# Jensen Huang | Does AI Revenue Justify Capex, '27 AI Bottlenecks, Singularity, Open vs Closed & more BG2 · 2026-10-15 · 100 min · https://www.youtube.com/watch?v=eS7qQ3WJSrw ## Transcript Jensen Huang This is an existential issue for us. We could easily stop, but if we stop today, we lose. Functionality, capability, safety, and security—they’re all in exactly the same place: priority number one. ### A Week at the White House & the National Medal of Science Brad Gerstner It’s our third pod back together. In the last 10 days, you’ve been at the she state dinner at the White House. You were at the White House summit where we did the accord on superintelligence, and just yesterday you were there with your parents when the president presented you with the national medal of science. You were the only investor at the table. Now that we’ve reflected on it—yeah, tell me: being there with your parents yesterday, it struck me that 5 out of the 6 people on stage were born outside of the United States. How did that feel? Jensen Huang Well, first of all, it’s a great honor all by itself, right? It’s a reflection of the contribution that the company has made. Obviously, we created—we invented—a new computing model. This is probably the most pervasive computing model in the world today. It’s the only computing architecture that’s in every cloud and every computer company, and it’s used in every field of science. It’s used by gamers. It’s used by creators and builders, and it’s used by scientists. It powers intelligence. If you take a step back and look at what the company has accomplished, we started out building a graphics chip for gamers, and I’m still proud of that, of course. We’ve now shipped 1.67 billion units. I mean, I raised generations of scientists, kids, and designers. Here we are: we’re the infrastructure that powers the intelligence of society. It’s an incredibly proud moment for me to receive the award on behalf of the company. Now, for my parents, I wanted them to go, and I gave them literally 24 hours’ lead time. They don’t travel that much, right? My dad didn’t even have a suit. Brad Gerstner Wow. Jensen Huang Yeah, so he ran out and bought a suit. Brad Gerstner He looked great. They looked great. Jensen Huang Yeah. It was a brand-new suit, and he showed up. I didn’t realize he had a suit. I thought he was just going to wear some shirt, and he went out and bought a suit. My mom got a new dress, literally in 24 hours. Brad Gerstner Yeah. Oh, I know how it works. Jensen Huang Oh, my gosh. It’s incredible. We chartered a plane for them and got them there, and they were so happy to meet the president. President Trump walked my mom because she has a hard time walking. The 2 of them were walking offstage, and he held her and walked her down the stage. They were incredibly proud to meet him. For them, it’s their American dream. I still remember my dad—I think I was about 5 years old—and he had just come back from the United States on a business trip. He was working for an air-conditioning company. Brad Gerstner And you were living in Taiwan. Jensen Huang We were at the time. We had just moved to Thailand. Brad Gerstner And— Jensen Huang Oh, no, no. We were just about to move to Thailand. We were in Taiwan. He had come back, and he was in the United States working for a company called Carrier, the air-conditioning company. He was here for training, and of all the places in the world, they landed him in New York. You can just imagine: a kid in his mid-to-late 20s, never having been outside of Taiwan, goes to the United States—New York City, of all places. He saw the future of the future, and he came back mind-blown. He brought back all the things that he bought at the airport. He brought chocolate. He brought that cheese that used to come in a tube, that strange white cheese. Brad Gerstner It’s great. Jensen Huang He brought back all kinds of strange things, and he said, “This is where we should go someday.” The first opportunity they had, they sent my older brother and me to the United States. It was his dream. Brad Gerstner I don’t think many people realize that you came to the United States ahead of your parents. Jensen Huang Yeah. Brad Gerstner They joined a few years later, but it hit me like a ton of bricks that 5 out of the 6 people on that stage yesterday were born outside the United States. You and I have talked about the need, with all the other stuff going on, for a real strategy and an immigration policy that makes certain we’re still the place that all the great builders in the world want to come and build. This is America’s superpower, right? In a lot of ways, the intellectual-property creation capability of our nation is unlike anything else in the world. The reason for that is that we attract the world’s best. Jensen Huang That’s right. ### America’s Superpower Brad Gerstner We are the place where they want to come and build their lives because of the opportunity it provides, the freedom that it offers, the spirit of invention and entrepreneurialism, and the can-do attitude. In the United States, everything is possible, right? Jensen Huang Right. Brad Gerstner We start with, “Of course it’s possible. Of course we could do that,” instead of, “Of course you can’t do that.” I do a lot of travel, and in other places, you hear, “Of course you can’t do that.” We take it for granted, but the spirit, the energy, and the culture that we’ve created here just can’t be understated. I think the inflow of intellectual property into this country is unlike anything else the world has ever seen. Jensen Huang It’s true. Brad Gerstner We want that. We want all of the world’s great minds to come here, bring their great ideas, and realize them here. Jensen Huang Realize them here. Brad Gerstner Let’s keep that flywheel going. It’s a delicate system. It’s a delicate balance, I get it. We have to do it lawfully. You have to do it rightfully, right? But we surely want to do it. Jensen Huang Yeah, for sure. ### Entering the Singularity: Where Are We Now? Brad Gerstner I want to start at a high level with where we are. Then I want to dig down a little bit into safety and regulation and getting this right, where we are in terms of the compute build-out. Finally, we’re going to finish up by drilling in a little bit to NVIDIA’s growth and what’s going on there. You’ve been really good at calling the future. Two years ago, we were sitting here, and you said that the market was dramatically underestimating the amount of compute build-out that was going to be required and was going to happen over the course of the next 5 years. You said it wasn’t going to be $1 trillion; it was going to be $2 trillion at a minimum in terms of cumulative compute build-out. Of course, now we’re over that. You said, before everybody was thinking about it, that inference compute was going to grow by 1 billion times because of inference-time reasoning and thinking. Then last year, before we saw Opus 4.6 and Claude Code, you said we were entering the age of agents, where we were going to unlock a lot of these capabilities. Of course, we saw Opus 4.6, and Claude Code really took off. So where are we now? Right, we have Elon and Sam and only three things that I said that came true. There's another 200. Why don't you just keep that? Let's just let that be the podcast. So our investor day in a couple weeks is on entering the singularity, this acceleration. We have Elon and Sam and Demis all saying, “Hey, we’re in the foothills or entering the singularity.” We have models solving Navier–Stokes, as you know. When you look into your crystal ball, where are we? Are we inflecting? Are we accelerating into a different phase of superintelligence? Jensen Huang Let’s unpack that first. It’s fairly logical why the amount of compute we’re going to need is 1 billion times more than where we were just recently—call it last year. The reason for that is because thinking and problem-solving take a lot more inference than repeating or regurgitating something that you memorized. When you use ChatGPT, you give it a prompt. Most of that knowledge was encoded into the large language model. Unfortunately, of course, it hallucinated, and so the next step was that it was generative, but it was also grounded or conditioned by ground truth. Brad Gerstner Yes. Jensen Huang It would have to go out and do a search, come back with the ground truth, and then, conditioned by that and augmented with what it had already learned in pretraining, it would generate a new answer. Most of the things that we really want to do that are interesting require you to reason step by step, come up with several options, and make a plan. You might decide, “Let’s see how these things all roll out.” Then, based on the answers to all of those, I reflect on the best answer so far and the best path so far. I might choose 1, 2, or 3 of them and roll them out. Brad Gerstner Right. Jensen Huang Then you prune the rest, and you just keep doing this. People do this. That’s the reason we say great strategists love strategic optionality. You don't want to prune things too fast, on the one hand, but you also have to optimize your resources. Well, AIs are doing that. They're keeping their strategic optionality. They're rolling things out, they're pruning, they're reflecting, and so on and so forth. All of that thinking, all of that exploration just takes a lot more compute. On top of that, because AI is now able to do productive things, we're going to use it for more things. The compounding is that it uses more compute to do its job, but it's doing a good job, and therefore we're going to hire it to do more jobs. Brad Gerstner Exactly. Jensen Huang Okay. And so that all compounds. At this point, I would say that the big observation from last year is that, on the one hand, we want the models to get better and better. We still want the models to get better, and the scaling laws of models are going to continue. However, the attachment of a harness—basically, the exoskeleton of the large language model, right? Brad Gerstner You know, kind of the body of the brain. Jensen Huang That allows it to manage working memory and manage knowledge. How do you go from working memory to short-term to long-term memory, to basically formulating your knowledge graph? That's what we all do. We think about, “What did I learn, and how does it fit into everything that I knew in the past?” When we go to bed, we do compaction. Things that don't make sense or don't matter, we eventually forget, but the things that matter somehow get reconnected. We're doing that kind of offline. We have working memory, offline knowledge extraction and creation, and knowledge graph creation. That whole part of memory is a big deal. The other thing is, of course, tool use. Brad Gerstner And we've taught these tools—these AIs—how to interact with the most important tools that humans have ever made. It's no different than if a robot came into our house. It's more likely that you're going to use all the tools that I have in the house than go off and build them from scratch. Jensen Huang SPICE is still going to be useful, and Verilog will be useful, and simulators will be useful, and design compilers, and Photoshop, and AutoCAD, and the cell kia, and all the Siemens tools. All of these different tools that we've been using to run our companies, build our companies, design things, and operate things are all going to be great tools for the agents to use. But the agents had to learn how to use them. These two areas—knowledge and working memory, and tool use—represent huge unlocks in