How Jensen Huang Built the World's Most Valuable Company with Stephen Witt, Author of "The Thinking Machine"

(Intro Narrator at 00:00:00) Today, we're talking to Stephen Witt, author of The Thinking Machine, about his most important takeaways from his six interviews with Jensen Huang. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:17) So tell me a little bit, what is this book, The Thinking Machine, and how did you get to write it with Jensen?

(Stephen Witt at 00:00:24) Yeah. So the book is just a history of NVIDIA, and that turns out to be a biography of Jensen because Jensen is NVIDIA. He founded the company in 1993. He's been the only CEO. It's his show. It's him. So I ended up kind of—I thought I was writing the history of a corporation. I inadvertently ended up writing a biography of Jensen. Well, that turned out to be even better because Jensen's just a fascinating person. He's an unbelievable person. He's unreal. I mean, he's arguably the most powerful man in AI. He might be the most powerful man alive. He tells Trump what to do. So that was great.

(Stephen Witt at 00:01:00) How did I get in with him? Well, you know, I cheated and lied and bluffed. First, I needed to get a magazine profile written of him. I wanted to write about AI. I wanted to write about NVIDIA, but I needed a way in. So I told the New Yorker magazine that Jensen had agreed to sit for a profile. I told Jensen that the New Yorker wanted to do a profile on him. I was actually lying to both sides. They'd never talked to each other or mentioned it to me. So I kind of manifested that, but it worked.

(Joel Beasley at 00:01:30) You brokered a deal, sir.

(Stephen Witt at 00:01:31) I brokered a deal. Yeah. Yeah. I learned that from sales, actually. I learned that from watching a video for entrepreneurs about how to land a sale. That works. So then, but I also knew that Jensen was worthy of a New Yorker profile. I could see that his stock was on the rise. And I also knew that the New Yorker was the place for Jensen because he was such a fascinating person, you know, this titan of industry. He needed a real magazine treatment.

(Stephen Witt at 00:02:05) Things that had been written about him before had either glazed him or alternatively were hatchet jobs. It was weird. Nobody had really peeled behind the scenes to see what was going on. So I felt both sides—there was a huge amount of value for it. For Jensen in particular, when I prepared for our first time, what was originally supposed to be our only interview, I only had one hour. You know, Jensen, even then, his time was one of the most valuable commodities in the world. One minute of Jensen's time is worth a million dollars, probably. And I knew I couldn't waste his time.

(Stephen Witt at 00:02:38) I like to do a lot of research before I ever interview somebody, but for Jensen, it was overdrive. I think I did three or four times the amount of research I normally do. I read every single New York Times article that mentioned NVIDIA going back to its IPO in 2001. It was 700 articles. And then I went on YouTube and I found every single thing I could find about Jensen going way back into the past. I mean, stuff from the first days of YouTube that was posted on there. Fourteen, fifteen year old videos. And I made a list of every question that every journalist I could find had ever asked Jensen. And then I made my list of questions which I was certain to have zero overlap with what he'd already been asked. So that when I sat down with him, every question I asked him was completely fresh and new, and he'd never heard them before.

(Stephen Witt at 00:03:23) In particular, I had been watching this Taiwanese program called Boss Talk where he'd been interviewed half in Chinese and half in English, way back in 2009, back in the gaming days. And he'd mentioned in passing that he'd once almost died in a car accident, but he didn't follow up on the story, and the interviewer didn't ask any more questions about it. So that was my first question. Tell me about the time you almost died in a car accident. He kind of perked up and looked at me. He was like, you heard about that? It showed him I'd done my homework. I didn't know this at the time, but this was perfect. Jensen loves it when people do their homework.

(Stephen Witt at 00:04:02) He's obsessed with doing his homework himself. He calls it his superpower. And all of the executives in NVIDIA, when they meet with him, have to be so prepared. They can't waste his time. So this turned out to be—almost inadvertently, I got lucky—the way to make Jensen like me. Because I wasn't asking him questions he'd ever heard before. I was advancing the conversation. And after that first hour, I was actually invited back for a second hour. And after that, it turned into six interviews total, plus, you know, kind of privileged access to a lot of his friends and coworkers and executives in NVIDIA. And this allowed me to draw a much richer and more kind of three-dimensional portrait than I thought I might get.

(Stephen Witt at 00:04:42) That was fantastic. It turned out to be the secret sauce of the book, because Jensen is just so different. He's so different than your average CEO. And I think that shows in the fact that, you know, he's built the most valuable company in the world. And I think, by some measures, the single most valuable company in human history. They're worth more than Standard Oil at their peak. That was a long answer.

(Joel Beasley at 00:05:05) No, no, no. It's good. This is a talk show. We're talking. That's fascinating. Six times.

(Stephen Witt at 00:05:12) Yeah.

(Joel Beasley at 00:05:13) What were you doing during these? Are we just hanging out eating grapes? Were you in a structured—

(Stephen Witt at 00:05:18) Formal. I wish we could have hung out. What I wanted to do was not interview him. What I wanted to do was shut up and follow him around. They wouldn't let me. I would have loved to just be a fly on the wall during his workday and say nothing. That would have been the dream. You can't imagine how this guy works. He's not a conventional boss. He has an office, but he never uses it. He just throws books in there. He spends all his time in the conference room in the center of the company. It's like a panopticon. It's like a brain stem.

(Stephen Witt at 00:05:45) He wants to be right up front of the action. You know, just go and talk to the most low-level employees in the company. I mean, invite them into meetings. College intern, a guy who's been working at NVIDIA for maybe six months—he'll treat him like he's a lifelong friend. It's fascinating to watch. Most bosses don't do this. They have a hierarchy of responsibilities so they can manage their workflow. Jensen has 65 direct reports. He's just got his finger on the pulse of the company at all times.

(Stephen Witt at 00:06:12) He has everyone in the company write him emails every week with the five things they worked on. He'll sit there at his desk and just pore through them. You know, he's got 30,000 employees. I mean, he doesn't read them all, but he samples them.

(Joel Beasley at 00:06:23) He uses AI for sure now.

