Jensen Huang：NVIDIA——4万亿美元公司与AI革命｜Lex Fridman Podcast #494

The following is a conversation with Jensen Huang, CEO of NVIDIA, one of the most important and influential companies in the history of human civilization. NVIDIA is the engine powering the AI revolution, and a lot of its success can be directly attributed to Jensen's sheer force of will and his many brilliant bets and decisions as a leader, engineer, and innovator. This is Lex Fridman Podcast. And now dear friends, here's Jensen Huang. You've propelled NVIDIA into a new era in AI, moving beyond its focus on chip scale design to now rack scale design. I think it's fair to say that winning for NVIDIA, for a long time, used to be about building the best GPU possible. You still do that, but now you've expanded it to extreme co-design of the GPU, CPU, memory, networking, storage, power, cooling, software, the rack itself, the pod that you've announced, and even the data center.

Let's talk about extreme co-design. What is the hardest part of co-designing a system with that many complex components and design variables?

Yeah, thanks for that question. First of all, the reason why extreme co-design is necessary is because the problem no longer fits inside one computer to be accelerated by one GPU. The problem that you're trying to solve is that you would like to go faster than the number of computers that you add. You added 10,000 computers, but you would like it to go a million times faster.

Then, all of a sudden, you have to take the algorithm, break it up, refactor it, shard the pipeline, shard the data, and shard the model. Now, all of a sudden, when you distribute the problem this way—not just scaling up the problem, but distributing the problem—everything gets in the way. This is the Amdahl's law problem, where the amount of speedup you have for something depends on how much of the total workload it is. If computation represents 50% of the problem, and I sped up computation infinitely—like a million times—I only sped up the total workload by a factor of 2.

Now, all of a sudden, not only do you have to distribute computation, you have to shard the pipeline somehow. You also have to solve the networking problem because all of these computers are connected together. Distributed computing at the scale that we do—the CPU is a problem, the GPU is a problem, the networking is a problem, the switching is a problem, and distributing the workload across all these computers is a problem. It's just a massively complex computer science problem.

We just have to bring every technology to bear. Otherwise, we scale up linearly, or we scale up based on the capabilities of Moore's law, which has largely slowed because Dennard scaling has slowed.

I'm sure there are trade-offs there. Plus, you have a complete set of disparate disciplines here. I'm sure you have specialists in each one of these: high-bandwidth memory, the network and NVLink, the NICs, the optics and copper that you're using, power delivery, cooling, and all of that. I mean, there are world experts in each of those. How do you get them in a room together to figure out—

That's why my staff is so large.

What's the process? Can you take me through the process of the specialists and the generalists? How do you put together the rack when you know the set of things you have to shove into a rack together? What does that process look like, designing it all together?

There's the first question, which is: What is extreme co-design? You're optimizing across the entire stack of software, from architectures to chips, to systems, to system software, to the algorithms, to the applications. That's one layer. The second thing that you and I just talked about goes beyond CPUs and GPUs and networking chips and scale-up switches and scale-out switches. Then, of course, you've got to include power and cooling and all of that because all these computers are extremely, extremely power-hungry. They do a lot of work and they're very energy-efficient, but in aggregate they still consume a lot of power.

That's the first question: What is it? The second question is: Why is it? We just spoke about the reason. You want to distribute the workload so that you can exceed the benefit of just increasing the number of computers. The third question is: How is it? How do you do it? That's kind of the miracle of this company.

When you're designing a computer, you have to have an operating system for the computer. When you're designing a company, you should first think about what it is that you want the company to produce. I see a lot of companies' organization charts, and they all look the same: hamburger organization charts, software organization charts, and car company organization charts. They all look the same, and it doesn't make any sense to me.

The goal of a company is for the company to be the machinery, the mechanism, the system that produces the output. That output is the product that we like to create. The architecture of the company should reflect the environment in which it exists. It almost directly says what you should do with the organization.

My direct staff is 60 people. I don't have one-on-ones with them because it's impossible. You can't have 60 people on your staff if you're going to get work done—

So you still have 60 reports. You still have across—

More, yeah.

More. Most of your staff at least have a foot in engineering.

Almost all of them. There are experts in memory, CPUs, optics, GPUs, architecture, algorithms, and design.

That's incredible. So you constantly have an eye on the entire stack, and you're having to have intense discussions about designing the entire stack?

And no conversation is ever one person. That's why I don't do one-on-ones. We present a problem, and all of us attack it. We're doing extreme co-design. Literally, the company is doing extreme co-design all the time.

So even if you're talking about a particular component, like cooling or networking, everybody's listening in?

Yeah, exactly.

And they can contribute, "Well, this doesn't work for the power distribution. This doesn't work for the memory. This doesn't work for this."

Exactly. And whoever wants to tune out can tune out. The reason for that is because the people who are on the staff know when to pay attention. If there's something they could have contributed to and they didn't contribute to it, I'm going to call them out: "Hey, come on, let's get in here."

So, as you mentioned, NVIDIA is this company that's adapting to the environment. At what point did the environment change, and did you begin adapting, sort of secretly, in the early days—from GPU for gaming, maybe through the early deep learning revolution—to where we're now going to start thinking of it as an AI factory? What does NVIDIA do? It produces AI; let's build a factory that makes AI.

I could reason through it systematically. We started out as an accelerator company. But the problem with accelerators is that the application domain is too narrow. It has the benefit of being incredibly optimized for the job. Any specialist has that benefit. The problem with intense specialization is that, of course, your market reach is narrower, but that's even fine. The problem is that the market size also dictates your R&D capacity, and your R&D capacity ultimately dictates the influence and impact that you can possibly have in computing.

When we first started out as a very specific accelerator, we always knew that was going to be our first step. We had to find a way to become accelerated computing. But the problem is, when you become a computing company, it's too general-purpose and it takes away from your specialization.

I connected 2 words that actually have a fundamental tension. The better computing company we become, the worse we become as a specialist. The more of a specialist we are, the less capacity we have to do overall computing. I connected those 2 words together on purpose because the company has to find that really narrow path, step by step by step, to expand our aperture of computing, but not give up on the most important specialization that we had.

The first step that we took beyond acceleration was that we invented a programmable pixel shader. That was the first step toward programmability, our first journey toward moving into the world of computing. The second thing that we did was put FP32 into our shaders.

That FP32 step—IEEE-compatible FP32—was a huge step in the direction of computing. It was the reason why all of the people who were working on stream processors and other types of data-flow processors discovered us. They said, "Hey, all of a sudden, we might be able to use this GPU that's incredibly computationally intensive, and it's now compliant with IEEE. I can take my software that I was writing previously on CPUs and see about using the GPU for that."

That led us to put C on top of FP32, which we called Cg. The Cg path eventually took us to CUDA. Putting CUDA on GeForce was a strategic decision that was very, very hard to make because it cost the company enormous amounts of our profits, and we couldn't afford it at the time. But we did it anyway because we wanted to be a computing company. A computing company has a computing architecture, and a computing architecture has to be compatible across all of the chips that we build.

Can you take me through that decision? Putting CUDA on GeForce—you could not afford to do it? Why boldly choose to do that anyway?

Yeah, excellent. I would say that was the 1st strategic decision that was as close to an existential threat.

For people who don't know, it turned out to be—spoiler alert—one of the most incredibly brilliant decisions ever made by a company. CUDA turned out to be an incredible foundation for computation in this AI infrastructure world.

Thank you.

Just setting the context: It turned out to be a good decision.

Yeah, it turned out to have been a good decision. Here's the way it went. We invented this thing called CUDA, and it expanded the aperture of applications that we can accelerate with our accelerator. The question is: How do we attract developers to CUDA? Because a computing platform is all about developers.

Developers don't come to a computing platform just because it could perform something interesting. They come to a computing platform because the install base is large. A developer, like anybody else, wants to develop software that reaches a lot of people. So the install base is, in fact, the single most important part of an architecture.

The architecture could attract enormous amounts of criticism. For example, no architecture has ever attracted more criticism than x86, a less-than-elegant architecture, yet it is the defining architecture of today. It gives you an example that, in fact, so many RISC architectures, which were beautifully architected and incredibly well-designed by some of the brightest computer scientists in the world, largely failed.

I've given you 2 examples where one is elegant and the other one is barely aesthetic, and yet x86 survived.

The install base is everything.

The install base defines an architecture. Everything else is secondary, okay?

There were other architectures at the time. CUDA came out, OpenCL was there, and there were several other competing architectures. The decision that we made that was good was that we said, “Hey, look, ultimately it’s about install base, and what is the best way we could get a new computing architecture into the world?”

By that timeframe, GeForce had become successful. We were already selling millions and millions of GeForce GPUs a year, and we said, “You know, we ought to put CUDA on GeForce and put it into every single PC, whether customers use it or not, and use it as a starting point for cultivating our install base.”

Meanwhile, we would go and attract developers. We went to universities, wrote books, taught classes, and put CUDA everywhere. At the time, the PC was the primary computing vehicle. There was no cloud, and we could put a supercomputer in the hands of every researcher in school, every scientist, every engineering school, or every student in school. Eventually, something amazing would happen.

The problem was that CUDA increased the cost of that GPU, which is a consumer product, so tremendously that it completely consumed all of the company’s gross profit dollars. At the time, the company was probably worth—I don’t know—$8 billion. Was it like $8 billion, or $6 or $7 billion? After we launched CUDA, I recognized that it was going to add so much cost, but it was something we believed in. Our market cap went down to around $1.5 billion.

We were down there for a while, and we clawed our way back slowly, but we carried CUDA on GeForce. I always say that NVIDIA is the house that GeForce built, because it was GeForce that took CUDA out to everybody.

Researchers and scientists discovered CUDA on GeForce because many of them were gamers. Many of them built their own PCs anyway. In a university lab, many of them built clusters themselves using PC components. That’s how we got going.

And then that became the platform and the foundation for the deep learning revolution.

That was another great observation. Yeah.

That existential moment—do you remember what those meetings were like? What were those discussions like, deciding as a company to risk everything?

I had to make it clear to the board what we were trying to do, and the management team knew our gross margins were going to get crushed. You could imagine a world where GeForce would carry the burden of CUDA, and none of the gamers would appreciate it or pay for it. They only pay a certain price, and it doesn’t matter what your cost is.

We increased our cost by 50%, and that consumed our gross profit dollars. We were a 35% gross-margin company, so it was quite a difficult decision to make. But you could imagine that someday this would go into workstations and supercomputers, and in those segments, maybe we could capture more margin. You could reason your way into being able to afford this, but it still took a decade.

But that’s more of a conversation with the board, convincing them. Psychologically, as NVIDIA has continued to make bold bets that predict the future and, in part—especially now—define the future, I’m almost looking for wisdom about how you’re able to make those decisions, to make leaps like that as a company.

First of all, I’m informed by a lot of curiosity. At some point, there’s a reasoning system that convinces me: Clearly, this outcome will happen. This will happen. I believe it in my mind, and when I believe it in my mind, you know how it is: You manifest a future, and that future is so convincing there’s no way it won’t happen. There’s a lot of suffering in between, but you’ve got to believe what you believe.

