Jensen Huang on Lex Fridman: Why CUDA almost sank NVIDIA

1. 0:00 – 0:33

   Introduction

   1. LFLex Fridman0:00

      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 the Lex Fridman Podcast. And now, dear friends, here's Jensen Huang.

2. 0:33 – 3:18

   Extreme co-design and rack-scale engineering

   1. LFLex Fridman0:33

      You've propelled NVIDIA into a, uh, new era in AI, moving beyond its focus on chip-scale design to now rack-scale design. And I think it's fair to say that, uh, winning for NVIDIA for a long time used to be about building the best GPU possible, and you still do. But now you've expanded that to extreme co-design of GPU, CPU, memory, networking, storage, power, cooling, software, the rack itself, the pod that you've announced, and even the data center. So let's talk about extreme co-design. What, uh, is the hardest part of, uh, co-designing a system with that many complex components and design variables?

   2. JHJensen Huang1:11

      Yeah, thanks for that question. So 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 you would like to go faster than the number of computers that you add. So you added, uh, you know, ten thousand computers, but you would like it to go a million times faster. Then all of a sudden, you have to take the algorithm, you have to break up the algorithm, you have to refactor it, you have to shard the pipeline, you have to shard the data, you have to shard the model. Now, all of a sudden, when you distribute the problem this way, not just scaling up the problem, but you're distributing the problem, then everything gets in the way. This is the Amdahl's law problem, where, uh, the amount of speed up you have for something depends on how much of the total workload it is. And so if computation represents fifty percent of the problem, and I sped up computation infinitely, like a million times, you know, I only sped up the total workload by a factor of two. Now, all of a sudden, not only do you have to distribute the com-computation, you have to, you know, shard the pipeline somehow, uh, you also have to solve the networking problem because you've got all of these computers are all connected together. And so 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 are a problem. It's just a massively complex computer science problem. And so we just gotta bring every technology to bear. Otherwise, we scale up linearly, or we scale up based on, uh, the capabilities of Moore's Law, which has largely slowed because Dennard scaling has slowed.

   3. LFLex Fridman3:16

      I'm sure there's trade-offs there.

3. 3:18 – 22:40

   How Jensen runs NVIDIA

   1. LFLex Fridman3:18

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

   2. JHJensen Huang3:34

      That's why my staff is so large.

   3. LFLex Fridman3:36

      [laughs] What's the pro-

   4. JHJensen Huang3:38

      Yeah.

   5. LFLex Fridman3:38

      Can you take me through the process of the specialists and the generalists? Like, how do you put together the rack when you know the s- the set of things you have to shove into a rack together?

   6. JHJensen Huang3:48

      Yeah.

   7. LFLex Fridman3:48

      Like, what does that process look like of designing it all together?

   8. JHJensen Huang3:51

      Yeah. There's the, the first question, which is, what is extreme co-design? You're-- We'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 t- just talked about is-- goes beyond CPUs and GPUs and networking chips and scale up switches and scale out switches. And then, of course, you gotta include power and cooling and all of that because, you know, all these computers are extremely, extremely power, power hungry. They do a lot of work, and they're very energy efficient, but they s-- in aggregate, still consume a lot of power. And so that's one-- the first question is, what is it? The second question is, why is it? And we just spoke about the reason, you know, you wanna distribute the workload so that you can exceed the benefit of just increasing the number of computers. And the, and then the third question is, how is it? How do you do it?

   9. LFLex Fridman4:50

      Mm-hmm.

   10. JHJensen Huang4:51

       And, and, uh, that's the, that's kind of the miracle of this company. You know, when you're designing a computer, you have to have operating system of computers. When you're designing a company, you should first think about what is it that you want the company to produce. You know, 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. You know, the goal of a compu- of a company is to be the machinery, the mechanism, the system that produces the output, and that output is the product that we like to create. It is also designed, the architecture of the company should reflect the environment by which it exists. It almost im- directly says what you should do with the o-organization. My direct staff is sixty people. You know, I don't have one-on-ones with them because it's impossible. You can't have, you can't have sixty people on your staff if you're, you know, gonna get work done and-

   11. LFLex Fridman5:51

       So you still have sixty reports. You still have-

   12. JHJensen Huang5:53

       More. Yeah.

   13. LFLex Fridman5:54

       More. [laughs]

   14. JHJensen Huang5:56

       Yeah.

   15. LFLex Fridman5:56

       And most are, at least have a foot in engineering.

   16. JHJensen Huang5:59

       Almost all of them.

   17. LFLex Fridman6:00

       [laughs]

   18. JHJensen Huang6:01

       There's experts in memory, there's experts in CPUs, there's ex-ex-experts in optical, all, all-

   19. LFLex Fridman6:06

       That's incredible.

   20. JHJensen Huang6:06

       Yeah, GPUs and-Architecture, algorithms, design.

   21. LFLex Fridman6:11

       So you constantly have an eye on the entire stack, and you're having do like intense discussions about the design of the entire stack.

   22. JHJensen Huang6:18

       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, you know, because we're doing extreme co-design, and literally the company is doing extreme co-design all the time.

   23. LFLex Fridman6:33

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

   24. JHJensen Huang6:40

       Yeah. Exactly

   25. LFLex Fridman6:41

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

   26. JHJensen Huang6:45

       Exactly

   27. LFLex Fridman6:46

       ... this, this doesn't work for the, for the memory. This doesn't work for this."

   28. JHJensen Huang6:49

       Exactly. And whoever wants to tune out, tune out.

   29. LFLex Fridman6:52

       [laughs]

   30. JHJensen Huang6:53

       You know what I'm saying?
4. 22:40 – 37:40

   AI scaling laws

   1. LFLex Fridman22:40

      Uh, so one of the things you've been a, a believer for a long time is, uh, scaling laws broadly defined. So are you still a believer in the, in the scaling laws?

   2. JHJensen Huang22:49

      Yeah, yeah. Yeah, we have more scaling laws now.

   3. LFLex Fridman22:51

      So I think, yeah, you've outlined four of them with pre-training, post-training, test time, and agentic scaling. What do you think when you think about the future, 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?

   4. JHJensen Huang23:12

      Well, we can go back and reflect on what people thought were blockers.

   5. LFLex Fridman23:15

      Mm-hmm.

   6. JHJensen Huang23:16

      So in the beginning, we were s- the first-- the pre s- pre-training scaling law. You know, people thought, uh, well, rightfully so, that the amount of data that we have, qu- high-quality data, data that we have, um, will limit the intelligence that we achieve, and that scaling law was an important, very important scaling law. The larger the model, the correspondingly more data, uh, c- results in a better-- with a-- results in a smarter AI. And so that was pre-training. And Ilya Sutskever, Ilya said, "We're out of data," or something like that. "Pre-training is over," or something like that. The, the industry panicked, you know, that this is the end of AI. And of course, of course, that's, that's obviously not true. Um, we're gonna keep on scaling the amount of data that we ha-have to, to train with. A lot of that data is probably gonna be synthetic, and that also confused people, you know? And, and what people don't realize is th-th-they've f-kind of forgotten that most of the data that, that we are training, uh, that we teach each other with and inform each other with is sy-synthetic. You know, I... It's synthetic becauseIt didn't come out of nature. You created it. I'm consuming it. I m-modify it, augment it, I regenerate it, somebody else consumes it. And so, so we've now reached a level where AI is able to take ground truth, augment it, enhance it, synthetically generate an enormous amount of data, and that part of post-training, um, continues to scale. And so 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, uh, train model, uh, uh, is gonna continue to scale to the point where we're no longer limited-- training is no longer limited by data. It's now limited by compute. And the reason for that is most of the data is synthetic. Then the next phase is, uh, test time. And, um, I, I still remember people, people telling me that inference, oh yeah, that's easy. Pre-pre-training, that's hard. These are giant systems that people are talking about. Inference must be easy. And so inference chips are gonna be little tiny chips, and-

   7. LFLex Fridman25:31

      Mm-hmm

   8. JHJensen Huang25:31

      ... you know, they're not, they're not like NVIDIA's chips. Oh, those are gonna be complicated and expensive, and, you know, we could make... and this is-

   9. LFLex Fridman25:39

      Mm-hmm

   10. JHJensen Huang25:39

       ... and, and the future inference is gonna be the biggest market, and it's gonna be easy, and we're gonna commoditize it, and, you know, everybody can build their own chips. And, and, and that was always illogical to me because inference is thinking, and I think thinking is hard. Thinking is way harder than reading.

   11. LFLex Fridman25:59

       Mm-hmm.

   12. JHJensen Huang25:59

       You know, pre-training is just memorization and generalization, you know, and looking for patterns and relationships. You're, you're reading and reading versus thinking, reasoning, solving problems, taking un-un-unexplored e-experiences, new experiences, and breaking it down into decom-decomposing it into, you know, solvable pieces that we then go off either through first principle reasoning or, you know, through, through, uh, previous examples, prior experiences. You know, or, or j- or just, uh, uh, exploration and, and search and, you know, trying different things. And that whole process of post-- of, of test time scaling, uh, inference is really about thinking. And, and it's about reasoning, it's about planning, it's about search, it's about... And so how could that possibly be compute light? And we were absolutely right about that, you know. So, so test time scaling is intensely compute intensive. Then the question is, okay, now we're at inference, and we're at test time scaling. What's beyond that? Well, obviously, uh, we have now created, you know, one agentic person, and that one agentic person has a large language model that we've now, we've now, you know, developed. But during test time, that agentic system goes off and does research and bangs on databases, and it goes on and, you know, uses tools. And one of the most important things it does is spins off and spawns off a whole bunch of sub-agents, which 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.

   13. LFLex Fridman27:43

       Mm-hmm.

   14. JHJensen Huang27:44

       And so the next scaling law is the agentic scaling law. It's kind of like multip-multiplying AI.

   15. LFLex Fridman27:51

       Mm-hmm.

   16. JHJensen Huang27:51

       Multiplying AI, we could spin off agents as fast as you wanna spin off agents. And so, you know, I-- you now have four scaling laws. And, and as we use the ag-agentic systems, they're gonna create a lot more data. They're gonna create a lot of experiences. Some of it, we're gonna say, "Wow, this is really good. We ought to memorize this."

   17. LFLex Fridman28:11

       Mm-hmm.

   18. JHJensen Huang28:12

       That data set then comes all the way back to pre-training. We memorize and generalize it. We then refine it and fine-tune it back into post-training. Then we enhance it even more with test time, you know, and the agents, agents, agentic systems, you know, put it out into the indust-industry. And so this loop, this cycle, is gonna go on and on and on. It kind of comes down to basically intelligence is gonna scale by one thing, and it's compute.

   19. LFLex Fridman28:41

       But there's a tricky thing there that you have to anticipate and predict, which is some of these components, it requires different kind of hardware to really do it optimally. So you have to anticipate where the AI innovation is going to lead. For example-

   20. JHJensen Huang28:57

       Perfect

   21. LFLex Fridman28:57

       ... mixture of experts with sparsity.

   22. JHJensen Huang28:59

       Perfect.

   23. LFLex Fridman29:00

       With hardware, you can't just pivot on a week's notice. You have to anticipate what that's going to look like.

