저자: JensenHuang인터뷰;진행자John/Becky제외.영상채널FinVid발언자미상 · 외국인
요약 보기
젠슨 황은 엔비디아가 모든 AI 기업과 협업하며 개방형 모델 전략을 유지한다고 밝힘. 루빈(Rubin)은 블랙웰 대비 토큰 비용 10분의 1, 토큰 처리량 10배, 연산 속도 4배로 제시되며 양산 중. 피지컬 AI(로보틱스·자율주행, 메르세데스 벤츠와 Q1 출시)를 강조하고 2030년까지 3~4조 달러 AI 팩토리 투자 전망. 진행자는 엔비디아 실적이 2월에 강할 것이라 전망하며 아이렌(IREN) 장기 강세를 언급.
다듬은 글
John, this is a big deal. What did what did Jensen have to say? Uh yeah, a lot of big news coming out here, Becky, and it's it's dark and early here in Vegas. I spoke with Nvidia CEO Jensen Huang after his keynote yesterday, after his talk with analysts, where they pressed him on China demand, overall demand, with both of which he said remain strong.
And I asked him about the rationale behind his Grok licensing deal. We actually ended up talking for about half an hour. Early in the conversation, he addressed open models as part of Nvidia's strategy. You listen. We're the only AI company in the world working with every AI company in the world. We run every model from OpenAI to This year, we announced Anthropic, and I'm very excited about that.
Of course, we have the benefit of working with Elon and xAI. We've always been working with with Google on Gemini. And so, Nvidia's the only company in the world that runs across every domain of science, every AI company, every AI model. And the only way to do that is to do it openly. If we were to be proprietary in some way, closed in some way, then we won't have the benefit of monetizing the entire layer and the entire stack, but it prevents our ability to be open.
And we I really love that our company has the ability to work with everybody. Robotics. It's been a dream part of CES for a long time, but it's also been elusive. I mean, we had iRobot kind of not go in a great direction. Um granted, we do have autonomous vehicles now, and those are really robots, I guess, right?
Cuz we're talking about physical AI. What makes this moment different? xAI robotics in particular, and maybe give me humanoid, because we tend to get very excited about it, but I wonder too excited. Excellent. Excellent. Timing is everything. We've been thinking about this area for a long time, waiting for that moment.
And and as in many of the things that we do, um whether it's digital biology and turning drug discovery from a from a discovery process to an engineering process, from a scientific process to a science and engineering process, to self-driving cars to even the work that we do in computer graphics, ray tracing is now completely done in real time.
It took us 30 years to do it in real time. And so, you have to pursue something for a long period of time, looking for that moment where that enabling technology is discovered. In the case of humanoid robotics, let me let me first say that a computer doesn't know and doesn't care what kind of tokens it's generating.
It could be generating a language token, it could be generating a video token, it could be generating a steering wheel activation token, or it could be generating a finger articulation grasp token. The computer, in the final analysis, is just a bunch of numbers. The moment that I saw generative video, us texting into a prompt, two people sitting having a conversation, the guy wearing a jacket reaches out, picks up a cup of water and drinks it.
I could describe that into a generative AI model today, and you and I both both know that Chris, for example, Yeah. right? With Runway. Runway, yeah. I could give him that prompt, and he will generate an amazing video. Isn't that right? Yeah. And that video has a person reaching out and picking up a cup.
Why is that model different than a generative model for a humanoid robot picking up a cup? The moment I saw that working that well, then the rest of it is still a whole bunch of research, a whole bunch of whole bunch of technology, but [snorts] you could tell that the enabling technology is just right around the corner.
Tell me how much more performant Vera Rubin is. Cuz I get questions, got one on Fast Money, how much does it cost? And I guess the flip side of that is how much value does it deliver? That's exactly right, yeah. you got some charts on the screen behind me, you know, showing the the difference in value between Hopper, Blackwell, Rubin.