(Stephen Witt at 00:06:24) Yeah, probably for sure. He has an AI digest now and he uses, you know, stuff from low-level employees. Right? Stuff from the guy on the front line, the sales guy, the guy who works in accounting. He's reading it. He wants to know what that guy's challenges and problems are. I've never met someone who had so close of a kind of affinity with everything that was going on. Jeff Bezos said the same thing. He wanted a completely flat organizational structure.

(Stephen Witt at 00:06:48) Or I think the rule at Amazon was that there could only be four levels of management between the lowest worker and Jeff Bezos. Okay? That was it. They couldn't have longer hierarchical chains because information would get lost. With Jensen, it's basically zero. Anyone in the company can email him at any time, and in fact, they're encouraged to.

(Joel Beasley at 00:07:10) You know, I've seen that behavior. And now that you've brought it up and I'm thinking about it, billionaires that I've met, interviewed, or met in person tend to have those qualities where they make you feel like they've known you forever, that you're the most important person in the room, that everything's on you, and they're just kind. And it's like, whoa. How did they become that type of person? It's almost admirable for you to strip the money away just to be that type of person.

(Stephen Witt at 00:07:38) It's a superpower. I emailed Mark Cuban once on a story I was writing. He wrote me back in four minutes. I was shocked. I was like, what? You know, normally, you try to contact somebody, you go through this slow army of PR agents and goldbrickers doing nothing. I think for these guys, they understand the value of speed and responsiveness so quickly, especially in the tech industry. If you lose six hours, it's like losing ten years. You know?

(Joel Beasley at 00:08:03) You're absolutely right. One of the first times I saw this was when the podcast was super new. Yeah, brand new, or three months into it or whatnot. And I emailed the CTO of Verizon. Direct emailed him. And I have a Verizon phone number, and it was in my signature. Within three minutes of me sending him the email, or whatever it was, five minutes, he called me. And he's like, hey. I was like, shut up. This is crazy. And he's like, I'm actually becoming the CEO next week. The new CTO is Kyle, and I'll introduce you to Kyle. Have him on your podcast.

(Joel Beasley at 00:08:35) We don't tell anybody. It's not public.

(Stephen Witt at 00:08:36) I mean, look. These guys need filters. They're bombarded by idiots and flatterers and, you know, people pitching dumb business ideas all day. So they do need some kind of frontline intake. But you'd be amazed. Jensen in particular, he said this, you know, he's so busy, and he is. But he often says, I have nothing but time. And what he means is if he's talking to someone, that person has a good idea, he will postpone the rest of his workday. You know, he's going to be flexible. Jensen spends, and this is really his secret, and the thing that I think really differentiates him—

(Stephen Witt at 00:09:07) He spends all his time talking to scientists. He's not really that interested in, you know, where the industry per se is going. He's interested in what some scientist wants to do. So he'll go to a quantum physicist or weather forecaster or, you know, who knows? Any kind of scientist. And he'll say, what could you do if you had a million times more computing power than you do right now? What might you accomplish? And then, he will take his R&D team and build that person a tool. It'll cost them tens of millions of dollars for one guy to use it. They lose money on it.

(Stephen Witt at 00:09:43) At least initially. But, the secret is, if you do this enough, you might unlock some new branch of science. And once that happens, a whole industry grows up around your tools. They have to use them. This is what happened with AI. I mean, it's why they dominate the space. When they were talking to the neural net guys in the early days, I mean, God, there was no interest at all. 2010, the amount of venture capital money going into AI was closer to zero than any other meaningful number. Even NVIDIA ignored the AI guys for a while. They didn't take them seriously. But once they saw what was happening, Jensen pivoted wholeheartedly to that sector, and basically said, okay, we're going to take all the money from the gaming profits that we're doing with these video game accessories, and build it into building AI hardware.

(Stephen Witt at 00:10:27) At the time, there was no AI hardware market. He invented it. It didn't exist. You know, so when people say, well, he has 90% of that market. Well, first of all, he made 100% of the market. It was the only such device in the world.

(Joel Beasley at 00:10:40) Was you writing this book and diving into NVIDIA your first time in the enterprise AI space? What's your—

(Stephen Witt at 00:10:48) Yeah. I mean, I had not—I have a funny history with AI because I thought I was going to be a computer scientist when I was young. And I just didn't—I don't have a mind for it. I mean, it's not even that I don't have a mind for it. I just didn't like it. I got bored sitting there programming a computer. It was the nineties too. Computers were nightmarish to use. It's not like today. But I interned at Argonne National Laboratory, and they put me—I wanted—I applied to be in the AI group, and I didn't get in.

(Stephen Witt at 00:11:20) This is a funny story. And the AI group was working on kind of automated symbolic logic, kind of what we would call expert systems. And that shit basically never worked. Pardon my French. Can I swear on the podcast?

(Joel Beasley at 00:11:32) Oh, you can do whatever you want.

(Stephen Witt at 00:11:32) That stuff basically never worked, but I was entranced by it. Meanwhile, they stuck me with this group that was doing something called automatic differentiation. And what we do here is we take the programming code and we compile it into low-level code, and then we run calculus on the low-level code to produce new results. I thought this was interesting, but I thought it was so far from AI that I ultimately just decided I don't want to do this anymore.

(Stephen Witt at 00:12:02) Well, guess what? I had it completely backwards. Symbolic logic went nowhere, and automatic differentiation is the backbone of back propagation. It's the backbone of how we do AI today. It's the technical way that AI is conducted. So I screwed that one up. Then I went into finance and did that for years and became a journalist and wrote about culture and music and technology. But AI was not anywhere near my mind. I will say, in the mid-2000s, I knew a number of people who were really into what we would call the board gaming scene, and in particular backgammon. One of my friends was trying to be a professional backgammon player, and he was using this program called, at that time, Snowy. Now they use something else. But anyway, it was a neural network. And that shocked me because I remembered from the brief amount of time I'd spent studying AI that neural networks were an obsolete and dead technology. But that turned out not to be true. That turned out to be just ideology. That was a mythology. In fact, they were great, and they worked really well for backgammon. And, you had to ask, well, maybe we could port or leverage the stuff that works for backgammon into some other zone. But at that time, the AI funders, and the PIs, particularly in the United States, they weren't having it. They hated neural networks. They hated them. It was weird. They had a bias against them. And so this technology was starved for money, starved for attention, even when you could provably do something with them. Right? It conquered the game of backgammon. That suggests that they're useful for something, right? They still got no funding at all. I'm off topic, but—

(Joel Beasley at 00:13:41) It's okay. I'm interested.