   24. JHJensen Huang29:06

       So good.

   25. LFLex Fridman29:06

       That's so, [chuckles] that's so scary and difficult to do, right?

   26. JHJensen Huang29:09

       For example, uh, these AI model architectures are being invented about once every six months.

   27. LFLex Fridman29:15

       [chuckles] Yeah.

   28. JHJensen Huang29:17

       Right? And, uh, system architectures and hardware architectures kind of every three years. And so you need to anticipate what likely is going to happen, you know, two, three years from now.

   29. LFLex Fridman29:31

       Mm-hmm.

   30. JHJensen Huang29:33

       And there's a couple of ways that you could do that. First of all, we could do research internally ourselves, and that's one of the reasons why we have basic research, we have applied research.
5. 37:40 – 39:23

   Biggest blockers to AI scaling laws

   1. LFLex Fridman37:40

      So you eloquently explained how we have a long history of blockers that we thought were gonna 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? So it's obviously we're gonna need compute. So what is going to be the blocker for that scaling?

   2. JHJensen Huang37:59

      Power is a concern, but it's not the only concern. But that's the reason why we're pushing so hard on extreme co-design, so that we can improve the tokens per second per watt orders of magnitude every single year. And so in the last ten years, Moore's Law would have progressed computing about a hundred times in the last ten years. We progressed and scaled up computing by a million times in the last ten years. And so we're gonna keep on, we're gonna keep on doing that through extreme co-design. Um, so energy efficiency per, per watt completely affects the revenues of a company, it affects the revenues of a factory, and we're just, we're just gonna push that to the limit so that we can keep on driving token costs down as fast as we can. You know, the-- our computer price is going up, but our token generation effectiveness is going up so much faster that token cost is coming down. It's just, it, it's coming down an order of magnitude every year.

   3. LFLex Fridman39:04

      So power, that's an interesting one. So the, the way to try to get around the power blocker is to try to, with the tokens per second per watt, try to make it more and more efficient.

   4. JHJensen Huang39:13

      Yeah.

   5. LFLex Fridman39:14

      Of course, there's the question of how do we get more power?

   6. JHJensen Huang39:16

      We should also get more power.

   7. LFLex Fridman39:17

      That's a really complicated one. You've talked about small module nuclear power plants. There's all kinds of ideas for energy.

6. 39:23 – 41:18

   Supply chain

   1. LFLex Fridman39:23

      Uh, how much does it keep you up at night, uh, the, the bottlenecks in the supply chain of AI, like ASML with EUV lithography machines, TSMC with advanced packaging like CoWoS, and, uh, SK Hynix with, uh, high bandwidth memory?

   2. JHJensen Huang39:38

      All, all the time, and we're working on it all the time. No company in history has ever grown at a scale that we're growing while accelerating that growth. It's incredible.

   3. LFLex Fridman39:50

      Yeah.

   4. JHJensen Huang39:50

      And it's hard for people to even understand this. In the overall world of AI computing, we're increasing share. And so supply chain, upstream and downstream, are really important to us. I spend a lot of time, um, informing all the CEOs that I work with, what are the dynamics that's gonna cause, uh, the growth to continue or even accelerate? It's part of the reasons why to the entire right-hand side of me were CEOs of practically the entire IT industry upstream and practically the entire infrastructure industry downstream.

   5. LFLex Fridman40:31

      Mm-hmm.

   6. JHJensen Huang40:32

      And they were all-- there were several hundred CEOs, and I don't think there's ever been keynotes where several hundred CEOs show up. And, [chuckles] and part of it is 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. And I'm also describing where are we gonna go next, so that they could use all of this information and all of the dynamics that are here to inform how they want to invest. And so, so I, I inform them that way like I inform my own employees. And then, of course, then I make trips out to them and make sure that, "Hey, listen, I want you to know this quarter, this coming year, this next year, these things are gonna happen." And, and

7. 41:18 – 47:24

   Memory

   1. JHJensen Huang41:18

      s- if you look at the CEOs of the DRAM industry, um, the number one DRAM in a, in the world was DDR memory for CPUs in data centers. About three years ago, I was able to convince several of the CEOs that even though at the time HBM memory was used quite scarcely, you know, and, and barely by supercomputers, um, that this was going to be a mainstream memory for data centers in the future. And at first it sounded ridiculous, but several of the CEOs believed me and decided to invest in building HBM memories. Another memory was rather odd to put into a data center is the low-power memories that we use for cell phones, and we wanted them to adapt them for supercomputers in the data center. And they go, "Cell phone memory for supercomputers?"

   2. LFLex Fridman42:11

      Mm-hmm.

   3. JHJensen Huang42:11

      And, and I explained to them why. "Well, look at these two memories, LPDDR5, HBM4. The volumes are so incredible. All three of them had record years in history, and these are, these are forty-five-year companies." And so-You know, I-- that's part of my job is to inform and shape, inspire, you know?

   4. LFLex Fridman42:36

      So you're not just manifesting the, the future and maybe inspiring NVIDIA, the, the, the different engineers of the company. You're [laughs] you're manifesting the supply chain of the future. So you're having conversations with TSMC, with ASML-

   5. JHJensen Huang42:50

      Upstream, downstream.

   6. LFLex Fridman42:51

      Upstream, downstream. So that's the thing-

   7. JHJensen Huang42:54

      GEV, Caterpillar.

   8. LFLex Fridman42:56

      [laughs]

   9. JHJensen Huang42:57

      Yeah, that's downstream from us.

   10. LFLex Fridman42:58

       Yeah.

   11. JHJensen Huang42:58

       See? Yeah, yeah. There you go.

   12. LFLex Fridman42:59

       Yeah, the whole thing. I mean-

   13. JHJensen Huang43:00

       Yeah

   14. LFLex Fridman43:00

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

   15. JHJensen Huang43:18

       Exactly. The deep science, the deep engineering, the incredible manufacturing, and so much of the manufacturing is already robotics. But we have a couple of hundred suppliers that contribute the technology that goes into our one point three million component rack.

   16. LFLex Fridman43:35

       Mm-hmm.

   17. JHJensen Huang43:36

       Each rack is one point three, one and a half million components. There are two hundred suppliers across the Vera Rubin rack.

   18. LFLex Fridman43:45

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

   19. JHJensen Huang43:49

       But I'm doing, I'm doing all the things necessary to-

   20. LFLex Fridman43:52

       Okay. Yeah

   21. JHJensen Huang43:53

       ... to see. I can go to sleep because I checked it off.

   22. LFLex Fridman43:55

       [laughs]

   23. JHJensen Huang43:55

       I said, "Okay." You know, I, I go, I, yeah, I can go to sleep when I, I go, "Well, let's see. What, um, re-let's reason about this. What's important for us?"

   24. LFLex Fridman44:03

       Mm-hmm.

   25. JHJensen Huang44:04

       Um, because l- okay, let's reason about this. Uh, because we changed the system architecture from the original DGX-1 that you remembered to, uh, NVLink seventy to rack scale computing-

   26. LFLex Fridman44:15

       Mm-hmm

   27. JHJensen Huang44:16

       ... what's gonna-- what does that, what does that mean? What does that mean to, uh, software? What does that mean to engineering? What does that mean, uh, to how we design and test? How-- and what does that mean to the supply chain? Well, one of the things that it meant was we moved, um, supercomputer, supercomputer integration at the data center into supercomputer manufacturing in the supply chain.

   28. LFLex Fridman44:40

       Mm-hmm.

   29. JHJensen Huang44:42

       If you're doing that, you also have to recognize you're gonna move one-- a-and, and if, if, if your, if your, you know, total footprint of whatever data center you're gonna build, let's say you would like to have, you know, fifty gigawatts of supercomputers that are running simultaneously, and it takes one week to manufacture that fifty gigawatts of supercomputers, then each week in the supply chain, the supercomputers are gonna need a gigawatt of power. And so, so we're gonna need the supply chain to increase the amount of power it has to build, test, to build and test the supercomputers in the supply chain before I ship it.

   30. LFLex Fridman45:25

       Mm.
8. 47:24 – 52:43

   Power

   1. LFLex Fridman47:24

      So maybe if we can just linger on the power for a little bit, uh, what are your hopes for how to solve the energy problem?

   2. JHJensen Huang47:30

      One of the areas, Lex, that I'm, um, that I would love, I would love, love us to talk about and just get the message out, you know, um, our, our, our power grid is designed for the worst case condition with some margin. Well, ninety-nine percent 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 around, call it sixty percent of peak. And so ninety-nine percent of the time, our power grid has excess power, and they're just sitting idle, but they have to be there sitting idle because just in case, when the time comes, hospitals have to be powered and, you know, infrastructure has to be powered, and airports have to run, and so on and so forth. And soThe question that I have is whether we could go and, um, help them understand and create contractual agreements and design computer architecture systems, data centers, such that when they need, um, the maximum power for infrastructure in society, that the data centers would get less.

   3. LFLex Fridman48:48

      Mm-hmm.

   4. JHJensen Huang48:49

      But that's in a very rare instance anyways. And during that time, we either have our backup generators for that little part of it, or we just have our computers shift the workload somewhere else, or we have the computers just run slower. You know, we could degrade our performance, reduce our power consumption, and provide for a, you know, slightly longer latency response, you know, when somebody asks for, you know, asks for an answer. And so I think that, that, that way of using computers, of building data centers, instead of expecting a hundred percent uptime and these contracts that are really, really quite rigorous, it's putting a lot of pressure on the grid to be able to... Now they're gonna have to increase from their maximum. I just wanna use their excess.

   5. LFLex Fridman49:34

      Mm-hmm.

   6. JHJensen Huang49:35

      It's just sitting there.

   7. LFLex Fridman49:36

      Yeah, that's not talked about enough. So what's, what's the s- what's stopping there? Is it regulation? Is it bureaucracy?

   8. JHJensen Huang49:43

      I think it's, it's a three-way problem. Uh, it starts with the end customer. The end customer puts, puts requirements on the data centers that they can never not be available, okay? So the, the end customer expects perfection. Now, in order to deliver that perfection, you need a combination of backup generators and your grid power supplier to deliver on perfection. And so everybody's gotta have six nines.

   9. LFLex Fridman50:12

      Mm-hmm.

   10. JHJensen Huang50:13

       Well, I think first of all, right now we ought to have everybody understand that when the customer asks for these things, you got somebody-- 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, and they go down to cloud service providers and the contract, the c- the two contract negotiators that are... You-- I could just see them now-

   11. LFLex Fridman50:46

       Mm-hmm

   12. JHJensen Huang50:46

       ... you know, negotiating these multi-year contracts. Both sides want, you know, the best contract. As a result, the CSPs then have to go down to the utilities, and they expect the nine, the six nines. And so I think, I think the first thing is just make sure that, that all of the customers, the CEOs of the customers realize what they're asking for. Now, the second thing is we have to build data centers that gracefully degrade. And so if the power, if the utility, the grid tells us, "Listen, we're gonna have to back you down to about eighty percent," we're gonna say, "That's no problem at all."

   13. LFLex Fridman51:21

       Mm-hmm.