the last year. So now, all of a sudden, you can put these large language models, encapsulated in their harness or exoskeleton, into your company. You can connect them to the information—the ground truth that you have—and connect them to the tools that they can use, and all of a sudden they start doing amazing things. The easiest part, frankly, was coding. Brad Gerstner Right. Jensen Huang That's kind of the first on-ramp to useful work. In a lot of ways, coding is easy because both the customer— Brad Gerstner And the work, the tools that they use— Jensen Huang —are fairly accustomed to things that have tremendous use but are not extremely easy to use, right? In the early days of coding, it wasn't extremely easy to deploy, but software engineers found a way. And so now I think the next step for us is—and you guys have probably been playing with Astra. OpenAI's Astra is really fantastic, and we're using it in all these different places. The next phase of these agents is that they're able to do design work and a lot of creative marketing work. We're using it to—you take a picture and give it to Astra, and it literally recreates it into CAD. Brad Gerstner Yeah. It recreates it into CAD, which is incredible because we need the CAD. Jensen Huang Yes. Brad Gerstner So that we can interact with the CAD, fine-tune it, and adjust it. Astra could really give us the first initial implementation incredibly productively. Jensen Huang Now you take Astra, the latest release, and you apply it to all of these different domains. I think that, besides software engineering, we're going to see designers and builders and creators, and this whole area of scientists. All of this is going to be unlocked. The challenge, of course, is how do we deploy agents safely and securely into the world's enterprises? Brad Gerstner This is exactly the same question again, with safety and security. Jensen Huang And we'll talk about— Brad Gerstner We'll come in on that. What I'm struck by is that all of this occurred this year with OpenAI and Anthropic having, I don't know, 2 or 3 gigawatts of compute each. That's a relatively small amount of compute relative to what's coming. Again, what I'm interested in is the way the economy generally works, or the way people model the economy or model a company: It's linear deceleration. They're not very good at modeling acceleration. But here we have this double exponential. We have a massive amount of new compute coming online. So if OpenAI and Anthropic are 3- to 5-timesing their compute next year, does it stand to reason that we solved Navier–Stokes, and we'll look back at this and say that was a tiny amount of compute and the worst intelligence we're ever going to have from here forward? We solved a Millennium Prize Problem. So when you look at what's coming online next year, do you expect that we're going to have major breakthroughs in long-running agentic problems that we can use these swarms of agents on? ### Navier-Stokes & the Falling Cost of Intelligence Jensen Huang Well, Navier–Stokes is really a great example. Remember, OpenAI did not solve Navier–Stokes from nothing. They solved it from the foundation of a lot of great scientists, a lot of great mathematicians, and a lot of great previous work. Earlier, I said we're at the tipping point of a lot of these different fields, because we're at that phase—that point—with Navier–Stokes, with a great deal of foundational work already having been done. If we could just apply the exhaustive exploration of AI— Brad Gerstner Yes. Jensen Huang —to the previously done work— Brad Gerstner Yes. Jensen Huang —we're going to see one breakthrough after another breakthrough rather quickly. But we do need a lot of compute to do that, and the reason is because a large amount of exploration was necessary. If it was simply somebody resolving some fundamental, first-principles theory in order to discover the breakthrough, if it didn't require a vast amount of exploration of the search space—search space, not space-space—then somebody would have found it earlier, right? People are pretty smart. I think there are many problems like this that are just on a precipice and really just want that massive amount of intelligent exploration to get us there. In the coming years, my expectation is that the cost of intelligence will fall fairly quickly. Brad Gerstner Right. Jensen Huang The reason for that is because, obviously, we're bringing more compute online. These models are getting smarter. When you're smarter, you can be more efficient. Brad Gerstner Right. Jensen Huang And so, with the combination of these two things, the cost of intelligence will come down, which is, of course, what they want, right? This is no different from Moore's law. If you go back 30-some-odd years, we have three decades of computing costs decreasing by a factor of 2 every single year. We want intelligence costs to decrease by a factor of 2 or 20 every single year, because the consumption of it is going to— Brad Gerstner More than make up for it. Jensen Huang That's right. More than make up for it. And so there's a massive empty space in the world that wants to be filled—an empty vacuum of opportunity—and we just have to get the cost down. Brad Gerstner I guess what stands out to me about Navier–Stokes is that they had 10,000 agents working together on this. To me, that sounds like a hyperscale company, basically pointing all its agents to work together toward a goal. In the early days, right now, it's a lot of parallel search and exploration, but I think the goal is that more of these agents can work together toward a common goal, and that we can solve a lot more. To your point, I think our math says it probably cost about $4 million of compute to solve this instantiation of Navier–Stokes. The cost to do something like that over the next few years is going to drop by another 5- or 10-times. The number of things that we're going to be able to do is rapidly expanding, and our ambition is going to go sky-high. The original GPT probably cost—let's pick a number—maybe $20 million to do. If you wanted to train GPT-1 now, it would probably cost you what, $1,000? $100, maybe? Jensen Huang It's like $100. Brad Gerstner Yeah. Jensen Huang Okay. And so you went from $20 million—let's say $10 million—to $100, and we did that in 10 years. Brad Gerstner Yeah. Jensen Huang Okay. And so I fully expect—meanwhile— Brad Gerstner Our ambition got so large— Jensen Huang —that the models that we're training now—3 trillion, 10 trillion, 30 trillion—we're talking about gigantic models now. And so our ambition for the capabilities of technology has gone up. Meanwhile, our use of it has gone up between 10 years ago and now, probably—I don't know—a million times. And so I don't think this trend is going to stop. I think there's a fairly insatiable demand for intelligence. ### What Will We Do with Abundance? The smarter you are, the more intelligent you are, the more ideas of applying intelligence you have. The more I've learned in life, the more I realize there's more to do, and that's because every time you learn something, it opens a new door. Every time you open a new door, you see more opportunity, more challenges, and more problems. That's the future, right? Brad Gerstner You know, people say we're going to have more time. I completely disagree. I think, for most people, this intelligence is going to unlock more opportunities. They're going to want to pursue them because they're ambitious. Jensen Huang Yes. Give me an example of a family—a father, a son, a husband, a wife, or a mother—who doesn't want more, who doesn't have ambition for their families to have more. They just don't have the opportunity, and these things are going to open up opportunity. Brad Gerstner If you go back to 1930, Keynes made a prediction about the Industrial Revolution. He said, “In 100 years, the economy will have increased 8x, but the workweek will go to 15 hours a week.” He was right. The economy did increase 8x, right? We had massive abundance and productivity gains from it. Jensen Huang Yeah, except we're working 80 hours. We didn't go to a 15-hour week. What he got wrong was the human spirit, right? He didn't realize that work is purpose. Discovery is purpose. It's deeply encoded in, I think, our DNA. I'm working harder than I ever have and having more fun than I ever have—not because I have to, like my grandparents did, but because I want to. And so I agree with you on that. I'm fascinated with this idea. The singularity postulates that we've now kind of got all these agents that can work together on long-running tasks. They can stay on task. I love your idea that they keep each other on task, working as a team: “Hey, you're off.” Brad Gerstner That's right. I love your idea that we build on top of all the things that have come before. If I think about full self-driving as an example, we talked about it for a decade and we really didn't get there, right? But all of a sudden, you go on Twitter and everybody's experiencing Tesla FSD, experiencing Waymo—we're here. Somebody asked me, “How was the drive here this morning?” I said, “I don't know,” because I got in my car, hit FSD, and didn't pay any attention until we pulled into the parking garage. I was working the whole way here. If I think about that in the fields of science, the breakthroughs that are going to come in pharma and genetics—we've been laying down the tracks for 30 years. We're going to break through on a lot of massive cures for the world. Of course, ChatGPT is already helping parents plan their day, solve their kids' health issues, and get better educated. Talk to us a little bit, though, as to why that's having such a hard time breaking through. Jensen, we have all of these things that are out there improving lives and that are right around the corner. That's curious to me—why isn't that breaking through? Jensen Huang I guess—let me take a shot at it. We ultimately won't really know. I've spoken to a lot of people, and they've told me what they thought. One of the things, of course, is that AI, or superintelligence, is soft matter. You can't look at it, you can't see it, and you can't touch it. You've got the whole world talking about this technology, but you can't see it or touch it. Even though people are already benefiting from it—when they search, when they go shopping, when they're using Netflix, when they go to Walmart—they're benefiting from it. The fact that FedEx is getting stuff to us—it's all leveraging superintelligence and accelerated computing today. But we don't do a very good job telling that story. Brad Gerstner So, life is—they're benefiting from it; it's just invisible. ### Physical AI & Robotics Jensen Huang Right. So if we shut off the data centers and shut off superintelligence for a weekend, I think everybody would notice. I know you had a question about robotics in particular, though. Brad Gerstner Yeah. You've notably said a few times this year that the ChatGPT moment for robotics is here, and I think you've also said that it could be your next $100 billion market. So what are you seeing now at the inflection point this year? Jensen Huang Yeah, that this is around the corner. I understated the $100 billion because, in the first 6 months of this year, we already did $10 billion in physical AI. Physical AI is basically AI that's somehow interacting with the physical world. It could be photon-counting CT machines, ultrasounds, new imaging radars, or 5G, the next generation of 5G. Not only does it understand radio waves, it's going to understand signals of all