(Stephen Witt at 00:13:42) This was my history. Anyway, so even with that, you know, I didn't think I was going to write about AI. Like everyone else, I really only paid attention when I saw ChatGPT. You know, but I will say, the moment I saw ChatGPT, I drank the Kool-Aid immediately. I was like, oh my God. This is the future. Actually, I thought it was cooked. I thought it was finished. You know, I was like, oh my God. This thing could write better than me. I'll never be employed again. I actually leaned into writing about AI as a hedge against that. That was the original reason. I was like, oh, God. This thing is fantastic.

(Stephen Witt at 00:14:19) And then, of course, once you see that, you're like, oh, how does it work? What made it possible? Part of it is the work of guys like Hinton, you know, the software side, back propagation, neural networks, transformer architecture from Google, and then, of course, the engineering genius of the people who worked at OpenAI. But the other part of it is these microchips. These parallel computing microchips.

(Stephen Witt at 00:14:43) It's Jensen out there talking to scientists. What could you do if I gave you a million times more computing power? And he was the only one asking that question. He was the only one doing that.

(Stephen Witt at 00:14:55) And not only was he asking that question, he was funding it to the tune of $1,000,000,000 a year, even when there were no profits in it. In retrospect, one of the greatest business decisions of all time. But I don't know if, maybe the greatest, but I don't know if it's easily replicable is the thing.

(Joel Beasley at 00:15:14) No. There's only one Jensen.

(Stephen Witt at 00:15:17) There's only one Jensen. But the other thing is, here's what made it so challenging. They were printing money with these video game accessories. I mean, the profit margins even on those were extremely high. But he was taking that money and plowing it into this bonkers research project that was going to take ten years to complete.

(Stephen Witt at 00:15:39) Wall Street hated it. They thought that the supercomputing, the scientific computing effort at Nvidia had negative value. Stock traded flat for twelve or thirteen years. It went nowhere. Investors were furious with him.

(Stephen Witt at 00:15:56) Customers didn't like it either. Video game guys, why am I paying for this? Why am I paying to fund your bonkers R&D program? I'm going to get an AMD card. They're cheaper. You're not paying attention to your customers. What are you doing? So his investors and his customers didn't like it. If you're the CEO of a company, it's very hard to do something if your investors and customers don't like it. You know, so it really took a very powerful force of will and conviction to make this work. It'd be hard to pull off anywhere else.

(Joel Beasley at 00:16:26) And that ten year research project, did it have a specific name?

(Stephen Witt at 00:16:28) Yeah. That was CUDA, which stands for Compute Unified Domain Architecture. That makes a lot of sense, right? No. I never understood why it's called that. I wrote a book on the company and it has ten people. But what it is, basically, is it's a platform for taking your video game card, and you flick a switch on the video game card and it turns into a low budget supercomputer. Right? Now, who is that for? Who needs that? Well, the gamers aren't going to use it. They're not doing high end physics simulations or medical tomography. Like, who is it for? Established research scientists, they have access to the Oak Ridge supercomputer and other supercomputers. They don't need a video game card. So who's it for? Well, it's for marginal scientists. It's for scientists who don't have much of a research budget, can't afford funding for what they want to do. It's for mad scientists, basically. Right? This is the mad science market. It's a science market that people don't believe in. They don't give a funding. And that's what neural nets were.

(Stephen Witt at 00:17:30) The pioneering neural net that really kind of broke open the modern paradigm of AI was something called AlexNet. It was trained on two $500 gaming cards that they bought online. The cards looked like parts of an alien spaceship. They were marketed to teenagers to play Call of Duty. And these guys built this pioneering neural net on this platform because Jensen had made it possible. Because Jensen had invested in this kind of low budget supercomputing effort that anyone could use. And finally, he found Atlantis. I mean, he found his golden customer. But that was after the project lost money for more than ten years. Almost. Almost ten years. I mean, it was crazy that it worked. The way you think of it, maybe, is like Jensen is making propellers. He's making ever bigger propellers and they go faster and faster. And the maritime industry is like, what are we going to use these propellers for? Our boats go fast enough. Who is this for? Meanwhile, you got another guy who's constantly pushing a glider off the top of a cliff and it falls down. And he's saying to himself, how do I get this glider to fly? Right? When you put the propeller and the glider together, you get the airplane. The propeller, in this case, is Jensen's platform. It's CUDA. It's these video game cards. And the glider is the neural net. They needed each other and they didn't know it.

(Joel Beasley at 00:18:43) Ain't that beautiful? Like, how that stuff comes together in life.

(Stephen Witt at 00:18:46) It's unreal. It's unreal. It's the kind of story you wish you could tell as a journalist, but it never actually is actually what happened. In this case, it is. You had two dead, cast off technologies that had never made money ever. Neither of them. AI had gone through multiple business failures. So had the parallel computing platforms, the supercomputing that Jensen did. That had also failed. Cray failed. Larry Ellison had an effort that failed. Road to success, as Jensen said, was littered with corpses. And so, to make this work, Jensen and his team have to wander past all these dead bodies on the road to success. It was amazing it worked. It was amazing they both worked. Those technologies needed each other, and nobody on either side knew it.

(Joel Beasley at 00:19:32) That's funny. About twelve years ago, so my background is software engineering.

(Stephen Witt at 00:19:36) Okay.

(Joel Beasley at 00:19:37) And about twelve years ago, I looked into, you know, when the Alexas were coming out, I looked into what would essentially become like a ChatGPT. Like, how would you be able to talk with it, and how could you backfill all this information? And one of the things I came across, this open source project, and I was reading the science papers on it and all of this. And basically, you could feed it a web page, and it would tell you new information through using a model that was not in the web page but about it. Like, it could consolidate and think to a degree. And it was really, really fascinating, and I thought, maybe this is it. But then I did more research into the neural networks, and the sentiment was so horrible about how they're not going to be scalable, how they don't work, all of this stuff, and the resolution was low on their. And so through all of that negativity, I was like, alright. Maybe I won't pursue this project because it seems like the underlying technology that looks pretty promising to me. The fact that I could feed it this information, it could come up with new information. Even though it looks promising, it was so negative that I just moved on.