So you envision the future and, essentially, from an engineering perspective, manifest it?

Yeah. You reason about how to get there. You reason about why it must exist. We all reason about it here. The management team reasons about it. We spend a lot of time reasoning about it.

The next part of it is probably a skill thing. Oftentimes, in leadership, leaders stay quiet or they learn about something, and then they issue some manifesto. It’s a brand-new year, and somehow, at the end of next year, we’re going to have a brand-new plan: a big, huge layoff this way, a big, huge organization change this way, a new mission statement, brand-new logos—that kind of stuff.

I’ve just never done things that way. When I learn about something and it’s starting to influence how I think, I’ll make it very clear to everybody near me that this is interesting, this is going to make a difference, and this is going to impact that. I reason about things step by step by step.

Oftentimes, I’ve already made up my mind, but I’ll take every possible opportunity—external information, new insights, new discoveries, new engineering revelations, new milestones developed—and use it to shape everybody else’s belief system. I’m doing that literally every single day.

I’m doing that with my board, I’m doing that with my management team, and I’m doing that with my employees. I’m trying to shape their belief systems such that, when I say, “Hey, let’s buy Mellanox,” it’s completely obvious to everybody that we absolutely should.

On the day that I said, “Hey guys, let’s go all in on deep learning—and let me tell you why,” I had already been laying down the bricks for different organizations inside the company. Every organization and everybody might have heard all of it; most of the company, of course, hears pieces of it. On the day that I announce it, everybody has bought into many pieces of it.

In a lot of ways, I like to announce these things, and I imagine that the employees are saying, “You know, Jensen, what took you so long?” I’ve been shaping their belief system for some time, and therefore, leadership sometimes looks like you’re leading from behind. But you’ve been shaping their belief system to the point where, on the day that you declare it, there’s 100% buy-in.

That’s what you want. You want to bring everybody along. Otherwise, you announce something about deep learning and everybody goes, “What are you talking about?” You announce, “Let’s go all in on this thing,” and your management team, your board, your employees, and your customers are all saying, “Where’s this coming from? This is insane.”

So, GTC—if you go back in time and look at the keynotes—I’m also shaping the belief system of my partners in the industry, and I’m using that to shape the belief system of my own employees. By the time that I announce something, like, for example, we just announced Grok, I’ve already been laying the foundation. I’ve been talking about the stepping stones for 2.5 years.

You go back and say, “Oh my gosh, they’ve been talking about it for 2.5 years.” I’ve been laying the foundation step by step by step, so when the time comes and you announce it, everybody’s saying, “You know, what took you so long?”

But it’s not just inside the company. You’re shaping the landscape—the broader global landscape of innovation. By putting those ideas out there, you really are manifesting reality.

We don’t build computers. We actually don’t build clouds. As it turns out, we’re a computing platform company.

And so nobody can buy anything from us. That’s the weird thing. We vertically design and vertically integrate to design and optimize, but then we open up the entire platform at every single layer to be integrated into other companies’ products and services, clouds, supercomputers, and OEM computers. The amazing thing is, I can’t do what I do without having convinced them first. Most of GTC is about manifesting a future that, by the time my product is ready, they’re going, “What took you so long?”

Yeah. One of the things you’ve been a believer in for a long time is scaling laws, broadly defined. Are you still a believer in the scaling laws?

Yeah. We have more scaling laws now.

I think you’ve outlined 4 of them: pre-training, post-training, test-time, and agentic scaling. When you think about the future—the deep future and the near-term future—what are the blockers that you’re most concerned about, that keep you up at night, that you have to overcome in order to keep scaling?

We can go back and reflect on what people thought were blockers. In the beginning, we were the first to recognize the pre-training scaling law. People thought, rightfully so, that the amount of high-quality data that we have would limit the intelligence that we achieve. That was an important scaling law. The larger the model, the more data you use, and that results in a smarter AI.

Ilya Sutskever said, “We’re out of data. Pre-training is over,” or something like that. The industry panicked that this was the end of AI. Of course, that’s obviously not true. We’re going to keep scaling the amount of data that we have to train with. A lot of that data is probably going to be synthetic, and that also confused people.

What people don’t realize is they’ve kind of forgotten that most of the data that we train on, that we teach and inform each other with, is synthetic. It’s synthetic because it didn’t come out of nature. You created it, I’m consuming it, I modify it, augment it, regenerate it, and somebody else consumes it. We’ve now reached a level where AI is able to take ground truth, augment it, enhance it, and synthetically generate an enormous amount of data.

That part of post-training continues to scale. The amount of data that we could use that is human-generated will be smaller and smaller and smaller. The amount of data that we use to train models is going to continue to scale, to the point where training is no longer limited by data. It’s now limited by compute.

Then the next phase is test-time scaling. I still remember people telling me, “Inference? Oh, yeah, that’s easy. Pre-training, that’s hard.” These are giant systems that people are talking about. Inference must be easy. Inference chips are going to be little tiny chips, and they’re not like NVIDIA’s chips. Those are going to be complicated and expensive.

In the future, inference is going to be the biggest market, and it’s going to be easy, and we’re going to commoditize it. Everybody can build their own chips. That was always illogical to me because inference is thinking, and I think thinking is hard. Thinking is way harder than reading.

Pre-training is just memorization and generalization, looking for patterns and relationships. You’re reading and reading versus thinking, reasoning, solving problems, taking new experiences, and decomposing them into solvable pieces that we then go off and solve, either through first-principles reasoning, through previous examples and prior experiences, or through exploration and search, trying different things. That whole process of test-time scaling—inference—is really about thinking. It’s about reasoning, planning, and search. How could that possibly be compute-light?

We were absolutely right about that. Test-time scaling is intensely compute-intensive. The question is, okay, now we’re at inference and we’re at test-time scaling—what’s beyond that?

Well, obviously, we have now created 1 agentic person, and that 1 agentic person has a large language model that we’ve now developed. During test-time, that agentic system goes off and does research, bangs on databases, goes out and uses tools, and one of the most important things it does is spin off and spawn a whole bunch of sub-agents.

That means we’re now creating large teams. It’s so much easier to scale NVIDIA by hiring more employees than it is to scale myself. The next scaling law is the agentic scaling law. It’s kind of like multiplying AI. We can spin off agents as fast as you want to spin off agents.

I have 4 scaling laws. As we use the agentic systems, they’re going to create a lot more data and a lot more experiences. Some of it we’re going to say, “Wow, this is really good. We ought to memorize this.” That data set then comes all the way back to pre-training. We memorize and generalize it, then refine it and fine-tune it back into post-training.

Then we enhance it even more with test-time scaling, and the agents—agentic systems—put it out to the industry. This loop, this cycle, is going to go on and on and on. It kind of comes down to 1 thing: intelligence is going to scale by 1 thing, and that’s compute.

But there’s a tricky thing there that you have to anticipate and predict, which is that some of these components require different kinds of hardware to really do optimally. You have to anticipate where the AI innovation is going to lead. For example, a mixture of—

Perfect.

—experts with sparsity.

Perfect.

With hardware, you can’t just pivot on a week’s notice. You have to anticipate what that’s going to look like. That’s so scary and difficult to do, right?

For example, these AI model architectures are being invented about once every 6 months, right? System architectures and hardware architectures come along about every 3 years. You need to anticipate what is likely going to happen 2 or 3 years from now.

There are a couple of ways that you could do that. First of all, we could do research internally ourselves, and that’s 1 of the reasons why we have basic research and applied research. We create our own models, and so we have hands-on life experience right here. This is part of the co-design that I’m talking about.

We’re also the only AI company in the world that works with literally every AI company in the world. To the extent that we can, we try to get a sense of the challenges that people are experiencing.

So you’re listening to the whispers across the industry, the AI labs.

That’s right. You have to listen and learn from everybody.

The last part is to have an architecture that’s flexible, that can adapt and move with the wind. One of the benefits of CUDA is that it’s, on the 1 hand, an incredible accelerator. On the other hand, it’s really flexible. That balance between specialization—otherwise we can’t accelerate the GPU—and generalization, so that we can adapt to changing algorithms, is really, really important.

That’s the reason why CUDA has been so resilient, on the 1 hand, and yet we continue to enhance it. We’re at CUDA 13.2, and so we’re evolving the architecture so fast that we can stay with the modern algorithms.

For example, when mixtures of experts came out, that’s the reason why we had NVLink 72 instead of NVLink 8. We could now take an entire 4-trillion- or 10-trillion-parameter model and put it in 1 computing domain, as if it’s running on 1 GPU.

People probably didn’t notice—I said it—but if you look at the architecture of the Grace Blackwell racks, it was completely focused on doing 1 thing: processing the LLM. All of a sudden, 1 year later, you’re looking at a Vera Rubin rack. It has storage accelerators. It has this incredible new CPU called Vera. It has Vera Rubin and NVLink 72 to run the LLMs. It also has this new additional rack called Rock.

This entire rack system is completely different from the previous 1, and it’s got all these new components in it. The reason for that is because the last 1 was designed to run MoE large language models—inference. This 1 is to run agents, and agents bang on tools—

Obviously, the design of the system had to have been done before Claude Code, Codex, and OpenClaw. You were anticipating the future, essentially. And that comes from what? From the whispers, from understanding what all the state of the art is about?

No, it’s easier than that. You just reason about it. First of all, you just reason. No matter what happens, at some point, in order for that large language model to be a digital worker—let’s just use that metaphor—let’s say that we want the LLM to be a digital worker. What does it have to do?

It has to access ground truth. That’s our file system. It has to be able to do research. It doesn’t know everything. I don’t want to wait until this AI becomes universally smart about everything, past, present, and future, before I make it useful. Therefore, I might as well let it go do research. It’s obvious that if it wants to help me, it’s got to use my tools.

A lot of people would say, “AI is going to completely destroy software.”

We don't need software anymore. We don't even need tools anymore. That's ridiculous. Let's use a thought experiment. You could just sit there, enjoy a glass of whiskey, and think about all these things, and it would become completely obvious.

If I were to create the most amazing agent that we can imagine in the next 10 years, let's say it would be a humanoid robot. If that humanoid robot were created, is it more likely that it comes into my house and uses the tools that I have to do the work that it needs to do? Or does this hand turn into a 10-pound hammer in one instance, turn into a scalpel in another instance, and, in order to boil water, beam microwaves out of its fingers? Is it more likely to just use a microwave?

The first time it goes up to the microwave, it probably doesn't know how to use it. But that's okay. It's connected to the internet. It reads the manual of the microwave and instantly becomes an expert, and so it uses it.

I think I just described almost all of the properties of OpenClaw: that it's going to use tools, that it's going to access files, and that it's going to be able to do research. It has an I/O subsystem. When you're done reasoning about it, you say, "Oh, my gosh, the impact on the future of computing is deeply profound."