kinds. It's a sensing network. Of course, there's robotics, self-driving cars, manufacturing systems, visual inspections, geospatial satellites imaging the world and reconstructing it into 3D. If you look at the entire electromagnetic spectrum, there's AI stuff going on from visible to invisible spectrums. And so we're doing work across that entire thing. As an AI system, there's the perception part of it, the reasoning part of it, and then the planning and actuation part of it. When you finally come to planning and actuation, that's self-driving cars and robotics—things like warehouse delivery vehicles and pick-and-place robots that you don't have to program as precisely because they learn by themselves. That goes all the way to people talking about consumer robots coming into your house, making your bed and things like that, or healthcare robots and hospital robots. There are lots of really good examples there, precisely delivering your prescription to the patient, having robotic care literally 24/7, every single second, to keep patients safe. There are lots of great technologies being deployed, including entire surgical rooms that are going to be completely robotic. There are lots of great examples. If you look at this whole thing, we're the world's largest physical AI company. We did $10 billion in the first 6 months. We did $10 billion in the last year. Brad Gerstner Mhm. Jensen Huang And that was our first $10 billion year. The year before that, we did $5 billion. Brad Gerstner Yeah. Jensen Huang Okay. So it went 5, 10. Let's say we did 20. Brad Gerstner Yeah. Jensen Huang Who knows? Brad Gerstner Yeah. Jensen Huang So maybe the $100 billion was a little understated. I think the reasoning part of it is everything that matters to us in the final analysis touches the physical world, and we live in the physical world. I think the market opportunity for bringing software, computing, and intelligence into an industry that historically was largely mechanical and electrical—that's a big opportunity. Look at what Elon has done to the automotive industry. He turned it into essentially a computing industry, a technology industry. I think every car company is going to be a technology company. Every shipping, receiving, and logistics company is going to be a technology company. I'm super excited for FedEx. They're obviously going to be an AI company and a technology company. Walmart is obviously going to be a technology company and an AI company, so I'm super excited for them. The entire telecommunications industry will be reinvented again into a technology industry. They were a technology industry a long time ago, then became a services industry. They're going to be a technology industry again. I'm delighted to work with AT&T and Verizon, and, you know, Europe and around the world, as they reinvent themselves. You take a look at every one of these industries that touches the physical world, and there's a great idea for reinventing them with AI. I was just in Washington. People can put us in chips. Yes, we're actually in warehouse and logistics. We're in healthcare. We're in automotive. We're in telecom—every industry on the planet. We're in life sciences and entertainment. For a long time, NVIDIA was really vastly an entertainment company, and each one of these verticals has a massive NVIDIA stack inside of it. We're very deep in the ecosystem, and we have excellent partnerships there. One of these days, people might not categorize us as a chip company anymore. The playbook would be to take the Alpameio automotive playbook and insert it into each one of these industries. Brad Gerstner Yeah, that’s the final contribution. Jensen Huang Yes. But in each one of these, it takes 3 computers to build a physical AI platform. You need the computer to create the brain—that’s the training system. But very importantly, and this is where you’re going to see the labs do this as well, they need to build a digital twin of the internet so that they don’t have to put their AI agents out onto the internet to learn how to use it. Just like we created Omniverse as the digital twin of warehouses, buildings, cities, and the physical world, we put our robot into the digital twin. The digital twin learns how to be a robot, and then we deploy it into the last computer, which is the computer inside, for example, a robot, a car, a surgical instrument, or whatever it is. It takes 3 robots, 3 computers, and 3 complete stacks, and we build complete stacks for each one of them. We’re the only company in the world that does that. If you’re a company that wants to build it end to end, you could use 2 or 3 or 1 of our computers. We don’t care. Each of our computers has its own software stack, and we do the same across all of our verticals. ### Building Responsibly Brad Gerstner But of course, these capabilities cut both ways, which is why we hear a lot of people talking about safety. You’ve been highly critical of the doomer narrative. You can only get there if you build the technology responsibly. Jensen Huang That’s right. The first thing is to build the technology responsibly. Brad Gerstner Right. Jensen Huang Notice the word “build.” It’s an engineering word, right? It’s about technology. It’s about engineering. Engineers, scientists, and researchers have to build this technology safely and responsibly. The question is, can you? The answer is absolutely yes, because if you can’t, then you shouldn’t build it at all. Brad Gerstner Of course. Jensen Huang We either have to believe we can, or we shouldn’t build it. I believe we can. Let me just reason about it with you. The first part of it is that some people say AI is uncontrollable because it’s impossible to align perfectly. Alignment means this: You tell somebody, “I want you to do this,” and so the mission is clear. Brad Gerstner The mission is clear, and it’s clear to you— Jensen Huang Because you said it in English. It could be, “I want you to solve this quadratic equation,” right? Brad Gerstner Right. Jensen Huang The mission is to solve the quadratic equation. Alignment is how you do it. There are many ways to do it. For example, you memorized it, right? Brad Gerstner Okay. Jensen Huang The answer is whatever it is. Or you saw the answer in a textbook, so you looked it up, right? Brad Gerstner Right. Jensen Huang You referred to a reference, or you went to a drawer, knowing that drawer had all the answers. You opened the drawer and picked out the answer. Some people call that cheating. I call that opening the drawer and retrieving the answer. Brad Gerstner Okay. Jensen Huang That’s what computers do. There are a lot of different ways you could solve that problem. Now, of course, that one is very explicit. Here’s another one: “Be successful in life.” The mission is easy to say, and in fact, we say it to each other all the time. We say it to our children. We say to our friends, “Let’s do our best to contribute to society.” Now the question is, how? There are so many ways. How do you reasonably align? Everybody’s culture and background are different. Some people think that contributing to society means being generous and kind. Some people might mean contributing to society by making a lot of money and then being philanthropic. Some people might think that building a company alone is a sufficient contribution to society. Creating a beautiful piece of art— all of it is true. Brad Gerstner Correct. Jensen Huang The question is, what’s the alignment? It depends on your background and your culture. Maybe it has something to do with where you live and where you came from. There’s an infinite number of ways. My point is this: Alignment is going to be a forever problem. Brad Gerstner Yes. Jensen Huang If we depend on perfect alignment in order to ensure safety, then we’re never going to get there. The answer is this: The way we think about babies when we’re raising kids, or when we have pets or employees, is that you start with containment. You give them access to tools, information, responsibility, capabilities, funds, and resources based on the job that you ask them to do. You start out with zero trust. When you hire somebody, you start out with zero trust. When you have a baby who’s 1 year old, you put them in a playpen. You assume zero trust. You give them minimum rights, and you expand that boundary as you learn more. Brad Gerstner Mhm. Jensen Huang Whenever we have a new AI, we have to assume the same thing. Whenever you’re testing an AI: zero trust, minimum boundaries, and minimum rights. You expand them as you test and give yourself confidence. So, very tight containment is really important to any autonomous system. The second thing you do is monitoring, and the monitoring has to be out of band. The reason for that is because if you tell an intelligent person, “I’m going to sit here and watch you do your work to make sure you’re doing it right,” and, “I’m going to sit here and watch to make sure you’re serving customers properly and working hard enough,” that autonomous system is going to behave differently than when you’re not watching. It’s like Brad’s in the room with me or not. Brad Gerstner Exactly. Jensen Huang The monitoring has to be out of band, and it has to have the ability to observe that the autonomous system is doing something inconsistent with the policy or inconsistent with alignment. You have the ability to either escalate or shut it down, right? Brad Gerstner Just turn it off. Jensen Huang Those capabilities have to exist. That’s engineering. That’s technology. It’s possible. We announced it recently; it’s called Open Shelf. We have over 100 companies joining us now, with lots and lots of activity around the world, because everybody wants the same thing. Brad Gerstner Starting at the highest level, when I looked around the table at the White House, everybody around that table—they’re parents; they have kids. I think there’s this view in the country that there are some leaders of superintelligence who are just YOLOing, who don’t really care. They’re being reckless. They just want to grab all the money. Yet when I looked around that table, there wasn’t a person in there that I saw who wanted to be reckless or wanted to YOLO. In fact, everybody there wanted to find the right balance between fast and safe. That was the president’s directive. We debated a little bit about what that meant, because there are different versions of how to implement it. What would you say to the American public that fears we have a reckless president or reckless leaders? Did you see any evidence of that? You have kids. Do you worry about the future that you’re bringing into existence with superintelligence? ### Safety, Security & the Airbags Analogy Jensen Huang Safety and security technology are part of AI, right? There’s some weird formulation that AI technology and safety and security are orthogonal. Brad Gerstner Yeah. Jensen Huang It’s kind of like the airbags, the braking system, the radars, the cameras, the ABS, the lane-keeping—all of that technology is somehow independent of the car. Do you understand what I’m saying? The only thing that matters to the car is the engine, right? Brad Gerstner Right. Jensen Huang In fact, the whole car is part of the car’s technology. When I say we need to accelerate AI technology, what I mean is that we need to accelerate the security technology. We have to accelerate the safety technology. They all have to be accelerated. If you ask me right now, I think everybody would say, “Let’s accelerate security. Let’s accelerate safety technology.” Nobody says, “No, let’s pace