(Stephen Witt at 00:20:45) It was all ideology. It all turned out to be BS. There was absolutely no factual basis to what they were saying, the opponents of the neural networks. I have the neural network textbook from 2011, 2010. Norvig and Russell. It's 1,100 pages long. It's huge. It's the size of a phone book. In that textbook, there are 16 pages on neural nets. Nothing. Nothing at all. 1% of the research efforts were neural nets. And if you studied this, people yelled at you. I mean, they literally yelled at you. It was all Marvin Minsky's fault. He was a professor at MIT. He actually built the first neural net, but then decided they could never work. For decades, he dominated U.S. research and funding and was just absolutely, like, he would scream at the neural net guys and call them idiots, call them names. Just bullied them out of researching. It's not an accident, actually, that all of the neural net pioneers are not from the United States. Bengio and LeCun from France and Canada and Hinton from Canada. In the U.S., they were ostracized by these jerks, these narcissists. Minsky ended up being an associate of Jeffrey Epstein. You know, and they had these bizarro beliefs. They all wanted to use this programming language called Lisp, which they believed had been not invented but discovered. They were obsessed with it. And when you went to MIT, you had to learn Lisp. You had to because they thought it was so beautiful and elegant and that it would someday produce pioneering AI. They were completely wrong. It was a dead end of history. You could see this because MIT abandoned Lisp in the mid twenty tens. Now they teach Python like everyone else. Set the field back decades. It's a whole story in and of itself. I tried to put it in the book, but it was too disruptive. I mean, it might disrupt the narrative. It was too distracting.

(Joel Beasley at 00:22:37) I get it. It might stand alone as a good article, though.

(Stephen Witt at 00:22:39) Yeah. It's a cautionary tale about what not to do and about, it's a cautionary tale of intellectual arrogance.

(Joel Beasley at 00:22:47) I learned that lesson.

(Stephen Witt at 00:22:48) It's been told in fragments. I talked to this guy, or I found this guy, Paul Werbos. Backpropagation, the basic mathematical method that we use to tune the AI, was discovered independently many times. Werbos was the first to propose using it for a neural net in the nineteen seventies. He took it to Minsky. He took it to Minsky at MIT. Minsky screamed at him and tossed him out of his office. We might be ten years ahead if that didn't happen.

(Joel Beasley at 00:23:15) Probably would be. Yeah.

(Stephen Witt at 00:23:17) Maybe not, though. The thing that was really missing was computing power. It's Jensen who put us ten years ahead.

(Joel Beasley at 00:23:24) Yeah. Of all the questions that you got to ask him in all the interviews that you did, what question was your favorite or favorite response?

(Stephen Witt at 00:23:38) I liked when he screamed at me. Oh, nice. I didn't see it coming. You know, I knew that Jensen could be a yeller. I knew that he could get angry, has a temper. Everyone talks about it. I never thought I'd see it myself. I wasn't even trying to provoke him. I showed him a clip of Arthur C. Clarke from 1964, talking about the future of mechanical brains. I tried to constantly get Jensen to talk about AI risk and speculate about the future. He hated these questions. Hated them. I got under his skin too because I would ask questions that no one had asked him. About his family, about his internal motivation, about why he did what he did. He hated those questions. Hated them. He told me he hated them. He never really answered them either. You know? He wasn't exactly a guy who spent a lot of time examining himself or his own motivations. Probably, it's good. Look what he's accomplished. Maybe we shouldn't examine ourselves at all.

(Joel Beasley at 00:24:34) You're right. Maybe we can wipe the—

(Stephen Witt at 00:24:36) Just do what comes naturally.

(Joel Beasley at 00:24:38) Shut up. Turn the therapy off and just get to work. Yeah.

(Stephen Witt at 00:24:40) He said that to me. I hate therapy. I'm not going to therapy. He's probably right. I had to go to therapy, but it's because my life was dysfunctional. His was not. You know, if you're as successful as him, I don't think you should go to therapy. What do you need it for? He's also, you know, look at his personal life. It's good. It's very rare.

(Joel Beasley at 00:24:58) He works with his boys.

(Stephen Witt at 00:24:59) He says his daughter and his son.

(Joel Beasley at 00:25:04) Daughter too?

(Stephen Witt at 00:25:04) Yeah. Madison. She works in the marketing department. He's, his wife loves him. They love each other, I think. I mean, it seems like that.

(Joel Beasley at 00:25:11) Yeah. As much as you can tell.

(Stephen Witt at 00:25:13) As much as I can tell, but they're still together. They've been together since they were 20. They met over electrical engineering. They met over a breadboard, basically. I mean, they were in the same—

(Joel Beasley at 00:25:22) That's so cute.

(Stephen Witt at 00:25:22) Class. He was randomly assigned as her study partner, and he fell in love with her. But he was so inexperienced in matters of women that it took him six months to screw up the courage to ask her out.

(Joel Beasley at 00:25:35) That's okay. There's people that have done worse. Yeah.

(Stephen Witt at 00:25:39) I can relate.

(Joel Beasley at 00:25:42) Yeah. Oh, man. So what did he yell at you about?

(Stephen Witt at 00:25:47) You know, I think, remember I said he doesn't like having his time wasted. I kept asking him the same question. I kept asking him, you know, in various ways, different takes on the same question. What the hell does the future look like? Not the near future. Not next quarter. What does twenty years from now look like when these systems are online? Is there going to be a place for the human in that world? Are we going to be left behind? Even if you don't think there's existential risk. Even if you think that we're heading towards utopia. What are we going to do in a world dominated by computers and robots? All physical and manual labor will be taken over by a robot. All cognitive labor will be taken over by a powerful AI. Are we just going to play tennis? You know? Like, what are we going to do? What's our role? He hated this question. He didn't like it. He didn't want to answer it.

(Joel Beasley at 00:26:35) Interesting.

(Stephen Witt at 00:26:35) Because it involved speculation. He doesn't like to speculate. I showed him Arthur C. Clarke, and he's like, I've never read Arthur C. Clarke. I've never read those fucking books. That's a direct quote. He's never read, you know, he's just not interested in science fiction. Which is crazy. So he says, crazy for a guy who named his headquarters after a Star Trek spaceship. Standing on stage with, you know, Star Wars androids. But I think he's never read a science fiction book, he says. I think it's true. I don't think he's a big reader of fiction. He absorbs business books at a voluminous rate. He's probably read them all, honestly. But I think he, he's only going to, he's, you know, again, he's like an engineer. He's going to optimize every moment of his day, every moment of his time to achieve what he wants, and that involves reading lots about business, I think. And I think there's more of course.