What what are some of the most compelling, you think, stats now that you've got these six chips in production and you're starting to really pump out some some best ones. down It's a ton of technology and a whole bunch of software, but it comes down to three things ultimately. The first thing is this.
The big picture is to think about these computers not as supercomputers, but as AI factories. These AI factories are producing what is called tokens, numbers. And so, you you have to use you have to use the factory for three things, three fundamental things. The first thing is to train your next frontier model, so that you could build the best AI, get to market first.
So, you want to train it as fast as you can. Comparing Blackwell to Rubin, Rubin is a 4x leap. And so, that 4x jump, you could either do something in 4 months or do it in 1 month. So, it's a very big difference. You could either do something with four times as many Blackwells or 1/4 as many. Save a ton of money.
You could decide to get to market sooner, or you could train your next generation model cheaper. You decide. And so, that's training the model. It's really about the next frontier. The second thing is that you have to generate the tokens as cost-effectively as possible, because every single generation, the token costs are coming down.
Okay. Now, when token cost comes down, you can use more of it. That's a very big deal. And so, we need to get the token cost to come down, and Rubin reduces our token cost by a factor of 10 over Blackwell. How much of that is because of energy efficiency? It's energy efficiency, it's algorithms, it's chips that go faster.
In the world of AI factories, going faster is the same as going cheaper. So, fast getting something done more quickly costs less to do. And the third thing that we have to do is the overall throughput of that AI factory has to increase, or the revenues of the company that builds that factory doesn't increase.
And so, we have to make sure that every single generation, we increase it tremendously. In the case of Rubin, the token throughput of an AI factory increases by a factor of 10 times. And so, the value is unquestionably incredible. And the reason why we're able to do that, even though the number of transistors we used was only 1.7 times, this is where Nvidia's ability to integrate and innovate across the CPU, GPU, the GPU, the networking, the networking, the switches, the scale-up switch, the scale-out switch, and the data processing engine, we literally innovated across every single chip of the data center.
This Vera Rubin this generation, every single chip changed. Six brand new chips, every one of them revolutionary. And it's by it co-designing it. It's called co-design. It's basically innovation innovating across everything at one time. Guys, what we see playing out here is the the classic uh ecosystem battle that you see when any major technology is on the cusp of trying to go mainstream.
We had Lisa Su from AMD giving her keynote last night as well. She'll be joining me on Squawk on the Street first on later today on CNBC. We had uh Qualcomm's uh CEO Cristiano Amon launching a larger robotics platform, physical AI, very much what what Jensen was talking about here, trying to make a play to take AI deeper into the overall economy, into every type of technology, and sort of assemble the ecosystem around them.
Of course, Nvidia is Nvidia. They got a bit a bit of a lead in everything, and Jensen not showing any signs that he's looking to release his grasp there. Yeah. You know, it's funny. We talk about Nvidia as a $4 trillion company, if you're not paying close attention, you haven't seen it continue to climb.
It's actually 4 and 1/2 trillion, 4.57 trillion, to be clear. Dan Ives, I know, has a note out where he thinks it's going to be a $5 trillion market cap, or it'll exceed that, and could even make it to 6 trillion. But how much I I guess the question is, does that rely on these new markets that he's talking about with robotics and autonomous vehicles and all of those things taking place?
And how much of it is just the chips' improved performance? It seems like the market the investors, at least for now, are kind of watching the chips themselves, and maybe everything else is potential gravy down the road. Well, Becky, I I I asked him about that. We were getting into it a little bit there when he was talking about all the different areas of the system that are sort of getting upgraded.
And there are a couple of people who companies that Nvidia mentioned that really put a light bulb on for me. One, you heard him there talking about Runway, Chris Valenzuela. He's a guy who I talked to on my Fort Knox series a couple of years ago. They're doing these video AI models, and essentially what he was saying is, if you can get AI to generate this video and do that stuff, then you can direct a robot in the real world to do those very same kinds of actions.