that,” right? We should accelerate that. It’s all part of the same thing. You can’t separate capability and functionality from safety and security. They go as one. When you deliver the service, nobody’s going to say, “I really like the functionality of your service, but boy, is it wreaking havoc.” I think it’s all one thing. We treat it as all one thing, and the industry has to treat it as one thing. When we say “accelerate,” we have to mean accelerating the entire package, including safety—especially safety. Brad Gerstner You mentioned pacing. Jensen Huang Especially safety. Brad Gerstner Some people read “Pacing the Frontier” by Dario Amodei and thought that meant slowing down. I thought a funny moment in the room was when the president called on his new best friend, Dario. He referred to him that way and said, “Explain to me your version of fast and safe.” And Dario Amodei started off by saying, “Mr. President, you said whoever wins AI wins, right?” He said that we’re all speeding up, that they’re speeding up, that they want to go fast, and that they want to win the race with China. But he also acknowledged that, along the way, just to your point, we have to make sure the safety systems are also speeding up and keeping pace. I thought that was an important moment because people have positioned him as though he’s in opposition, but it seemed to me that you were all very much on the same page. You went to the Cabinet Room with him and Greg and the other Frontier Labs to sign this agreement. Was there any chatter in that room about differences of opinion, or was everybody—you obviously got it signed—did anybody have to be put in a headlock, or did everybody go willingly? Jensen Huang No, everybody was very constructive and very productive. I think that, without exception, everybody is dedicated to advancing the industry safely and securely. The reason for that is because if it’s not safe, it’s not secure, and the functionality will never get a chance to speak. The capability will never get a chance to shine. We know that the priorities are, at all times, safety and security. Of course, we also have to recognize that this industry came through a transition just recently, right? It wasn’t but 6 months ago that this technology barely worked. Brad Gerstner Exactly. Jensen Huang Without functionality and capability, what’s the point in making it safe and secure? Nobody even cares about it. Obviously, the 1st job of AI is to demonstrate that it can do something useful. Brad Gerstner Right. Jensen Huang It really reached that point 6 months ago. Now it really wants to be deployed and broadly deployed. There’s no question that the entire industry now has a strategic imperative—an incredible imperative and great urgency—to make sure that whatever we deploy is safe and secure. I think everybody is completely convinced: capability, functionality, safety, and security—they’re all at exactly the same place. ### What “America Winning” Means Brad Gerstner Priority number 1. And so, the question is winning. I would reframe winning—not as the technology industry winning, and not as just the AI industry winning. Jensen Huang America winning means every industry in America winning: healthcare. Brad Gerstner That’s right. Jensen Huang Nothing would give me more joy. The definition of America winning is Lilly winning— Brad Gerstner Yes. Jensen Huang Walmart winning, FedEx winning, every single one, Boeing winning, the U.S. military winning— Brad Gerstner Every citizen winning. Jensen Huang Every citizen winning. Our universities winning, our high schools winning—that’s the definition of winning. Brad Gerstner There is no winning if it’s just a technology that won. Jensen Huang Exactly. But nobody used it. Brad Gerstner Right. Jensen Huang The greatest risk is, of course, that there will be no technology industry if we don’t build it safely and securely. This is an existential issue for us. If we make it an existential industry for everybody else, that’s not going to happen, because it’ll be an existential issue for us first. Having said that, the definition of winning for us—and you know this is the part I always have to change a little bit—I want us to pivot. Brad Gerstner I like that. Jensen Huang I want us to pivot. When I see America winning AI, I’m looking down from the air. These days, we spend a lot of time in the air, and we’re looking down. Every state is winning, every county is winning, every community is winning, and every company is winning. Everybody won because we could afford to spend more on healthcare. We could afford to spend more on education. We could afford to spend more on interstate commerce, highways, and all of the infrastructure of the country. Brad Gerstner Jensen, what I hear you saying is that the definition of winning is that when we look back 25, 30, or 40 years from now, the economy needs to be 10 times as large. Everybody’s lives need to be better, and existential and primary to all of that is that it has to be done in a way that’s aligned with human interest. ### The Three-Legged Stool: America’s Superintelligence Framework This is the last question I have on this, and I know we have a lot to get to. Most people don’t understand what our security framework is in the country. Jensen Huang I’ve said we have the most robust superintelligence security framework in the world. We’re clearly the only country on the planet that has actually sanctioned a model and pulled it off the shelf with Mythos earlier this year. I describe it as a 3-legged stool. We have the frontier-model review of every new frontier model by Casey and the Department of Commerce. They do all their work. If a new model is coming out, they review it and test whether they think it’s going to cause a problem in the marketplace. Now we have the White House Accord on Superintelligence that you helped to architect: internal and external audits, board oversight, and all of the liability that comes with that. Then, 3rd, we have the SI Force, now led by Jay Clayton, of course, with Emil, Scott, and Andrew. That’s doing critical industry coordination, cyber, and so forth. These 3 legs on the stool—there are people out there, and I heard Gates recently saying, “We’re not doing enough. We need new laws on the books,” and everything else. That 3-legged stool seems like a lot to me in a very short period of time. As a leader of one of the most important companies in superintelligence, do you feel comfortable that, for the future we’re trying to usher in, this is a good step forward and where we need to be right now? The best way to get to the future is to take action quickly. If we argue about what’s perfect versus what’s 1st, we’ll end up arguing about perfection and never get there. I love the fact that these 3 legs of the stool are actionable. They’re tangible. They have real teeth. They’re happening now. We’re doing it right now. We’re taking action now. The reason we’re taking action now is because it’s good for our company, it’s good for our industry, it’s good for our country, and it’s good for our society. It’s good all the way around. If we don’t do these things, the outcome is less desirable. We don’t get to pursue that optimistic future that we all dreamed about. If you go back and look at another industry, I don’t remember another industry that, in its current state, has now formulated all of this cross-company, cross-industry, government-to-industry coordination to make sure that we advance this in a responsible and optimistic way. Brad Gerstner I just think it’s so important for people to understand, because when I hear senators talking about it on television, on both sides, to be perfectly honest, they don’t— Jensen Huang They have to learn about what we’re doing here. Brad Gerstner They don’t even know what we’re doing, right? We are doing an extraordinary amount. And, by the way, these aren’t voluntary things. There’s not going to be a frontier model that gets released that Casey and the Commerce Department aren’t reviewing, and they’ve demonstrated that they have the authority under export controls to pull models off the shelf. Jensen Huang Do 2 things. 1st, come and spend some time and learn what actually has to be done. What’s the problem, right? If you want a solution, you have to at least understand what the problem is. Come and learn about what we’re doing. We have plenty of time. I’ve got all the time in the world for this kind of stuff. Come talk to us. Come talk to our industry and learn. The 2nd thing is to learn about the solution that we put in place and try to see if we can make it better. Of course, we’re all completely open-minded, but be constructive. Brad Gerstner I thought that was part of the 4th point on the Accord on Superintelligence: this is not static, it’s dynamic. This group is going to get together routinely, and we’re going to update the framework as required when the technology evolves. ### 2027 AI Bottlenecks: The Compute Build Out I want to move forward now to compute. Of course, long-running agents that work in teams of thousands or millions to solve complicated tasks consume massively more compute. Dylan, our friend at SemiAnalysis, has said we’re going to bring on board 43 gigawatts of new AI compute next year. To contextualize it, we’ll share a chart with people online. That’s more than all of the AI compute that exists in the world today. We’re going to bring that on board next year. The year after that, another 40 gigawatts of compute. You’re the one who knows where every single GPU is allocated. As you sit here, does that sound right to you? Do you think we’re going to bring on that much compute? Do you think some of the things we’re hearing about permitting issues, bottlenecks in the supply chain, or local protests mean that the number is going to be materially different from that, based on what you’re seeing? Jensen Huang I think the demand is definitely there—and more. The challenge, of course, is that there are a lot of obstacles when you move from digital to the physical. There are a lot of obstacles there: land permitting and power generation that you have to bring online. Of course, you need construction workers, outfitters, and plumbers. We’ve created a million new jobs just building these things. Their skills are not fungible. They have a lot of skills, and particularly the ones who are building these $50 billion data centers—you’re going to want some people with trade skills. There’s a critical part of every single one of these buildouts where you need some very specific skills. Then there are, of course, whatever moratoriums are happening and whatever issues we’re seeing around the country. My guess is that land, power, shell, and buildout resources will be one of the most important obstacles to reaching that. Our company, as you know—now speaking specifically for us—there are several things that we’re doing to mitigate that. One, because we’re the only company that serves every CSP, we serve every NCP, and we have great enterprise businesses through Dell, HP, Lenovo, and others. Because we serve every single industry and we’re across the entire life cycle of AI—from data processing to training, post-training, inference, agentic processing, and all of the tool use that we accelerate—our exposure to the go-to-market is the largest of all. Every contract manufacturer in the world is working on my behalf. Every go-to-market is connected to our architecture. If one particular state, one particular community, or one particular data center has a hard time, our demand is so much greater than our supply at