저자: JimCramer · 외국인
요약 보기
AI 데이터센터 논쟁 속 엔비디아 주가 하락에도 Cramer는 보유 유지. 중국 등 저가 오픈모델 확산 우려에 대해 Cramer는 Anthropic 등 B2B 강자와 사이버보안주에 긍정적. 오픈모델 확산도 결국 컴퓨팅 수요를 늘린다는 논리로 엔비디아에 우호적 입장. 엔비디아가 최고의 대차대조표를 보유해 경쟁을 버틸 수 있다고 평가.
다듬은 글
Let's move on to the Stop Trading section with Jim Cramer. One of the most difficult parts of this data center story is the new open model situation being promoted by Nvidia. Where all the different hyperscalers and other big companies can use cheaper Chinese models instead of using, say, the Anthropic model.
I now consider Anthropic... as it is a nonsensical line of thought. Anthropic is an incredible B2B software development company. You wouldn't sell it if it were public, but here's what happens : people say, " Come to your senses." Now we have to be aware of open models and maybe not as concerned as I thought about cybersecurity, maybe keeping...
Well, that doesn't... Doesn't that explain some of the pricing dynamics of Crowd and... Explains, and I think it's ongoing... Of course it explains, and I know Crowd and Palo Alto both... Crowd was one of the early signatories. So I think cybersecurity companies will feel very good. I also want to say at this point: look, I know Nvidia is falling, but I still own it, I'm not trading it.
I find it absurd, but I understand. I think people feel they will have to turn to these reserves. They have the best balance sheet of any company in the world. If they want to use it that way, I think they can succeed, especially since the government is also a kind of hidden guarantor in all of this.
Our government, this government. Yes, because they don't want China to win. This is important. Let's not forget that this is also geopolitics. Very serious geopolitics. Without a doubt. You know what's scary: the Secretary of Commerce controls a lot of what happens in Washington. Well, certainly this deal in Ohio that we talked about.
Yes, let's not lose sight of this . You know, I don't know how much the government will earn this year, but I think they might have to increase the quarterly figures. Jim, get ready. There's still a busy week ahead. See you tonight on Mad Money. At 6:00 PM Eastern Time with CRH . OK. The person who conducted this interview is a fascinating person.
Zennie, I think you did an extraordinary job, and we were all, you know, glued to our screens. I'm curious. I mean, you touched on so many topics. You touched on the topic of AI , the whole idea that he really thinks that in 5 years AI will be even smarter, maybe even smarter than him, and what that will mean for his own power.
You mentioned some things that happened with DOGE and its role in the government, USAID. What was that? What was the moment during the conversation that made you think, “Wow, that really surprised me”? Oh, there were so many of them. I think he's an extraordinary person, and there were many moments when I was simply speechless, at least mentally.
I was struck by his vision of the future, namely the fact that he now sees it as inevitable that, as you say, in 5 years AI will be smarter than humans, and in 10 years he paints a picture where we essentially become something like labrador retrievers for AI. Money will become irrelevant. It will be an era of such abundance, but control will be in their hands.
And you know, Andrew, you know his story, this is a person who said on a podcast in 2015 that this is one of the scariest things she could imagine. In 2023, he wanted to stop the development of AI. And now he's, you know, he's right on the forefront of this process. And I said to him, "It looks like you've given up on your fears and are just going with the flow." And he's like, "Yes, yes, I'm going with the flow." He admitted that he can go from excitement to horror in one day.
But I was really struck by the inevitability with which he believed this technology and the speed of its development would mean that this extraordinary future would be here within five years. I mean, I was really impressed by it . And at the same time, he seems to be, for the first time—and you showed this in your video— expressing a willingness to collaborate with other advanced model developers to create some kind of— and he wants it to be informal, Elon-style, right?
He wants them all to get together, talk to each other, and test each other's models for a week, and you know, if they find any security holes, let them report it to the developer of that model, and only if the developer doesn't do anything, then go to the government. But I was struck by how similar this is to Demis Hassabis' proposal last week for a FINRA-style public- private partnership , and I think it now really has promise.
I think this will lead somewhere. He seemed to hint that he would even be open to talking to Sam Altman on some level. Ahem. Certainly. I mean. He hinted at it, considering what he said. He said it. No, he absolutely said that, because I said, " Look, you're not just big competitors. A lot of you don't like each other, and frankly, you throw dirt on each other on social media quite a bit." And at the end, he said, "Well," he explained to me why he does n't like Sam Altman, but then he added, "You know, maybe we should talk for the good of humanity." So I think there's a sense of responsibility, he realizes that the leading modelers have to do this.
And I said, " Does it have to happen quickly?" "He said, 'Yes.' And I said, 'Well, yes, I'm told it has to be done within the next 6 months.'" He said, ' 6 months—that's a very long time.' It has to happen now." Okay. I want to show a snippet of another part of this interview that's going viral. It's because you seem to have used the word " far-right" politics at one point, and he did n't react in the best way.
Let's look at that for a second. These are normal people. And here are the principles. Tell me, which of these sounds terrible? That we should have secure borders, that we should have safe cities, that we should have sensible spending. Which of these three things is far-right radicalism? I would just like to rebuke you and the media for your absurd characterization of the " far-right." It's false, misleading, and nonsense.
Please leave that part out. We'll leave all of it in. Don't cut that part out. Let me object. Let me object because you have nonsense. You know how many of you have there? Almost 250 million followers on X. Yes, because I You were saying things like I'm a popular guy. quote: "A civil war in Britain is imminent." Just a question.
Hi everyone, and welcome to my channel. Before we get into Nvidia's huge deal, , you definitely don't want to miss thibout $12.93 billion. And this isn't just another Nvidia purchase of community of developers who are get to this point? started over the summer. The Hugging Face team had getting more powerful, and the community was getting broader.
And the company fel push open AI further. belief in open AI. The company had previously suffered a cyberattack using cutting-edge AI capabilities. What struck the team was that some of the models involved were not publicly available. They were closed systems. And that away in a proprietary system, what happens when the people trying to protect themselves from such systems don’t have access to comparable technology?
. And the debate around AI is increasingly leaning toward just access their models through an API and pay for the service. And you let someone else control the underlying intelligence. The other is an open ecosystem. Developers can access the models, modify them, and build their own systems. Ultimately, having .
But Jensen Huang isn’t saying that closed AI will go away. ’re already built, they’re already build everything themselves. But there are a million situations where a business can’t just outsource their intelligence to an outside vendor. Or think about scientific research. Or think about companies with their own unique datasets.
Or think about industries with tight regulation. Or think about governments that want to control their own AI infrastructure. In those at’s where open AI comes in. AI huge I and closed AI. It can support both. If companies build conventional closed models, Nvidia sells the computing infrastructure to train and run them .
If companies increasingly move to open models, those models also require a huge amount of computing power.
Comments 0