So, when he saw Runway doing that, he figured it's the time for robotics. Another entrepreneur who he mentioned yesterday, Vast Data, Ren and Hallock. They're really doing some things with large amounts of data, and that's what's necessary, re-architecting the system for agentic AI to work. So, it it's it's a part of Nvidia trying to remake all of IT in the whole IT stack.
There's value there if really they're influencing what overall technology is doing and and yeah, changing everything as well. So, it's not just the chips, Becky. All right, I hope you're all doing great today. Today Nvidia's CFO spoke at a fireside chat with JP Morgan. CFO Colette Kress told us that demand has increased since the $500 billion figure that Jensen shared on stage at GTC back in October.
Colette also told us that the US government is working on licenses for Nvidia to ship the H200 to China. Jensen indicated in an interview with Bloomberg today that they're seeing demand from companies in China, which he believes indirectly means that the Chinese government is likely to allow the sales to go through.
Also today Nvidia and Siemens announced they're expanding their partnership to build the industrial AI operating system. Under this collaboration, the two companies are reinventing the entire industrial value chain. Everything from design and engineering to manufacturing and production operations and supply chains.
This fits right in with Nvidia's theme at this year's CES, which is physical AI, bringing artificial intelligence into the real world. I'll talk more about that in a moment. Yesterday after Jensen delivered his CES keynote, he held a short Q&A session with analysts. And so really quick, let me mention a few points that stuck out to me from that Q&A.
Jensen was asked about the Groq deal and said that Groq's technology excels at extremely low latency inference, which is going to be essential in the future when demand for low latency inference rises from AI-powered devices. So, the massive low latency inference demand in the future from consumer devices appears to be the main reason why Nvidia did the deal with Groq.
Jensen also told us that in the future Nvidia expects to become the world's largest company in storage processors, similar to how they are currently the world's largest in networking. Nvidia will establish a dominant position in storage processing like they already have with networking. That is a big deal.
Jensen was asked about how tokens are growing at 5x and when a re-acceleration or slowdown would occur. Jensen said that the 5x growth driver is reasoning and that the next re-acceleration factor is agentic systems. Token usage will surge. It could grow 50x. This is exactly what I've been telling you for months.
. And Jensen is clearly confirming this. When asked about Groq's technology being integrated into Nvidia's product portfolio, Jensen said that the product portfolio will remain mostly centered on Grace Blackwell and Vera Rubin, but some could take new forms specialized for extreme low latency. This is in line with what I've been suggesting in recent videos, hypothesizing that we may see a new low latency inference-focused variant of Rubin from Nvidia in the future following the Groq deal.
Jensen seems to be hinting at that potentially happening in the future. Maybe there will be another variant of Rubin or maybe it will be a new type of accelerator altogether. That remains to be seen. All right now, once again, I'm going to rapid-fire multiple points that stood out to me from Jensen's CES keynote last night in case you missed Monday night's video.
We learned a lot of important details. First, Jensen made it very clear that agentic systems are going to take off from here. I've been saying in recent months that I expect . And I think 2026 will be the year we see agentic AI at scale go mainstream. Jensen also spoke about the importance of open-source models and directly confirmed something I've been saying over the past months.
In December Nvidia released the Nematron 3 family of open models. I repeatedly made the point that this is why Nvidia stopped offering DGX Cloud capacity to customers back in September and struck a deal to buy additional capacity from Coreweave. Moving forward, Nvidia wants to be the leader in open-source models.
Jensen directly confirmed this on stage. Jensen told us that Nvidia has built up massive amounts of DGX capacity in order . Jensen shared a lot of exciting details about Nvidia's physical AI business. First, Jensen reiterated that Nvidia offers the full robotics stack equipped with three computers that make physical AI possible.