The reason for that is I think we've just reinvented the computer. And then you say, "Okay, when did we reason about that? When did we reason about OpenClaw?" If you take the OpenClaw schematic that I used at GTC, you'll find it from 2 years ago. Literally, 2 years ago at GTC, I was talking about agentic systems that exactly reflect OpenClaw today.

Of course, the confluence of many things had to happen. First of all, we needed Claude and GPT and all of these models to reach a level of capability. Their innovation, breakthroughs, and continued advances were really important. Then, of course, somebody had to create an open-source project that was sufficiently robust and sufficiently complete that we could all put it to work.

I think OpenClaw did for agentic systems what ChatGPT did for generative systems. I just think it's a very big deal.

Yeah, it's a really special moment. I'm not exactly sure why it captured so much of the world's attention, but it did—more than Claude Code and Codex and so on.

Because consumers could reach it.

Sure, yeah. But there's also so much of this that is vibes. Peter—I had a podcast with him—he's a wonderful human being. Part of it is also the humans that represent the thing.

Yeah, no doubt.

Part of it is memes, because we're all trying to figure it out. There are really serious and complicated security concerns. When you have such powerful technology, how do you hand over your data so it can do useful things? But then there are scary things associated with that. We, as individual people and as a civilization, are figuring out how to find that right balance.

Yeah, we jumped on it right away and sent a bunch of security experts this way. We did this thing called OpenShell. It's already been integrated into OpenClaw.

And NVIDIA put forward NemoClaw.

Yep, exactly.

They install super easily, and they make sure that it's secure.

We give you 2 out of 3 rights. Agentic systems can access sensitive information, execute code, and communicate externally. We could keep things safe if we gave you 2 out of those 3 capabilities at any time, but not all 3.

Out of those 2 out of 3 capabilities, we also give you access control based on whatever rights you're given by the enterprise. Then we connect it to a policy engine that all these enterprises already have. We're going to try to do our best to help OpenClaw become a better claw.

You eloquently explained how we have a long history of blockers that we thought were going to be blockers, and we overcame them. But now, looking into the future, what do you think might be the blockers, now that it's clear that agents will be everywhere? Obviously, we're going to need compute. What is going to be the blocker for that scaling?

Power is a concern, but it's not the only concern. That's the reason why we're pushing so hard on extreme co-design, so that we can improve tokens per second per watt by orders of magnitude every single year.

In the last 10 years, Moore's Law would have progressed computing about 100 times. We progressed and scaled up computing by a million times in the last 10 years. We're going to keep on doing that through extreme co-design.

Energy efficiency and performance per watt completely affect the revenues of a company. They affect the revenues of a factory. We're just going to push that to the limit so that we can keep driving token costs down as fast as we can.

The price of our computers is going up, but our token-generation efficiency is going up so much faster that token cost is coming down. It's coming down by an order of magnitude every year.

Power—that's an interesting one. The way to try to get around the power blocker is to make it more and more efficient with tokens per second per watt. Of course, there's the question of how we get more power.

We should also get more power.

That's a really complicated one. You've talked about small modular nuclear power plants. There are all kinds of ideas for energy. How much does it keep you up at night—the bottlenecks in the supply chain of AI, like ASML with EUV lithography machines, TSMC with advanced packaging like CoWoS, and SK hynix with high-bandwidth memory?

All the time, and we're working on it all the time. No company in history has ever grown at the scale that we're growing while accelerating that growth. It's incredible, and it's hard for people to even understand this.

In the overall world of AI computing, we're increasing our share. Supply chain, upstream and downstream, is really important to us. I spend a lot of time informing all the CEOs that I work with about the dynamics that are going to cause the growth to continue or even accelerate.

That's part of the reason why, to my entire right-hand side, were CEOs of practically the entire upstream IT industry and practically the entire downstream infrastructure industry. There were several hundred CEOs. I don't think there's ever been a keynote where several hundred CEOs showed up.

Part of it is that I'm telling them about our business condition now. I'm telling them about the growth drivers in the very near future and what's happening. I'm also describing where we're going to go next, so that they can use all of this information and all of the dynamics that are here to inform how they want to invest.

I inform them that way, like I inform my own employees. Then, of course, I make trips out to them and make sure that they know, "Hey, listen, I want you to know that this quarter, this coming year, and next year, these things are going to happen."

If you look at the CEOs of the DRAM industry, the number-one DRAM in the world was DDR memory for CPUs in data centers. About 3 years ago, I was able to convince several of the CEOs that even though, at the time, HBM memory was used quite sparsely and barely by supercomputers, it was going to be a mainstream memory for data centers in the future.

At first, it sounded ridiculous, but several of the CEOs believed me and decided to invest in building HBM memories. Another memory that was rather odd to put into a data center was the low-power memory that we use for cell phones. We wanted them to adapt it for supercomputers in the data center, and they said, "Cell phone memory for supercomputers?"

I explained to them why. Look at these 2 memories: LPDDR5 and HBM4. The volumes are so incredible. All 3 of them had record years in history, and these are 45-year companies. That's part of my job: to inform, shape, and inspire.

So you're not just manifesting the future and inspiring NVIDIA's different engineers; you're manifesting the supply chain of the future. You're having conversations with TSMC and ASML.

Upstream, downstream.

Upstream, downstream. So that's the thing.

GE Vernova, Caterpillar. That's downstream from us. Yeah, there you go.

Yeah, the whole thing. I mean, there's so much incredibly difficult engineering that happens in the entire semiconductor industry. It feels scary how intricate the supply chain is and how many components there are, but it works somehow.

Exactly—the deep science, the deep engineering, the incredible manufacturing. So much of the manufacturing is already robotics, but we have a couple hundred suppliers that contribute the technology that goes into our 1.3-million-component rack. Each rack has 1.3 to 1.5 million components. There are 200 suppliers across the Vera Rubin rack.

So it's interesting that you don't list that as the thing that keeps you up at night among the blockers.

But I'm doing all the things necessary to—

Okay.

Yeah, see? I can go to sleep because I checked it off. I say, "Okay, I can go to sleep." Then I say, "Well, let's reason about this. What's important for us?"

Because we changed the system architecture from the original DGX-I that you remembered to NVL72 rack-scale computing, what's that going to mean? What does that mean to software? What does that mean to engineering? What does that mean to how we design and test? And what does that mean to the supply chain?

One of the things that it meant was we moved supercomputer integration at the data center into supercomputer manufacturing in the supply chain.

If you're doing that, you also have to recognize this: if the total footprint of whatever data center you're going to build—let's say you would like to have 50 gigawatts of supercomputers that are running simultaneously—and it takes one week to manufacture that 50 gigawatts of supercomputers, then each week in the supply chain, the supercomputers are going to need 1 gigawatt of power. So we're going to need the supply chain to increase the amount of power it has to build and test the supercomputers in the supply chain before I ship them.

Well, NVL72 literally builds supercomputers in the supply chain and ships them 2–3 tons at a time per rack. They used to come in parts, and we used to assemble them inside the data center. But that's impossible now because NVL72 is so dense.

That's an example. I would have to fly into the supply chain and meet my partners, saying, “Hey, guess what? This is the way we used to build our DGXs. We're going to build them this way. This is going to be so much better because we're going to need them for inference.”

The market for inference is coming. The inflection point for inference is coming. It's going to be a big market. So I first explain to them what's going on and why it's going to happen, and then I ask them to make several billion dollars of capital investments each. Because they trust me, and I'm very respectful of them, I give them every opportunity to question me. I spend time explaining things to people, and I reason about it. I draw pictures and reason about it from first principles. By the time I'm done with them, they know what to do.

A lot of it is about relationships and building a shared view of the future. But do you worry about certain bottlenecks? What are the biggest bottlenecks in the supply chain? Are you worried about ASML's EUV tooling? Are you worried about the CoWoS packaging at TSMC and how fast it could scale? Like you said, you're not only growing incredibly fast, you're accelerating your growth. So it feels like everybody in the supply chain—and those are certainly bottlenecks—would have to scale up. Are you having conversations with them about how they can scale up faster? Do you worry about it?

No.

Okay.

Because I told them what I needed. They understood what I needed. They told me what they were going to do, and I believe them.

Interesting. That's great to hear. Maybe we can just linger on the power for a little bit. What are your hopes for how to solve the energy problem?

One of the areas, Lex, that I would love us to talk about and just get the message out is that our power grid is designed for the worst-case condition with some margin. Well, 99% of the time, we're nowhere near the worst-case condition, because the worst-case condition is a few days in the winter, a few days in the summer, and extreme weather. Most of the time, we're nowhere near the worst-case condition, and we're probably running at around 60% of peak.

So 99% of the time, our power grid has excess power, and it's just sitting idle. But it has to be sitting idle because, just in case, when the time comes, hospitals have to be powered, infrastructure has to be powered, airports have to run, and so on and so forth.

The question that I have is whether we could help them understand this, create contractual agreements, and design computer architecture systems and data centers such that when they need the maximum power for infrastructure in society, the data centers would get less. But that's a very rare instance anyway. During that time, we either have a backup generator for that little part of it, we have our computers shift the workload somewhere else, or we have the computers just run slower.

We could degrade our performance, reduce our power consumption, and provide a response with slightly longer latency when somebody asks for an answer. I think building data centers with these really quite rigorous contracts that expect 100% uptime is putting a lot of pressure on the grid. Now they're going to have to increase from their maximum. I just want to use their excess. It's just sitting there. We should go after it.

Yeah, that's not talked about enough. What's stopping that? Is it regulation? Is it bureaucracy?

I think it's a 3-way problem. It starts with the end customer. The end customer puts requirements on the data centers that they can never not be available, so the end customer expects perfection.

In order to deliver that perfection, you need a combination of backup generators and your grid power supplier to deliver on perfection. Everybody's got to have six nines. I think, first of all, we ought to have everybody understand that when the customer asks for these things, you have somebody in your data center operations team disconnected from the CEO. I bet the CEO doesn't know this.

I'm going to talk to all the CEOs. The CEOs are probably not paying any attention to the contracts that are being signed, and so everybody wants to sign the best contract, of course. They go down to cloud service providers, and the 2 contract negotiators are negotiating these multiyear contracts. Both sides want the best contract.

As a result, the CSPs then have to go down to the utilities, and they expect the six nines. I think the first thing is just to make sure that all of the customers and CEOs realize what they're asking for.

The second thing is we have to build data centers that gracefully degrade. If the grid tells us, “Listen, we're going to have to back you down to about 80%,” we're going to say, “That's no problem at all.” We're just going to move our workload around. We're going to make sure that data is never lost, but we can reduce the computing rate and use less energy. The quality of service degrades a little bit.

For the critical workloads, I shift them somewhere else right away so I don't have that problem.

How difficult of an engineering problem is that, that smart, dynamic allocation of power in a data center?

As soon as you can specify it, you can engineer it.

Beautifully put. So long as it obeys the laws of physics from first principles, I think we're good. What was the third thing you were mentioning?