the moment that we’ll just adjust our supply chain accordingly. The other thing that we have—and I don’t love this answer as much—but as you know, NVIDIA is also the infrastructure for the planet. Whether it’s in Europe or in Africa—we’ve got early developments now in Africa, Central Asia, Southeast Asia, Latin America, the Middle East, and so on. I don’t love that the United States can’t keep up. For our business specifically, we have such a global reach, and there’s buildout demand all around the world. If we don’t build it out here, I think we all have to recognize that every country in the world would love to have it built out there. Brad Gerstner Yeah. Jensen Huang I think our business specifically is probably the most resilient. The reason for that is all the things that I mentioned: We cover the entire spectrum of AI. We cover the entire spectrum of computing. We’re in every single cloud, every single region, and so the resilience of our company is much greater. But I do think that land, power, and shell are going to be a challenging part next year. ### 50 GW of Chips, 35 GW of Power Clark Tang And I guess what I hear from the street is there’s this idea that maybe there are 50 gigawatts of announced chips underway or talked about, and maybe 35 to 40 gigawatts of compute—or land, power, and shell—that might come online. So what compute fits into that land, power, and shell, and on what timeline, is a question up in the air. But I guess if we think through the entire— Jensen Huang And just, Clark, let me interrupt you there. As you know, NVIDIA’s architecture is the most productive. It generates the most tokens per watt, and so if the world’s amount of power and energy reduces, it shifts their choices to the most productive version. The reason for that is that they still want to make their revenues, right? All of my CEO colleagues have high expectations for revenues next year, and people need to deliver it. Their shareholders are counting on them to deliver it, and all their partners need them to deliver it. When the amount of power reduces, the quality of the compute you put into it has to be the most extraordinary, or the most productive computing resource, you could put in the world today. We’re the most rentable. We’re fungible and durable. The forces around the world, because of the shortage of resources, are shifting interest toward us. Our market share was already very high. But surprisingly to everybody, not only are we a large company and our market share is high, our market share is actually increasing for these reasons. Brad Gerstner Mhm. Jensen Huang Because we offer so much more per gigawatt than we used to, you’re going to see us continue to grow. We used to be a GPU company; then we became the largest and fastest-growing networking company. We’re now going to be one of the largest CPU companies, surely the fastest-growing one. Because we’re offering more and more, and we’re in computing for storage, we’re in computing for the knowledge infrastructure, and now we’re in security with open shell and Sentry. The offering that we provide into one gigawatt went from about $18 billion or so per gigawatt, to $20-something billion per gigawatt, to now $40-plus billion per gigawatt. My sense is that’s going to keep going up. Not only are we growing share because people are shifting toward our computing platform, we’re also growing share because we offer more to the world’s capex and to the world’s dollar share of the market. Brad Gerstner Yeah. Jensen Huang If you go back just 6 years, our percentage of the world’s capex was in the single digits. Brad Gerstner Yes. Jensen Huang Then I think right now, with the world’s CSPs, we’re probably running about 40%. Brad Gerstner Mhm. Jensen Huang But we’re also expanding the market beyond the CSPs. Outside of the CSPs, whether it’s on-prem or neoclouds, half of our business—our percentage of capex is probably, what, 80%? Brad Gerstner Yeah. Jensen Huang Between 40% for half of our business and 80% for the other half, and both of them growing, doubling every year, you can see that our percentage of the world’s capex is, in fact, growing. ### Elon, Colossus & SpaceX Brad Gerstner Yeah. Well, one of those buyers of NVIDIA that we talked about 2 years ago is Elon, when he started building Colossus. We were blown away a couple of years ago to see him stand up a couple hundred thousand Hoppers in Memphis. I remember you saying, “Nobody can do it like him.” Now he’s raising another $40 billion, going to buy maybe 1 million-plus Vera Rubin. He’s moved from the Ks to the Ms. He says he wants to have 10 gigawatts by the end of 2027. That’s hundreds of billions spent on Vera Rubin. It’s an unprecedented ramp at scale. You were with him again yesterday, when he was also presented with a medal. Is Elon that much better at building compute today as he was then? Is it now obvious to you? Is it a foregone conclusion that SpaceX will be the largest AI hyperscaler in the world? Jensen Huang Well, in the two parts of our chart, Elon used to be under Enterprise. That’s the way we reported our numbers. Now he’s part of our CSPs, of course. Brad Gerstner Yeah. Yeah. Of course. Jensen Huang And so they’re massive. I’m delighted that I love working with Elon. He’s using our technology for his own AI in training models and deploying models. Of course, Grockbot is super cool, and the demand for Grockbot—the consumption for Grockbot—is really skyrocketing. Then, of course, he rents it. Brad Gerstner Yeah. Jensen Huang The combination of these 3 things is fairly unique. I guess Google does that, right? Brad Gerstner Right. That would be another example. ### Data Centers in Space Jensen Huang I think he also has the advantage that he loves the intersection between computing and the physical buildouts. He loves that whole area, and nobody knows more about energy than he does. One particular new market that we’re going to explore together is putting data centers out in space. Our 2 engineering teams are working super closely together to create a new type of server, a new type of architecture that might be super good in space and on Earth. Brad Gerstner Well, we saw Starship launch last week. It had its first orbital rendezvous and was launching those V3 satellites that literally captured the imagination and kind of broke Twitter for a moment. People were mesmerized watching them launch into space. I can’t wait to be a payload—not a person. Do you have increasing confidence that we will have a material amount of compute in space, via Starship, by, let’s call it, 2028 or 2029? Elon has put some big numbers out there, but do you think we can get to material inference in space in that time frame? Jensen Huang Only NVIDIA compute. Let me tell you why. It’s not just about the technology. NVIDIA’s performance per unit energy is the world’s best, and so we call it performance per watt. The most important thing is the productivity of NVIDIA’s compute. The return on invested capital was so good. Brad Gerstner Yes. Jensen Huang That you could put it out in space, and the launch cost— Brad Gerstner —is a small percentage of it. Okay. Jensen Huang The launch cost is not inconsequential. Brad Gerstner And so you’re not going to put low-cost computing out there. Correct? Jensen Huang You’re going to put high-value computing out there. We’re the highest-value computing system you could imagine, the highest computing system in the history of computing. We’re the perfect fit, really, to launch into space right now. Irrespective of the launch cost, it’ll still have excellent ROI. Brad Gerstner So do we get 500 megawatts in space by ’29? Jensen Huang Well, I’ll let him tell you, but I think— Brad Gerstner Impossible. Jensen Huang I—you know, what’s Elon launching? 100 launches a year right now? I don’t know exactly what it is—multiple a day. Brad Gerstner Yeah. Right. Jensen Huang And so you could put Vera Rubin in there, and it’ll justify it. The demand is there. His return on investment is so good on it, and so the productivity is excellent. It’s fungible, meaning you put a satellite out in space. If you don’t have a customer to rent it, it sits idle, and you certainly don’t want that. NVIDIA compute is going to be fully fungible, and there’ll be lots and lots of renters. We work with every single closed-model company in the world. We run every single open model in the world. We have thousands of startups working with us and enterprises in every single industry, and so the computing is completely fungible and it’s durable. Brad Gerstner Yeah. Jensen Huang It’ll last. It’ll sit up there, and we’ll keep it up there. The reason for that is because we keep upgrading CUDA and the software. We’re the only hardware company in the world that you pay eight years ago and we’re still updating your software with new algorithms. Brad Gerstner Right? ### Does AI Revenue Justify the Capex? Jensen Huang We’ll update the software for as long as the satellite is up there, and so it’s going to be really durable. I think the combination of these 3 traits, along with the technology, makes it the perfect payload for SpaceX. Brad Gerstner It costs trillions of dollars to build all this capex. You guys have leveraged your own balance sheet to help folks like the neoclouds build that out. In order for finance people to get it to pencil out, they need to see payback, right? You talked about some companies seeing a 1-year payback. $50 billion in the first year. Some spot pricing out of SpaceX, perhaps. But if I look at the total incremental offtake—the revenue of all the labs this year—if I add up the 2 frontier labs—oh, yeah, get yourself a snack—and we add up some of the inference labs, it’s about $200 billion. We need to see that double next year to $400 billion. We need to see it double the year after that to $800 billion in order to get in the zip code of the payback on trillions of dollars of capex. We had this massive unlock this year with Claude Code that led to a 10× increase in Anthropic’s revenues. What’s your confidence that the labs can continue to 2× to 3× their revenues every year in order that we get the payback that’s going to be required to build out trillions of dollars of capex? You’re investing in these companies. You have big investments in these companies, right? Do you see that level of growth continuing in these companies? Jensen Huang Oh, yeah. People misunderstand what’s going on. First of all, NVIDIA represents, call it, 80% of the revenue share of AI infrastructure in the world today, and so my description of the universe of computing and NVIDIA computing matters here. There are basically 3 large segments—probably slightly more than that, but let me categorize them into 3 large segments. The first segment of our computing is just baseline computing. You use it for data processing, molecular dynamics, physics simulation, fusion simulation, fission simulation—you name it, right? Quantum chemistry, quantum physics, everything. Brad Gerstner Right? Jensen Huang Okay. The world of simulation and data processing, not to mention computer graphics—all of that processing is done on our GPUs. If you just said there was no AI at all, just classical machine learning—XGBoost at Visa and Mastercard for fraud detection, recommender systems to figure out what news to recommend to you—just all of that stuff combined, classical computing, not AI, not deep learning, not agentic systems, is, call it, a third of our business. Brad Gerstner Yeah. Jensen Huang And that business is growing, call it, 30% to 40% per year. Brad Gerstner