DGX for training, RTX Pro for simulation and AGX Thor, . Physical AI is a very broad industry and part of that industry is autonomous vehicles. Jensen announced Nvidia Alpaca IO, the world's first thinking, reasoning autonomous vehicle AI. Alpaca IO is trained end-to-end. Not only does it receive sensor input and take action, it also reasons about what action it's about to take, tells you the reasons by which it came about that action, and then of course the trajectory.
Nvidia started working on self-driving technology eight years ago. Jensen announced that the first autonomous car from Nvidia is going to be on the road in Q1 in partnership with Mercedes-Benz. Jensen told us that the Mercedes-Benz CLA was just rated by Euro NCAP as the world's safest car. Jensen told us that Nvidia's vision is that someday every single car and every single truck will be autonomous.
The ecosystem of companies working with Nvidia on robotaxi and self-driving technology is growing rapidly. And Jensen expects this to be a giant business for Nvidia. This is going to be the first large-scale mainstream physical AI market and Jensen believes that over the next 10 years a very large percentage of the world's vehicles will be autonomous.
And this is only part of Nvidia's physical AI business. I'm telling you as I've been for months, Nvidia's physical AI business does not get enough attention and I seriously think head of Jensen's keynote, I made the point that I would be surprised if Jensen made any big announcements related to data center since CES is geared toward consumers, but that I would be happy to be wrong about that.
And now, I am very happy to have been wrong about that because Jensen gave us an update on Rubin. Jensen told us that Vera Rubin is in full production right now. This is very important and this makes me very optimistic about Nvidia's earnings report and guidance in February. Think about this. Foxconn published their December revenue report and looking at this graphic, we can see that Foxconn's revenue jumped sharply starting in September.
That's when Blackwell Ultra started ramping in substantial volumes. And Foxconn had a very strong fourth quarter in the months following. This bodes very well for Nvidia's next earnings report in February. Now, couple that with the fact that Rubin is already in full production and on track to launch seamlessly this year.
Plus we that should be coming online very soon in the early part of this year. All of this put together makes me very optimistic that we will get both a very strong earnings report and very strong guidance from Nvidia in February, even stronger than usual. And that makes me optimistic that we may see a very positive reaction to Nvidia's next earnings report.
Now, of course, if the stock rallies a lot ahead of the earnings report, then that decreases the likelihood of a positive reaction after the earnings report. But you get what I'm saying. Foxconn's revenue report indicates that Nvidia should report very strong earnings in February. Large-scale sovereign AI projects coming online very soon increases the likelihood that we will get very strong next quarter guidance.
And Vera Rubin being in full production means we are likely to get very as well. Putting all of it together, that is a very positive setup in my opinion. And this evening we get to hear from Jensen again. This time at Lenovo Tech World starting at 5:00 p.m. Pacific, 8:00 p.m. Eastern. I probably won't be able to get a video out in time after that presentation.
So, I'll try to cover what Jensen says at Lenovo Tech World in Wednesday's videos. So, be on the lookout for those. Very briefly, I'll mention that I'm very bullish on Iron in 2026 and I'm optimistic about the year ahead. Last I looked, Iron had a very positive green day yesterday and then a notable red day today.
I wouldn't worry too much about the short-term volatility. company. There will be pullbacks along the way and I view pullbacks in Iron as opportunities to responsibly add to long-term positions at lower prices. At the time of recording, I'm not aware of any Iron-specific news and I continue to be very bullish on Iron this year as the company starts to deliver on their AI Cloud forecasts.
Again, ignore the short-term noise and focus on the long-term fundamentals of the company. Now, back to Nvidia. I'm glad that Nvidia is starting to really shine the spotlight on the company's physical AI business. As I've been saying for months, this part of Nvidia's business does not get enough attention today, but I think it will start to get a lot of attention throughout this year and in 2027.
Jensen is clearly telling us what is coming next. Physical AI stands to benefit massively as they have embedded themselves as the underlying infrastructure powering all of it. .
Comments 0