The second thing is the data centers. The third thing is that we need the utilities to also recognize that this is an opportunity. Instead of saying, “Look, it's going to take me 5 years to increase my grid capability,” they could say, “If you're willing to take power at this level of guarantee, I can make it available for you next month and at this price.”

If utilities also offered more segments of power delivery promises, then I think everybody would figure out what to do with it. But there's just way too much waste in the grid right now. We should go after it.

You've highly lauded Elon and xAI's accomplishment in Memphis in building the Colossus supercomputer, probably in record time, in just 4 months. It's now at 200,000 GPUs and growing very quickly. Is there something that you could speak to, to help us understand about his approach that's instructive broadly to all data center creators, that enabled that kind of accomplishment? His approach to engineering, his approach to the whole management of construction—everything?

First of all, Elon is deep in so many different topics. Yet he's also a really good systems thinker. He's able to think through multiple disciplines, and he obviously pushes things and questions everything: 1, is it necessary? 2, does it have to be done this way? And then, 3, does it have to take this long?

He has the ability to question everything to the point where everything is down to its minimal amount that's necessary—you can't take anything else out. And yet the necessary capabilities of the product remain. He's as minimalist as you could possibly imagine, and he does it at a system scale.

I also love the fact that he is present at the point of action. He'll just go there. If there's a problem, he'll just go there and say, “Show me the problem.”

When you do all of this in combination, you overcome a lot of previous excuses: “This is just the way we do it.” “I'm waiting for them.” Everybody has a lot of excuses.

The last thing is, when you act personally with so much urgency, it causes everybody else to act with urgency. Every supplier has a lot of customers and projects going on, and he makes it his business that he's the top priority of everybody else's projects. He does that by demonstrating it.

Yeah, I've been in a bunch of those meetings. It's fun to watch, because really, not enough people ask the question, “Can this be done a lot faster, and how? Why does it have to take this long?”

That becomes an engineering question often. I remember one of the times I was hanging out with him, he literally was going through the entire process of how to plug cables into a rack.

He's working with an engineer on the ground who's doing that task, and he's just trying to understand what that process looks like so it can be less error-prone. Just building up that intuition from every single task involved in putting together a data center, you start to immediately get a sense—at the detailed scale and at the broad systems scale—of where the inefficiencies are, and so you can make it more efficient. Plus, you have the big hammer of being able to say, “Let's do it totally differently—”

Yeah, that's right.

“...and remove all possible blockers.”

Is there a parallel in the NVIDIA extreme systems co-design approach that you see in the way Elon approaches systems engineering?

Well, first of all, co-design is the ultimate systems engineering problem. And so we approach the work that we do from that first-principles perspective. The other thing that we do—and this is a philosophy, a thought, a state of mind, a method that I started 30 years ago—is called “the speed of light.”

The speed of light is not just about speed. The speed of light is my shorthand for the limit of what physics can do. Everything that we do is compared against the speed of light: memory speed, math speed, power, cost, time, effort, number of people, and manufacturing cycle time.

When you think about latency versus throughput, cost versus throughput, and cost versus capacity, you test all of these things against the speed of light to achieve the different constraints separately. Then, when you consider them together, you know you have to make compromises, because a system that achieves extremely low latency and a cheap system that achieves very high throughput are architected fundamentally differently. But you want to know: What's the speed of light for a system that achieves high throughput? What's the speed of light for a system that achieves low latency?

And then, when you think about the total system, you can make trade-offs. I force everybody to think about what the first principles and the physical limits are for everything before we do anything. We test everything against that, and so that's a good frame of mind.

I don't love the other method, which is continuous improvement. First of all, you should engineer something from first principles, with speed-of-light thinking, limited only by physical limits—the limits of physics. After that, of course, you would improve it over time.

But I don't like going into a problem and somebody saying, “It takes 74 days to do this today. Right now, we can do it for you in 72 days.” I'd rather strip it all back to zero and say, “First of all, explain to me why it takes 74 days in the first place. Let's think about what's possible today. If I were to build it completely from scratch, how long would it take?”

Oftentimes, you'd be surprised. It might come to 6 days. Now, the rest of the 74 days could be very well-reasoned compromises, cost reductions, and all kinds of different things. But at least you know what they are. And now that you know 6 days is possible, the conversation from 74 to 6 is surprisingly much more effective.

In such incredibly complex systems that you're working with, is simplicity sometimes a good heuristic to reach for? The Vera Rubin pod that you announced is just incredible. We're talking about 7 chips, 7 chip types, 5 purpose-built rack types, 40 racks, 1.2 quadrillion transistors, nearly 20,000 NVIDIA dies, over 1,100 Rubin GPUs, 60 exaflops, and 10 petabytes per second of scale bandwidth. That's all just one pod.

That's just one pod.

Yeah, that's just one pod.

And then even the NVL72 rack alone is 1.3 million components, 1,300 chips, and 4,000 pods crammed into a single 19-inch wide rack.

We're probably going to have to crank out about 200 of these pods a week, just to put it in perspective.

The amount of different components, I suppose simplicity is impossible, but is that a metric that you reach for in trying to design things?

The phrase that I use most often is, “We need things to be as complex as necessary, but as simple as possible.” The question is: Is all that complexity there necessary? We ought to test for that. We have to challenge that. After that, everything else above it is gratuitous.

But it's still almost incredible. The semiconductor industry broadly, but what NVIDIA is doing, is some of the greatest engineering in history. These systems are truly, truly marvels of engineering.

It is the most complex computer the world has ever made.

Yeah, the engineering teams—I don't know, it's not a competition, but if it were like an Olympics of engineering teams, TSMC does incredible engineering. Like I said, ASML, at every scale, is incredible. But NVIDIA is going to give them a run for their money. Just incredible teams.

Well, it's gold medalists in every single sport, all assembled right here.

And they have to work together and report directly to you. This is wonderful.

You recently traveled to China, so it's interesting to ask you: China's been incredibly successful in building up its technology sector. What do you understand about how China has been able to, over the past 10 years, build so many incredible world-class companies, world-class engineering teams, and this technology ecosystem that produces so many incredible products?

A whole bunch of reasons. First of all, let's start with some facts. 50% of the world's AI researchers are Chinese, plus or minus, and they're mostly in China still. We have many of them here, but there are amazing researchers still in China.

Their tech industry showed up at precisely the right time. At the time of the mobile-cloud era, their way of contributing was software. This is a country of incredible science and math, with really well-educated kids. Their tech industry was created during the era of software, and they're very comfortable with modern software.

China is not one giant economic country. It's got many provinces and cities with mayors all competing with each other. That's the reason why there are so many EV companies. That's the reason why there are so many AI companies. That's the reason why there are so many of every company you could imagine. They all create some of them, and as a result, they have insane competition internally. What remains is an incredible company.

They also have a social culture where it's family first, friends second, and company third. The amount of conversation that goes back and forth between them is enormous. They're essentially open source all the time.

The fact that they contribute more to open source is so sensible because they're probably asking, “What are we protecting?” My engineers' brothers are in that company, their friends are in that company, and they're all schoolmates. The schoolmate concept is that if you're schoolmates, you're brothers for life.

They share knowledge very quickly, and so there's no sense in keeping technology hidden. You might as well put it on open source. The open-source community then amplifies and accelerates the innovation process.

You get rapid, incredible talent, rapid innovation because of open source, the nature of friendship, and insane competition among the companies. What emerges is incredible stuff. This is the fastest-innovating country in the world today.

Everything that I've just said is fundamental to how the kids were raised: the fact that they have excellent education, the fact that their parents want them to do well in school, and the fact that their culture is that way. These are just the things about their country, and they showed up at precisely the time when technology is going through that exponential growth.

Plus, culturally, it's pretty cool to be an engineer. It connects to all the components that you're mentioning.

It's a builder nation.

Yeah, it's a builder nation. Our country's leaders are incredible, but they're mostly lawyers. Their country's leaders—and because we're trying to keep us safe through rule of law and governing—their country was built out of poverty. Most of their leaders are incredible engineers, some of the brightest minds.

To take a small tangent, because you mentioned open source, I have to go to Perplexity here, which you've been a fan of for a long time. Thank you for releasing open-source Nemotron 3 Super, which you can also use inside Perplexity to look things up. It's a 120-billion-parameter open-weight MoE model. What's your vision with open source?

You mentioned China, with DeepSeek and MiniMax, and all these companies really pushing forward the open-source AI movement. NVIDIA is really leading the way in close-to-state-of-the-art open-source LLMs. What's your vision there?

First, if we're going to be a great AI computing company, we have to understand how AI models are evolving. One of the things that I love about Nemotron 3 is that it's not just a pure transformer model; it's a transformer and SSMs.

We were early in developing conditional GANs, then progressive GANs, which led step-by-step to diffusion. The fact that we're doing basic research in model architecture and in different domains gives us visibility into what kind of computing systems would do a good job for future models. It is part of our extreme co-design strategy.

Second, I think we rightfully recognize that, on the one hand, we want world-class models as products, and they should be proprietary.

On the other hand, we also want AI to diffuse into every industry and every country, every researcher, and every student. If everything is proprietary, it's hard to do research and it's hard to innovate on top of, around, and with it. Open source is fundamentally necessary for many industries to join the AI revolution. NVIDIA has the skills, scale, and motivation to build and continue to build these AI models for as long as we shall live. And so, therefore, we ought to do that.

We can open up and activate every industry, every researcher, and every country to be able to join the AI revolution. The third reason is recognizing that AI is not just language. These AIs will likely use tools, models, and sub-agents that were trained on other modalities of information. Maybe it's biology or chemistry or the laws of physics, or fluids and thermodynamics, and not all of it is in language structure. Somebody has to make sure that weather prediction, biology AI, AI for biology, physical AI—all of that stuff—stays at the frontier and can be pushed to the limits.

We don't build cars, but we want to make sure every car company has access to great models. We don't discover drugs, but I want to make sure that Lilly has the world's best biology AI systems so that they can use them for discovering drugs. And so these three fundamental reasons are recognizing that AI is not just language, that AI is really broad; that we want to engage everybody in the world of AI; and also the co-design of AI.

Well, I have to say, once again, thank you for open-sourcing—really, truly open-sourcing—Nemotron 3 and ...

Yeah, I appreciate you saying that. We open-sourced the models, we open-sourced the weights, we open-sourced the data, and we open-sourced how we created it. It's pretty amazing.

It's really incredible. You're originally from Taiwan and have a close relationship with TSMC. So I have to ask: TSMC, I think, is also a legendary company in terms of its engineering teams and the incredible engineering work that they do. What do you understand about TSMC's culture and approach that explains how they're able to achieve this singular, unmatched success in everything they're doing with semiconductors?

First of all, the deepest misunderstanding about TSMC is that their technology is all they have. That somehow they have a really great transistor, and if somebody shows up with another transistor, game over. It's the technology—and, of course, I don't mean just the transistor: the metallization systems, the packaging, the 3D packaging, the silicon photonics, and all of the technology that they have. That technology is really what makes the company special. Their technology makes the company special.

