원문 보기
and dealt him a blow by calling their government scum. The president seemed genuinely offended by the numerous Iranian attacks on shipping in the strait despite the so- called ceasefire. So, we have no idea what he will do next. Of course, we've seen this movie before and know that the end result is usually just higher oil prices.
Today's market reaction reflects this dilemma. The Dow fell 577 points, the S&P sank. NASDAQ, however, rose by 2%. Although it started the day almost as horribly as the other two indices. Look, the president's tirade has dragged down interest-rate-sensitive consumer stocks, wiped out a great rotation in food and pharmaceutical stocks, and crushed anyone who dared to invest in travel, leisure, and aerospace .
And listen, I can't deny the short-term significance of Trump's belligerent rhetoric. But at least when it comes to the stock market, I'm much more concerned about something else. I'm worried about supply, namely the flow of new stocks and bonds that have flooded this market, absorbing a lot of the capital that was in reserve.
This worries me much more than the traffic in the Strait of Hormuz. Let me explain. You know, I've always said that oversupply—too many stocks than the market can handle—can kill any bull market. And I'm afraid it's getting too much now. If issuers and their investment bank assistants don't slow down, don't step back a bit , I think the bull will be hurt.
After the blow that the market has taken, especially the technology sector , we need to talk about the balance and whether we have completely upset it . We need to know this because after equilibrium comes a sudden irreversible surplus. And this is exactly what we saw during the dot-com crash. Before I take you down this trail of tears, let me say that we are not there yet.
So far, the market is handling the new offering like a champ. Especially considering the almost constant rattling of weapons by the president. So what I'm saying is: if anyone tries to misinterpret me, I'll refer to this paragraph or to what I'm saying now, which is—this is not a call to run away. helping the debt market recently.
No. And, uh, nice to be with you, Sam. I went a little overboard with my World Cup cheering. So I apologize for the voice. Um, I think this is a natural progression at this point in the AI cycle. What we are seeing is a kind of cutting edge of skepticism about the return on investment in AI. However, there is no shortage of demand.
So it's a kind of shuffling of cards . I wouldn't worry in the long run. I believe that out of these three, Nebius currently has the best value, but we will see a lot of ups and downs in the near future. Why is Nebius the best value? I think if you look at the ratio of real orders to the cost of capital and the amount of funds raised, compared to others, they are not chasing such lofty dreams, and therefore, there are fewer risks.
Okay, I understand. And in terms of overall performance, as we mentioned, Nebius has been performing well recently, but cloud technologies in general are under pressure right now. What do you think is the main driver of this, besides the rotation you mentioned? There are examples where a large consumer of your products opens their own business to sell the same thing they bought from you.
In this case, I'm talking about Meta. This does not indicate that you have monopolized the market . I think we will see this more and more often in the future, as hyperscale cloud providers will be able to sell their excess capacity. I repeat, we still have no problems with demand, but there are already the first signs that unlimited demand is not something guaranteed.
We don't see unlimited demand. Got it. What do you think about a moratorium on data centers here in New York State? I think we are only at the beginning. It will be interesting to see if this spreads as a movement across the US. I can say that some concerns about water consumption are actually unfounded .
But I don't think it matters . I believe that the negative reaction to AI in general, as people are concerned about the impact of the technology on society, along with concerns about resource allocation and electricity rates for consumers—is mostly emotional, although there are also well-founded fears.
But you know what? These emotions can drive the markets. Certainly, they can. I'm interested in what you said about some of the environmental impacts. I mean, I've heard this discussion literally this week, especially with the moratorium that everyone's talking about, and also with TSMC's plans to expand into Arizona—a state that some argue also has water problems.
Why do you think these arguments are not valid? Well, without going into deep detail, when it comes to water consumption, we are mostly talking about older data centers that do not have a closed loop. So, if you think about a car radiator, your car doesn't waste water to cool the engine. And most of the new large data centers that are planned to be built are closed- loop systems that consume almost zero water.
Something very unusual is happening in the field of artificial intelligence right now. Companies at the center of AI development continue to talk about the hug. AI labs continue to compete to create better models, and the need for memory remains unchanged. Data centers continue to expand, but shares of have suddenly come under heavy selling pressure.
It is important for investors to understand this gap, as the market begins to question whether the extraordinary boom in AI infrastructure has reached a point where expectations have become too high, and recent AI developments from China have only exacerbated this debate. A new generation of increasingly capable Chinese AI models is once again challenging the notion of how expensive artificial intelligence should be .
The market reaction was quite predictable. If AI models become significantly more efficient , companies may not need so many GPUs. If companies don't need GPUs, hyperscalers are building too many data centers. And if too many data centers are built, overcapacity can eventually arise throughout the semiconductor supply chain .
This is the argument of the " bears" (pessimists ). But there is a serious problem with this conclusion. We still see evidence that the demand for computing power is extremely high, and this creates one of the biggest contradictions in the market today. Investors are worried about a future overcapacity , while companies in the industry continue to talk about a shortage.
Memory is vast amounts of advanced memory. As demand increases, supply cannot keep up . This has allowed memory manufacturers to raise prices and systematically improve profitability. Naturally, the market is now asking how long this can last. And this is a completely valid concern. No shortage lasts forever.
High prices attract investment, and investment creates new capacity. New capacity eventually increases supply, and if it grows faster than demand, the price advantage disappears. This is how the traditional semiconductor cycle has worked for decades. But the question investors need to answer today is whether this cycle is fundamentally different because the source of demand has changed.
Historically, demand for memory has been significantly influenced by consumer electronics . For example, smartphones or traditional servers— these markets have experienced significant fluctuations. Today, artificial intelligence is creating a completely new source of demand. Hyperscalers, AI labs, cloud providers, enterprises, governments, data centers, operators.
These customers are making investments that are designed to last for many years, which could potentially make demand more stable than in previous cycles. Micron is at the heart of this debate. The company's shares have shown tremendous volatility, despite high expectations for demand for AI memory. Optimists argue that supply remains limited.
Customers are booking capacity, and spending on AI infrastructure will sustain demand for years. Pessimists argue that memory remains memory. Eventually, capacity will increase, prices will normalize, and the historic semiconductor cycle will return. Both sides have valid arguments, but what I think is most important is time.
댓글 0