   14. JHJensen Huang51:21

       We're just gonna move our workload around. We're gonna make sure that data's 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 that somewhere else right away, so I don't have that problem. And so, you know, whoever, whichever da-data center still has a hundred percent uptime. And so-

   15. LFLex Fridman51:44

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

   16. JHJensen Huang51:49

       As soon as you could specify, you could engineer it. [laughs]

   17. LFLex Fridman51:52

       Beau-beautifully put. [laughs]

   18. JHJensen Huang51:54

       So long as it obeys the laws of physics on first principles, I think we're good. [laughs]

   19. LFLex Fridman51:58

       What was the third thing you were mentioning? Um-

   20. JHJensen Huang52:00

       So the second thing is the, the data centers. And the third thing is we need the utilities to also recognize that this is an opportunity.

   21. LFLex Fridman52:09

       Mm.

   22. JHJensen Huang52:10

       And, and instead of, instead of saying, "Look, um, it's gonna take me five years to increase my grid capability, uh, if you, if you have-- if you're willing to take power of this level of guarantee, I can make them available for you next month and at this price." And so if u-utilities also offered more segments of power delivery promises, then I think everybody will figure out what to do with it. Yeah, but there's just way too much waste in the, in the grid right now. We sh- we should go after

9. 52:43 – 56:11

   Elon and Colossus

   1. JHJensen Huang52:43

      it.

   2. LFLex Fridman52:44

      Uh, you've, uh, highly lauded Elon's and, uh, xAI's accomplishment in Memphis in building, um, Colossus supercomputer probably in record time, in just four months. It's now at two hundred thousand GPUs and growing very quickly. Is there something that you could speak to the, uh, understand about his approach that's instructive to the broadly to all the data center creators that, um, that enabled that kind of accomplishment, his approach to engineering, his approach to the whole management of construction, everything?

   3. JHJensen Huang53:15

      First of all, Elon is deep in so many different topics. Um, uh, yet he's also a really good systems thinker.

   4. LFLex Fridman53:24

      Mm-hmm.

   5. JHJensen Huang53:24

      And so he's able to think through multiple disciplines and, and, um, uh, he obviously, uh, pushes things, questions everything, whether, number one, is it necessary? Number two, does it have to be done this way? And number th-- you know, does it have, does it have to take this long? And, and so, so he, he has, he has the a-- he has the ability, uh, to question everything, uh, to the point where everything is down to its minimal amount that is necessary. N- you can't take anything else out. And, and yet, yet the, the, uh, the, the, the, the necessary, um, capabilities of the product retains, you know? And so he's, he is as minimalist as you could possibly imagine, and he does it at a system, system scale. Um, I, I th- I also love the fact that he, he is, um, he is represented. He, he is, he is present at the point of action.

   6. LFLex Fridman54:24

      Mm-hmm.

   7. JHJensen Huang54:25

      You know, he'll just go there. A-if there's a problem, he'll just go there, and he'll show me the problem. You know, when you do all of this in combinationYou overcome a lot of previous, "This is just the way we do it."

   8. LFLex Fridman54:38

      Mm-hmm.

   9. JHJensen Huang54:38

      Um, you know, "I'm, I'm waiting for them." Uh, you know, I mean, it's just everybody has a lot of excuses. A-and so-- And then, and then the last thing is when, when you act personally with so much urgency, uh, it causes everybody else to act with urgency, you know. And, and every supplier has a lot of customers going on. Every supplier has a lot of projects going on. And he, he make it-- he made it-- he makes it his business that he's the top priority of everybody else's, you know, projects. And so he, he does that by demonstrating it.

   10. LFLex Fridman55:09

       Yeah, I've been in a bunch of those meetings. It's, it's fun to watch 'cause really not enough people ask the question like, "Okay, so, uh, can this be done a lot faster, and how? Why does it have to take this long?"

   11. JHJensen Huang55:21

       Yeah. Right.

   12. LFLex Fridman55:22

       And then that becomes an engineering question often. And yes, I think when you get the ground truth of actually... I remember, um, one of the times I was hanging out with him, he literally is going through the entire process of how to plug in cables into a rack. Uh, he's, was, was working with an engineer on the ground that's doing that task, and he's just trying to understand what does that process look like so it can be less error-prone. And just building up that intuition from every single task in-involved in, uh, putting together the data center-

   13. JHJensen Huang55:51

       Mm-hmm

   14. LFLex Fridman55:52

       ... you start to immediately get a sense at the detailed scale and then at the broad system scale of where the inefficiencies are, and so you can make it more and more and more efficient. Plus, you have the big hammer of being able to say, "Let's do it totally different"-

   15. JHJensen Huang56:08

       Yeah. That's right

   16. LFLex Fridman56:08

       ... "and remove all possible blockers."

   17. JHJensen Huang56:10

       That's right.

10. 56:11 – 1:01:37

    Jensen's approach to engineering and leadership

    1. LFLex Fridman56:11

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

    2. JHJensen Huang56:18

       Well, first of all, co-design is a ultimate systems engineering problem.

    3. LFLex Fridman56:22

       Yeah.

    4. JHJensen Huang56:22

       And so we approach, we approach the work that we do, um, from that first-- from that principle. Um, the other thing that we do, uh, and this is, this is a, a philosophy that a, a thought, a, a, um, a, a state of mind, I guess, a method, uh, that I started, uh, thirty years ago, and it's called the speed of light. The speed of light is not just about the speed. Speed of light i-is my, my shorthand for, um, what's, what's the limit of what physics can do. And so every single-- everything, everything that we do is compared against the speed of light. Um, memory speed, uh, math speed, uh, power, cost, time, effort, number of people, manufacturing cycle time. And, uh, when you think about latency versus throughput, uh, when you think about cost versus throughput, cost versus capacity, all of these things, uh, you test against the speed of light to achieve all of these different constraints separately. And then when you consider it together, you know you have to make compromises because a system that achieves extremely low latency versus achieve-- a system that achieves very high throughput are architected fundamentally differently. But you want to know what's the speed of light of a system that achieves high throughput? What's the speed of light of a system that achieves low latency? And then when you think about the total system, you can make trade-offs. And so I, I force everybody to think about what's this-- what the fir-- the first principles, the limits-

    5. LFLex Fridman58:01

       Mm-hmm

    6. JHJensen Huang58:01

       ... the physical limits, um, for everything before we, you know, before we, uh, do anything. And, and we test everything against that. And so that's a good frame of mind. I don't love the other methods, which is continuous improvement.

    7. LFLex Fridman58:18

       Mm-hmm.

    8. JHJensen Huang58:19

       The, the problem with continuous improvement, it, it-- First of all, you should engineer something from first principles at the speed, you know, with speed of light thinking, limited only by physical limits and, and physics limits. And, um, after that, of course you would improve it over time. Um, but I don't like going into a problem and somebody says, "Hey, you know, it takes seventy-four days to do this today"-

    9. LFLex Fridman58:45

       Mm-hmm

    10. JHJensen Huang58:45

        ... "um, right now, and, um, we can do it for you in seventy-two days." You know, I'd rather strip it all back to zero.

    11. LFLex Fridman58:51

        Mm-hmm.

    12. JHJensen Huang58:52

        And say, "First of all, explain to me why it's seventy-four days in the first place. And let's know-- let's think about what's possible today. And if I were to, to build it completely from scratch, you know, how long would it take?" Oftentimes you'd be surprised. It might come to six days. Now, the rest of the six days to seventy-four could be very well-reasoned and compromises and, you know, cost reductions and all kinds of different things, but at least you know what they are. And then now that you know that six days is possible, then the conversation from seventy-four to six, surprisingly much more effective.

    13. LFLex Fridman59:30

        In, uh, such incredibly complex systems that you're working with, is simplicity sometimes a, a good heuristic to, to reach for? I mean, if I can just-- I mean, the pod, the Vera Rubin pod that you announced is just incredible. Uh, we're talking about seven chips, seven chip types, five purpose-built rack types, forty racks, one point two quadrillion transistors, nearly twenty thousand NVIDIA dies, over eleven hundred Rubin GPUs, sixty exaFLOPS, ten petabytes per second of scale bandwidth. Uh, that's all just one-

    14. JHJensen Huang1:00:03

        That's just one pod.

    15. LFLex Fridman1:00:04

        That's just one pod. [laughs]

    16. JHJensen Huang1:00:06

        Yeah. That's just one pod.

    17. LFLex Fridman1:00:07

        I mean, and so you have the...

    18. JHJensen Huang1:00:09

        Yeah.

    19. LFLex Fridman1:00:09

        And, and then even the, the NVL72 rack alone is one point three million components, thirteen hundred chips, four thousand pounds crammed into a single nineteen-inch wide rack.

    20. JHJensen Huang1:00:19

        And Lex, we'll probably kind of crank out about two hundred of these pods a week, just to put in perspective.

    21. LFLex Fridman1:00:25

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

    22. JHJensen Huang1:00:35

        You know, the phrase, the phrase that I use most often is we, we need things to be as complex as necessary but as simple as possible.And, and so the question is, is all that complexity there necessary? And we ought to test for that.

    23. LFLex Fridman1:00:50

        Mm-hmm.

    24. JHJensen Huang1:00:50

        And we ought to challenge that. And then after that, everything else above it, you know, is gratuitous.

    25. LFLex Fridman1:00:56

        But some of the most incredible semiconductor industry broadly, but what NVIDIA is doing, uh, s-s-some of the greatest engineering in history. So these systems are just truly, truly marvels of engineering.

    26. JHJensen Huang1:01:10

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

    27. LFLex Fridman1:01:13

        Yeah, the engineering teams, I mean-

    28. JHJensen Huang1:01:14

        Yeah

    29. LFLex Fridman1:01:14

        ... I don't know. It's not a competition, but I don't know if, if it was like an Olympics of, uh, engineering teams. I mean, TSMC does incredible engineering, like I said, ASML at every scale, but NVIDIA is gonna give them a run for their money.

    30. JHJensen Huang1:01:27

        Yeah.
11. 1:01:37 – 1:09:50

    China

    1. LFLex Fridman1:01:37

       Uh, you've recently traveled to China. Uh, so it's interesting to ask you, uh, China's been incredibly successful in building up its technology sector. What do you understand about, um, how China is able to, over the past ten years, build so many incredible world-class companies, world-class engineering teams, and just this technology ecosystem-

    2. JHJensen Huang1:02:01

       Mm-hmm

    3. LFLex Fridman1:02:01

       ... that produces so many, um, incredible products?