But their ability to orchestrate the demands—the dynamic demands—of hundreds of companies in the world as they're moving up, shifting out, increasing, decreasing, pushing out, pulling in, changing from customer to customer, wafer starting, wafer stopping, emergency wafer starts—all of these dynamics of the world's complexity, as the world is shape-shifting all the time—and somehow they're running a factory with high throughput, high yields, really great costs, and excellent customer service.

They take their work and their promises seriously. When your wafers were promised to show up, the wafers show up, because they know that they're helping you run your company, so that you can run your company appropriately. Their manufacturing system is completely miraculous, I would say.

Then the second thing is their culture. This culture is simultaneously technology-focused on one hand, advancing technology, and customer-service-oriented on the other hand. A lot of companies are very customer-service-oriented, but they're not very technologically excellent. They're not at the bleeding edge of technology. There are a lot of companies that are at the bleeding edge of technology, but they're not the best customer-service-oriented company. Somehow, they've balanced these two, and they're world-class at both.

Probably the third thing is the technology that I most value in them: They created this intangible called trust. I trust them to put my company on top of them. That's a very big deal.

With trust, I mean, there's a really close relationship there that you've established. That trust is established based on many years of performance, but there are human relationships involved there as well.

Three decades. I don't know how many tens or hundreds of billions of dollars of business we've done through them, and we don't have a contract. That's pretty great.

Amazing. Okay, there's this story that in 2013, the founders of TSMC, Morris Chang, offered you the chance to become TSMC's chief executive, and you said you already had a job. Is this story true?

The story is true. I didn't dismiss it. I was deeply honored, and of course, I knew then, as I know now, that TSMC is one of the most consequential companies in history. Morris is one of the highest-regarded executives and a business and personal friend that I've had in my life. For him to ask was—I was humbled and really honored.

But the work that I'm doing here is really important, and I've seen, in my mind's eye, what NVIDIA was going to be and what impact we could have. It was really important work. It's my responsibility, my sole responsibility, to make this happen. And so I declined it—not because it wasn't an incredible offer. It's an unbelievable offer, but I simply couldn't take it.

I think NVIDIA and TSMC are two of the greatest companies in the history of human civilization. Running either one, I'm sure, is an incredibly complicated effort and takes—you have to truly be all in. Everybody, at every scale, not just at the CEO level, is really truly all in—

Yeah. Yeah, no doubt.

—to accomplish this kind of complexity.

So now I can help both companies.

Exactly. So NVIDIA is now the most valuable company in the world. I have to ask: What is NVIDIA's biggest moat, as the folks in the tech sector say—the edge you have that protects you from the competition?

Our single most important property as a company is the install base of our computing platform. Our single most important thing today is the install base of CUDA.

The reason is that 20 years ago, of course, there was no install base. If somebody came up with a GUDA or TUDA it wouldn't make any difference at all. It's never been just about the technology. The technology, of course, was incredible and visionary. But it's the fact that the company was dedicated to it, stuck with it, and expanded its reach.

It wasn't 3 people that made CUDA successful. It was 43,000 people that made CUDA successful, and the several million developers that believed in us, that trusted we were going to continue to make CUDA 1, 2, 3, 13. They decided to port and dedicate their software—their mountain of software—on top of it. The install base is the number one most important advantage.

That install base, when you amplify it with the velocity of our execution at the scale that we're talking about—no company in history had ever built systems of this complexity, period. And then to build it once a year is impossible.

From the developer's perspective, if I support CUDA, tomorrow it'll be 10 times better. I just have to wait 6 months on average. Not only that, if I develop it on CUDA, I reach a few hundred million computers. I'm in every cloud, I'm in every computer company, I'm in every single industry, and I'm in every single country. So if I create an open-source package and put it on CUDA first, I get both of these attributes simultaneously.

And not only that, I trust 100% that NVIDIA is going to keep CUDA around, maintain it, improve it, and keep optimizing the libraries for as long as they shall live. You could take that to the bank. That last part—trust. You put all that stuff together, and if I were a developer today, I would target CUDA first. I would target CUDA most.

That's the reason that, in the final analysis, CUDA is our first core advantage. Our second one is our ecosystem: the fact that we vertically integrated this incredibly complex system and integrated it horizontally into every single company's computers.

We're into Google Cloud, we're into Amazon, we're in Azure. We're ramping up AWS like crazy right now. We're in new companies like CoreWeave and Nscale. We're in supercomputers at Lilly. We're in enterprise computers. We're at the edge in radio base stations. It's just crazy. One architecture is in all these different systems.

We're in cars, we're in robots, we're in satellites, and we're out in space. The fact that you have this one architecture and the ecosystem is so broad basically covers every single industry in the world.

Well, how does the CUDA install base evolve into the future with AI factories as a moat? Do you think it's possible that NVIDIA of the future is all about the AI factory?

The unit of computing used to be a GPU to us. Then it became a computer, then it became a cluster, and now it's an entire AI factory. When I see what NVIDIA builds, in the old days, I would visualize the chip. When I announced a new product, a new generation, like, “Ladies and gentlemen, we're announcing Ampere today,” I'd pick up the chip.

That was my mental model of what I was building. Picking up the chip is still kind of adorable, but it's not my mental model of what I'm doing.

My mental model is this giant gigawatt thing that has power generation connected to the grid. It's got cooling systems and networking—an incredible monstrosity. 10,000 people are in there trying to install it, hundreds of networking engineers in there, and thousands of engineers behind it trying to power it up. Powering up one of those factories, as you know, is not somebody going, “It's on now.” It takes thousands of people to bring it up.

So mentally, when you're thinking about a single unit of compute, you're literally thinking, when you go to bed at night, about a collection of racks, pods, not individual chips.

Entire infrastructure. I'm hoping my next click is when I'm thinking about building computers, it's planetary scale. That'll be the next click.

What do you think about the space angle that Elon has talked about, doing compute in space for solving some of the energy issues? It makes some of the energy issues in terms of scaling energy easier.

Cooling issues are not easy.

Cooling. Well, there's a large number of engineering complexities involved with that. NVIDIA has also announced that you're already thinking about that.

Yeah, we're already there. NVIDIA GPUs are the first GPUs in space. I didn't realize it; it was so interesting. I would have declared it, maybe: “We're in space.” A little astronaut suit on one of our GPUs. But we've been in space.

It's the right place to do a lot of imaging, because those satellites have really high-resolution imaging systems, and they're sweeping the Earth continuously now. You want centimeter-scale imaging that's done continuously for the world, so that you'll basically have real-time telemetry of everything. You don't want to beam that back down to Earth. It's just petabytes and petabytes of data.

You've got to do AI right there at the edge, throw away everything you don't need, everything you've seen before or that didn't change, and then just keep the stuff that you need. AI had to be done at the edge. Obviously, we have 24/7 solar if we put it at the poles.

There's no conduction, no convection. You're pretty much just radiation. Space is big. I guess we're just going to put big, giant radiators out there.

How crazy of an idea do you think it is? Is this 5 years out, 10 years out, 20 years out? We're talking about blockers for AI scaling.

I'm just so much more practical. I look for where my next bucket of opportunities are first. Meanwhile, I'm cultivating space. I send engineers to go work on the problem. We're starting to learn a lot about it.

How do we deal with radiation? How do we deal with degrading performance? How do we deal with continuous testing and attestation of defects? How do we deal with redundancy? How do we degrade gracefully and things like that?

We could start doing a lot of engineering exploration upfront. But what about software? How do you think about software, redundancy, and performance out in space? Make it so that the computer never breaks; it just gets slower.

So we could start doing a lot of engineering exploration upfront. But in the meantime, my favorite answer is: eliminate waste. We've got all that idle power. I want to evacuate it as fast as possible.

Yeah. There's a lot of low-hanging fruit here on Earth that we can utilize for the AI scaling.

Quick pause. Quick 30-second thank you to our sponsors. Check them out in the description. It really is the best way to support this podcast. Go to lexfridman.com/sponsors. We got Perplexity for curiosity-driven knowledge exploration, Shopify for selling stuff online, LMNT for electrolytes, Fin for customer service AI agents, and Quo for a phone system, like calls, texts, contacts, for your business. Choose wisely, my friends. And now, back to my conversation with Jensen Huang. Do you think NVIDIA may be worth $10 trillion at some point? Let's ask it this way: What does the future of the world look like where that's true?

I think that NVIDIA's growth is extremely likely and, in my mind, inevitable. Let me explain why.

We're the largest computer company in history. That alone should beg the question: Why? The reason, of course, is two foundational technical reasons. The first reason is that computing went from being a retrieval-based, file-retrieval system.

Almost everything is a file. We pre-write something, we pre-record something, we draw something, we put it on the web, and we put it in a file. We use a recommender system, some smart filter, to figure out what to retrieve for you. We were a human pre-recording and file-retrieving system. That's what a computer is, largely.

Now AI computers are contextually aware, which means that they have to process and generate tokens in real time. We went from a retrieval-based computing system to a generative-based computing system. We're going to need a lot more processing in this new world than in the old world. We need a lot of storage in the old world; we need a lot of computation in this new world.

We fundamentally changed computing and the way computing is done. The only thing that would cause it to go back is if this way of computation—this way of computing, generating information that's contextually relevant and situationally aware, grounded on new insight before it generates information—this computation-intensive way of doing computing would only go back if it's not effective.

For the last 10 or 15 years, while working on deep learning, if at any single moment I had come to the conclusion that, “You know what? This is not going to work out. I think this is a dead end,” or, “It's not going to scale, it's not going to solve this modality, it's not going to be used in this application,” then, of course, I would feel very differently about it. But I think the last 5 years have given me more confidence than the previous 10 years.

The second idea is that computers, because they were a storage system, were largely a warehouse. We're now building factories. Warehouses don't make much money. Factories directly correlate with a company's revenues.

The computer did 2 things. Not only did it change the way it did things, but its purpose in the world changed. It's no longer a computer; it's a factory used for the generation of revenue.

We're now seeing not only that this factory is generating products—commodities that people want to consume—we're seeing that these commodities are so interesting, so valuable to so many different audiences, that the tokens are starting to segment, like iPhones.

Mm-hmm.

You have free tokens, you have premium tokens, and you have several tokens in the middle.

Yeah.

Intelligence, as it turns out, is a scalable product. There are extremely high-intelligence products, tokens that you could use for specialized things that people would be willing to pay for. The idea that somebody's willing to pay $1,000 per million tokens is just around the corner. It's not if; it's only when.

Now we're seeing that the commodity that this factory makes is actually valuable, revenue-generating, and profit-generating. The question is: How many of these factories does the world need? How many tokens does the world need? How much is society willing to pay for these tokens?

What would happen to the world's economy if productivity were to improve so substantially? What would happen? Are we going to discover new drugs, new products, new services?

When you take these things in combination, I am absolutely certain that the world's GDP is going to accelerate in growth. I'm absolutely certain the percentage of that GDP that will be used for computation will be 100 times more than in the past, because it's no longer a storage unit. It's a product-generation unit.

