저자: DanIves · 외국인
요약 보기
AI 인프라 수요가 여전히 공급을 크게 초과하며 AI 슈퍼사이클은 초기 단계라는 주장. 스페이스X 발사 차질은 단기 노이즈로 보고 장기 데이터 수요를 강조. 마이크로소프트·아마존·알파벳의 실적과 삼성·SK하이닉스의 다년 장기계약이 메모리 수요 지속을 뒷받침한다고 봄. 엔비디아·반도체 급락은 2분기 실적 발표를 계기로 반전될 수 있다고 언급. 중동 긴장과 유가 변동성은 불확실 요인으로 제시.
다듬은 글
Nice to see you again. here. We'll talk about the new firm in a minute, because that's a separate story, but for now, we need to discuss this : Remember, before you left your previous job, you had just started doing analytics on SpaceX and were one of many analysts who were bullish on the stock. But why isn't there the growth from analysts that Leslie just talked about?
So, look, I think when you look at SpaceX, which obviously has some significant launches coming up, the most important thing, I think, is that long-term investors will see these as minor setbacks, because ultimately you have to look at this as the story that we've been talking about. This is a story about data for AI.
Space will increasingly become a huge part of the AI revolution. In my opinion, covering technology for decades, if you look at Facebook or, say, the IPO, what it did, look at Amazon, it's very easy to become short-sighted. You know, it's a 26-mile marathon. I think we're only on the first or second mile.
Yes, I hear you. I mean, there's a lot of commentary around this, but the fact remains: it's hard to assess. And in fact, this is one of the problems. Normally you look at companies, you can look at all the different metrics and determine a price you think they should be trading at, right? Valuation of cost relative to profits.
But this is a different case. This is difficult. See, companies like this one won't have " evaluation support." And I think that's part of the problem. When you get into it, you know, the domino effect starts. But as we said, whether it's SpaceX , or Tesla, or many other transformative technology stocks, you have to be able to see the forest for the trees and look 3-5 years ahead.
And we're in , it's hard to imagine it without mentioning SpaceX. What do you think about what continues to happen with chips and what megacorporations are doing in terms of trading today and over the last month? Well , they definitely woke up. Yes. But the chips remain very, very difficult to analyze.
It all comes down to SK Hynix . These are the golden children of this AI trade. You've seen a massive sell-off since SK became a public company. But if you look at Nvidia and many other chipmakers, it all comes down to second-quarter reports. This will truly be a ray of light in a dark tunnel. We've already talked about this.
Demand exceeds supply by 15 times. This AI revolution is still in its infancy, but hyperscalers are the ones funding the AI party and the capital expenditures. In fact, they are the ones who are in the most difficult situation right now. That's why it's so important for Alphabet, Microsoft, and Amazon to enter this reporting season, which I see as a turning point.
The story of demand is still relevant. to talk to you about a number of these companies. Let's take Alphabet, because it's so relevant given Warren Buffett's words. And that was my choice. Mm. And he believes they will be winners. What do you think about this? Look, Buffett talks about valuation; Look at Alphabet a year ago— even taxi drivers in New York were pessimistic.
And now , given their successes in search and cloud technology, it looks like a smart valuation in the AI space for investors like Buffett. You mention SpaceX and other companies, perhaps on the other side of the spectrum. This is a perfect example that fits in well, and they are at the epicenter of the build, especially when it comes to search.
Intel reported earnings, as you mentioned. They are in the black, but not as much as before, and, you know, some macroeconomic indicators, like the price of oil, are lower, but there is no good news from there. Iran rejected the US ceasefire proposal, which was conveyed by Iraq, about an hour ago, so we don't have any good news yet.
The dollar rose slightly, but bond yields fell slightly along with oil prices. So it's definitely a mixed morning, Nicole, after a pretty big drop yesterday. Futures have been rising all morning, but are now gradually approaching the zero mark at the start of the day. Yes, and yesterday Marco Rubio said that Iran is not ready for a deal, and I said that the memorandum of understanding really added optimism to the market.
People thought the end was near, but now it doesn't look like that because of the escalation in the Red Sea and the Houthi intervention. We have added a new element to this war in the Middle East, and that has brought some uncertainty. But it was interesting to watch USO and oil grow. Carly Garner of DeCarley talked yesterday about the short covering that was happening in oil, and that changed oil trading, assuming that the price would fall again, which is what actually happened.
Today it fell to 60 after jumping more than 5% yesterday. Your thoughts? Yes, you know, fundamentally oil is in very good shape , but destructive factors are emerging all around it. Some good news is now trickling into the market, such as Saudi Arabia starting to use other routes to transport millions of barrels through the region.
But now the Houthis are attacking these routes as well , so we can be sure that people will find a way to get the oil where it needs to be. But all of that is broken, and yes, oil futures are down this morning, but they are barely below $90. They are still at 89.60, or as I see it now, 89.67. So, they're still incredibly high compared to where they were—below $70 a few weeks ago.
So serious problems remain. As I said, the rhetoric is escalating, not subsiding. In fact, it is only getting worse. President Trump is talking about bigger steps. So the pressure is there, right? Something strange is happening on Wall Street. The world's largest AI companies are reporting incredible performance .
Microsoft, Amazon, Alphabet, Meta, even Samsung and SK Hynix. However, Nvidia and many semiconductor manufacturers' stocks crashed. So what's going on? Has the artificial intelligence boom reached its peak? Is the semiconductor cycle finally coming to an end ? Or is Wall Street making the same mistake it makes during every major technological revolution?
Today, we'll break down everything that happened and why I believe this reporting season could completely change the history of AI investment. For a month, investors have been asking one question. Can companies actually make money from all this spending on AI? Microsoft just gave a very strong answer.
Azure growth accelerated, the number of paid Copilot users increased astronomically. Management said it expects to maintain positive free cash flow despite the huge investment in AI. This is incredibly important as investors have begun to fear that spending on AI will simply destroy profitability. Instead, Microsoft showed us something different.
They are monetizing AI faster than many expected. This changes the whole discussion. Amazon shocked Wall Street. Amazon reported, and honestly, these numbers surprised almost everyone. AWS revenue has accelerated to one of the fastest growth rates in recent years. Operating margins exceeded expectations, and management again increased capital spending from about $200 billion to about $220 billion .
Typically, higher capital spending has hurt stocks recently, but not this time. Why? Because Andy Jassy explained what investors wanted to hear. Amazon doesn't build data centers in the hope that customers will one day show up. The demand already exists. New AI capacity is being leased almost instantly.
Even more surprising is that Amazon has stated that demand for 2028 capacity is already extremely high. Think about it. Companies prepare infrastructure years before it becomes available. This doesn't seem like a slowdown in AI development. The shortage of computing power is not going away. This is where many investors get confused.
People are constantly talking about competition, Chinese AI models, new startups, cheaper alternatives , but all of these models still require one thing—computation. Whether you use OpenAI, Anthropic, Meta, DeepSeek, or Amazon, somewhere servers are running, GPUs are running, memory is being consumed, storage is being filled , and network equipment is transmitting data.
Any AI model ultimately depends on physical infrastructure. And, according to almost every hyperscaler, it's still not enough. Samsung and SK Hynix sent another powerful signal. The memory business delivered another important message. Samsung has announced multi-year long-term agreements with major cloud companies.
Many of these contracts are designed for a minimum of 5 years, include minimum prices, and some require upfront payments from customers. And this has fundamentally changed the traditional memory business.
Comments 0