    4. JHJensen Huang1:02:05

       Whole bunch of reasons. Fir- well, first of all, let's, let's start, let's start with some facts. Fifty percent 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's amazing researchers still in China. Um, they-- their tech industry showed up at precisely the right time. At the time of the mobile cloud era, uh, their way of contributing was software. And so this is a country's in-incredible science and math, uh, really well-educated kids. Um, uh, their tech industry was created during the era of software. 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's so many EV companies. That's the reason why there's so many AI companies. That's the reason why there's so many-- every company you could imagine, um, they all create some of them. And, and, um, as a result, they have insane competition internally, and, you know, what remains is an incredible company. Um, they also have a, um, social culture where, where f-it's family first, friends second, and company third. And so, um, the amount of conversation that goes back and forth between-- They're essentially open source all the time. So the fact that they contribute more to open source is so sensible because they're probably, "What are we protecting?" You know, my engineers, their brothers are in that company, their friends are in that company, and they're all schoolmates. You know, the schoolmate concept. It's a, you know, one schoolmate, you're brother for life. And, um, and so they, they, they share knowledge very, very quickly. And so there's no sense keeping technology hidden. You might as well put it on open source. And so the open source community then amplifies, accelerates the, the innovation process. So you get this rapid, incredibly great talent, rapid innovation because of open source and just, you know, the, the nature of friends and, and, um, insane competition among comp-- among the company. What emerges is incredible stuff. And so this is the fastest innovating country in the world today, and this is something that has everything that-- everything that I just said is fundamental to just how the kids were grown, the fact that they have excellent education, the fact that they-- parents want them to do well in school, the fact that they-- their culture is that way. These are f- you know, these are just the thing about their country, and they showed up at precisely the time when technology is going through that exponential.

    5. LFLex Fridman1:05:09

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

    6. JHJensen Huang1:05:16

       It's a, it's a builder nation.

    7. LFLex Fridman1:05:18

       It's a builder nation.

    8. JHJensen Huang1:05:19

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

    9. LFLex Fridman1:05:43

       To take a small tangent, because you mentioned open source, I have to, uh, go to Perplexity here, who you have been a, a fan of a long time.

    10. JHJensen Huang1:05:51

        I love it, yeah.

    11. LFLex Fridman1:05:52

        And thank you for releasing open source Nemotron 3 Super, which you can also use inside Perplexity to look stuff up-

    12. JHJensen Huang1:05:59

        Yeah

    13. LFLex Fridman1:05:59

        ... now, which is 120 billion parameter open weight, uh, MoE model. Uh, what's your vision with open source? So you've mentioned China with, with DeepSeek, with Minimax, with all these companies really pushing forward the open source, uh, AI movement, and NVIDIA is really leading the way in, um, close to state-of-the-art open source LLMs. What's your vision there?

    14. JHJensen Huang1:06:28

        First, if we're gonna be a great AI computing company, we have to understand how AI models are evolving.

    15. LFLex Fridman1:06:36

        Mm-hmm.

    16. JHJensen Huang1:06:36

        One of the things that I love about Nemotron 3 is it's, it's not a, just a pure transformer model. It's transformer and SSMs.And, uh, we were early in, uh, developing the, the, uh, conditional GANs, which that progressive GANs, which led step-by-step to diffusion. And so, um, the fact that we're doing basic research in model architecture and in different domains gives us visibility into, you know, what kind of computing systems would do a good job for future models. And so it is part of our extreme co-design strategy. Second, um, I think we, we rif-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, every student. And if everything's proprietary, it's hard to do research and it's hard to innovate on top of, around, with. And so open source is fundamentally necessary for many industries to join the AI revolution. NVIDIA has the scale, and we have the motives to not only 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, we can activate every industry, every researcher, you know, every country to be able to join the AI revolution. There's the third reason, which is from that-- to recognizing that AI is not just language. These AIs will likely use, uh, tools and models and sub-agents that were trained on other modalities of information. Maybe it's biology or chemistry or, um, you know, laws of physics or, you know, fluids and thermodynamics, and not all of it is in language structure.

    17. LFLex Fridman1:08:49

        Mm-hmm.

    18. JHJensen Huang1:08:50

        And so somebody has to go make sure that weather prediction, biology, AI, AI for biology, physical AI, all of that stuff stays-- can be pushed to the limits and pushed to the frontier. We don't build cars, but we want to make sure every car company has access to great models. We don't, 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 go use it for discovering drugs. And so these three fundamental reasons, both in, in recognizing that AI is not just language, that AI is really broad, that we want to engage everybody into the world of AI, and then also co-design of AI.

    19. LFLex Fridman1:09:32

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

    20. JHJensen Huang1:09:39

        Yeah, I appreciate you for saying that. We open source the models, we open source the weights, we open source the data.

    21. LFLex Fridman1:09:44

        Mm-hmm.

    22. JHJensen Huang1:09:44

        We open source how we created it. Yeah. Pretty amazing.

    23. LFLex Fridman1:09:48

        [laughs] It, it's really, it's really incredible.

12. 1:09:50 – 1:15:04

    TSMC and Taiwan

    1. LFLex Fridman1:09:51

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

    2. JHJensen Huang1:10:19

       You know, first of all, the deepest misunderstanding about TSMC is that, that, um, their technology is all they have. That somehow they, they have a really great transistor, and if somebody shows up another transistor, game over.

    3. LFLex Fridman1:10:40

       Mm-hmm.

    4. JHJensen Huang1:10:41

       It's the technology, uh, and of course, you know, I, I don't mean just the transi-transistor and metallization systems, the packaging, the three-D packaging, the silicon photonics, the, you know, 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, the demands, the, the dynamic demands of hundreds of companies in the world-

    5. LFLex Fridman1:11:11

       Mm.

    6. JHJensen Huang1:11:11

       -as they're moving up, shifting out, you know, increasing, decreasing, push-pushing out, pulling in, um, changing from customer to customer, uh, wafer starting, wafer stopping, uh, emergency wafer starts, you know. All of this 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 cost, excellent customer service. They, they take their work ser-- they take their promises seriously. When your wafer-- because they know that you're help-- they're helping you run your company. When the wafers, when the wafers were promised to show up, the wafers show up, you know, so that you could run your company appropriately. And so their system, their manufacturing system is completely miraculous. I would say then the second thing is their culture. This culture is, uh, simultaneously, uh, technology-focused on one hand, advancing technology, simultaneously customer service-oriented on the other hand. A lot of cust-cus-companies are very customer service-oriented, but they're not very technology excellent. They're, they're not at the bleeding edge of technology. Or a lot of companies who are tech-- at the bleeding edge of technology, but they're not the best customer service-oriented company. And so it just depends on somehow they've, they've balanced these two, and they're world-class at both. Um, and then probably the third thing is the technology that I most value in them, uh, that they created this, you know, this, this, uh, intangible called trust. I trust them to put my company on top of themThat's a very big deal.

    7. LFLex Fridman1:12:55

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

    8. JHJensen Huang1:13:05

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

    9. LFLex Fridman1:13:15

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

    10. JHJensen Huang1:13:30

        Story is true. I didn't, I didn't dismiss it.

    11. LFLex Fridman1:13:33

        Yeah, yeah.

    12. JHJensen Huang1:13:33

        Um, uh, but I was, I was deeply honored, and, and of course, of course, um, uh, I knew then, as I know now, TSMC is one of the most consequential companies in history.

    13. LFLex Fridman1:13:43

        Yeah.

    14. JHJensen Huang1:13:45

        And, and Morris is one of the, uh, the highest regarded executive and, and, um, business and personal friend that I've, that I've had in my life. And, um, uh, for him to ask is, uh, uh, um... I wa-- I was humbled and, and, um, really honored. Um, but, but the work that I'm doing here is really important, and I've seen, you know, in my mind, in ways, in my mind's eye, what NVIDIA was going to be and what the impact that we could have. And, um, uh, it was really important work. Uh, and it's my responsibility, you know, my sole responsibility to make this happen. And so I, I, um, uh, I declined it, uh, you know, n-not, not because it wasn't an incredible offer. Uh, it, it's an unbelievable offer, um, but, but I simply couldn't take it.

    15. LFLex Fridman1:14:38

        I think NVIDIA, both NVIDIA and TSMC are two of the greatest companies in the history of human civilization, and running either one, I'm sure, is incredibly complicated effort, and it takes... Y-you have to truly be all in.

    16. JHJensen Huang1:14:52

        Yeah.

    17. LFLex Fridman1:14:52

        Uh, everybody at every scale, not just at the CEO level, everybody is really truly all in-

    18. JHJensen Huang1:14:57

        Yeah. Yeah, no doubt

    19. LFLex Fridman1:14:59

        ... uh, to, to accomplish this kind of complexity.

    20. JHJensen Huang1:15:00

        See, now I can help both companies.

    21. LFLex Fridman1:15:02

        Exactly. [laughs] Um,

13. 1:15:04 – 1:20:41

    NVIDIA's moat

    1. LFLex Fridman1:15:04

       so NVIDIA is now the most valuable company in the world. I have to ask, what is the-- NVIDIA's biggest moat, as the folks in the tech sector say?

    2. JHJensen Huang1:15:15

       Mm-hmm.

    3. LFLex Fridman1:15:15

       The edge you have that protects you from the competition.

    4. JHJensen Huang1:15:20

       Our single most important, uh, property as a company is the install base of our computing platform. Our single most important thing is the invol-- today is our, is the install base of CUDA. Now, the reason why, uh, twenty, twenty years ago, of course, there was no install base. But what makes-- And if somebody, if somebody came up with, with a gouda or Tuda, uh, it wouldn't make any difference at all. And the reason for that is because, because it's never been just about the technology. The technology, of course, was incredible visionary. Um, but it's the fact that the company was dedicated to it, stuck with it, expanded its reach. Um, it wasn't three people that, that made CUDA successful. It was forty-three thousand people that made CUDA successful. And the several million developers that believed in us, um, that trusted that we were going to continue to make CUDA one, two, three, thirteen, that they decided to port and dedicate their software on top of it, their mountain of software on top of it. And so 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. And, and that velocity combined with the install base, in the developer's mind, you just gotta now take a developer's mind. From the developer's perspective, if I support CUDA, tomorrow it'll be ten times better. I just have to wait six months on average. Not only that, if I develop it on CUDA, I reach a few hundred million people, computers. I'm in every cloud, I'm in every computer company, I'm in every single industry, I'm in every single country. So if I decreate an open source package, and I put it on CUDA first, I get these both attributes simultaneously. And not only that, I trust one hundred percent that NVIDIA is going to keep CUDA around and maintain it and improve it and keep optimizing the libraries for as long as they shall live. You could take that to the bank, and that last part, trust, you put all that stuff together, if I were a developer today, I would target CUDA first. I would target CUDA most. And that's the reason. That, that I think in, in the final analysis is our first-- that's even our first-

    5. LFLex Fridman1:18:16

       Mm-hmm

    6. JHJensen Huang1:18:17

       ... core advantage. Our second one is our ecosystem. The fact that we vertically integrated this incredibly complex system, but we integrated horizontally into every single, every single company's computers.

    7. LFLex Fridman1:18:30

       Mm-hmm.

    8. JHJensen Huang1:18:30

       We're in the Google Cloud, we're in Amazon, we're in Azure. You know, 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. You know, I mean, it's, it's just crazy. One architecture is in all these different systems. We're in cars, we're in robots, we're in satellites.We're out in space. And so, so the fact that you have this one architecture, and the ecosystem is so broad, it basically covers every single industry in the world.

