저자: JensenHuang · 외국인
요약 보기
젠슨 황은 AI 수요가 폭발적으로 늘고 있으며 토큰 생성 수요는 1년 새 25배, 공개 토큰 비중은 30%에서 70%로 증가했다고 주장. AI 공장 1기가와트당 비용 500~600억 달러, 임대료 연 500억 달러로 투자수익률이 약 1년이라고 설명. 엔비디아가 개방형·폐쇄형 모델을 모두 지원하는 유일한 플랫폼이며 컴퓨팅 자원을 담보로 활용 가능하다고 언급. 최근 GPU 임대가격(GB 기준 시간당 16달러, 1년 전 5달러)이 올라 투자 수익이 늘고 있다고 강조. 엔비디아 주식 보유를 권유하는 진행자 멘트 포함.
다듬은 글
beer, but I don't think you would miss it for anything. This is incredible. This...this is the global epicenter of sales. Yes. Commerce. Why not? Do you know? Well, let me ask you something . Creating opportunities for the whole world. I...before I saw you, I went to my trusts and estates lawyer and said, " Listen, if I know I'm going to die in 2030, is there anything I can do to help my children?" You question this .
Will we all die in 2030? Maybe it won't matter. We're not...we're not going to die in 2030. There are so many people in the world who are going to build AI correctly. We will have all kinds of safeguards, we will invent all kinds of technologies for safety and protection. All this is happening all over the world.
These are not just two companies building something. There are thousands of companies building something. So, we're going to...we're going to create an AI that's going to be incredibly useful. It will be incredibly helpful and it will be incredibly safe . We can have it all at the same time. Because this is an engineering problem.
And this... Safety is an engineering problem. Testing is an engineering problem. We must build products and test them properly . And if they're not ready for release, just hold them, keep testing and improving them until they are . And that's just common sense . I...I...I believe in it. You told me two years ago that companies would be able to buy something from you and get four times more than they thought.
It was conservative, as usual, right? You lowered expectations because they are already making money now. They just don't like to talk about it. Well, the really incredible part is that in the last 6 months, AI has become super useful. And because it's super useful, people are building these computing systems, Nvidia AI factories , like crazy.
Each gigawatt of Nvidia's AI factories costs about $50-60 billion . OK. However, the rent for this gigawatt AI factory is about $50 billion per year. For a year. Of course. Yes, the return on invested capital is currently about 1 year. And they can use it because the interchangeability and durability of Nvidia is so high that they can use it for many years, much more than 5 or 6.
So the return on that invested capital is incredible. That's why he...he's building it. No, it's difficult, because a closed model, so to speak, Frontier can probably bring in three times as much. But at the same time, we need both open and closed, right? You should rent and use the best models available as a product or service, these closed models.
However, every country, every company, every industry must also build its own intelligence. How can you completely outsource your intelligence to someone else? And that's why all these companies are here. This is the reason for our partnership with Salesforce: Salesforce will use NVIDIA NeMo and our agent technology to build their own agents that run on Salesforce.
Also, because these agents are going to be so fast and there are going to be so many of them, the Salesforce data processing layer, it's on top, it's in Salesforce, it's in...Databricks, it's in Snowflake. We are accelerating data processing in all of this. So, you will get lightning- fast data processing .
You will have many agents to help you with your work. And all of this is based on NVIDIA's NeMo . Thus, Salesforce also becomes an AI company. Well, then it will be the same as in all previous industrial revolutions. In the beginning, everyone said, "Listen, the loom will put a lot of seamstresses out of work." And it was a difficult situation for them.
True, but after 4 years the number of jobs increased tenfold. You know, these stories are annoying, unnecessary, and, frankly, just wrong. The fact is that when a company becomes more productive thanks to artificial intelligence, it becomes more ambitious. Their ambition allows them to imagine and strive for more, so if you look at what's happening at Salesforce, they were the data layer and the application layer.
Now they are also a level of agents on top of that. They are much more ambitious about the future of Salesforce today than I have ever seen Mark, and he is always enthusiastic. And today he is simply charged. So AI helped his company raise its ambitions. As a result, they will be able to reach new heights.
This will happen not just at Salesforce, but across all Salesforce customers, across every single industry. It is completely illogical to believe that if we had agents to write code, engineering would become unnecessary. I grew up in a generation where engineers didn't have software. We used all sorts of other tools and built prototypes.