Yeah. Jensen Huang Okay. In the absence of any AI labs, NVIDIA is not zero, right? We’re quite a large company. The second part is the AI labs. People see 2, but we work with every one of them, right? Brad Gerstner Right. Jensen Huang MetaMuse is doing fantastic, right? Brad Gerstner Incredible. Jensen Huang And it’s largely NVIDIA. There were 0 last year, or a year before that it was just recommender systems, and now they’re killing it with MetaMuse. Of course, Groq didn’t exist 3 years ago. We didn’t run Gemini before; we run Gemini now. Starting this year, we started working with Anthropic. If you look at our compute, people see a couple of labs, but I see more than that. I see 5 or 6 labs. These labs are all doing great. Brad Gerstner Correct. Jensen Huang If you told me that represents not quite one-third of our business and, let’s say, a quarter of our compute, fantastic. Brad Gerstner Okay. Jensen Huang So let’s say that’s a quarter. When people are looking at 80% NVIDIA and we now run the infrastructure of the world, I just said 25%—maybe 30%—of it is the frontier labs. A third is just processing in general, right? Brad Gerstner Right? Jensen Huang And another third—this is the part that’s really great—is that we’re at the beginning of a new computing platform shift. A vast number of new startups are coming out. Brad Gerstner So many great companies. Jensen Huang Some of them are, of course, doing consumer stuff. One of them is really doing great—Noah Instinct is doing great. There’s all kinds of physical AI, biological AI, materials-science AI. Jeff Bezos’s new company, Prometheus, is doing fantastic. The number of companies across the spectrum of industries that have been started is in the thousands. What do they do? They rent NVIDIA compute. What is a third of our business? It’s kind of cloud computing for the next generation. Brad Gerstner Right? Jensen Huang It’s kind of like AWS for today, right? Brad Gerstner And so it’s not AWS for the future, right? Jensen Huang It represents about a third of our business altogether. My sense is that we’re in the early innings of the buildout of this new computing architecture. It’s a very big mistake to recognize just a couple of companies and think that they’re funding the infrastructure. ### Open vs. Closed Models One observation: open models. I’ve been a great advocate of open models, and the reason for that is because open models open up new opportunities. If you look at all these startup companies, 80% of them use open models. Brad Gerstner That’s right. Jensen Huang They’re not using them to compete with the frontier labs. They’re using them to unlock a new opportunity somewhere else. It could be in physical AI, manufacturing, healthcare—they’re doing something completely different. They’re taking these open models, and many of them are taking Neotron. These models are then customized into superintelligence. Brad Gerstner Yeah. Jensen Huang I see the general closed labs as artificial general intelligence, and it gets super good. Then you have all these vertical companies that are focused on a particular domain and have very proprietary data. They teach it to do something very specific, and they achieve superintelligence for their domain. Brad Gerstner Right? So again, if we go to your compute math, which I think a lot of people are using to forecast the future revenues of the labs, you said if you stand up a gigawatt of this compute, you can get $40 billion out the other end. If we took the number we talked about earlier—40 gigawatts of compute coming online next year—and took $40 billion out the other end, that would be $1.6 trillion of revenues coming out the other end of that factory. But of course, there’s something—you know, it doesn’t work that way, right? At the end of the day, you’ve got to look at the demand curve and the rate of diffusion to determine the revenue that’s going to come out the other end of the factory. What I hear a lot is that people are nervous. They’ve seen, over the course of the year, maybe the labs’ revenue slowing down a little bit, and they’re nervous that at some point you’re going to have the proverbial bubble popping, where we have more supply coming online than we have demand at a moment in time—that AI is more like these step functions rather than just this parabolic line straight up. I guess the question to you is: Are you worried about a moment in time when everybody gets nervous about financing the buildout, and then we don't get the buildout? Jensen Huang Well, you just have to break all these different things down—back down to first principles. The part that people are trying to extrapolate—the $100 billion of ARR, or $125 billion—is a part of the overall infrastructure buildout. It's a small part—not a small part; it's just a part. And so the question is: Are people going to continue shopping online? Brad Gerstner Right. You know, that's—so that third of the buildout, let's call that a third. You say it's a third of your business. Jensen Huang Take that to the bank. Brad Gerstner Yeah. Jensen Huang Right. That's not even a debate. I said a couple of years ago that, as we exit this decade, infrastructure buildout will probably be somewhere between $3 trillion and $4 trillion a year. Obviously, it's closer to $4 trillion than it is to $3 trillion. Brad Gerstner By the end of the decade. Jensen Huang Yeah. And if you said that $1.5 trillion of that was just compute—just compute—why is it going to be so big? Well, it was half a trillion before AI started. If I told you that over the course of 6, 7, or 8 years it became $1.5 trillion because the world is just consuming more compute, that's not illogical. Brad Gerstner Right. Jensen Huang There was a time when, if you wanted twice as many transistors, you just waited 1 year. If you wanted twice as many transistors, you waited another year. Same price. But if you want twice as many transistors now, it doesn't matter how long you wait; you have to pay twice as much. Brad Gerstner Okay. Jensen Huang Because Moore's law is over. Whatever computing capacity the world wants every single year, if you want twice as much, you're going to pay twice as much. Brad Gerstner Right. Jensen Huang And so, on top of that half a trillion dollars that was growing naturally because the world was using more and more digital technology, you're going to have to give it an X factor just to account for the fact that Moore's law is over. Without AI, the semiconductor industry was on its path to becoming the largest industry in the world. I've mentioned that long ago, but people forgot. It's just one of those things that I said. Yes, yes, yes, yes—I just said that 2 years ago. Brad Gerstner Yeah. Jensen Huang And so that's just one part. The only thing that we—and don't forget, because we're in a new computing-platform shift, there are going to be a whole lot of startups. Innovation is part of the economy. New companies—we said it's accelerating. You know, if you go back to Moore's law and general-purpose compute, it was $5 billion per gigawatt. Today, it's $50 billion for a Vera Rubin. So the question is: Does Vera Rubin deliver 10× more value to the world? The punch line is, I don't think it's necessary to extrapolate any of the labs into multiple trillions of dollars. It's unnecessary to do that. Brad Gerstner Yeah. I think that's an important one, because people—we just saw yesterday, even in the market, some chatter about whether OpenAI's revenue was $70 billion or $50 billion. I think you and I— Jensen Huang They're going to be insanely successful companies. ### Too Much Pressure on Two Labs Brad Gerstner Correct. Yeah, but you saw the pullback in a lot of the infrastructure companies on the fear that, “Oh, my gosh, OpenAI doesn't have the money, or won't have the money, to pay for the infrastructure.” I think the point that you're hammering home is that, when we look at the 40 gigawatts of infrastructure buildout next year, at least a third of it is just going to general-purpose compute replacement—the basic stuff out of the hyperscalers. A third of it is going to neoclouds, sovereigns, enterprises, and everybody else who's using this—all the startups. And a third of it is going not to 2 labs, but to many labs, including Groq, Thinking Machines, and all the others yet to come. Jensen Huang We're putting way too much pressure on 2 labs. I'm delighted to partner with them and support them. They're going to be 2 of the most consequential technology companies ever built in history. They're going to be magnificently successful. Brad Gerstner Yeah. Jensen Huang But they don't have to be anywhere near $1 trillion in revenues to be successful. Brad Gerstner Yeah. I think we're putting way too much pressure on that. And then the rest of the industry—we still do have to build out a few trillion dollars of computing infrastructure. Just to keep going, let's shift a little. How good is this, you guys? By the way, what are these called? Jensen Huang These are dried plums. Brad Gerstner Dried plums. The old dried plums. Jensen Huang Yeah. This one particular store in Taiwan is one of my favorites. Brad Gerstner Kind of licoricey. Jensen Huang There is licorice. Yeah, licorice. A little licoricey. Salt, sugar, cinnamon. Brad Gerstner Yeah. It has a little twang to it. Jensen Huang Yeah, yeah, you got it. Brad Gerstner Yeah. Okay. You said that, as to next year, demand for NVIDIA's chips is well over 100%. You guided to 70% revenue growth. You gave a year-forward look because you said supplies were constrained. Jensen Huang That's all I had at the time. Brad Gerstner That's right. Yeah. You said, “I got 18 months to work here.” Jensen Huang Right. ### AI Supply and Demand Dynamics Brad Gerstner You said you were very confident. Jensen Huang Don't come to work every day. Brad Gerstner You said you were very confident and had plenty of time to work with the supply chain to do even better. You expected to do even better and not disappoint customers. So are things going even better? Has the order book continued to grow? Because the other part is that you took a snapshot at a moment in time, and of course people are continuing to place orders. Do you think there's enough slack in the supply chain that you can squeeze more out? Is the order book continuing to grow as more of these capabilities are unlocked? Jensen Huang First, let me tell you why I told everybody. The challenge that we have is that we had giant order books. Every single CSP, every single NCP, and every single OEM had already placed orders with us. Startups, big companies, small companies—they had already placed orders with us. The reason for that is because we work with them, depending on the size of the customer, for at least 1 year—for the largest ones, 2 years of forecasts. We already have a very large order book, and because I was in the process of allocating—and what I was allocating to all of our customers was a lot less than— Brad Gerstner They wanted. Jensen Huang They wanted. If I didn't publicly say what we had, everybody would be second-guessing: Are they being shortchanged? Are we being— It was my way of essentially handling every single conversation where I'm disappointing people. Brad Gerstner Right. Jensen Huang So pretty much every customer would buy more from you if you could deliver more. Brad Gerstner By far. Jensen Huang And then, since then, of course, we have new orders almost weekly and daily. In addition, we're getting more supply, and some of the supply that's limiting us is land, power, shell, and downstream. Brad Gerstner