    9. LFLex Fridman1:19:03

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

    10. JHJensen Huang1:19:16

        Well, the, the unit of computing used to be GPU t-to us, then it became a computer, then it became a cluster. Now it's an entire AI factory. When I see a computer, when I see what NVIDIA builds, in the old days, I would, you know, I visualize the chip.

    11. LFLex Fridman1:19:31

        Mm-hmm.

    12. JHJensen Huang1:19:32

        And then, and then when I announced a new product, you know, a new generation, like, "Ladies and gentlemen, we're announcing Ampere today," I pick up the chip.

    13. LFLex Fridman1:19:39

        Yeah.

    14. JHJensen Huang1:19:40

        That was my mental model-

    15. LFLex Fridman1:19:41

        Yeah

    16. JHJensen Huang1:19:41

        ... of what I was building. Today, I w- I wouldn't-- Picking up the chip is kind of still adorable.

    17. LFLex Fridman1:19:47

        Yeah.

    18. JHJensen Huang1:19:47

        But it's adorable. It, it, it's not, it's not my mental model of what I'm doing. My mental model is this giant gigawatt thing that has power generation. It's connected to the grid. It's got cooling systems and networking of incredible monstrosity. You know, ten thousand people are in there trying to install it, hundreds of networking engineers in there, thousands of engineers behind it trying to power it up.

    19. LFLex Fridman1:20:13

        Mm-hmm.

    20. JHJensen Huang1:20:14

        You know? Powering up one of those factories, as you know, is not somebody going, "It's on now." [laughs] Takes thousands of people to bring it up.

    21. LFLex Fridman1:20:22

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

    22. JHJensen Huang1:20:33

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

14. 1:20:41 – 1:24:30

    AI data centers in space

    1. JHJensen Huang1:20:41

       click.

    2. LFLex Fridman1:20:42

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

    3. JHJensen Huang1:20:56

       Cooling issues is not easy. Yeah.

    4. LFLex Fridman1:20:58

       Cooling. Well, there's a large number of engineering complexities involved with that.

    5. JHJensen Huang1:21:03

       Yeah. Yeah.

    6. LFLex Fridman1:21:03

       So, so what-- You know, NVIDIA has also announced that you're already thinking about that.

    7. JHJensen Huang1:21:09

       Yeah, we're already there. Uh, NVIDIA GPUs are the first GPUs in space. And, um, I, I didn't realize it, it was, it was so interesting to-- I would have declared it maybe [laughs] were in space. You know, little, little astronaut suit on one of our GPUs.

    8. LFLex Fridman1:21:25

       [laughs]

    9. JHJensen Huang1:21:25

       Uh, [laughs] um, but, but we've been in space. Uh, it's the right place to do a lot of imaging-

    10. LFLex Fridman1:21:31

        Mm-hmm

    11. JHJensen Huang1:21:32

        ... you know, because those satellites have, have really high-resolution imaging systems, and they're sweeping the Earth, you know, continuously now. And, um, uh, y-you want, you know, centimeter scale, you know, imaging that is done continuously, uh, for the world so that, you know, you'll basically have real-time telemetry of everything. Uh, you don't wanna beam that back down to Earth. It's just, you know, petabytes and petabytes of data. You gotta just do AI right there at the edge, throw away everything you don't need, you've seen before, didn't change, and then just keep the stuff that, that you need. And so AI ought to be done at the edge. Um, obviously, we have, we have, uh, twenty-four/seven solar if we put it at the polars. And, um, uh, but, you know, there's no conduction, no convection, and so, you know, you're pretty much just radiation. And, um, uh, but, you know, space is big. I guess, you know, we're just gonna put big, giant radiators out there.

    12. LFLex Fridman1:22:31

        How, how crazy of an idea do you think it is? Like, is this, is this five years out, ten years out, twenty years out? So, uh, we're talking about blockers for AI scaling.

    13. JHJensen Huang1:22:41

        You know, I'm just so much more practical. I, I look for where, where, um, uh, my next, next bucket of opportunities are first. Uh, w- meanwhile, I'm cultivating space. And so I send, I send engineers, uh, to go work on the problem. We're st- we're starting to-- we're learning a lot about it. Um, how do we do radiation? How do we do, uh, degrading performance? How do we deal with, um, uh, continuous, uh, testing and attestation of, of, um, de-defects? And, and, um, you know, how do we deal with redundancy, and, uh, how do we de-degrade, uh, gracefully, and things like that. And so we could, we could do, uh, what, what about software? How do you think about software and, and redundancy and performance out in space? Uh, make it so that, so that the computer never breaks. It just gets slower, you know. [laughs] And, um, uh, so we could start doing a lot of engineering exploration up front. But in the meantime, my, my favorite answer is g- eliminate waste. You know, we've, we've got all that idle power. I wanna evacuate it as fast as possible. [laughs]

    14. LFLex Fridman1:23:47

        Yeah. There, there, yeah, there's a lot of low-hanging fruit here on Earth-

    15. JHJensen Huang1:23:50

        Yeah

    16. LFLex Fridman1:23:50

        ... uh, that we can utilize, uh, for the AI scaling. Uh, quick pause. Quick thirty-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.

15. 1:24:30 – 1:34:39

    Will NVIDIA be worth $10 trillion?

    1. LFLex Fridman1:24:30

       Do you think NVIDIA may be worth ten trillion at some point? W- let's, let's ask it this way. What does the future w-of the world look like where that, where that's true?

    2. JHJensen Huang1:24:45

       I think that NVIDIA's growth is, is, um, uh, extremely likely and in my mind, inevitable. And let me explain why. We're the largest computer company in history.That alone should beg the question why. And the reason for, of course, uh, two reasons. First, two foundational technical reasons. The first reason is that computing went from being a retrieval-based file retrieval system. Almost everything is a file re-- we, we pre- prewrite something, we pre-record something, you know, we, we draw something, we put it on the web, we put it in a file, and we, we use a recommender system, some smart filter to figure out what to retrieve for you. And so we were pre-recording, human pre-recording and file retrieving system. That's what a computer is largely. To now, AI computers are contextually aware, which means that it has to process and generate tokens in real time. So we went from a retrieval-based computing system to a generative-based computing system. We're gonna need a lot more processing in this new world than in the old world. We need a lot of storage in the ol-ol-old world. We need a lot of computation in this new world. And so, so that's, that's the first part of it. We fundamentally changed computing and the way how computing is done. The only thing that would cause it to go back is if this way of computation, th-this way of computing, generating information that's contextually relevant, situationally aware, that is grounded on new insight before it generates information. This computation-intensive way of doing computing would only go back if it's not effective. So if for the last ten, 15 years while working on deep learning, if at any single moment I would've come to the conclusion that, that, "You know what? This is not gonna work out. I think this is a dead end," or, "It's not gonna scale, it's not gonna solve this modality, it's not gonna be used in this application," then of course I would feel very differently about it. But I think the last five years has given me more confidence than the last ten year-- the previous ten years. The second idea is computers, because it was a storage system, it was largely a warehouse. We're now building factories. Warehouses don't make much money. Factories directly correlates with a company's revenues. And so the computer did two things. Not only did it change the way it did it, its purpose in the world changed. It's no longer a computer, it's a factory. It's a factory, it's used for generation of revenues. We're now seeing not only is this factory generating products, commodities that people want to consume, we're seeing that the commodities are so interesting, so valuable, so-- to so many different audiences that the tokens are starting to segment like iPhones. Mm-hmm. You have a f-free tokens, you have premium tokens, and you have several tokens in the middle. Yeah. And so intelligence, as it turns out, you know, is a scalable product. There's extremely high intelligence products, tokens that you could-- that are used for specialized things. People would be willing to pay, you know, 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. And so, so now we're seeing that the commodity that this factory makes is actually valuable and is revenue generating and profit generating. How-- Now the question is, how many of these factories can-- does the world need? How much-- how many tokens does the world need? And, um, how much is society willing to pay for these tokens? And what would happen to the world's economy if the productivity were to improve so substantially? What would happen? Are we, are we gonna discover new drugs, new products, new services? And so 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 a hundred times more than the past- Mm-hmm ... because it's no longer a storage unit, it's a product generation unit. And so when you look at it in that context, and then you back into what is NVIDIA's, what does NVIDIA sh-- what does NVIDIA do, and how much of that new economics, new industry would we have to benefit to, to address? I think we're gonna be a lot, lot bigger. And then the rest of it to me is, um, you go, is it possible for NVIDIA to be a, you know, $3 trillion revenues company in the near future? The answer is of course yes, and the reason for that is because it's not limited by any physical limits. There's nothing that I see that says, you know, gosh, um, $3 trillion is not possible. And as it turns out, NVIDIA's supply chain is the burden is shared by 200 companies. And the fact that we s-scale out on the backs of with the partnership of this ecosystem, the question is, do we have the energy to do so? And surely we will have the energy to do so. And so all of these things combined, that number is just a number, you know. And I still remember NVIDIA was a, NVIDIA was a w-- the first time we crossed a billion dollars, I was reminded of, of a CEO who told me, you know, "Jensen, it's theoretically impossible for a fabless semiconductor company to exceed a billion dollars."And, and, um, I won't bore you with why, but, but the an- of course it's illogical, and there's a lot of evidence we're not. And then there-- somebody told me, "You know, Jensen, you'll never be more than twenty-five billion dollars because of some other company." Somebody told me that you'll never be, you know, because... And then so, so the, the, those, those aren't principled, first principle reason thinking. And the simple, 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 don't exist.

    3. LFLex Fridman1:31:47

       Mm-hmm.

    4. JHJensen Huang1:31:48

       That's the part that's hard. You know, if NVIDIA was a, was a, was a ten billion dollar company trying to take NVIDIA's share, then it's easy to, to see for shareholders that, oh yeah, if they could just take ten percent share, they could be 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.

    5. LFLex Fridman1:32:12

       Mm-hmm.

    6. JHJensen Huang1:32:13

       You know? And so, so I think that that's one of the challenges for the world is, is, um, the imagination of the future. But I got plenty of time, and I'll keep reasoning about it, and I'll keep talking about it, and every s-single GTC will become more and more real.

    7. LFLex Fridman1:32:27

       Mm-hmm.

    8. JHJensen Huang1:32:27

       You know, and it-- and, and then more and more people will talk about it, and one of these days, you know, we'll, we'll get there. But I'm a hundred percent we'll get there.

    9. LFLex Fridman1:32:34

       Yeah, this view of, uh, you know, token factories, essentially, this token per second per watt and every token having value. Like it's an actual thing that brings value, and it brings different kinds of value, different amounts of value to different people, but it's value. That's the actual product. It's really could be loosely thought of as the token. And so you have a bunch of token factories, and then it's very easy, first principles, to imagine a future, given all the potential things that AI can solve, that you're going to need an exponential number more of token factories.

    10. JHJensen Huang1:33:05

        Yeah. [laughs] And, and what's really interesting, the reason why I was so excited about it, the iPhone of tokens arrived.

    11. LFLex Fridman1:33:11

        What do you call it? Wait, are you saying OpenClaw is the iPhone?

    12. JHJensen Huang1:33:13

        Yeah.

    13. LFLex Fridman1:33:14

        That's interesting. Uh-

    14. JHJensen Huang1:33:15

        Agents.

