저자: LarryFink · 외국인
요약 보기
BlackRock 마진이 12개월간 260bp 상승했고 AI/기술 도입으로 인력 증가 없이 자산 1조 달러를 늘렸다며 향후 12개월 시장에 매우 낙관적이라고 밝힘. 레버리지 우려는 제한적(2008년 대비 낮음, 비트코인 레버리지는 과거 우려). 자막 후반부는 Oracle 118달러 부근 매수 여부를 두고 신용등급 리스크와 AI 성장 베팅을 논하며 양면적 입장을 보임. 메모리(SK하이닉스) 관련 발언은 HBM 장기계약으로 수익 반영이 지연될 수 있으나 사이클 리스크(경쟁·증설)가 있다는 취지.
다듬은 글
There is not as much leverage in this system compared to 2008-2009. Perhaps the Korean market has such built-in leverage. Certainly, as I said before, when I was here, I was always worried about leverage in Bitcoin and cryptocurrency. There were too many leveraged players . That's why we saw the crash, and I think things are more stable at these levels .
But, um, no, we don't see that much hidden leverage. For the scale of today's markets, leverage is not that great. I mean, that doesn't mean there aren't individual cells. But, no, as I said in my morning speech, I am very optimistic about the markets for the next 12 months. I believe that the technological revolution will contribute to better margins for many companies.
I mean, look at BlackRock. We have increased, our margin has increased by 260 basis points in the last 12 months. This is largely due to the use of an increasing number of technologies. Will you be able to maintain that margin? You said that you don't see 45-46% margins as limits. You mentioned 2021 .
Yes. However, your business is very different from what it was in 2021. So I'm wondering what mechanism will allow you to get back above 47%? Continue to drive more growth in our private markets sector, where we are seeing incredible momentum. Continue to create opportunities in the pension space, which I believe will also contribute to better margins.
And just making sure we use technology as quickly as possible to do more with less. I mean, our staffing levels, David, haven't changed. And we, you know, increased our assets by a trillion dollars. Okay, here's what you see. We can use technology to handle more operations, to handle more actions. So, we are able to do this.
We can effectively utilize our human capital through technology by working together with them. We use AI, and the amount of code we write with our programmers has accelerated significantly. So we can do more and more, and as the AI and computing capabilities that I talked about earlier become faster and cheaper, that will drive even higher margins not just for BlackRock but for other firms as well.
From September to the present. It is at annual lows. And, you know, they obviously have the infrastructure and equipment that companies need. I mean, a company like OpenAI or Amazon, the Pentagon, the Navy— they all invest , spend money on Oracle for a reason. And that's because they have everything they need to run these companies.
So that's a pretty broad spectrum, right? This is not just a crazy AI startup or Open AI. There are many customers with legacy systems , larger and more well-known customers, including the government, who rely on this equipment and these solutions. So I think it's worth asking yourself: what if everything goes well?
Yes, it's a little confusing right now. I do n't like it, I don't like their credit rating. It seems to me that it is only a step above " trash", in my opinion . You know, if it drops even lower, a forced sale will begin . I don't like everything you listed. I do n't want to go through all this again .
Everything you listed just makes me a little nervous. But if they can overcome this barrier, it could be the genius move of the year. And it might look like this: how could you not buy Oracle at 118? They create everything for everyone, and of course it will work. This is what it might look like in the future, looking back.
But we live in the present, so you need to make a bet, weighing the risks and rewards. And so this is a difficult decision for us. I think, for me personally, I don't like that their data is at -11, but their order book is very, very healthy. They have big profits ahead of them. They build their success.
And until AI collapses, we can look back on those 118 and say, "Wow, that was an incredible opportunity." So, it's almost a binary bet, which is something you rarely expect from a company of this size. Typically, binary options betting is typical of much smaller companies that bet everything on one side.
But I feel like Oracle is really more of a binary bet. If AI lives up to expectations and everyone progresses as far as we expect, then this could be a very cheap stock price, looking ahead or looking back. Yeah, and Landon, when I looked at the chart, like you said about Oracle , an 11% drop in enterprise software, with IBM, Salesforce, and ServiceNow seeing traffic growth—it's, you know, kind of makes you scratch your head because maybe now is the time to buy.
When I look at the three-year chart, I think about this level of 118, we are at 52-week lows right now. This stock was worth $99 three years ago, at the end of 2024 . Okay, it was under 100 bucks, then it went up to 345, and now it's down to 118 again. So, any time, let's begin our video. Many investors are observing somewhat unusual developments in the semiconductor market.
AI spending continues to rise , and demand for cloud technologies remains extremely high. Demand for storage is steady, and Microsoft just reported higher-than-expected Azure growth. Meta raises the lower end of its capital expenditure forecast . Yet many companies providing the infrastructure for this AI expansion are under pressure.
This apparent contradiction is exactly what investors need to understand, as two completely different stories could be happening simultaneously in the semiconductor market right now. The first is a fundamental industrial AI boom, and the second is a serious reset of investor expectations . The second is a serious reset of investor expectations , and SK Hynix may be one of the most striking examples.
The company just reported extraordinary profitability, and memory prices have skyrocketed. Demand for high-bandwidth memory remains extremely strong, and SK Hynix continues to play a critical role in the AI infrastructure supply chain. However, the market remained unsatisfied. One reason seems to be the average selling price, as investors have seen huge price increases for traditional DRAM and NAND memory.
And, of course, it was expected that these increases would be immediately reflected in SK Hynix's financial results. But high-bandwidth memory works differently. HBM prices can be negotiated through long-term contracts. This means that today's rapid price increases for conventional memory do not necessarily immediately translate into higher HBM selling prices.
They may be delayed, and since SK Hynix has a stable share of HBM, its blended average selling price may behave differently than investors expect when looking at just spot memory prices. This is an extremely important distinction. The market may have approached this earnings report with the expectation of immediate financial benefit from today's memory shortage.
Instead, some of these benefits may appear later, as new HBM products enter the market and previously concluded contracts are reviewed. This does not eliminate risks. Historically , the memory market is one of the most cyclical in technology. High prices encourage investment, investment over time creates supply , and supply catches up with demand.
Prices fall, margins collapse, and the cycle begins again. This is a traditional scenario for the memory market. Today, the huge question arises: has artificial intelligence fundamentally continued this cycle? There is reason to believe so. AI doesn't just consume more memory. Modern artificial intelligence architecture is increasingly dependent on memory performance.
Large models require moving huge amounts of data between processors. Data output workloads are expanding. Contextual windows are getting bigger and AI agents are getting more complex. Coding systems consume an extraordinary amount of tokens, and future applications in robotics, autonomous transportation, and physical AI may create additional demand for memory.
In other words, memory is becoming increasingly closely tied to computation. GPUs may be getting faster, but without sufficient memory bandwidth, some of that processing power can become a bottleneck. This is why high- bandwidth memory has become strategically important to the Nvidia ecosystem, and SK Hynix currently holds a significant share of this market.
The most important argument, however, deserves attention: such a high margin always attracts competition. Samsung is not standing still. Micron is also not standing still . Chinese semiconductor manufacturers are not standing still, and over time, there will be more capacity.
Comments 0