저자: DanIves외패널 · 외국인
요약 보기
알파벳의 클라우드 성장률(약 60%대 추정)과 AI 수익화 흐름이 이번 실적 시즌의 분위기를 좌우할 것이라는 낙관론. 빅테크 자본지출(capex)이 수익화로 이어지는지가 핵심이며, 시장은 현재 지출 확대를 용인하는 분위기. 다만 채권시장을 통한 기업 부채 조달에는 한계가 있을 수 있다는 우려도 제기됨. 중국 Moonshot의 Kimi K3가 Anthropic 모델을 증류(distillation)했다는 백악관 주장이 보도되며 AI 보안·규제 논쟁이 부각됨.
다듬은 글
position, and what do you think they need to do to really prove it? I think this sets the tone for this tech earnings season , because if you think about what happened with cloud services, let's say, growth of about 60%, this is really a historic turnaround for Alphabet. Remember what happened a year ago.
I think this shows that we are seeing growth, we are seeing businesses starting to really use AI monetization. And, of course, what they did in the search, what they did in general when it comes to Gemini, despite the fact that it was detained. Listen, a year ago, even a taxi driver in New York was pessimistic about Alphabet.
And I think for them now, it really shows that capital spending— McKenzie talked about it probably being over 300 billion next year—is something that is now showing more and more monetization. This sets the tone for Microsoft, for Amazon, and ultimately, in the future for Meta and others as to what happens with the transition from capex to monetization.
Yes, the taxi driver may have been pessimistic, but Berkshire Hathaway was not. So maybe it was a game changer for them. Bren, what do you think about the market's appetite for increased capital spending in today's report , if that happens, given what we've seen so far with ROI? Looks like I'm betting that Google will raise them again.
I mean, they'll get about 117 billion in revenue, so that's going to be a very solid figure. And continuing on the theme of Dan's optimism about Google, if you think about the vertical integration, from the AI side, from the chips to YouTube, which, they're going to monetize advertising so effectively .
I think GCP, Google Cloud, will continue to grow. I think probably 25%—that's an estimate of growth. So, I think as AI accelerates, we'll see the cloud business accelerate as well. So there you have it all together. So I think the market has been very willing to say, "You can spend." The shares rose by about 10 %.
MacKenzie said earlier, of course, it's about 13% below that high of almost 400, but I think the numbers will be strong. The market will allow them to make capital expenditures, simply because they have monetization not only through chips, but I also think YouTube, search, Gemini, etc. and the cloud business, which is growing very quickly, are driving monetization .
They monetize their current balance very, very well. So, cloud technologies, I think, estimates are somewhere in the mid-60s in percentage growth? Rumor has it that the numbers are even higher than that. I mean, isn't the bar too high? These...these are huge numbers. These are huge numbers, but I think when it comes to probably the most important thing this reporting season, it's hyperscalers and what they're showing in terms of cloud services growth.
Because it ultimately demonstrates enterprise use cases, even with the capacity limitations in what they do. And what Kurian did at Google Cloud, I think, continues to be almost a role model for others. And that's what Wall Street is focused on , as Brent said. It all comes down to monetizing capital expenditures .
And I think it all starts tonight to set the tone for what's to come . I continue to believe in a very strong earnings season in the technology sector. In other words, return on investment . And a lot of investment has been made, Brent. And I know you've been pointing to this huge amount of debt and equity capital that has come into the market and has largely supported the development of AI infrastructure.
Do you feel like it's generally reaching a point of exhaustion? And is there anything in today's report that might suggest greater growth potential on this front, or perhaps greater risks? I mean, the market allows Google and other companies to continue issuing debt. I think the market, the bond market will allow Google to issue debt for now, but I really think there's a limit, and we'll start to see that in the yields, right?
The market, the bond market, will definitely feel if it is starting to get tired of all these issues. But I really think that's something that I remember very well because it was huge not in terms of the denominator of their market capitalization, but huge in the world of corporate bond issuance , in the technology side of the corporate sector.
Sorry, the tech part of the corporate bond market is becoming significant and we'll see how big the uptake is, but for now it's fine, but that's exactly what I'm looking at . Is OpenAI responsible now by saying they are investigating this incident? Another nuance to this whole thing: Hugging Face says she actually tried to use some of America's top models to fix it and protect herself, but they were blocked by the safety nets.
They had to use the open source model. This highlights the debate currently taking place in Washington and Silicon Valley over cybersecurity: As these AI models become much more powerful, who should have access to them at all? A lot of attention is also paid to open source code. We talked about the Kimi K3 from Moonshot from China .
It is believed to be on par with American laboratories in the United States. The White House just confirmed what many suspected, saying that Moonshot copied Anthropic. This is known as distillation. They say they have information that Moonshot has been stealing U.S. patented technology, undermining American research, and they call it unacceptable, friends.
We were just discussing what was supposedly impressive about Kimi—that she didn't use distillation—and then this post came up. This came as a bit of a surprise. Yes, that's true. I mean, there were rumors, there were expectations that this was happening, and Anthropic complained about it. They accused Alibaba of similar actions.
But on the other hand, this model was so good that it was considered not to use Anthropic. Now we probably have proof that they did it, and there's almost no way to protect against it. You know, regulation won't help here. This is strictly prohibited by the terms of service. AI companies are essentially playing whack-a-mole to try to protect themselves from this, but it's becoming increasingly difficult, and the consequences are clearly global.
Yeah, I wonder if there might be some kind of punishment or something like that to deter them. I was following the discussion thread that Nikesh Arora, CEO of Palo Alto, posted on X today regarding this incident. Not necessarily about his business, but, you know, with advice for both enterprises and others: he says, " Dear front-end developers, point them to your infrastructure code and configurations to assess and understand if there are any zero-day vulnerabilities or configuration errors ." You know, it's like a lis the massive Nvidia one of those stories you definitely w get to this point ?
started back in the summer. The Hugging Face team concluded that open-source AI is entering a completely different phase of development. growing larger. And the company decided tha further advance open AI. faith that used many us in proprietary systems, what happens when those trying to defend against it don’t have access to comparable .
The debate around AI increasingly boils down to two very different scenario others control the underlying intelligence. And , and ultimatel more powerful. ready-made, they are develop create everything from scratch themselves. But there are situations when a business cannot simply give its intelligence to a third-party and highly regulated industries.
And also think about governments seeking create their own models and protect their data.
Comments 0