    15. LFLex Fridman1:33:16

        Yeah, agents. True.

    16. JHJensen Huang1:33:18

        Agents in general.

    17. LFLex Fridman1:33:19

        [laughs]

    18. JHJensen Huang1:33:19

        The iPhone of tokens arrived. Uh, it is the fastest growing application in history. It went straight up.

    19. LFLex Fridman1:33:25

        Yeah.

    20. JHJensen Huang1:33:25

        Went straight up.

    21. LFLex Fridman1:33:26

        That says something.

    22. JHJensen Huang1:33:27

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

    23. LFLex Fridman1:33:31

        Yeah, there's something truly, as you know, something truly special happening from about December, where people really woke up to the power of Claw code, of Codex-

    24. JHJensen Huang1:33:42

        Mm-hmm

    25. LFLex Fridman1:33:42

        ... of OpenClaw. Um, I mean, I've-- I'm embarrassed to admit that on the way here in the airport, I've-- [laughs] This is the first time I've done this in public. I was programming, quote unquote, by talking to my laptop.

    26. JHJensen Huang1:33:59

        [laughs] Exactly.

    27. LFLex Fridman1:33:59

        And I was embarrassed because I was pretending like I'm talking to a human colleague.

    28. JHJensen Huang1:34:02

        Mm-hmm.

    29. LFLex Fridman1:34:04

        Uh, I'm not sure how I feel about the future where everybody-

    30. JHJensen Huang1:34:06

        Mm-hmm
16. 1:34:39 – 1:48:25

    Leadership under pressure

    1. JHJensen Huang1:34:39

       [laughs]

    2. LFLex Fridman1:34:39

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

    3. JHJensen Huang1:35:38

       I'm conscious about the fact that, um, NVIDIA's success is very important to the United States. We generate enormous amounts of tax, tax revenues. Uh, we establish technology leadership for our nation. Technology leadership is important for national security. National security, not just in one aspect of national security, all aspects of national security. When our country is more prosperous, we could do a better job with domestic policies and helping social, social benefits. Because we're generating so much re-industrialization in the United States, we're creating mountains of jobs. We're helping shift, um, how we, how we, how we build things, uh, back to the United States in so many different plants, chips, computers, and of course, these AI factories. I'm completely aware that, that, um, and I have, I have the benefit, and this is a real, real, um, a real gift, uh, with, with, uh, mainstream investors, teachers, policemen who have somehow, for whatever reason, invested in NVIDIA or because they watched Jim Cramer, um, bought some stock and now are millionaires.

    4. LFLex Fridman1:36:57

       Mm-hmm.

    5. JHJensen Huang1:36:57

       And, um, uh, I am completely aware of that circumstance. I'm aware of the circumstance that, that NVIDIA, uh-Is central to a very large network of ecosystem partners behind us and downstream from us. And so the way, the way I deal with that is exactly what I just did. I reason about what is it, what is it that we're doing? Um, what is it causing? What's the impact it has on other people benefit, you know, positively or even, even, um, uh, through great burden, for example, to supply chain. Uh, and, and the question is, uh, therefore, what are you gonna do about it? In almost everything that I feel, I break it down. I reason about, okay, what's the circumstance? What is, what has changed? What's hard? Um, and what am I gonna do about it? And I'm-- I break it down, decompose the problem. And the de-de-decomposition of these circumstances turns it into manageable things that I can do. And the only thing that I-- after that I could do is, did you do it? Did you either do it or did you get somebody else to do it? And if you didn't do it, you, you reasoned that you need to do it, and you didn't do it, and you get it-- didn't get anybody else to do it, then stop crying about it, you know? And so, and so [chuckles] -

    6. LFLex Fridman1:38:27

       Yeah

    7. JHJensen Huang1:38:27

       ... so I, I'm, I'm fairly, I'm fairly, uh, uh, tough on myself, and but I also break things down so that, so that, um, uh, I don't panic. Uh, I can go to sleep because I've made the list of things that needed to be done, and I've made sure that everything that could put our company in harm's way, could put my partners in harm's way, put our industry in harm's way, I've told somebody. Everything that I feel could put anybody in harm's way, I've told someone. And I've told that someone who could do something about it. And so I've gotten it off my chest, or I'm doing something about it. And so after that, Lex, what else can you do?

    8. LFLex Fridman1:39:10

       So given all the in-insane, intense amount of suffering on y- the journey of building up NVIDIA, you-- Have you hit low points psychologically?

    9. JHJensen Huang1:39:22

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

    10. LFLex Fridman1:39:27

        And there-

    11. JHJensen Huang1:39:27

        All the time

    12. LFLex Fridman1:39:27

        ... you just break down the problem into pieces.

    13. JHJensen Huang1:39:31

        Yeah. You know-

    14. LFLex Fridman1:39:31

        See what you could do about it. [laughs]

    15. JHJensen Huang1:39:33

        And, and part of, and, you know, Lex, part of it, part of it is forgetting. One of the most important attributes of AI learning, as you know, is, right, systematic forgetting. You, you need to know when to forget some things. You can't memorize everything. You can't keep everything in, in, you know, you wanna ca- you don't wanna carry everything. One of the things that I do very quickly is I decompose the problem, I reason about the problem, and I, I share the load with it. When I say I tell everybody, I'm essentially sharing that burden-

    16. LFLex Fridman1:40:03

        Yeah

    17. JHJensen Huang1:40:04

        ... as quickly as possible.

    18. LFLex Fridman1:40:05

        Mm-hmm.

    19. JHJensen Huang1:40:06

        Whatever worries me, tell somebody else. Don't just keep it. You know, decompose-- Don't, don't freak them out. Decompose the problem into smaller parts and get people to so-- and, a-and inspire them to be able to go do something about it. But part of it is just, just forgetting. You know, I-- A lot of it is you gotta be tough on yourself. You know, you just, "Come on, stop crying about it. Let's get going." You know? And, and then you get out of bed. And then the other part is, is, um, y-you, you, you, you're attracted to the next shiny light, the next future, you know, the next opportunity, the next, "Okay, that's behind us. Let-- What's next?" And it's a lot-- I think, you know, you watch this with great athletes. They, they, um, just worry about the next point.

    20. LFLex Fridman1:40:49

        Mm-hmm.

    21. JHJensen Huang1:40:50

        The last point is behind them. The embarrassment, the, you know-

    22. LFLex Fridman1:40:55

        [laughs]

    23. JHJensen Huang1:40:55

        ... the setback.

    24. LFLex Fridman1:40:56

        Yeah. All of it. [laughs]

    25. JHJensen Huang1:40:56

        You know, and, and then and because I do so much of my job publicly. You know, Lex, you do a fair amount of your job publicly too. And so, so I do a lot of my job publicly. And so, um, you know, I, I say a lot of things that, that seem sensible at the time or funny at the time. Mostly it's just because it's funny to me at the time. And then y-you know, you reflect on it, it's less funny. But, but [laughs]

    26. LFLex Fridman1:41:20

        Yeah. No, trust me, I know.

    27. JHJensen Huang1:41:22

        [laughs]

    28. LFLex Fridman1:41:22

        But you basically allow yourself to be pulled by the light of the future.

    29. JHJensen Huang1:41:26

        Yeah.

    30. LFLex Fridman1:41:26

        Forget the past and just keep-
17. 1:48:25 – 1:55:16

    Video games

    1. LFLex Fridman1:48:25

       I'm glad you maintained that same spirit of Denny's, um, the, the work. I mean, that, that was beautiful. Your whole journey from starting from Denny's is a beautiful one. Uh, let me ask you about video games. So I'm a big gaming fan.

    2. JHJensen Huang1:48:40

       Yeah.

    3. LFLex Fridman1:48:41

       So I have to say thank you to NVIDIA for many years of incredible graphics. Um-

    4. JHJensen Huang1:48:47

       By the way, it, it is, GeForce is our still, to this day-

    5. LFLex Fridman1:48:51

       Yeah

    6. JHJensen Huang1:48:51

       ... our number one marketing strategy, right? People learn about NVIDIA while they're in their teenage years.

    7. LFLex Fridman1:48:59

       Mm-hmm.

    8. JHJensen Huang1:49:00

       And then they go to college, and they know who NVIDIA is, and they, um, and then in the beginning it's just, you know, playing Call of Duty, you know, you know, Fortnite, and then later they're using CUDA, and then later they're using NVIDIA and, you know, Blender and Dassault and Autodesk.

    9. LFLex Fridman1:49:16

       Yeah. I mean, I should say I, I, I mentioned to a friend that I'm... uh, talking with you. He said, "Oh, they make [laughs] great gaming GPUs."

    10. JHJensen Huang1:49:25

        Yeah, exactly.

    11. LFLex Fridman1:49:26

        [laughs] It's like-

    12. JHJensen Huang1:49:26

        Exactly

    13. LFLex Fridman1:49:27

        ... you know, there's-

    14. JHJensen Huang1:49:28

        [laughs]

    15. LFLex Fridman1:49:28

        ... there's more to it, but-

    16. JHJensen Huang1:49:29

        [laughs]

    17. LFLex Fridman1:49:30

        But yeah, yeah, people really love the-- It really brought a lot of joy to a lot of people. The, the, the hardware really brings these worlds to life. Uh, there was some controversy around this, uh, with DLSS 5.

    18. JHJensen Huang1:49:43

        Yeah.

    19. LFLex Fridman1:49:44

        Can you explain to me the drama around this? Uh, I guess people, gamers online were concerned that it makes games look like AI slop.

    20. JHJensen Huang1:49:53

        Yeah.

    21. LFLex Fridman1:49:54

        Uh, what do you think of this drama?

    22. JHJensen Huang1:49:56

        Yeah. Uh, I think their, their perspective makes sense, and I can see where they're coming from because I don't love AI slop myself. You know, all, all of the, the AI-generated content increasingly, um, looks similar, and they're all beautiful, and, and I can-- So I can-- I, I'm empathetic towards what they're, what they're thinking. Um, that's just not what DLSS 5 is trying to do. I showed several examples of it. But DLSS 5 is 3D-conditioned, 3D-guided. It's ground truth structured data-guided, and so, so the artist determined the geometry. We are completely truthful to the geometry, maintain so in ev-every single frame. Um, it's, uh, conditioned by the textures, the artistry of the artist, and so every single frame, it enhances, but it doesn't change anything. Now, the question is, the question about enhancing, DLSS 5 also lets, because it's, the system is open, you could train your own models to determine, and you could even, in the future, prompt it. You know, I want it to be a toon shader. I want it to look like this kind of... You know, so you can give it even an example, and it would generate in the style of that, all consistent with the artistry, you know, the style, the intent of the artist. And so all of that is done for the artist so that they can create something that is more beautiful, um, but still in the style that they want. I think that they got the impression that the, the games are gonna come out the way the games are, ship the way they do, and then we're gonna post-process it. That's not what DLSS is intended to do. DLSS is integrated with the artist, and so it's, it's about giving the artist the tool of AI, the tool of generative AI. They could decide not to use it, you know.

    23. LFLex Fridman1:52:01

        I think people are very sensitive to human faces.