(Joel Beasley at 00:27:29) I think there's more Jensens out there in the world, and I think as we start automating a lot of this stuff, and I'm an optimist, so I'm going towards in the direction of utopia. But it'll give other people that ability. For example, when you say he doesn't like to speculate, I can almost put my feet in his shoes and imagine if I had the resources that he had, why would I, why would I consume science fiction? Why would I, it's so interesting what's happening today. It's so interesting I could put $10,000,000 there and advance that. It's like, I would be so focused, close to the metal in the sense of what is not what culture thinks, but what is scientifically possible today, and what can we move forward that—

(Stephen Witt at 00:28:14) Most of the pioneers love science fiction, though. Elon Musk loves science fiction. I think Sam Altman's a science fiction guy. I think Demis Hassabis is a science fiction guy. It's a very unusual trait. And if he wasn't building science fiction, I mean, he was building video game accessories for a long time. You know, the fundamental question about Nvidia is what if AI, you know, hadn't been profitable? What if there was no business case? Would he still pursue it? You know, is he doing it because it's so valuable? Or is he doing it because, you know, he believes in AI? Jensen, the way he's wired, I don't think he can do things that don't make business sense. Like, he's just not, if AI was there, but it didn't make business sense, I think he would leave it to the scientists and the academics. You know? This is the question about him, right? Like, what if the most profitable thing they did was, I don't know, make orange juice? Would he go in the orange juice business? Probably. But I think he got lucky. But maybe the question is stupid because maybe the most profitable thing you can do ultimately is unlock some new field, and you have to be an N of one, and you have to be a pioneer. And so those things are not in conflict. As I think out loud, I think that's probably true.

(Joel Beasley at 00:29:30) Yeah. Because in one case, he's actually out there funding scientists to find out what the next breakthrough can be.

(Stephen Witt at 00:29:38) That's right. And I think in a technology company in particular, if he doesn't innovate, his company will die. I mean, they don't own any manufacturing facilities. That's all done in Taiwan. That's all TSMC.

(Joel Beasley at 00:29:50) Mhmm.

(Stephen Witt at 00:29:50) You know, it's basically just a giant R&D laboratory. And if they just do video games all the time, knock-off competitors in Asia will catch up to them and destroy their margin. Jensen was terrified of this. He was obsessed with it. That's actually the reason he went into science as much as he saw the potential.

(Stephen Witt at 00:30:06) He also was afraid. He was afraid that, you know, his business would be commoditized and his profit margins would go to zero. In a sense, this was his true motivation and still is. He's obsessed with that problem. Here at the Innovator's Dilemma by Clayton Christensen, which describes in great detail how this happens again and again.

(Stephen Witt at 00:30:25) What happens is that established companies ignore small, low margin players, and then those low margin players take over. Clayton Christensen's example, it's not necessarily high-tech. People misinterpret what the term disruption means. His canonical example was a Honda dirt bike. You know, they were selling it in the early nineteen sixties to teenagers in The United States.

(Stephen Witt at 00:30:46) GM looked at the business and said, why would we do that? We'd make no money. It's a small market. It's a low margin product. We'd rather sell a Cadillac to a businessman.

(Stephen Witt at 00:30:54) It's a waste of our time. But it gave Honda a foothold in the market, and they leveraged that expertise to build an economic compact car and then raided GM from below. This happens again and again, and the reason it happens, or at least historically the reason it happened, is it's very hard if you're in a high profit midget, a business to deliberately lower your profit margins by pursuing small niche ideas. But, Christensen says this explicitly. There are times when it is correct to pursue small rather than big markets, and low margin opportunities rather than high margin ones.

(Stephen Witt at 00:31:29) Jensen internalized that. He took it to the extreme. He built products with zero margin, no customers. He called it the $0 market. It's true.

(Joel Beasley at 00:31:40) But the hope is— So this was a strategy.

(Stephen Witt at 00:31:42) Oh, for sure. He was obsessed with Christensen. He assigned the book to everyone in Nvidia, and he hired him as a consultant.

(Joel Beasley at 00:31:48) So this could be part of the backbone of, or the driving motivation of CUDA. Yeah.

(Stephen Witt at 00:31:54) CUDA project. 100%. It's all in the book, by the way. Okay.

(Joel Beasley at 00:32:01) Alright. Let's do the plug for the book. For people who are better than me, who are going to read the book, where do they download it? Where do they buy it?

(Stephen Witt at 00:32:08) Yeah. So you can anywhere. Amazon, obviously, Barnes and Noble, any bookstore you can think of. It's called The Thinking Machine. It's about NVIDIA.

(Stephen Witt at 00:32:16) I'm Stephen Witt. I'm the author of the book. 60% of my sales come through the audiobook format. So that's an excellent way to access the book. You know, we then have a very niche number of people called, oh, what is the name for them? They're like intense readers or something. But, essentially, what they do is this is a very small crowd, but they buy the book as an audiobook, and then they also buy the physical copy of the book. You do this.

(Joel Beasley at 00:32:44) That's me. Oh, yeah. All my books are— that's what I— that's the— So what I do is I buy the audible book.

(Stephen Witt at 00:32:48) Uh-huh.

(Joel Beasley at 00:32:49) And if it's good, I buy the physical book because every time I see it on my shelf, it reminds me of the content inside. And so all the memories I have here of all these books are all things that I really want to remember.

(Stephen Witt at 00:33:02) Yeah. It works great. I mean, so they— it's crazy. They follow along with the audiobook, and they're reading with their finger on the page.

(Joel Beasley at 00:33:09) Oh, I don't do that.

(Stephen Witt at 00:33:10) No. It's— I mean, but it's great. I love it. You get to sell the book twice, first of all.

(Stephen Witt at 00:33:14) The problem is if the audiobook differs in any way from what's on the printed page, you get letters, and people start yelling at you.

(Joel Beasley at 00:33:19) I'm assuming yours does a little bit.