So, engineering existed before software. Engineering will exist even after the advent of AI. There are no such clips. This is "no". It just doesn't make sense that Yes, you were with Mike on that popular All-in podcast. You got a call in the middle of a recording. That was a little strange. Was this planned?
No. No. I was on stage and out of the corner of my eye I saw my team waving their phones in the air backstage, and while I was trying to figure out what was happening, they came closer and said, "Mr. President." And, of course , you can't, you can't not answer this call. Yes, yes. He is smart. I did n't even pick up the phone.
When the president calls, you answer. And so, he just wanted to talk to me about some tweets that he posted. And I told him I was right on stage with the guys, and he said, "Let me on air. Let me talk to everyone." And so I turned it on. Here's what happened. OK. Good. That was good. So, it's completely unplanned.
He used the term "mystification" . That caught a lot of people off guard, but the consensus of that panel was that there are things that are made up. You would agree with the term "fictional". Is n't it? How to stop such things? AI safety is a real thing. OK. Um, designing products to be safe for use in the world—that's a real thing .
And companies should innovate as quickly as possible, but they should never do so quickly that they release unsafe products. This is a real thing. However, the fact is that we need new laws, new antitrust laws or new regulations so that these companies can do fundamental design and do it properly before releasing products.
It's simply not necessary at all. And I think that's what the president means, that we don't need new laws. We don't need, yes, we don't need new ones, we have plenty of laws. We have a lot of rules that govern the reliability and functionality of products. And you know, there are many, many. Yesterday was a strange day .
I mean, your stock opened down 5.5 percent because everyone is saying OpenAI won't go public , so you won't close the quarter, and so on, and so on. When we realize that you're selling at 14 times earnings for the '28 fiscal year, which starts in January, and that you have so many advantages, and the best investment in the world is what?
Nvidia. Well , to work. Let's buy a quadrillion. Jim, listen. This is incredible. We are the only platform that runs any closed model. Yes. Each individual closed model. And we are increasing our share in the closed model segment. We are the only platform that runs every open model, including Nvidia with our own NeMo Triton models.
Demand for token generation has increased 25 times in less than a year. And the share of public tokens generated has increased from 30% to almost 70%. So, the demand for AI, because it is now useful, is growing exponentially. Nothing that happened this week will slow down this process. In fact, what's likely to happen is this: we used to have learning systems and inference systems, giant systems.
Now we will have a third system. This third system is essentially a testing environment. And these testing systems will be extremely large so that we can have products that are functional, capable, safe and reliable. And testing them requires huge data centers. So, we just created another new demand for AI.
This is brilliant. What about whether we have enough computing power? And when will we force Wall Street financiers to start trading computing power? Well, Jim, this is a big breakthrough in the last few years. People now realize that NVIDIA computing is the only system that is interchangeable and durable.
The reason we are interchangeable is because we run everything. We launch each individual model, closed or open. We cover every stage of the AI lifecycle: from data processing to training, post-training, inference, and of course, testing. Every company uses us, and as a result, it is now the world's first computing system that can be used as an asset.
So, you can use computers as collateral . And we are seeing more and more evidence of this . If so, this will be a huge revelation for all companies that do not have the ability to finance this level of scale, as the asset itself will be collateral. So this is a moment in time that we will all remember as the one where everyone who hired you and bought your product got a break-even point and a profit for those 6, 7, 8, 9 years.
The probability will be extraordinary. Can you help explain this to people? I'm tired of talking about it . The strange thing is that so many of the first tenants signed long-term agreements, long-term leases , long-term contracts. And now all these cloud providers are regretting it, because due to demand, spot prices, rents, have become simply unbelievable.
You can now rent Nvidia Grace Blackwell for $16 per hour of GPU work . And a year ago, some of them were rented for just $5. So now that these contracts are expiring , everyone is just raising prices; they are raising their revenue forecasts. The return on investment is increasing, they are buying more equipment, they want to build more infrastructure, and I think this flywheel is really spinning at full speed now.
Well, I'll say something you may have already heard. Nvidia —own it, don't trade it. Own it, don't trade it. This is what I do. Thank you very much, Jensen. Nice to see you, CEO, co-founder and all . Everyone, Jim Cramer.
댓글 0