    24. JHJensen Huang1:52:03

        Yeah.

    25. LFLex Fridman1:52:04

        And we're now living in this moment, which I think is a, is a beautiful one, which is people are sensitive to AI slop.

    26. JHJensen Huang1:52:10

        Yeah.

    27. LFLex Fridman1:52:11

        It, it puts a mirror to ourselves to help 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.

    28. JHJensen Huang1:52:24

        Mm-hmm.

    29. LFLex Fridman1:52:25

        And that's beautiful.

    30. JHJensen Huang1:52:25

        Yeah.
18. 1:55:16 – 1:57:29

    AGI timeline

    1. LFLex Fridman1:55:16

       what's, uh-- You've said, I think accurately, that the AGI timelineQuestion rests on your definition of AGI. So l-let's, let me ask you about possible timelines here. Let's-- this ridiculous definition perhaps of what AGI is, but a, an AI system that's able to essentially do your job. So run... No. Start, grow, and run a successful technology company that's worth-

    2. JHJensen Huang1:55:52

       A good one or a one?

    3. LFLex Fridman1:55:54

       No.

    4. JHJensen Huang1:55:54

       [laughs]

    5. LFLex Fridman1:55:54

       It has to, it has to be worth more than a billion, m-more, more than a billion dollars. So, you know, you know how hard it is to do all of those components. So how far are we away from that? So we're, we're talking about Open Claw that does all the incredibly complex stuff that are required to sc-- to, first of all, innovate, to find customers, to sell to them, to, to manage, to build a team of some agents, some humans, all that kind of stuff. Is this five, ten, fifteen, twenty years away?

    6. JHJensen Huang1:56:31

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

    7. LFLex Fridman1:56:35

       You think you can have a company run by an AI system like this?

    8. JHJensen Huang1:56:37

       Possible. And the reason for that is this. You said a billion, and you didn't say forever. And, and so for example, uh, it is not out of the question that, uh, a claw was able to create a web service, some interesting little app that all of a sudden, you know, a few billion people used for fifty cents, and then it went out of business again shortly after. Now, we saw a whole bunch of those type of companies during the internet era, and most of tho-those websites were not anything more sophisticated than what Open Claw could generate today.

    9. LFLex Fridman1:57:20

       Interesting. Ach-achieve virality and monetize that virality.

    10. JHJensen Huang1:57:23

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

19. 1:57:29 – 2:11:01

    Future of programming

    1. JHJensen Huang1:57:29

       you know? And-

    2. LFLex Fridman1:57:30

       You're gonna get a lot of people excited with that statement. [laughs]

    3. JHJensen Huang1:57:32

       Yeah, I know. [laughs]

    4. LFLex Fridman1:57:33

       They're just like, "What do you mean? I can, I can just, uh, launch an agent and, um, make a lot of money."

    5. JHJensen Huang1:57:38

       Well, by the way, it's happening right now, right? You know that when, when you go to China, uh, you're gonna see, you're gonna see, um, a whole bunch of people, uh, teaching their... getting their claws to try to go out and look for jobs and, you know, [laughs] do work, make money. And, and I, I'm not, I'm not actually... I wouldn't be surprised if some social thing happened or somebody created a, a digital influencer, super, super cute, um, or some social application that, you know, feeds your little Tamagotchi or something like that and, and it become an out of the blue, an instant success. A lot of people use it for a couple of months, and it kind of dies away. Now, the odds of, of, of, you know, a hundred thousand of those agents, um, building NVIDIA is zero percent. And, and then, and then the, the one part that I will, I will do, um, and I, and I, I wanna make sure we all do, is to recognize that people are really worried about their jobs. And, and, um, I just want to remind them that the purpose of your job and the tasks and the tools that you use to do your job are related, not the same. I've been doing my job for thirty-three years. I'm the longest running tech CEO in the world, thirty-four years. And the tools that I've used to do my job has changed continuously in the last thirty-four years, and sometimes quite dramatically, you know, over the course of couple, two, three years. And, and the o- the, the one story that I, I, I really wanna make sure that everybody hears is the story the, the first job that ev- that computer scientists said, AI researchers said was gonna go away was radiology because computer vision was going to achieve superhuman levels, and it did. CV-- Computer vision was superhuman in twenty nineteen, twenty... maybe sh- maybe a little bit l-later, twenty twenty.

    6. LFLex Fridman1:59:38

       Mm-hmm.

    7. JHJensen Huang1:59:39

       Okay? And so it's been a long time since computer vision has been superhuman. And so the prediction was radiologists would go away because studying radiology scans was thing of the past. AI will 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. And so the question is why? And we now have a shortage of radiologists in the world. And so, one, the alarmist warning went too far, and it scared people from doing this profession that is so important to society. And so it did harm. Now, why was it wrong? The reason why is because the purpose of a radiologist, the purpose is to diagnose disease and help patients and doctors diagnose disease. And because we're able to study scans so much faster now, you could study more scans, you could diagnose better, you could, you could, um, inpatient faster. We can see people more. The hospitals are making more money. You have more patients in the hospital. You need more radiologists. I mean, the, the amazing thing is it's so obvious this was gonna happen. The number of software engineers at NVIDIA is gonna grow, not decline. And the reason for that is because the purpose of a software engineer and the task of a software engineer for 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, you know? But their job, their purpose of their job didn't change. Solving problems, working as a team, diagnosing problems, evaluating the result.Looking for new problems to solve, innovation, connecting dots, you know, none of that stuff is gonna go away.

    8. LFLex Fridman2:01:39

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

    9. JHJensen Huang2:01:45

       Yes. And the reason for that is this. What is the definition of coding? I believe that the definition of coding as of today is simply specifying specification, and maybe if you want to be rather directive, you could even give it an architecture of the software that you're-- you wanted to write. So the question is, how many people could do that? Describe a specification for a computer to go-- telling the computer what to go build. How many people? I think we just went from thirty million to probably one billion. And so every, every carpenter in the future will be a coder, except a carpenter with AI is also an architect. They just increased the value that they could deliver to the customer. Their, their artistry just elevated tremendously. I believe that every accountant is, you know, also your financial analyst, also your financial advisor. So all of these professions have just been elevated. And if I were a carpenter, I sees AI-- I see AI, I would just completely go berserk. You know? The services I can bring to my clients, if I were a plumber, completely go berserk.

    10. LFLex Fridman2:03:04

        And the, the people that are currently programmers and software engineers, I think they're 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.

    11. JHJensen Huang2:03:20

        That's right. Exactly.

    12. LFLex Fridman2:03:20

        So the, the-- over time, they'll converge, but I think, uh, there's still value in getting, I think, uh, learning how to program, like learning what programming languages [chuckles] are, uh, the old, the old kind of programming. Uh, what, what are good practices for programming languages? What are design principles for programming-

    13. JHJensen Huang2:03:39

        That's right

    14. LFLex Fridman2:03:40

        ...uh, languages for large software systems?

    15. JHJensen Huang2:03:43

        And, and the reason for that, Lex, and you, you know that I was just saying for the audience, I think the goal of, the goal of specification, the artistry of specification, the goal and the artistry of it, um, is going to depend on what problem you're trying to solve. When I'm thinking, when I'm thinking about giving the company strategies and, um, formulating corporate directions and things that we should do, um, I describe it at a level that is sufficiently specific that people generally understand the direction, and it's actionable. They-- It's so specific enough that they can take action on it. But I under-specify it on purpose so that enable forty-three thousand amazing people to make it even better than I imagined. And so when I'm working with engineers and when I'm working with people, um, I think about who-- what problem am I trying to solve? Who am I working with? And the level of specification, the level of architecture definition relates to that. And, and so everybody's gonna have to learn how-- where in the spectrum of coding they wanna be. Writing a specification is coding. And so you might decide to be quite prescriptive because there's a very specific outcome you're looking for. You might dec-decide that, you know, this is an area you wanna be much more exploratory. And so you might under-specify and enable you to go back and forth with the AI to even push your own boundaries of creativity. And so this artistry of where you are in the spectrum, this is the future of coding.

    16. LFLex Fridman2:05:31

        But just to linger on it outside of coding, I think a lot of people, rightfully so, uh, are worried about their jobs, have a lot of anxiety about their jobs, especially in the white-collar sector. Um, I don't think any of us know what to do, uh, with tumultuous times that always come when automations and new technology arrives. And I just-- First of all, I think, um, we all need to have compassion and the responsibility to feel sort of the burden of what the actual suffering feels like for individual people and families that lose their job. I think whenever you have transformative technology like that's coming with, uh, with artificial intelligence, there's going to be a lot of pain. And I don't know what to do about that, uh, pain. Hopefully, it creates much more opportunities for those same people, uh, for the same kind of job as, uh, the tooling evolves and makes them more productive and makes it more fun, hopefully, as it does in the programming. I've, I have-- I've been having so much fun programming, I have to say. Like, I've never had this much fun. So hopefully it makes their job-- automates the boring parts and makes the creative parts, uh, the ones that the, the human beings are responsible for. But still there's going to be a lot of pain and suffering.

    17. JHJensen Huang2:06:51

        So my first recommendation before-- And this is now how I deal with anxiety. In fact, we just talked about it earlier.

    18. LFLex Fridman2:06:58

        Mm-hmm.

    19. JHJensen Huang2:06:59

        Enormous anxiety about the future, enormous anxiety about the pressure, enormous anxiety about uncertainty. I first break it down, and then I'm gonna tell myself, "Okay, there are some things you can do something about. There's some things you can't do anything about. But for the stuff that you can do something about, let's reason, reason about it, and let's go do it."

    20. LFLex Fridman2:07:19

        Yeah.

    21. JHJensen Huang2:07:20

        If we were to hire s-- a new college graduate today, and I have a choice between two, one that have-- that has no clue what AI is and one that is expert in using AI, I would hire the one who's expert in using AI.If I had an accountant, a marketing person, the one that is expert in using AI, supply chain, customer service, a salesperson, business development, a lawyer, I would hire the one who is expert in using AI. And so I would, I would advise that every college student, every, every teacher should encourage their student to be, to go use AI. Every college student should graduate and be an expert in AI. And every-everybody, if you're a carpenter, if you're, you know, electrician, go use AI. Go see what it can do to transform your current job. Elevate yourself. If I were a farmer, I would absolutely use AI. If I were a pharmacist, pharmacist, I would use AI. I wanna see how, what it could do to elevate my job so that I could be the, I could be the innovator to revolutionize this industry myself.

    22. LFLex Fridman2:08:35

        Mm-hmm.

    23. JHJensen Huang2:08:36

        And so that, that would be the first thing that I would do. And, and then I would also, I would also help them, um, it, it is the case that the technology will dislocate and will eliminate many tasks if-- and because it will automate it. If your job is the task, if your job is the task, then you're very highly going to be disrupted. If your job's purpose includes u-certain tasks-

    24. LFLex Fridman2:09:08

        Mm-hmm

    25. JHJensen Huang2:09:08

        ... then it, it's vital that you go learn how to use AI to automate those tasks. And then there's the world of spectrum in between.