(Stephen Witt at 00:33:21) No. It doesn't. I have to admit. No.

(Stephen Witt at 00:33:22) This is the thing. It dovetailed so perfectly with my workflow. Perhaps you've noticed I'm a talker. I read everything out loud anyway. Everything I write is basically a script for what I wanna say.

(Stephen Witt at 00:33:35) You know, so the way I'm terrified— the lesson of the book, the lesson of Jensen's life is to take your insecurities, use them as fuel to accomplish what you want. It's a profound lesson. I do this myself. I'm terrified that what I'm writing is boring and stupid, and that no one would ever wanna read it for any reason. This voice constantly talks to me and tells me that.

(Stephen Witt at 00:33:59) So when I'm writing, I have to read it out loud and ask myself constantly, is this the part where the reader puts the book down never to return? You know, because it's my experience with most books. I put them down and don't finish them, unless they're really compelling, unless they really grab me by the nose and pull. So, I have to do that too. And one way to tell if you're doing it is to read it out loud.

(Stephen Witt at 00:34:20) It's fantastic. I've done this my entire career. As it turns out, it's perfect for the modern day. I got very lucky because so much, especially in the business and technology side, is done via audiobook now. I mean, I basically turned into a podcaster.

(Stephen Witt at 00:34:34) You look at somebody like Acquired, you know, they're doing twelve hour podcasts on video and people listen to them. My book only takes ten hours to read, you know. So we're in the same zone. You could argue we're even competitors. I'm lucky they don't charge.

(Stephen Witt at 00:34:48) Those guys are great, by the way. I talked to them for this book. I mean, god, they're good. Who did you talk to? Do you know Acquired?

(Stephen Witt at 00:34:56) It's, uh, oh god. What are their names? David and Ben. I'm blanking on their last names. They're great.

(Stephen Witt at 00:35:01) They run this podcast called Acquired. It's addictive. They do it rarely. They do, I think, maybe six episodes a year or six stories a year where they just do these deep, intense dives on the history of businesses, Coca Cola and the NFL, NVIDIA. It's so captivating.

(Stephen Witt at 00:35:20) They're really good at what they do. They're former venture capitalists. I recommend it.

(Joel Beasley at 00:35:24) Interesting. I like that comment that you made about thinking at the end of the page, is this the page where they stop reading? I took a master class with James Patterson on writing, and he had some advice related to that. He goes, I read at the end of every page, and I ask myself the question, will the reader turn the page? Yep.

(Joel Beasley at 00:35:43) And as he's going through his edits, I thought, oh, that's brilliant.

(Stephen Witt at 00:35:47) And for this stuff, it's so hard. I'm writing about freaking microchips. It's not, you know—

(Joel Beasley at 00:35:51) I'm inter— yes. No.

(Stephen Witt at 00:35:52) I'm interested in audience, but most people, their eyes glaze over the moment you start talking about it. Trying to explain parallel computing to a layperson in terms that won't make them turn off is impossible. I would read my work to people, and they'd be like, what? So boring.

(Joel Beasley at 00:36:09) How are the sales doing of the book?

(Stephen Witt at 00:36:11) They're decent. In China, it hit the bestseller list. My Chinese publisher is superb. They really leveraged it, and they built— I mean, god, they flooded the zone with it. In China, Jensen's celebrity is as famous as Steve Jobs was at his peak.

(Stephen Witt at 00:36:25) Taiwan too. Oh, god. Yeah. They love him. I mean, look.

(Stephen Witt at 00:36:30) If Jensen was a white guy from Ohio, they probably wouldn't love him so much because he's Taiwanese. Right? So Asian ancestry, and I think they identify. When I was there, I kind of wrote the book. I wouldn't say it was a cautionary tale, but I wanted to show all sides of Jensen, including the more negative sides.

(Stephen Witt at 00:36:45) Instead, I have teenagers coming up to me being like, how can I be more like Jensen? People were buying the book for their kids to read so they could be like Jensen. I was shocked.

(Joel Beasley at 00:36:53) Bad. No. No. No. That's— I had that strategy.

(Joel Beasley at 00:36:55) When I was at a point in my life where I wasn't a very good person and I wasn't getting the results that I wanted, I did two things. First, I found the personal development field, which I had previously ignored because I thought it was stupid. Mhmm. And I looked at a list of things I'd ignored because I thought they were stupid, and I gave them a shot. And personal development was one of them, and it was helpful.

(Joel Beasley at 00:37:13) And the second thing was I bought three books by three different billionaires, Branson, Bezos, and Musk. Mhmm.

(Stephen Witt at 00:37:19) And—

(Joel Beasley at 00:37:19) I read those three books after having read many business books. And I learned more about business and decision making and how to evaluate opportunities from reading the life stories of these three billionaires than I did from all the business books, all the formulas and all the things. It was so valuable for me to be able to think like them after reading about them.

(Stephen Witt at 00:37:41) Me too. I read Brad Stone's book on Bezos. Bias for action was such an important concept for me. I would sit there and waffle for years about whether to do something. Who cares?

(Stephen Witt at 00:37:52) It takes ten minutes to write and fire off to a publisher. What do they say? No. Big deal. Didn't cost you anything.

(Stephen Witt at 00:37:58) Completely changed the way I thought about my career and everything. You know, and also personal development in my case. I mean, I felt so much shame about everything I wrote, even when it was praised in the press. I just I felt so ashamed of it. It felt so empty.

(Stephen Witt at 00:38:13) I went to see a therapist, and I was like, listen, I am blocked. I can't write. This was a five year period. I produced almost nothing. I'm so ashamed of what I write.

(Stephen Witt at 00:38:21) It's so dark. No. I can't even look at it. And I worked with a guy and ultimately, I came to realize that it didn't matter. I'm still ashamed of everything around you in this book.

(Stephen Witt at 00:38:34) Who cares? You know, you're in the habit of writing. It doesn't matter. I never fixed the core problem, but I became very productive.

(Joel Beasley at 00:38:41) What other books have you written?

(Stephen Witt at 00:38:43) I wrote one other book called How Music Got Free. It was a history of the MP3. It did well. You know, it was translated into 12 languages and excerpted in the New Yorker. But then I wrote for the New Yorker for a while, but I just got total writer's block.