When you look at it in that context and then you back into what NVIDIA does, and how much of that new economics, that new industry, we would benefit from and address, I think we're going to be a lot, lot bigger.

The rest of it, to me, is: Is it possible for NVIDIA to be a $3 trillion-revenue company in the near future? The answer is, of course, yes. The reason for that is because it's not limited by any physical limits. There's nothing that I see that says, “Gosh, $3 trillion is not possible.”

As it turns out, NVIDIA's supply chain—the burden is shared by 200 companies. The fact that we scale out on the backs of this ecosystem, with the partnership of this ecosystem, the question is: Do we have the energy to do so? Surely we will have the energy to do so.

All of these things combined—that number is just a number. I still remember, the first time NVIDIA crossed $1 billion, I was reminded of a CEO who told me, “You know, Jensen, it's theoretically impossible for a fabless semiconductor company to exceed $1 billion.” I won't bore you with why, but in the end, of course, it's illogical, and there's a lot of evidence we're not.

Then somebody told me, “You know, Jensen, you'll never be more than $25 billion because of some other company.” Those aren't first-principles ways of thinking. The simple way to think about that is: What is it that we make, and how large is the opportunity that we can create?

Now, NVIDIA is not in the market-share business.

Almost everything that I just talked about doesn't exist. That's the part that's hard. If NVIDIA was a $10 billion company trying to take market share, then it's easy for shareholders to see that, if it could just take 10% share, it could become this much larger. But it's hard for people to imagine how large we could be because there's nobody I could take share from.

I think that's one of the challenges for the world: the imagination of the future. But I have plenty of time, and I'll keep reasoning about it and talking about it. Every single GTC will become more and more real, and then more and more people will talk about it. One of these days, we'll get there. I'm 100% sure we'll get there.

Yeah, this view of token factories—essentially, tokens per second per watt, with every token having value. It's an actual thing that brings value, and it brings different kinds of value and different amounts of value to different people. The actual product could really be loosely thought of as the token. You have a bunch of token factories, and it's very easy, from first principles, to imagine a future, given all the potential things that AI can solve, where you're going to need an exponentially larger number of token factories.

What's really interesting, and the reason why I was so excited about it, is that the iPhone of tokens arrived.

What do you call it? Wait, are you saying OpenClaw's iPhone? That's interesting.

Agents.

Yeah, agents. True.

Agents in general. The iPhone of tokens arrived. It is the fastest-growing application in history. It went straight up.

That says something.

Yep. There's no question OpenClaw is the iPhone of tokens.

Is there something truly special happening starting around December, where people have really woken up to the power of Claude Code, Codex, and OpenClaw? I'm embarrassed to admit that on the way here, at the airport, I've—it's the first time I've done this in public—been programming, quote unquote, by talking to my laptop. I was embarrassed because I was pretending like I was talking to a human colleague. I'm not sure how I feel about the future where everybody is walking around talking to their AI, but it's such an efficient way to get stuff done.

It's more likely that your AI is bothering you all the time. The reason for that is because it's getting stuff done so fast. It's reporting back to you, "I got that done. What do you want me to do next?" That's the part that I think most people don't realize: the person who's going to be chatting with them and texting them the most is their Claude or lobster.

What an incredible future. I read that you attribute a lot of your success to your ability to work harder than anyone and withstand more suffering than anyone. We can list many of the things that entails: dealing with failure, the cost and engineering problems we've talked about, human problems, uncertainty, responsibility, exhaustion, embarrassment, and the near-death company moments that you've mentioned. But also the pressure. Now, as the CEO of this company that economies and nations strategize around, plan their financial allocations around, and plan their AI infrastructure around, how do you deal with this much pressure? What gives you strength, given how many nations and peoples depend on you?

I'm conscious of the fact that NVIDIA's success is very important to the United States. We generate enormous amounts of tax revenue. We established technology leadership for our nation, and technology leadership is important for national security—not just one aspect of national security, but all aspects of national security. When our country is more prosperous, we can do a better job with domestic policies and social benefits.

Because we're generating so much reindustrialization in the United States, we're creating mountains of jobs. We're helping shift how we build things back to the United States in so many different plants: chips, computers, and, of course, these AI factories.

I'm completely aware that NVIDIA is central to a very large network of ecosystem partners behind us and downstream from us. I also have the benefit—and this is a real gift—of mainstream investors, teachers, and policemen who have somehow, for whatever reason, invested in NVIDIA, or who watched Jim Cramer, bought some stock, and now are millionaires. I am completely aware of that circumstance.

The way I deal with that is exactly what I just did. I reason about what we're doing, what it's causing, and what impact it has on other people—whether it benefits them positively or even places a great burden, for example, on the supply chain. The question is, therefore, what are you going to do about it?

In almost everything that I feel, I break it down and reason about it: "Okay, what's the circumstance? What has changed? What's hard? What am I going to do about it?" I decompose the problem, and the decomposition of these circumstances turns them into manageable things that I can do. The only thing after that I can do is ask, "Did you do it? Did you either do it or get somebody else to do it?"

If you reasoned that you needed to do something but didn't do it and didn't get anybody else to do it, then stop crying about it. I'm fairly tough on myself, but I also break things down so that I don't panic. I can go to sleep because I've made the list of things that needed to be done.

I've made sure that everything that could put our company in harm's way, put my partners in harm's way, or put our industry in harm's way, I've told somebody about. Everything that I feel could put anybody in harm's way, I've told someone—someone who could do something about it. I've gotten it off my chest, or I'm doing something about it. After that, Lex, what else can you do?

So, given the insane, intense amount of suffering on the journey of building up NVIDIA, have you hit low points psychologically?

Oh, yeah. Oh, yeah. Sure. All the time. All the time.

Do you just break down the problem into pieces and see what you could do about it?

Yeah. Part of it, Lex, is forgetting. One of the most important attributes of AI learning, as you know, is systematic forgetting, right? You need to know when to forget some things. You can't memorize everything. You can't keep everything, and you don't want to carry everything.

One of the things that I do very quickly is decompose the problem, reason about the problem, and share the load. When I say I tell everybody, I'm essentially sharing that burden. As quickly as possible, whatever worries me, I tell somebody else. Don't just keep it. Don't freak them out.

Decompose the problem into smaller parts, get people to do something, and inspire them to go do something about it. Part of it is just forgetting. A lot of it is that you have to be tough on yourself. Come on, stop crying about it. Let's get going. Then you get out of bed.

The other part is that you're attracted to the next shiny light—the next future, the next opportunity. "Okay, that's behind us. What's next?" I think you see this with great athletes. They just worry about the next point. The last point is behind them: the embarrassment, the setback.

Because I do so much of my job publicly—and, Lex, you do a fair amount of your job publicly too—I say a lot of things that seem sensible at the time or funny at the time. Mostly, it's just because it's funny to me at the time. Then you reflect on it, and it's less funny.

Yeah. Trust me, I know. But you basically allow yourself to be pulled by the light of the future, forget the past, and keep working toward that.

That's right.

You did say this famous thing: If you knew how hard it would be to build NVIDIA—what it turned out to be, a million times harder than you anticipated—you wouldn't do it.

Yeah, right.

But isn't that probably true about everything worth doing? When I hear that, that's probably true about everything worth doing, right?

Exactly. That is what I was trying to explain. There's an incredible superpower in having the mind of a child. I say to myself oftentimes when I look at something—almost everything—my first thought is, "How hard can it be?"

You get yourself into that mode: How hard could it be? Nobody's ever done it. It looks gigantic. It's going to cost hundreds of billions of dollars. It's going to take all this time. You just go, "Yeah, but how hard could it be?" How hard could it be? You have to get yourself into that state of mind.

You don't want to simulate everything in advance—all the setbacks, all the trials and tribulations, and all the disappointments. You don't want to know that. You want to go into a new experience thinking it's going to be perfect, it's going to be great, and it's going to be incredibly fun.

Then, while you're there, you need to have endurance and grit, so that when the setbacks actually happen—and those setbacks are going to surprise you—the disappointments are going to surprise you, the embarrassments are going to surprise you, and the humiliations are going to surprise you.

Now you just gotta turn on the other bit, which is just: forget about it. Move on. Keep moving. To the extent that my assumptions about the future and why the future is gonna manifest don't change materially, then I should expect that the output won't change. And so my simulated output of the future is still gonna happen. If it's still gonna happen, I'm still gonna go after it.

I believe it's gonna happen. There's a combination of two or three human characteristics: the ability to go into an experience fresh-minded, the ability to forget the setbacks, and the ability to believe in yourself—to believe what you believe and stay true to that belief. But you're constantly reevaluating. This combination of three, four, five things, I think, is really important for resilience.

I'm fortunate that whatever life experiences led to this, I've got those four or five things. I'm always curious, always learning. I'm always learning from everybody. Because I'm humble about everything, I'm always thinking, “Gosh, they did that so nicely. They did that so wonderfully. I wonder what they're thinking through. How do they...?”

I'm simulating everybody. In a lot of ways, I'm emulating almost everybody I watch. You're empathetic toward everything that they're doing and observing and respecting. And so you're constantly learning.

You're now one of the wealthiest people on Earth, one of the most successful humans on Earth. Do you feel the effect of money, power, and fame in making it harder for you to be wrong in your own head, enough to hear out an opinion of somebody else when they disagree with you and learn from them? Those kinds of things.

Surprisingly, no. I would actually go the other way. Because I do so much of my work publicly, when I'm wrong, pretty much everybody sees it.

You get humbled. Fair enough.

And when I'm wrong, or it didn't turn out that way—most of the things that I say publicly, I'm fairly certain about. The reason for that is because it's gonna impact somebody else, and I want to be quite concerned and circumspect about that.

For stuff that I'm reasoning about inside a meeting, a lot of things could turn out differently. But it doesn't ever stop me from reasoning. The way that I manage and lead, I'm constantly reasoning in front of people. Even when I'm talking to you, you can see me reasoning through things.

I want to make sure that you understand what I'm saying, not because I told you, but because I'm so humble about what I'm about to tell you. I show you the steps that I got there, and then you can decide whether you believe what I said in the end. I'm doing that all day long in meetings.

With all of my employees, I'm constantly reasoning through, “Let me tell you how I see it.” Then I reason through it. It gives everybody the opportunity to intercept and say, “I disagree with that part.” The nice thing about reasoning through things and letting people interact with it is that they don't have to disagree with your outcome. They can disagree with your reasoning steps.

They can pull me in different directions, and then we can reason forward. We're kind of a collective path-searching method. It's really fantastic.

You have this way about you where, when you're explaining stuff, I can feel you actually reasoning on the spot about it, with a constant open-mindedness. I feel like I could steer your thinking. That's really beautiful, that you've been able to maintain that after so many years of success and pain.

I think sometimes pain makes you close down a bit.

Mm-hmm. Yeah. Tolerance for embarrassment, I think, is—

Yes, that's a real thing. It's many years of embarrassing yourself. Even in those meetings, when there are people around you, you declared one idea and it was shown to be wrong, and being able to admit that and grow from it—that's very difficult on a human level.

