TECH002: Jensen Huang & NVIDIA w/ Seb Bunney - Review of The Thinking Machine by Stephen Witt

25:12But when you look in the early days, they came out with this NV1 chip was their first thing that they brought to market. And it sold actually pretty well considering their size and whatnot. But then Microsoft came out with a new graphics protocol and killed this thing overnight. This thing had only been on the market for, call it a year. And the competition had already stepped in and just obliterated them. They came out with this NV2 chip, which was a total flop because of just the sheer timing and all the competitors that are bringing other things to market. And like I said, there was like 30 or 40 competitors against them at this point in time.

25:50And this was a really cool part of the book is they talk about then they came out with their NV3 chip and to try to stay in business. And not only did they come out with this chip, but they couldn't even build a prototype of it to stay in business. They had to do the entire thing through simulation with the hope and the prayer that there wouldn't be any mistakes when they actually went to the foundry to produce the chip and materialize it. They had no idea. They hadn't tested it on any type of real material before. It was complete simulation. And the book does a really good job showing the company was failing.

26:28It was going to fail. It was pretty much assured that it was going to fail. And this was the final Hail Mary pass into the end zone and the NV3 chip came out and it was successful and it didn't have issues and it kept the company on life support for their next thing that they had to do. And I'm looking at this and I'm just thinking the amount of stress that Jensen and anybody else that was participating in the company, I just can't even imagine the cycle time of producing this hardware and going to the foundries and having no clue. And the final thing that I would say, I find it crazy ironic that they emulated this in a simulated environment to create the hardware.

27:14And when you look at what they do right now, they create the hardware to create these simulated intelligent simulations of reality. And I find that really just mind-blowing that that's what saved them back in the very early days. We're talking mid-90s, that they were really on the cusp of death for quite a while. So I think that's probably why he's always looking at the company. And this is another theme that kind of touches on what Seb was talking about, this idea that he's always looking at the company as if it's going to die tomorrow. And that is culturally huge at NVIDIA. I'm assuming if you work at NVIDIA that you are very well aware of this idea that he's constantly looking at the company like it could die tomorrow.

27:58A lot of it goes back to these early days of like, that's what reality was for the company. Do you want to talk about his personality, Seb? There's one point before we jump on his personality that really, you very briefly touched on it, which is this idea of simulating these environments. There was a point that he mentions at the end of the book that I thought was really, really fascinating. It's this idea that the challenge with robots today is that if you want to go and train robots, it takes a ton of time. If the robot falls over, damages itself, cool. You're back to square one. You've got to rebuild the robot and such.

28:30Now, in order to be able to train robots, they've built this thing called Cosmos. Cosmos is basically just this hyper realistic environment that abides by similar laws to the real world. You've got physics, fluid dynamics, gravity, cause and effect, physical permanence. If you look at an object and then you look away from it, it's still there. The idea is it's meant to be training robots in a digital environment, hyper-realistic digital environment. By the time they actually enter the real world, they're far more proficient than trying to train them in the physical world. This is where he really is ahead of his time in the way that he thinks about things.

29:06To give an example, if you had a warehouse and you wanted to train your robot on all of the various different paths it could take to go and say pick packages throughout this warehouse. If you were to do that in real time, it would take a ton of time. But with a digital hyper realistic environment, you're able to within a split second, have it map out every potential possible path throughout their warehouse without ever having to set foot in the real warehouse. So by the time it sets foot in the real warehouse, it's good to go. And so I think that the world which we're moving into is one where we're able to get so far in whether it's our robotics, or whatnot, understanding the world in these hyper-realistic digital environments before ever touching the physical world.

29:48And that's something that's kind of really coming into existence today. And from my understanding, Cosmos is free. Anyone can go and play around in Cosmos because he wants to support science, advancements in technology, advancements in robotics, advancement in AI, which I think is really, really cool. But again, it just kind of goes back to the fact that I think when it comes to the way that he thinks about things, he really is like a zero to one. He thinks about things in a way that there's no one else doing this thing. I want to go into this space and create. And his focus is not necessarily on perfection.

30:19His focus is on, let's just test it, iterate, iterate, iterate. There's a quote in the book. And it basically says someone who came into Nvidia and started looking at the code, he was just like, this thing is like cancer. Like, what is this thing? This is so poorly written, but it does what it's meant to be doing. And he ultimately ends with the saying, there was a brilliance to it all. Just iterate, iterate, iterate, execute, execute, execute. And so rather than its competitors that were trying to create this super clean, professional looking code, NVIDIA was overtaking them because it was not about clean professional code.

30:55It was just like, let's try and minimize our execution times and getting this to market. They changed it from one year, two year cycles to six month cycles, and they were just trying to get it over out into the market. And they really dominated the market as a result of that strategy. Yeah. Yeah. Almost like just get in the room and start sensing the room so you can come up with a mental model or map it as fast as possible. We can iterate faster. We can get to what we think the truth is if we can just start sensing the environment is really kind of the approach. Yeah. One other thing that I wanted to kind of hit at with what you were saying there as far as simulation and where it's going, I find LiDAR so fascinating and important to just how quickly and how accurately we can model things in physical reality for the simulation.

31:45So for people that aren't familiar with LiDAR technology, you can go out, you can get an emitter and sensor, a LiDAR sensor. It's almost like it is light. It's light energy. It's just in a certain frequency that you can't see with your own eyes. But you could go into a room, you emit and you sense the return of the light energy as it comes back to the emission. And you can map in 3D with super high precision down to millimeters of depth. And you could go into this room I'm in right now. You could come in with a LiDAR sensor. You could shine it around. You could, and again, you can't see it, but you can emit the energy into the room and then the feedback, and you can get a pristine mapping, depending on how much energy you emit and how much you collect back, you can get a really hyper-realistic depth map of everything in the room.