(Stephen Witt at 00:38:56) I was convinced that everything I wrote was worthless. I just didn't believe fundamentally in what I was producing. I mean, not that— I don't know where it came from because the people seemed to like what I was writing. You know, I didn't get negative feedback and it was, you know, people were— I was getting published, you know, in prestigious organs, but inside, it just felt like I wasn't doing a good job. Even this book.

(Stephen Witt at 00:39:21) Yeah. Jensen's the same way. It's so relatable. He built this billion dollar headquarters. It's immaculate.

(Stephen Witt at 00:39:28) It's beautiful. He spent all his time complaining about where he located the bathrooms. I get upset. Serious. I talked to the architect who built it.

(Stephen Witt at 00:39:36) He's never satisfied. Two days after the IPO happened at Nvidia in the nineties. I mean, the guy's now gone from, you know, spitballing at a Denny's in 1993, now has several $100,000,000. Two days after the IPO, you think he'd be happy. He's reading an email to his employees.

(Stephen Witt at 00:39:52) He's obsessed with the competition. You'd think his business was about to go bankrupt reading this email. It's unbelievable. So, you know, that's the lesson. You can take these— it's kinda negative.

(Stephen Witt at 00:40:04) He's so neurotic, Jensen. He experiences more negative emotions in a day than most people do in a lifetime. But he uses it. He needs it. It's his fuel.

(Stephen Witt at 00:40:14) And the more powerful and successful Nvidia becomes, the more intensely he feels these things. He's terrified his business is gonna collapse at any moment. It's what makes him work, it's what makes him succeed. It's the lesson of life because ultimately we all feel these things in different ways. And I think for a lot of people, they just end up being suppressed or they try to eliminate them.

(Stephen Witt at 00:40:36) But actually, you can use them.

(Joel Beasley at 00:40:38) Yeah. Instead of— keeping that edge, and there there is definitely a fine line between using it as fuel and letting it overtake you. Yeah. For sure. Being able to dial that in is is what they do.

(Stephen Witt at 00:40:54) I will say, Jensen balances it with optimism about AI. I was talking to— he says this— he said this to me. His first job was a waiter, a dishwasher, Denny's, low level, low paid job. He loved it, though. He loved working at Denny's.

(Stephen Witt at 00:41:11) And his favorite time was the dinner rush, when the restaurant got really busy. He told me that when the rush hour happened, he thought that his heart rate actually went down and he became more relaxed, because he could spring into action. He felt, you know, capable. When nothing was going on, he would get bored and antsy.

(Stephen Witt at 00:41:29) Forty years later, I'm talking to someone at the company immediately after DeepSeek premiered. When DeepSeek came out, the Chinese AI, the market misinterpreted what had happened and Jensen lost something. I can't remember the exact number, but basically, NVIDIA's market capitalization plunged by about $1,000,000,000,000 in a few days. And I was talking to someone inside the company. I was like, how's it going over there?

(Stephen Witt at 00:41:50) And they're like, Jensen seems so relaxed. He seems happy for the first time in years.

(Joel Beasley at 00:41:58) Dude, I get it. There is calm in the chaos. I like Jensen more now after talking with you about your experience with him than I did just from listening to the Rogan interview.

(Stephen Witt at 00:42:07) When things are going down, he's a better leader. He's more compassionate. It's when things are doing well that he becomes obsessed with these ideas, these fears of complacency, and this is what drives him. So he hounds everyone when things are going well because he thinks they're relaxing.

(Joel Beasley at 00:42:24) But they do.

(Stephen Witt at 00:42:25) It's true.

(Joel Beasley at 00:42:26) They do.

(Stephen Witt at 00:42:27) Tendency. I would.

(Joel Beasley at 00:42:28) Yeah. It— I notice it. Yeah. It happens. See, the thing is—

(Stephen Witt at 00:42:32) What what are the NVIDIA modules? Probably their key NVIDIA module is the company is thirty days from going out of business. Now several times in NVIDIA's history, it's been literally true. It's been literally true that they were thirty days from going out of business, but it hasn't been true for a long time. It's certainly not true right now.

(Stephen Witt at 00:42:48) But Jensen still says it. He wants everyone at the company to act like Nvidia's teetering on the verge of bankruptcy, and they have to act out of desperation to save the company. It's the way that he kind of fuels them constantly, propels them towards greatness. It's constant war against complacency. It works, but it makes it a hard place to work, I think.

(Stephen Witt at 00:43:09) You know, the people who work there, I mean, they've sacrificed everything to be there. Family lives, conventional lives, they don't have them. You know, they're at Jensen's beck and call, and there's this constant sense of falling behind, this constant sense of you're about to lose and disgrace, you know, be disgraced. It's tough. I had, what, eight, nine months to write this book.

(Stephen Witt at 00:43:31) It's so fast. I had five years to write my last one. I adopted this kind of Chinese inspired nine nine six work schedule. I worked from 9AM to 9PM every day, six days a week. Except in my case, it was nine nine seven.

(Stephen Witt at 00:43:44) I just worked every day. Got the work done, but it broke me. I was completely depleted by the end. I couldn't do it one more day. I had to take four months off.

(Stephen Witt at 00:43:57) I don't know how people do it. I went to China. They're all that way all the time. Same in Nvidia. I don't have it in me.

(Stephen Witt at 00:44:03) It takes a certain kind of person.

(Joel Beasley at 00:44:05) Oh, their culture is designed around it too, which helps. They got—

(Stephen Witt at 00:44:08) That's true.

(Joel Beasley at 00:44:09) Odds and ends.

(Stephen Witt at 00:44:10) They do all seem a little unhappy. But I think also—

(Joel Beasley at 00:44:15) With the nets. Did the nets around the building? Did that give it away?

(Stephen Witt at 00:44:19) But I think they also, you know, in America, we—I think for them, happiness is not their core mission. The sense of duty that China must succeed and that their families have to be made proud and that your personal happiness is a kind of irrelevant variable. I think Jensen has some of this in him too.

(Joel Beasley at 00:44:39) It makes sense that that's his mentality because knowing the story of how the company got started, that's what worked for him early on. Yeah. I mean, he's all in on business.