    26. LFLex Fridman2:09:14

        And by the way, the beautiful thing about AI, so the, the, the chatbot versions, is you can break down, you have anxiety, and you can break down the problem by talking to it.

    27. JHJensen Huang2:09:27

        [laughs]

    28. LFLex Fridman2:09:28

        Like I've, I've recently-- It's really just incredible how much you can think through your life's problems and through-- And I don't mean like therapy problems. I mean like very practically, "Okay, I'm worried about my-- Literally, 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 can literally ask, and it's going to give you-

    29. JHJensen Huang2:09:47

        Yeah

    30. LFLex Fridman2:09:47

        ... a point-by-point plan. I mean, it's just a great life coach, period. This-
20. 2:11:01 – 2:17:22

    Consciousness

    1. LFLex Fridman2:11:01

       Uh, do you think there are some things about human nature, about human consciousness that is fundamentally non-computational? Maybe something a chip, no matter how powerful, uh, can never replicate?

    2. JHJensen Huang2:11:18

       I don't know if the chip will ever get nervous. And that's the, you know, of course, the conditions by which, uh, that causes anxiety or nervousness or whatever emotion, um, I believe that AI will be able to recognize those and understand those. I don't think my chips will feel those. And therefore, the-- how, how that anxiety, how that feeling, how that excitement, how that, how that, you know, all of those feelings manifest in human performance. For example, extremely am-amazing human performance, athletic performance, you know, average or lesser than average. Um, that ex- that entire spectrum of human performance that comes out of exactly the same circumstances for different people manifesting in different outcome, manifesting in different performance, I, I don't think there's anything about anything that we're building that would suggest that two different computers being presented with all of exactly the same context would perfo- of course, it would produce statistically different outcomes, but it's not because it felt different.

    3. LFLex Fridman2:12:34

       Yeah, the subjective... Boy, there's something truly special about the subjective experience that we humans feel.

    4. JHJensen Huang2:12:43

       Yeah.

    5. LFLex Fridman2:12:43

       Like I mentioned to you, I was, I was, I was pretty nervous talking to you, like I mentioned to you, that 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-

    6. JHJensen Huang2:13:08

       Yeah

    7. LFLex Fridman2:13:09

       ... think AI being able to, a computational device being able to do that. But there are so many mysteries about this whole thing that we're yet to uncover that I am open to be surprised. [laughs]

    8. JHJensen Huang2:13:21

       Yeah.

    9. LFLex Fridman2:13:21

       I've been surprised a lot over the past-

    10. JHJensen Huang2:13:23

        Yeah

    11. LFLex Fridman2:13:23

        ... few months and few years. Scaling can create some incredible miracles in the space of intelligence.

    12. JHJensen Huang2:13:29

        Mm-hmm.

    13. LFLex Fridman2:13:31

        Ha-has been truly marvelous to watch.

    14. JHJensen Huang2:13:32

        And-

    15. LFLex Fridman2:13:33

        So I'm open to surprise.

    16. JHJensen Huang2:13:34

        And, and it's just really important to un- to, to break down what is intelligence. And, you know, the word, that word we use all the time, it's not a my-mysterious word. Intelligence has a meaningYou know?

    17. LFLex Fridman2:13:46

        Mm-hmm.

    18. JHJensen Huang2:13:46

        And, and it's a system that, you know, it's, it, it's something that we do that in-includes perception and understanding and reasoning and the ability to plan and, you know, that, that loop, that loop is, is, um, the-- fundamentally what intelligence is. Intelligence is not one word that is exactly equal to humanity, and that's, I think it's really important to separate the two. We have two words for that. I'm not-- I don't overfantasize about, and I don't over-romanticize about intelligence. Intelligence is-- a-and, uh, 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 intelligent people more intelligent than I am in each one of the s-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-- in any w- in the sp-fields that they're in, all of them. I have sixty of them. They're all superhuman to me.

    19. LFLex Fridman2:14:58

        Mm-hmm.

    20. JHJensen Huang2:14:59

        And somehow I'm sitting in the middle orchestrating all sixty of them. And so you gotta ask yourself, what is, what is it about a dishwasher that allows that dishwasher to sit in the middle of superhumans?

    21. LFLex Fridman2:15:13

        [chuckles]

    22. JHJensen Huang2:15:13

        Does that make sense? And so-

    23. LFLex Fridman2:15:15

        Yeah.

    24. JHJensen Huang2:15:15

        But that's my point. My point is intelligence is a, is a functional thing. Humanity is not a, not specified functionally. It's a much, much bigger word. And, and our life experience, our tolerance for pain-

    25. LFLex Fridman2:15:34

        Mm.

    26. JHJensen Huang2:15:34

        Our determination, those are, those are different words than intelligence.

    27. LFLex Fridman2:15:38

        Mm-hmm.

    28. JHJensen Huang2:15:39

        And so the, the thing that I, I wanna help the audience understand, if I could give them one thing, is, is intelligence is a word that we've elevated to very high form over time.

    29. LFLex Fridman2:15:50

        The, the word we should really elevate is humanity.

    30. JHJensen Huang2:15:53

        Character, humanity-
21. 2:17:22 – 2:25:55

    Mortality

    1. LFLex Fridman2:17:22

       so much of the success of NVIDIA and, um, the lives of millions of people that I mentioned, uh, depend on you, uh, 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?

    2. JHJensen Huang2:17:42

       I really don't wanna die. Um, I have a great life. I have a g-great family. I, I have really important work. Uh, this is, this is not a once in a, once in a lifetime experience suggests that it has been experienced by many people, just not one person. Uh, this is a once in a humanity experience, what I'm going through. Uh, NVIDIA is one of the most consequential technology companies in history. We're doing very important work. I take it very seriously. Um, and, and so some of the, some of the things that, that of course are, are practical things, like how do we think about succession planning? And, and, um, I, I'm famous in saying that I don't believe in succession planning.

    3. LFLex Fridman2:18:35

       [laughs] Man.

    4. JHJensen Huang2:18:36

       And, and the reason, the reason for that, the reason for that isn't because I'm immortal. Um, the reason for that is because if you're worried about succession planning, if you're worried-- all that anxiety of succession planning, then what should you do about it? Then 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, experience as often and continuously as you can, which is the reason why I continuously reason about everything in front of my team. Every single meeting is about a reasoning meeting. Every moment I spend inside a company, outside a company, is about passing on knowledge to people as fast as I can. Nothing I learn ever sits, sits on my desk longer than, you know, a fraction of a second. I'm passing that information, that know-- 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 gonna wanna, you're gonna wanna learn this." And so y- I'm constantly passing knowledge, empowering people, elevating the capability of everybody around me so thatUm, the outcome that I, that I seek, that I hope for is that I die on the job, you know? And, and hopefully I die on the job instantaneously, you know, [laughs] and, and there's no long periods of suffering, you know? And so [laughs]

    5. LFLex Fridman2:20:06

       Well, from a fan perspective, uh, given your, your, uh, uh, extremely, um, your enormous positive impact on civ- on civilization, of course, I hope you keep going. But also it's just fun to watch [laughs] what NVIDIA's doing in your keynote. It's just the rate of innovation, and I'm a huge fan of engineering. It's so much incredible engineering is continuously being done by NVIDIA. It's just fun to watch. It's a celebration of humanity. It's a celebration of great builders. It's a celebration of great engineering. So it represents something special. Uh, so I hope, uh, you and NVIDIA keep going. What gives you hope about this whole thing we got going on, about humanity, about the future of humanity? When you look out-- and you think about the future quite a bit. When you look out ten, twenty, fifty, hundred years from now, what gives you hope?

    6. JHJensen Huang2:20:56

       I, I've always had, I've always had, uh, uh, great confidence in, in the, in the kindness, uh, the generosity, uh, um, the compassion, the human capacity. I've always been extremely confident of that. Sometimes, um, more so than I should.

    7. LFLex Fridman2:21:25

       Mm-hmm.

    8. JHJensen Huang2:21:25

       And, and I, I get taken advantage of, but it doesn't, it doesn't ever cause me not to. I start with always, uh, that, that people want, want to do good. People want to, um, uh, help others. And, uh, vastly I am proven right, constantly proven right, and, and often, uh, exceeds my expectations. And, and so I have complete confidence in the human capacity. I think the, the, the thing that-- the things that give me incredible hope, uh, is what I see as, as I extrapolate, as I-- what I see now is p-possible, and as I extrapolate, um, based on the things that we're doing, what will very likely happen.

    9. LFLex Fridman2:22:21

       Mm-hmm.

    10. JHJensen Huang2:22:22

        And, and, um, uh, and that there's so many things that we want to solve. There's so many problems we want to solve. There's so many things that we want to build. There's so many good things that we want to do that are now within our reach and within the reach of my, my lifetime. You just can't possibly not be romantic about that.

    11. LFLex Fridman2:22:44

        [laughs]

    12. JHJensen Huang2:22:45

        You know what I'm saying?

    13. LFLex Fridman2:22:46

        Yeah. What an exciting time to be alive.

    14. JHJensen Huang2:22:48

        Yeah.

    15. LFLex Fridman2:22:48

        Like, truly, truly so.

    16. JHJensen Huang2:22:50

        How can you not be romantic about, about, about that? The, the, the fact that, that there is a, there-- i-it's a reasonable thing to expect the end of disease. It's a reasonable thing to expect. 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.

    17. LFLex Fridman2:23:16

        Mm-hmm.

    18. JHJensen Huang2:23:17

        And then, you know, f- not, not for long distances, but short distances. You know, and people ask me how. And well, first of all, very soon w- I'm gonna put a humanoid on a spaceship, and it's gonna be, you know, my humanoid, and, and we're gonna send it out as soo- you know, as soon as possible, and it's gonna keep improving and enhancing along the flight.

    19. LFLex Fridman2:23:36

        Mm-hmm.

    20. JHJensen Huang2:23:36

        And then when it's time, all of the-- all of my consciousness has already been, you know, uh, so much of my life has been uploaded in the internet. Take all my inbox, take everything that I've done, everything I've said, you know, it's been coll- and b-becoming my AI, and, um, I'm gonna, you know, when the time comes, you know, we'll just send that at the speed of light, catch up with my robot. [laughs]

    21. LFLex Fridman2:23:59

        [laughs] Oh, that's brilliant. I, I mean, but for me, that's sort of application focus. [laughs]

    22. JHJensen Huang2:24:04

        [laughs]

    23. LFLex Fridman2:24:04

        But also for me, the curiosity-

    24. JHJensen Huang2:24:07

        [laughs]

    25. LFLex Fridman2:24:07

        -uh, maxing perspective, I just-- all of those mysteries. There's so much-

    26. JHJensen Huang2:24:12

        Yeah

    27. LFLex Fridman2:24:12

        ... fascinating scientific questions there.

    28. JHJensen Huang2:24:14

        Understanding the biological machine is a r-- is right around the corner. It's, it's not ten years. It's five years probably.

    29. LFLex Fridman2:24:20

        And the neurobiological machine, the, the human mind, and cracking physics, theoretical physics open. It's so exciting.

    30. JHJensen Huang2:24:26

        Explaining consciousness, that one would be awesome.

Episode duration: 2:25:58