저자: 미상 · 외국인
요약 보기
AI 데이터센터 수요가 공급을 초과하며 막대한 자본지출이 이어지고 있다. 엔비디아는 최근 분기 85% 성장, 다음 분기 90% 이상 성장이 예상되며 실적 발표를 앞두고 기대치가 높다. 브로드컴은 구글 TPU 관련 마벨 수주 소식에 압박을 받았다. 메모리 업종은 80% 이상 마진과 공급 부족 속에 있으나 추가 공급 가능성이 있다. 중소기업의 AI 컴퓨팅 비용 부담에 대한 우려도 제기됐다.
다듬은 글
are elected at certain state levels because they are concerned about electricity prices for consumers, and that is a perfectly valid question. But other states that have more capacity and ability to transmit energy without driving up prices for consumers will be the main winners in economic growth. Yes.
So, last quarter, we contracted almost 1 gigawatt of capacity in Pennsylvania. Do you understand? We are funding another large data center in another state, and in my conversations with all the hyperscalers, one thing rings true: demand exceeds supply . And yes, you are right . Previously, these large companies were light in terms of assets.
And now, their business has changed. This is precisely the role of capital markets. And I promise you, it's going to be a great investment for individuals, if we can get the underlying business model to eventually generate enough revenue and profit to pay for all these data centers. They are not cheap .
Yes, they cost 50, 60 billion dollars. dollars for 1 gigawatt. Dave, I'm not concerned with the demand side. I'm concerned about whether all of society can benefit from AI? Because computing power is very expensive now. My question to all hyperscalers is not about whether they have enough demand. And how quickly will they be able to reduce the cost of computing?
I'm not worried about BlackRock paying for AI, because we've been a big investor in this space , and our systematic equity investment team has really benefited from the flows and returns. But I am worried about small and medium-sized businesses. How will they be able to compete in this new world of AI technology?
So the question I'm asking behind the scenes is: How quickly can we reduce the cost of computing? And they tell me that it's like Moore's Law, and how fast they can go, an important component here—is that the cloud platform is playing a bigger role in this build, trying to provide sufficient computing power.
And this comes amid news that Marvell may receive a large order from Google Alphabet to build their TPUs. That's what put pressure on Broadcom this week. But it's no surprise, Jenny, that this financial deal could reach almost a hundred billion dollars at this point. But is the market comfortable with a company that is showing excellent results taking on such a large amount of debt?
It seems like the market is comfortable with this for now. Yes, and I think you're right . So, I think the main catalyst is not so much market concern about Broadcom, but rather Google losing critical market share and some of their spending on AI infrastructure to Marvell, which of course announced a deal with Google earlier this week, which also led to Broadcom shares falling.
However, next Wednesday after the close of trading, Nvidia will report its earnings, which I'm sure we'll be discussing almost every day until then . But [ __ ] is now starting its analytical coverage with a more optimistic outlook. So, in your opinion, what exactly does Nvidia need to prove to keep up with this AI trend?
Because the last few earnings reports from this company haven't been a "bullish" catalyst for the stock, let's say. Yes, if you look at Nvidia's latest reports, there were stock sell-offs after them. Not big, but at the same time, you know what else they can do? Jensen Huang, CEO , has also mentioned this on several recent earnings reports.
That is, 85% growth in the last quarter alone. This was better than expectations by more than 5%, but the market still wants more at this point. Growth rates this quarter are expected to exceed 90% year-on-year. So I think that's what the market will focus on as the stock recently broke out of the 190 to 210 range, but we're already down over 3% this week before this report.
So I think this is going to be a key factor for the markets because it's the largest component of the S&P 500 and NASDAQ 100 indexes. So, it's going to impact the markets. Today I received a note from [ __ ] Capital where they reaffirm their “outperform” rating, setting a price target of $340. The analyst noted that their integrated technology stack is a leading support for large language model ecosystems.
It's also worth considering that they might be diversifying some of the money they're putting into this AI infrastructure funding cycle. This is already bearing fruit, but I think the market will focus on their growth rate. They effectively sold out the entire Blackwell volume for the next 12 months.
Uh , what kind of projections they're going to make for the future, whether they can sustain that growth rate, is going to be , uh, really a key point here, Jenny, uh, for Nvidia and the entire market. I know, if not support, I would even say accelerate, because it seems like a quarter of exceeding expectations and raising forecasts is not enough right now.
You have to, I mean, show massive acceleration in growth, which, in Nvidia's defense, I mean, they've been doing for several quarters. The market does not always perceive this as positively as the actual numbers indicate. But thank you very much, Tom White, for mentioning this. Nvidia is approaching its next earnings report.
But this time, the main the same . The price depended on supply and demand. Manufacturers expanded capacity during boom periods, and eventually the excess supply again reduced profitability. more information. And this is where the supply problem gets takes years, setting up equipment takes time, and manufacturing advanced memory requires high technical expertise .
So, even though memory manufacturers are currently enjoying extremely high margins, additional supply is not coming in fast enough to immediately eliminate the shortage. That's one reason, and this is another year of massive AI-related capital spending by margin that's capitalism doing its thing. A margin of over 80% doesn't exactly whisper, "Please ignore me." She tells .
Therefore, the question is not whether competition will appear. She will almost certainly appear. The question is when exactly this additional supply will aroming to an end. Despite the huge profit growth and limited supply, this creates a strange situation. A company can show extraordinary earnings growth rates, while its shares will still trade at a relatively low P/E ratio because investors do not believe in the sustainability of these earnings.
This is the main limitation. Is memory powerful cycle, or has AI fundamentally changed the structure of long-term demand? I think the more accurate answer lies in the beat itself. Memory is still fundamentally driven cycles . Demand comes not only from modern AI I-based enterprise software, AI infrastructure , more advanced reasoning models.
All of these applications require computation, and computation requires memory, so memory shortages will become one of the defining constants of the next stage of AI infrastructure development. Now let's get back to Nvidia. Nvidia shares have been rising recently on renewed optimism about AI spending and Elon Musk's comment about using Nvidia architecture for SpaceX technology.
However, the stock is still below its all-time high, and investors are awaiting the upcoming earnings report with extremely high expectations.
Comments 0