Correct. Jensen Huang A couple of weeks ago, we announced 2 gigawatts in Australia. That's a very big deal. Two gigawatts to us is like $80 billion. Brad Gerstner Huge. Jensen Huang And I'm hoping that, when I go to Europe with UTC, we'll be able to announce some new partnerships there and unlock more land, power, and shell. This land, power, and shell is clearly one of the obstacles, and this is an area where NVIDIA has a huge advantage. Based on what we build, if there's a neocloud—like CoreWeave or EBS—and they would like to scale out their business, NVIDIA has the entire, full-stack, complete AI infrastructure. We could support them as they scale out. We can build in places and with partners and customers that nobody else can. I think this gives us a lot more resilience. Nonetheless, as we continue to increase supply upstream and downstream, one of the critical areas is substrate. Substrates are getting larger, and there are more chips, so it's working against them in 2 ways. Every single chip that's being shipped uses more of the material. These organic materials are really hard to make at the precision that we need. As the packages get larger, the thermal coefficients and the way we stack things make it really hard to handle things like warpage and various manufacturing defects. At the scale that we do, we have to be very precise, and we need really great partners. That's supply-constrained. Silicon photonics is supply-constrained. Everything is supply-constrained, so we're going to keep working on it. I have 18 months to do so. Every single week that goes by, we increase supply, but nonetheless, as we keep increasing supply, the order book keeps going up. Brad Gerstner Right. Jensen Huang It's going to be hard to close that gap for a few years. ### Review of Wall Street Consensus Brad Gerstner Take a look at this chart. You know, this is like a deposition— Jensen Huang No, right? Brad Gerstner Excuse me, now it usually goes like this. Okay, so the gray line was the consensus at the start of 2025. The blue line was Altimeter’s forecast way back then, so we were a lot higher than the consensus. The line is now where the new consensus is based on your guidance. One of the things that blows my mind about this is that the 25 people who get paid to forecast what this company is going to do—the consensus of all their ideas—were off by a couple hundred percent. Right at the beginning of 2025—oh, no, this time last year. Jensen Huang At this time last year. Brad Gerstner Yes, that was the consensus at this time last year. Okay, one year ago. Jensen Huang Last week. Brad Gerstner Okay. So, in a year, the consensus forecast has shifted up by 100% to 200% for this company. Jensen Huang Mm-hmm. Brad Gerstner Extraordinary. In my 20-plus years of doing this, you don’t often see consensus being off by that much, and yet the stock is only up 20%. You literally doubled what the revenue forecast and the free cash flow forecast were for the company. You’re now trading at 13 to 14 times earnings, depending on who you’re looking at. What do you think the market is saying about Nvidia, or about the confidence in the market’s expectations, to cause that to be the case? Because usually, if you see expectations shift up by that much, the stock would have shifted up by a lot more. Jensen Huang First of all, for a company to be as large as we are, gaining share tells you something about our resilience. Brad Gerstner Yes. Jensen Huang It tells you something about the size and scale of our advantage. Brad Gerstner Right. Jensen Huang Is that advantage increasing or decreasing? Is your moat increasing or decreasing? Brad Gerstner Increasing. Jensen Huang Your competitive moat is increasing at scale. Brad Gerstner Yeah. ### Vera Rubin Ultra Orders Jensen Huang Let me give you an example. Let’s just do a thought experiment right now. If you want to get in line for Vera Rubin Ultra, it’s close to order-placement time. Brad Gerstner Wow. Jensen Huang Sight unseen. Brad Gerstner Yeah. Right. Jensen Huang But they have the confidence. Let’s say you’ve got to place a $100 billion order with somebody. How do you know it’s going to be any good? Brad Gerstner Right. Jensen Huang Because Nvidia builds it, you know it’s going to be better than Vera Rubin. Brad Gerstner Right. Jensen Huang And Vera Rubin is awesome. Brad Gerstner Right. Jensen Huang The reason why everybody has confidence in Vera Rubin is because Grace Black was awesome. Brad Gerstner Yeah. Jensen Huang In the world of computing, the workloads that these computers do are so complex. You can’t just measure them with one benchmark. Brad Gerstner Right. Right. Jensen Huang You know, it’s $50 billion. Brad Gerstner Yes. Jensen Huang The number of technologies that have to come together to deliver the performance is insane. The question is: sight unseen, who can you place a $50 billion order with? Brad Gerstner Yeah. Jensen Huang We’re literally the only one. Brad Gerstner Yeah. Jensen Huang Back when it was placing a billion-dollar purchase order, that was challenging enough. But if you had placed a $100 million purchase order and the system was sitting right in front of you, you could benchmark the chips. Brad Gerstner Right. Right. Jensen Huang How do you benchmark these systems? Just building it is a miracle. Brad Gerstner Right. Jensen Huang If somebody wants to do an evaluation, I’m happy to support it, but you’re going to need to give me a giant data center, and we’re going to need a couple of years to retrofit it. You can’t even benchmark these things anymore. Brad Gerstner Interesting. And that’s exactly what happened with TSMC. Jensen Huang The reason why we give 100% of our business to TSMC is because they’re the only company in the world that I can plan five years out with and know, sight unseen, that that transistor is going to be really good because it’s better than that one, and better than that one. Brad Gerstner Yeah. Jensen Huang They’ll have capacity for me, and they know how to work with my business team. They have the agility to handle all of my ups and downs. For example, they can deal with the fact that I went from gray to blue to green. Brad Gerstner Yeah. Exactly. Okay. Jensen Huang Exactly. They have to be big enough to help me go from gray to blue to green. Brad Gerstner Exactly. Jensen Huang There’s no other company in the world that could help me do that. Brad Gerstner That’s right. Jensen Huang They also have all these different process technologies, from silicon photonics to logic to memory to push-pull, push-pull packaging. The number of things that we do is so dynamic. Brad Gerstner And how do you possibly do that with a company that is one-tenth their size? It’s impossible. Jensen Huang I think the big idea is this: We’re building infrastructure for the planet, and Nvidia has become an infrastructure company. Brad Gerstner Mm-hmm. Jensen Huang These infrastructures cost so much money, and technology is moving so fast. We’re practically the combination of trust, the fact that we run everything, and the fact that we started out with some GPU models. Look at the FlashAttention mechanisms that were invented, the way that we parallelized the models and sharded them across NVL72, and how all of that has changed. Now people are talking about transformer diffusion models, hybrid SSMs, transformer hybrids, and all these different technologies. The evolution of all that—we know that it all runs on CUDA. The only thing we don’t know is whether you need to do more pre-training, post-training, or just agentic inference. It turns out we run it all anyway. Brad Gerstner Right. Jensen Huang You could build a $50 billion data center and know that the work is fungible. If, for whatever reason, you wanted to sublease half of your capacity to somebody, Nvidia is completely fungible. Brad Gerstner Right. Jensen Huang We’re fungible in workloads, fungible across models, and fungible across customers. We’re also deployable everywhere in the world. ### Global Reach: Europe, Japan & Beyond You start a company in Europe or the United States, but you want to serve customers in Europe or Japan. It turns out we’re already there. Nvidia’s infrastructure is already there. OpenAI wants to go to Europe? We’ll take them there in a heartbeat. If they want to go to India, in a heartbeat. If Anthropic wants to go to Japan or Korea, in a heartbeat—we’re already there. Brad Gerstner Right. Jensen Huang All of these architectures are already there. If you take Nvidia as a whole and then come back and say, “Okay, I want to do the thought experiment. I want to place $100 billion on somebody,” how do you do it? Brad Gerstner A year ago, we were sitting here, and the consensus forecast was that you were going to be flat or shrink in 2027. Now the consensus forecast is that you’re going to grow 74%. I want to let people sit with that, because a lot of people follow these consensus forecasts, and they were really wrong. Jensen Huang Did they go back and listen to your podcast from last time? Brad Gerstner Right. You told them. Jensen Huang Yeah, I explained everything. Listen, ladies and gentlemen: I did explain everything. Brad Gerstner And so I look out at 2029. Jensen Huang Wasn’t there an ROI question? Brad Gerstner Yes. Jensen Huang And then there was a question about custom versus general purpose. Brad Gerstner Custom versus general purpose. Jensen Huang The reason why Nvidia is general purpose is so that we can have all this fungibility. Brad Gerstner That’s right. Jensen Huang The benefit of fungibility is low risk. Low risk really comes into play when you’re placing big dollars. When you’re building out $50 billion to $100 billion worth of infrastructure, you need to know that the infrastructure is going to be really good for a long time. Brad Gerstner No matter what. Jensen Huang Specialization is the enemy of fungibility. Specialization is the enemy of durability. Brad Gerstner Right? Jensen Huang So long as I can prove that our performance is extraordinary—and we prove that every single day—fungibility and durability translate to rentability, which translates to productivity and financeability, and therefore leads to the lowest total cost of ownership. Brad Gerstner Right. Right. Exactly. Well, now you’re buying back the largest amount in history. On my math, you could probably buy upwards of 10% of the company back within the next 2 years. At Goldman, you called Nvidia the first growth-value company. Jensen Huang Sounds a lot like Apple’s pitch back then. Brad Gerstner Okay. Is that right? Jensen Huang Yeah, or more—the idea that they could return the capital. Brad Gerstner It worked for sure. Jensen Huang It worked. It went from $30 at the time to $300. Brad Gerstner Yeah. Jensen Huang $300-plus. In a lot of ways, we’re going through a similar transition. The thing I forgot to mention—and, Brad, I think you were about to ask me about that—is that most of our investment and financing partnership strategies are trying to set up 2 business models, 2 new business models. One business model we’ve been working on for some time and will accelerate in the near future is software licensing. You know, the body of software that we have in our company is just extraordinary, and that data center is so large. Imagine all the software that runs in it, and we provide the entire stack. Brad Gerstner Well, our entire stack is completely industrial-strength for the CSPs. They want to run their own stack, right? Jensen Huang And so we license or give them our software as open source, and then they take it, modify