Yeah. Well, my first job was cleaning toilets, so.

I'm glad you maintained that same spirit of Denny's work. I mean, that was beautiful. Your whole journey from starting at Denny's is a beautiful one.

Let me ask you about video games. I'm a big gaming fan, so I have to say thank you to NVIDIA for many years of incredible graphics.

By the way, GeForce is still, to this day, our number-one marketing strategy. People learn about NVIDIA while they're in their teenage years. Then they go to college, and they know who NVIDIA is. In the beginning, it's just playing Call of Duty or Fortnite. Later, they're using CUDA, and later they're using NVIDIA, Blender, Dassault, and Autodesk.

I should say, I mentioned to a friend that I'm talking with you. He said, “Oh, they make great gaming GPUs.”

Yeah, exactly.

You know, there's more to it, but people really love it. It brought a lot of joy to a lot of people. The hardware really brings these worlds to life.

There was some controversy around this with DLSS 5. Can you explain to me the drama around this? I guess people, the gamers online, were concerned that it makes games look like AI slop. What do you think of this drama?

Yeah. I think their perspective makes sense, and I could see where they're coming from, because I don't love AI slop myself. All of the AI-generated content increasingly looks similar, and it's all beautiful, so I can be empathetic toward what they're thinking. That's just not what DLSS 5 is trying to do. I showed several examples of it.

DLSS 5 is 3D-conditioned, 3D-guided, and guided by ground-truth structural data. The artist determined the geometry. We are completely truthful to the geometry and maintain it in every single frame. It's conditioned by the textures and the artistry of the artist. Every single frame, it enhances, but it doesn't change anything.

Now, the question about enhancing is that DLSS 5 also lets you, because the system is open, train your own models to determine the style. In the future, you could even prompt it: “I want it to be a toon shader. I want it to look like this kind of thing.” You can give it an example, and it would generate in that style, all consistent with the artistry, the style, and the intent of the artist.

All of that is done for the artist, so that they can create something that is more beautiful but still in the style that they want. I think people got the impression that the games are gonna come out the way they're shipped, and then we're gonna post-process them. That's not what DLSS is intended to do. DLSS is integrated with the artist, and it's about giving the artist the tool of AI—the tool of generative AI. They could decide not to use it.

I think people are very sensitive to human faces. We're now living in this moment, which I think is a beautiful one, where people are sensitive to AI slop. It puts a mirror up to ourselves and helps us realize that what we seek is imperfections. What we seek is sometimes not perfect graphics. It helps us understand what we find compelling in the worlds we create. That's beautiful. As long as they're tools that help us create those worlds, it's wonderful.

That's right. Yet another tool. They want the generative models to generate the opposite of photorealism. It'll do that, too. It's just yet another tool.

I think the gamers might also appreciate that, in the last couple of years, we introduced skin shaders to game developers. Many of those games have skin shaders that include subsurface scattering, which makes skin look more skin-like. The industry—game developers—are looking for more and more tools to express their art.

This is just one more tool, and they get to decide what to use.

Ridiculous question: What do you think is the greatest or most influential game ever made? Maybe from NVIDIA's perspective?

Doom.

Doom, unquestionably. That was the start of 3D.

I would say Doom, from an artistic perspective, at the intersection of the cultural implication as well as the industry. It turned a PC into a gaming device. That was a very important moment.

Of course, flight simulation companies were before it, but they just didn't have the popularity that Doom did to make the industry turn the PC from an office automation tool into a personal computer for families, gamers, and things like that. Doom was really impactful there.

From an actual game-technology perspective, I would say Virtua Fighter. We're great friends with both of them.

And then there are games more recently, like Cyberpunk 2077, with really nice GPU-accelerated graphics.

Fully ray-traced.

Fully ray-traced. Also, I personally am a huge fan of Skyrim, the Elder Scrolls. It's been released a long, long time ago, but people release mods and—

We love mods.

They create these incredible—it's like a different game, and it just allows me to replay the game over and over. It makes you realize that you can reexperience, in a totally new way, the world you already love.

I do that all the time. One of my favorite things is just to walk across Skyrim.

We created this thing called RTX Mod. It's a modding tool.

Awesome.

It allows the community to inject the latest technology into an old game.

Of course, what makes a great video game is not just graphics; it's also story and character development, but—

That's right.

Beautiful graphics can add to the immersion—the feeling that it's another place you're transported to.

What you said, I think accurately, is that the AGI timeline question rests on your definition of AGI. So let me ask you about possible timelines here. Let's use this ridiculous definition, perhaps, of what AGI is: an AI system that's able to essentially do your job—start, grow, and run a successful technology company that's worth more than $1 billion.

A good one or a bad one?

No. It has to be worth more than $1 billion. You know how hard it is to do all those components. So how far are we away from that? We're talking about OpenClaw doing all the incredibly complex stuff that's required to, first of all, innovate, find customers, sell to them, manage and build a team of agents and humans—all that kind of stuff. Is this 5, 10, 15, or 20 years away?

I think it's now. I think we've achieved AGI.

Do you think you could have a company run by an AI system like this?

It's possible, and the reason for that is this: You said $1 billion, and you didn't say forever. So, for example, it is not out of the question that a Claw was able to create a web service, some interesting little app that all of a sudden a few billion people used for 50 cents, and then it went out of business again shortly after. We saw a whole bunch of those types of companies during the internet era, and most of those websites were not anything more sophisticated than what OpenClaw could generate today.

Interesting. Achieve virality and monetize that virality.

Yeah. It's just that I don't know what it is, but I couldn't have predicted any of those companies at the time either.

You're going to get a lot of people excited with that statement. It's like, what do you mean? I can just launch an agent and make a lot of money.

Well, by the way, it's happening right now, right? You know that when you go to China, you're going to see a whole bunch of people teaching their Claws to try to go out and look for jobs, do work, and make money.

I wouldn't be surprised if some social thing happened, or somebody created a digital influencer that was super, super cute, or some social application that feeds your little Tamagotchi or something like that, and it became, out of the blue, an instant success. A lot of people use it for a couple of months, and then it kind of dies away.

Now, the odds of 100,000 of those agents building NVIDIA are 0%.

The one part that I won't do, and I want to make sure we all do, is recognize that people are really worried about their jobs. I just want to remind them that the purpose of your job and the tasks and tools that you use to do your job are related, not the same.

I've been doing my job for 33 years. I'm the longest-running tech CEO in the world—34 years. The tools that I've used to do my job have changed continuously in the last 34 years, and sometimes quite dramatically over the course of 2 or 3 years.

The one story that I really want to make sure that everybody hears is the story that the first job that computer scientists—AI researchers—said was going to go away was radiology. Computer vision was going to achieve superhuman levels, and it did. Computer vision was superhuman in 2019, maybe a little bit later—2020. It's been a long time since computer vision has been superhuman.

The prediction was that radiologists would go away because studying radiology scans was a thing of the past. AI would do that. Well, they were absolutely right. Computer vision is completely superhuman. Every radiology platform and package today is driven by AI, and yet the number of radiologists grew. So the question is why? We now have a shortage of radiologists in the world.

The alarmist warning went too far, and it scared people from going into this profession that is so important to society. So it did harm.

Why was it wrong? The reason is because the purpose of a radiologist is to diagnose disease and help patients and doctors diagnose disease. Because we're able to study scans so much faster now, you could study more scans, diagnose better, admit patients faster, and see more people. The hospitals are making more money. You have more patients in the hospital. You need more radiologists.

The amazing thing is that it's so obvious this was going to happen. The number of software engineers at NVIDIA is going to grow, not decline. The reason for that is because the purpose of a software engineer and the task of a software engineer—coding—are related, not the same.

I wanted my software engineers to solve problems. I didn't care how many lines of code they wrote. Their purpose didn't change: solving problems, working as a team, diagnosing problems, evaluating the result, looking for new problems to solve, innovating, and connecting dots. None of that stuff is going to go away.

Do you think it's possible that, even if we take coding, the number of programmers in the world might increase, not decrease?

Yes, and the reason for that is this: What is the definition of coding? I believe the definition of coding, as of today, is simply specification. If you want to be rather directive, you could even give it an architecture of the software that you wanted to write.

The question is, how many people could do that—describe a specification for a computer to build, telling the computer what to build? How many people? I think we just went from 30 million to probably 1 billion.

Every carpenter in the future will be a coder, except a carpenter with AI is also an architect. They've just increased the value that they could deliver to the customer. Their artistry has been elevated tremendously.

I believe that every accountant is also a financial analyst and also a financial advisor. All of these professions have just been elevated. If I were a carpenter and I saw AI, I would just completely go berserk. The services I could bring to my clients if I were a plumber—I would completely go berserk.

The people who are currently programmers and software engineers are at the cutting edge of understanding intuitively how to communicate with the agents using natural language in order to design the best kind of software.

That's right, exactly.

So over time they'll converge, but I think there's still value in learning how to program—learning what programming languages are, the old kind of programming, what good practices for programming languages are, and what design principles for programming languages for large software systems are.

And the reason for that, Lex, as you're saying for the audience, is that I think the goal of specification—the artistry of specification—is going to depend on what problem you're trying to solve.

When I'm thinking about giving the company strategies and formulating corporate directions and things that we should do, I describe it at a level that is sufficiently specific that people generally understand the direction and it's actionable. It's specific enough that they can take action on it, but I underspecify it on purpose, so I enable 43,000 amazing people to make it even better than I imagined.

When I'm working with engineers and when I'm working with people, I think about who I'm working with and what problem I'm trying to solve. The level of specification, the level of architecture definition, relates to that.

Everybody's going to have to learn where in the spectrum of coding they want to be. Writing a specification is coding. You might decide to be quite prescriptive because there's a very specific outcome you're looking for. You might decide that this is an area where you want to be much more exploratory, and so you might underspecify it and enable yourself to go back and forth with the AI to push your own boundaries of creativity.

This artistry of where you are in the spectrum is the future of coding.

But just to linger on it outside of coding, I think a lot of people, rightfully so, are worried about their jobs and have a lot of anxiety about their jobs, especially in the white-collar sector. I don't think any of us know what to do with the tumultuous times that always come when automation and new technology arrive.

First of all, I think we all need to have compassion and the responsibility to feel the burden of what the actual suffering feels like for individual people and families who lose their jobs. Whenever you have transformative technology like what's coming with artificial intelligence, there's going to be a lot of pain, and I don't know what to do about that pain.

Hopefully, it creates many more opportunities for those same people, for the same kind of jobs, as the tooling evolves and makes them more productive and makes them more fun, hopefully, as it does in programming.

I've been having so much fun programming, I have to say. I've never had this much fun. So hopefully, it makes their job—automating the boring parts and making the creative parts the ones that human beings are responsible for.

But still, there's going to be a lot of pain and suffering.

