leave you with today is not simply which is better to choose for investment: Palantir or Cerebras.
No, the real question you should ask yourself as an investor is: where in the AI value chain will the greatest economic value ultimately accrue? Will it be in the flashy applications we use every day, or in the powerful infrastructure that quietly runs in the background? Keep this in mind as you watch the markets this week. Thanks for joining this review, and keep learning.
Comments 0