it, integrate it into theirs, and operate their services. But for a lot of the world’s companies, they don’t have the IT department or the software department to do it, so they need me to maintain it. And so, because AI in the last several years was largely focused on training, but now they’re moving into deployment— Brad Gerstner Right? Jensen Huang They need to operate these as factories, and they need to know that I’m on the hook. I’m on the SLA to support and maintain it, keep it secure, and keep it safe for as long as we shall live. And so, software licenses for enterprise businesses will be very, very large. If our software license is, just for simple math, a couple of thousand dollars per GPU per year, and we have millions of GPUs—we ship, say, 20 million GPUs a year. I’m growing, and half of them go into the world’s enterprises. It adds up. Brad Gerstner Yeah. Jensen Huang Okay. And so I’m fairly certain this is going to be a very big part of our business: software licenses. And then, of course, there are some clouds that rely on NVIDIA for our full stack. Brad Gerstner They rely on us for our offtake and our ecosystem. They also rely on us, maybe, as an investment partner to help them secure their financing and capital. Jensen Huang They’ve secured land, power, and shell. They’re regional champions. They’ve got all kinds of things working to their advantage. But they need our support in these 3 areas. And so, in those kinds of relationships, it’s very likely that we’ll move to rev share. We’ll make an investment and get the benefit of the sale up front, but we’ll also have the benefit of rev share going forward. We’re not a cloud service provider, but we’re going to be providing offtake and software support for them for 4, 5, or 10 years. We ought to get some economic return on that. And so, rev share— Brad Gerstner Notice I just described 2 new growth vectors independent of the hardware sales of our company, and they’re much more related to the installed base of our platform. What’s the benefit of software licensing if you have no installed base? And how could you have rev share if there’s no offtake? Jensen Huang Okay. And so, the installed-base element, the richness of our software stack, the offtake, the richness of our offtake, and the ecosystem that we have around the world are going to fuel that. Brad Gerstner One of the ways—you know, we’re talking about this trading at 13 times. When Tim and Luca went to see Warren Buffett, one of the reasons Buffett has said he doesn’t like technology companies is that they produce an enormous amount of free cash flow, and rather than ever giving any of it back to the shareholders, they go and spend it on something else. Oftentimes, the something else isn’t as good as the thing that caused them to generate the cash flow in the first place. You guys are now spending over 50% of your free cash flow buying back your stock, effectively returning it to shareholders, while you have this exponential growth at the same time. As we’ve already discussed— Jensen Huang We said 50, right? Brad Gerstner I think this last quarter we did 60-some-odd. Jensen Huang Correct. Yeah. We’re going to increase it. We invest in really high-quality companies, as you guys know, and the last couple of years, whoever NVIDIA buys from—whoever Jensen buys from—you’ve got to invest in that company. Brad Gerstner That’s right. And now the new word on the street is, whoever Jensen invests in, you invest in that company. Jensen Huang And it’s true. The reason for that is because, like you, we’re in the ecosystem. We’re in the flow of information. We know those investments together. Brad Gerstner Yeah. And we know a lot of smart people, and we want to back these people when they want to do something great. So we’re kind of on the inside. We have a lot of information, and our investments are well-informed. Jensen Huang Right. Brad Gerstner And not to mention, NVIDIA is also in the ecosystem, so we could help them succeed. Jensen Huang That’s right. And so there are a lot of different reasons why the investments turn out great. But as a percentage of our total cash generation, it’s going to decline very rapidly, partly because OpenAI and Anthropic aren’t going to come around again. That opportunity isn’t going to come around again, frankly. And secondarily, our cash-flow generation is growing quite fast. Brad Gerstner Right. And to that point, it seems to me that if you’re buying back that much, and the durability of the revenues, I think, as you’ve described here today, is longer than people expect, that’s the one place where people are still hanging too much on this only lasting so long. Ezra asked you this on the pod the other day: How long until the bubble bursts? And I think you said something about 2 to 3 years. You have good visibility that the demand is there, and who knows what happens after that. I heard some people saying, “Oh, it may only be 2 years before this bursts.” But when I look at this, this is fundamentally a rearchitecture of all global compute forever. Jensen Huang Forever. So unless there’s a global depression— Brad Gerstner Yeah. Jensen Huang Right. Then every single year, people are going to have to consume an enormous amount of NVIDIA compute. Brad Gerstner Yeah. Jensen Huang And so, the way to think about that is, the CSPs were running about half a trillion. The whole compute industry—the whole world—is running, call it, $1 trillion of build-out each year. Brad Gerstner Right. Jensen Huang And every year more gets built, and every 5 years or so, some of it gets depreciated, turned off, taken offline, and so on. Some of it lasts longer. That old world was in the retrieval way of doing computation. In the new way of doing computation, I’ve predicted that going out of the decade—so, 3, 4, let’s say 4, let’s say 5—I have a sense that we’re going to continue to grow from there. If you told me that the annual capacity building, infrastructure building, and refreshing—the combination of new and old—runs around $10 trillion, I’m not even a little bit surprised. Brad Gerstner Yeah. Right. I mean, just again for people listening, you think $4 trillion of annual spend by the end of the decade because every time you consume tokens, every time you use intelligence, you have to run the machine. The factory has to run. And I think you’ve described that as about 50%. If you assume that $50 trillion of global GDP is related to knowledge work, and we get a 20% productivity gain over 5 years on that—you’ve seen much bigger productivity gains inside NVIDIA—if we see that sort of productivity gain, then we get $10 trillion of incremental economic activity by 2029 or 2030. That seems very reasonable. And you’re just saying we’ll spend half of that— Jensen Huang On capex, on the intelligence to produce it. Brad Gerstner And so maybe that’s a good place for us to take a bit of an offramp. You’ve been incredibly generous. I could go on forever, but I want to wrap where we started. When we look back— Jensen Huang If this was one of those eating-and-drinking podcasts— Brad Gerstner We could keep going. We could keep going. Jensen Huang Yeah. Brad Gerstner But it wasn’t so long ago that the human condition was described as nasty, brutish, and short. Jensen Huang Was that last week, or— Brad Gerstner Innovation has really driven all human progress, mostly in the last 200 years: more time with family, longer lives, less time doing drudgery that humans really didn’t want to do, and safer lives. Jensen Huang We’re the first generation where 63 years old is considered young. Brad Gerstner Exactly. Exactly. Jensen Huang People call me “young man.” Brad Gerstner Safer, better lives. All of this is despite all of these concerns that you and I hear out there. By the way, those same concerns were expressed at the beginning of the Industrial Revolution, right? ### The Decade Ahead: 99.99% Confident What’s your level of confidence that, in 5 years from now, we will have made the right decisions and the world will be that much better a place because of the investment and all that we’re doing? Are you more confident as we sit here today than you were 12 months ago? Jensen Huang I’m 99.99%—just in service to, and in honor of, a good friend, not using 0 and 100—I’m 99.99%, 99% confident that too many of us care deeply that the outcome turns out well. Brad Gerstner Yeah, that we will dedicate our might, our intellect, and our resources to ensure we get there responsibly and optimistically. Jensen Huang And so there are just so many people doing the right things. Oftentimes, there are people who pontificate about these things, and they’re informed by several things, but they don’t realize that there are a lot of people, industries, incredible researchers, and engineers outside of their sphere of information who are doing amazing things all by themselves, independent of them. Just because they didn't say it doesn't mean a lot of smart people aren't doing smart things. I say this all the time inside the company: Even without my leadership or strategies, the company is doing smart things all by itself. Smart people do smart things. They do more impactful things when you organize them, but amazing people are doing smart things all by themselves. And so, when somebody tells me that they're worried, it's good that we worry, right? We should express our worries. We should express our concerns. But we shouldn't predict the end of the world. Because they're underestimating the human condition. They're underestimating the human capacity. And for some reason—maybe it's just hubris, maybe it's arrogance—they think that unless they do it, it won't get done. Brad Gerstner Right. Right. Jensen Huang Unless they do it, it won't get done. Brad Gerstner Right. Jensen Huang The fact of the matter is, this company is doing amazing things whether I say it or not. Brad Gerstner Okay. Jensen Huang I get all the accolades and the blame, but in both cases, I'm fair—especially the accolades. The reason for that is that when you surround yourself with amazing people and create the conditions under which everybody wants the same outcome—the good outcome, the good version of it—I think most people do. That's the reason why I say we need to hold ourselves accountable and responsible. Brad Gerstner But we should express a future that's optimistic. Jensen Huang For sure. And I am 99.99% certain— Brad Gerstner Right? Jensen Huang —that, surrounded by all the amazing people in the world, we're going to get there together. Brad Gerstner Yeah, agreed. And I think you're a national treasure. We're lucky to have you helping to lead the charge. The nation's indebted to you for getting everybody together, getting us all on the same page, and building the framework. There's going to be more work to do. But as a father of 2—and I know you have kids, and you're going to have grandkids—we don't care about anything more than actually building a better world for them, right? We could easily stop. Jensen Huang That's right. Brad Gerstner But if we stop today, we lose, right? We lose the opportunity for that better world. So we'll wrap it there. Thank you, Jensen. Jensen Huang Thank you, guys. And Brad, thank you. I've never met a greater patriot—somebody who cares so much about the industry, but even more so about the country. Your gift, your enthusiasm, and your tireless energy in coordinating and pulling together a lot of people in order to do something great like what you did—it's incredible. As a reminder to everybody, these are just our opinions, not investment advice.