My first recommendation—and this is now how I deal with anxiety—is to break it down. In fact, we just talked about it earlier. When I have enormous anxiety about the future, enormous anxiety about the pressure, enormous anxiety about uncertainty, I first break it down, and then I tell myself, “Okay, there are some things you can do something about, and there are some things you can’t do anything about. But for the stuff that you can do something about, let’s reason about it and let’s go do it.”

If we were to hire a new college graduate today, and I had a choice between 2 people—one who has no clue what AI is and one who is an expert in using AI—I would hire the one who’s an expert in using AI. If I had an accountant, a marketing person, someone in supply chain, customer service, a salesperson, someone in business development, or a lawyer, I would hire the one who is an expert in using AI.

I would advise that every college student use AI, and every teacher should encourage their students to use AI. Every college student should graduate and be an expert in AI. And everybody—if you’re a carpenter or an electrician—should go use AI. Go see what it can do to transform your current job and elevate yourself. If I were a farmer, I would absolutely use AI. If I were a pharmacist, I would use AI. I want to see what it could do to elevate my job so that I could be the innovator to revolutionize this industry myself.

And then I would also help them understand that the technology will dislocate and eliminate many tasks because it will automate them. If your job is the task, then you’re very likely to be disrupted. If your job’s purpose includes certain tasks, then it’s vital that you go learn how to use AI to automate those tasks. And then there’s the whole spectrum in between.

And by the way, the beautiful thing about AI—the chatbot versions—is that you can break things down. You have anxiety, and you can break down the problem by talking to it. It’s really just incredible how much you can think through your life’s problems. I don’t mean therapy problems; I mean very practical things: “Okay, I’m worried about my job. What are the skills? What are the steps I need to take? How do I get better at AI?” Everything you just said, you could literally ask, and it’s going to give you a point-by-point plan. I mean, it’s just a great life coach, period.

“I don’t know how to use AI,” and the AI goes, “Well, let me show you.”

Exactly. It’s very meta, but it’s kind of incredible. People definitely should—

You can’t walk up to Excel and say, “I don’t know how to use Excel.” You’re done.

I mean, that’s really what AI has done for me in all walks of life: that initial friction of being a beginner, of using a thing for the first time. I can literally ask about any single thing, “What are the first steps I need to take?”

That’s right.

And that handholding—removing the friction from all the experiences that the world offers—is beautiful. Like I mentioned to you offline, you said, “I’m going to China and Taiwan.”

So awesome.

So excited for you.

When you go to Taiwan, just ask AI, “What are Jensen’s favorite restaurants in Taiwan?” And it’ll actually—

You don’t know?

Oh, yeah.

Is it accurate? Okay.

Yeah.

All right.

It’s all over Taiwan.

Well, you’re a rock star over there. And, like we also mentioned offline, maybe our paths will cross, which would be really wonderful in computing.

COMPUTEX. NVIDIA GTC Taiwan.

Do you think there are some things about human nature, about human consciousness, that are fundamentally noncomputational? Maybe something a chip, no matter how powerful, can never replicate?

I don’t know if the chip will ever get nervous. Of course, I believe AI will be able to recognize and understand the conditions that cause anxiety or nervousness or whatever emotion. I don’t think my chips will feel those. And therefore, how that anxiety, how that feeling, how that excitement, how all of those feelings manifest in human performance—for example, extremely amazing human performance, athletic performance, average or lesser-than-average performance—is something I don’t think computers will experience.

That entire spectrum of human performance comes out of exactly the same circumstances for different people, manifesting in a different outcome and manifesting in a different performance. I don’t think there’s anything about anything we’re building that would suggest that 2 different computers, being presented with exactly the same context, would perform differently. Of course, they would produce statistically different outcomes, but it’s not because they felt different.

The subjective experience—there’s something truly special about the subjective experience that we humans feel. Like I mentioned to you, I was pretty nervous talking to you. The hope, the fear, the anxiety, and just life itself—the richness of life. How amazing everything is. How deeply we fall in love, how deeply our hearts get broken, how afraid we are of death, and how much pain we feel when our loved ones pass away. All of that—the whole thing.

I know it’s very hard to think of AI, a computational device, being able to do that. But there are so many mysteries about this whole thing that we have yet to uncover that I am open to being surprised. I’ve been surprised a lot over the past few months and few years. Scaling can create some incredible miracles in the space of intelligence. It has been truly marvelous to watch, so I’m open to surprise.

It’s really important to break down what intelligence is. The word we use all the time—it’s not a mysterious word. Intelligence has a meaning. It’s something that we do that includes perception, understanding, reasoning, and the ability to plan. That loop is fundamentally what intelligence is.

Intelligence is not one word that is exactly equal to humanity. I think it’s really important to separate the 2. We have 2 words for that. People have heard me say it before: I actually think intelligence is a commodity.

I’m surrounded by intelligent people, and I’m surrounded by people more intelligent than I am in each one of the spaces that they’re in. And yet, I have a role in that circle. It’s actually kind of interesting. They’re more educated than I am. They went to better schools than I did. They’re deeper than I am in any of the fields that they’re in. All of them. I have 60 of them. They’re all superhuman to me.

Somehow, I’m sitting in the middle orchestrating all 60 of them. And so you’ve got to ask yourself: What is it about a dishwasher that allows that dishwasher to sit in the middle of superhumans? Does that make sense?

That’s my point. My point is that intelligence is a functional thing. Humanity is not specified functionally. It’s a much, much bigger word. Our life experience, our tolerance for pain, and our determination are different words than intelligence.

The thing that I want to help the audience understand, if I could give them one thing, is that intelligence is a word that we’ve elevated to a very high form over time.

The word we should really elevate is humanity.

Character, humanity.

All those things.

All of those things. Compassion, generosity—all of the things that you said just now. I believe those are superhuman powers. And now intelligence is going to be commoditized.

Because we’ve spoken about it, the most important thing is your education. Even when they said the most important thing is your education, when you went to school, there was more than just knowledge that you gained. Unfortunately, our society has put everything into one single word, and life is more than one word.

I’m just telling you, my life would suggest that being lower on the intelligence curve than everybody around me doesn’t change the fact that I’m the most successful. And I think that’s what I’m trying to do: hopefully inspire everybody else not to let this democratization of intelligence, this commoditization of intelligence, cause you anxiety. You should be inspired by that.

Yeah. I think AI will help us celebrate humans more, and certainly humanity and humans first. I think what makes this world incredible is humanity, and AI is this incredible tool that makes us—

That’s exactly right.

—humans more powerful.

That’s exactly right.

So much of the success of NVIDIA and the lives of millions of people that I mentioned depend on you. But you’re just one human, like we mentioned—a mortal like all of us. Do you think about your mortality? Are you afraid of death?

I really don’t want to die. I have a great life. I have a great family. I have really important work. This is not a once-in-a-lifetime experience, which suggests that it has been experienced by many people, just not one person. This is a once-in-humanity experience, what I’m going through.

NVIDIA is one of the most consequential technology companies in history. We’re doing very important work. I take it very seriously. Some of the things that are, of course, practical are: How do we think about succession planning? And I’m famous for saying that I don’t believe in succession planning.

Man.

The reason for that isn't because I'm immortal. The reason is that if you're worried about succession planning, if you're worried about all that anxiety of succession planning, then what should you do about it? You break it all the way back down. The most important thing you should do today, if you care about the future of your company post you, is to pass on knowledge, information, insight, skills, and experience as often and continuously as you can.

That's the reason why I continuously reason about everything in front of my team. Every single meeting is a reasoning meeting. Every moment I spend inside a company or outside a company is about passing on knowledge to people as fast as I can. Nothing I learn ever sits on my desk longer than a fraction of a second.

I'm passing that information, that knowledge—oh my gosh, this is cool. Before I even finish learning all of it myself, I'm already pointing it to somebody else: "Get on this. This is so cool. You're going to want to learn this." I'm constantly passing knowledge, empowering people, and elevating the capability of everybody around me, so that the outcome that I seek, that I hope for, is that I die on the job. Hopefully, I die on the job instantaneously, and there's no long period of suffering.

From a fan perspective, given your enormous positive impact on civilization, of course I hope you keep going. But it's also just fun to watch what NVIDIA is doing—the rate of innovation. I'm a huge fan of engineering, and there's so much incredible engineering being continuously done by NVIDIA. It's just fun to watch. It's a celebration of humanity, a celebration of great builders, and a celebration of great engineering. NVIDIA represents something special, so I hope you and NVIDIA keep going.

What gives you hope about this whole thing we have going on—about humanity and the future of humanity? When you look out, when you think about the future quite a bit, and when you look out 10, 20, 50, or 100 years from now, what gives you hope?

I've always had great confidence in the kindness, the generosity, the compassion, and the human capacity. I've always been extremely confident in that, sometimes more so than I should. I get taken advantage of, but it never causes me not to believe in it. I always start with: people want to do good. People want to help others. By and large, I am constantly proven right, and often it exceeds my expectations. I have complete confidence in the human capacity.

The things that give me incredible hope are what I see as I extrapolate—what I see now is possible and, based on the things that we're doing, what will very likely happen. There are so many things that we want to solve. There are so many problems we want to solve, so many things we want to build, and so many good things we want to do that are now within our reach and within the reach of my lifetime. You just can't possibly not be romantic about that. You know what I'm saying?

What an exciting time to be alive. Truly so.

How can you not be romantic about that? The fact that it's a reasonable thing to expect the end of disease. It's a reasonable thing to expect that pollution will be drastically reduced. It's a reasonable thing to expect that traveling at the speed of light is actually in our future—not for long distances, but for short distances.

People ask me how. First of all, very soon I'm going to put a humanoid on a spaceship, and it's going to be my humanoid. We're going to send it out as soon as possible, and it's going to keep improving and enhancing along the flight. When the time comes, so much of my life will already have been uploaded to the internet. Take all my inbox, take everything that I've done, everything I've said—it's been collected and is becoming my AI. When the time comes, we'll just send that at the speed of light to catch up with my robot.

Oh, that's brilliant. But for me, that's sort of application-focused. Also, from a curiosity-maxing perspective, all of those mysteries—there are so many fascinating scientific questions there.

Understanding the biological machine is right around the corner. It's not 10 years; it's probably 5 years.

And then there's your biological machine, the human mind, and cracking theoretical physics open. It's so exciting.

Explaining consciousness—that one would be awesome.

And it's all within our reach. Jensen, thank you so much for everything you've done over the years. Thank you for everything you're doing for the world. Thank you for being who you are. I can tell you're a great human being, and I wish you incredible success this year. I can't wait. As a fan, I can't wait to see what you do next, and hopefully I'll see you in Taiwan. Thank you so much for talking today.

Thank you, Lex. I had a great time. And also, if I could just say one more thing.

Yes.

Thank you for all the interviews that you do—the depth, the respect that you bring to them, and the research that you do to reveal, for all of us, the amazing people that you've interviewed over the years. I've enjoyed them immensely. As an innovator, to have created this long-form format is unbelievable, and yet it's just captivating. Anyways, thank you for everything you do.

It means the world. Thank you, Jensen.

Thank you, Lex.