(Stephen Witt at 00:44:50) Even before NVIDIA, he was this way. Yeah. I mean, Jensen, this is the thing that I think is difficult to replicate. He's brilliant. I mean, even when he was like 10, people were like, this guy is brilliant.

(Stephen Witt at 00:45:02) I talked to his childhood friend from sixth grade, Ben Baze. I was like, are you surprised that Jensen has all this money and all this success? Baze was like, no, not at all. I knew he was gonna be a star from the moment I met him. He was just brilliant from the moment I met him. Multiple people throughout Jensen's life told me this. By his early twenties, mid-twenties, he was leading a division at LSI. He had 200 employees. He was a microchip engineer kind of being groomed to be the CEO of that company.

(Stephen Witt at 00:45:30) So he was always a star. His ability to learn information is just almost superhuman. He started playing table tennis at the age of 15. He had no formal experience, and by the end of the year he was a nationally ranked player competing in tournaments. His coach wrote a letter to Sports Illustrated calling him the most promising junior he'd ever seen.

(Stephen Witt at 00:45:50) But he can do that with anything. It's not just table tennis. His ability to learn new stuff happens at a superhuman rate. Partly, it's his drive and dedication, you know, but he's just smart. You know?

(Joel Beasley at 00:46:05) Alright. Let's tell people to go buy the book again.

(Stephen Witt at 00:46:10) The Thinking Machine. Yeah. By Stephen Witt. Available everywhere. Check Amazon.

(Stephen Witt at 00:46:14) Buy it on Audible.

(Joel Beasley at 00:46:16) Did you do the voice over on Audible?

(Stephen Witt at 00:46:18) Of course.

(Joel Beasley at 00:46:19) You have a very nice voice.

(Stephen Witt at 00:46:20) I still—I've worked on it. I've worked on it. I saw it coming. You know, I could tell that other people would hire voice actors. Like sure, they're better at it.

(Stephen Witt at 00:46:31) But, you know, the modern author needs to be a voice over artist. They need to. Because especially nonfiction. Because once you write the book, you're on 300 podcasts. You have to create continuity between your voice and the presence that you create as a media identity on podcast and television, and what the reader ultimately gets if they buy the audiobook.

(Stephen Witt at 00:46:55) I'm obsessed with audiobooks. I wish 100% of my sales were audiobook. It's a great way to access the product, and you can be sure it's compelling. It's also kind of a new format. It's dynamic.

(Stephen Witt at 00:47:08) It doesn't feel fusty and old, like old book. I mean, I love books. I love literature, obviously, but I'm not gonna worship at the altar of irrelevancy. I wanna create a modern product.

(Joel Beasley at 00:47:20) I don't have the ability to sit down and, like, read the book.

(Stephen Witt at 00:47:27) Oh, okay. That's fine. Get the audiobook. But that's what I do.

(Joel Beasley at 00:47:30) But people are like—

(Stephen Witt at 00:47:30) That's how most people—

(Joel Beasley at 00:47:31) That's what I do. I go for walks. I hate the people that are—and, yes, I know hate's a strong word—who are like, it's not reading unless you're reading on the page. I was like, okay.

(Stephen Witt at 00:47:41) Forget it. They're pseudo-intellectuals. They understand nothing about literature. Yeah. You know?

(Stephen Witt at 00:47:45) Shakespeare was meant to be performed out loud. Homer was meant to be sung with musical accompaniment. Those are song lyrics that you're reading. Those people don't understand anything.

(Joel Beasley at 00:47:54) Yeah. But I can—I walk. I do 10,000 steps a day, so I listen to stuff then. When I have to drive, I always have an Audible or Spotify podcast, something long form for the drives.

(Stephen Witt at 00:48:08) Yeah.

(Joel Beasley at 00:48:08) And so that's where you're gonna end up in my queue.

(Stephen Witt at 00:48:10) Oh, fantastic. Good. If you get bored, just skip to the next chapter.

(Joel Beasley at 00:48:16) You know, ten hours sounds long. If you drive.

(Stephen Witt at 00:48:19) It's not a—

(Joel Beasley at 00:48:19) long time. It's not a long time.

(Stephen Witt at 00:48:21) This is the thing. I mean, you're fighting it. Right. So my constant fear, the thing I was constantly thinking about was, at what point—not only is the reader—my competition is, in some sense, not other books. It's TikTok.

(Stephen Witt at 00:48:34) Right? It's Instagram. Like, I live in this universe of constant distraction. How do I get people to focus on this long piece of this long story that I'm writing?

(Joel Beasley at 00:48:44) I think it's like being hungry.

(Stephen Witt at 00:48:45) Serial television does it and podcasts do it. So you have to modernize what you're doing. You have to adapt to the forms that they use. Every chapter of my book, I end on a cliffhanger. I did that on purpose.

(Stephen Witt at 00:48:55) I structured it like a potboiler. You have to, you know, otherwise, they won't read the next chapter. It seems cheap. It seems like a cheap trick, but then you look at the greats and they do it too.

(Joel Beasley at 00:49:07) I don't think it's a cheap trick to pay attention to how people's attention is wired and then to build content that helps keep it.

(Stephen Witt at 00:49:18) Patterson is excellent at this. He's really good at the pot, the thriller. I don't—look, I'll be honest with you. I don't read a lot of James Patterson.

(Stephen Witt at 00:49:25) I mean, I've read one book of his, and I didn't really like it because I didn't think it was—

(Joel Beasley at 00:49:28) Not either, written. Yeah.

(Stephen Witt at 00:49:29) But, you know, he understands his audience really well. I think I watch so many movies and television shows and books that fail. And they fail because they don't establish the stakes quickly enough. They don't get the reader invested, the viewer or the listener invested. Especially in the modern media landscape, you have about a minute to do that if that.

(Stephen Witt at 00:49:56) Otherwise, the thumb is going to somewhere else on the dial. That's an important lesson too. You really gotta sink your hooks in right away and then not let go. I compare it to just grabbing the reader by the nose and pulling them. You have to do it.

(Stephen Witt at 00:50:09) You have to do it, or your book's just gonna sit on the shelf and, you know, you have to do it. You know?

(Joel Beasley at 00:50:17) I'm so glad that we got to do this. I like you, Stephen.

(Stephen Witt at 00:50:20) It's my pleasure.

(Joel Beasley at 00:50:22) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.