32:35And so then you can take these models and you can apply it to AI to train, call it a robot that you wanted to be in the room with you. So some of the technology, as you look at the convergence of it all, is beyond exciting, beyond just mind-melting of where this is going to go as you start applying it to call humanoid robots and whatnot. Okay. You want to go to Jensen? Let's do it. Let's go to Jensen. Okay. He's really interesting, right? I don't really know how else to describe it other than I'll watch an interview of him on YouTube or wherever, an interview just within the past couple of years.

33:12And he's crazy humble, almost attributes nothing to his skill. It's almost always like, well, I don't know. Maybe I'm good at it. Maybe I'm not good at it. And it's like that, I find that really fascinating. And then when I read the book, I was taken back a little bit in this idea of him dressing down employees and like just lambasting people in public, but not all the time. It was kind of sparingly here and there, but had this side to him where he could be almost explosive and just, and is that how you read it too, Seb? Is that kind of your take on how the book laid out his personality? because I mean, that was kind of my takeaway.

33:52I was a little surprised because I wasn't expecting that from all the public interviews that I had seen him do. I would have never expected that kind of side of him. I had the exact same takeaway. It was my background. I studied to be a somatic therapist. So there's a part of me, and I never want to say I'm psychoanalyzing people, but I'm always curious. I'm like, where does this come from? His ability to see the future, but at the same time, it sounds like he has a bit of a temper at times and will unleash. But he wants to make sure, and I think it's strategic because it very much makes the point in the book that he believes in public feedback.

34:30And so, he wants to turn one person's kind of mistake into a collective learning and like builds this kind of shared wisdom. And there's a few times where he made it a very like, pertinent point to do it in front of the team, which would be hard to work in that kind of environment. But I understand to a certain extent why he's doing it. So think about this. What is his creation with all of this? It's parallel processing, right? And so when you look at him doing this in public in front of everybody, what is he doing? He's making sure every other person that's standing there can learn all at the same time.

35:03And I'm not trying to promote this in the public workspace or anything like that. All I'm trying to do is what he created, which is parallel processing, is also the way he operates just as a human in the way he leads. And when you look at his staff, I'm sure you are familiar with the way he runs his staff. There isn't one. It's him and his interactions with everybody and anybody in the organization, regardless of what level you're at. And so he's leading through almost a similar scheme as the GPUs that he's making, which is just, it's all happening in parallel all the time, which as a person out of the military leadership side, that's just like beyond comprehension for me to think of like, how would you manage that?

35:49Especially with a company. I mean, how many employees do they got? Hundreds of thousands of employees. I just couldn't imagine how you manage that. It seems like chaos. Oh, man. I couldn't agree more. And one thing that I thought was really interesting is that he rarely fires. And I think I've got a quote that says he tortures to greatness. There's almost this idea. And actually, I'm trying to remember, there was a book I read a few years back. I think it was Fooled by Randomness by Nassim Taleb. And one of the things he talks about is, if an employee makes a mistake, the last thing you should do in that moment is fire them.

36:22Because it is from that point on that they've now understood, oh man, I should not be doing that thing. It's kind of like the best time to invest is right after a recession. That's not the worst time to invest. It's the best time because the probability of that happening again is very low. And so I think that he sees this. He sees when an employee makes a mistake, the instinct might be to fire them. But in doing so, you're just letting go of someone who just learned the lesson that they'll never repeat again. And so I think he sees that very much and it ends up building that cohesivity in the team when they see these are lessons for all of us to learn.

36:57Yeah. He paid for that learning lesson and now he's going to make sure he gets his money's worth in the future, right? Yeah. I was really surprised by that in the book too, in that people that want to stay there can stay there what seems like forever. He just doesn't get rid of people, but he may throttle you from time to time is really kind of the takeaway. And the other thing that they talked a lot about at the end of the book was just how people love working for him. He is a celebrity within the company itself where the people love him to death. But where I think it's hard to kind of understand the why is it also talks about how everybody that's working there that has stock in the company gets another zero added to their net worth on what feels like every two to four years.

37:47And I'm wondering how much of the love is just because he's made a lot of people there fabulously rich and how much of it is because they actually respect him as a leader or whatever other factor. It's kind of hard to know. whether like this type of leadership is repeatable or what. I don't know. I kind of left the book not really feeling like I could have an opinion on any of that. I felt almost more confused before I started the book. I was very similarly, I was kind of confused at the end because it's just like, is this to your point, is it repeatable? And the thing that's interesting is that I think when companies grow to a certain size, you almost need hierarchy.

38:32Otherwise, it's really hard to sort through the signal versus the noise. And I think to that point, he very much throughout the book, it talks about kind of this flat system as opposed to this hierarchical system. And there was one point that I thought was really interesting. And in order to be able to constantly support the individual, he doesn't really have executive-only meetings. Like junior engineers can sit in on these meetings too. So anybody can share their opinion. He even has this thing, and I may butch this, so correct me if I'm wrong. But at the end of each week, everyone in the business sends an email to him, and he very much supports this idea of conciseness, but tell me the top five things you're interested in and working on right now.

39:14And he then goes and picks at random a whole bunch of these emails and reads through them. And this is where a lot of his inspiration comes from. And so he wants this flat organizational structure. He doesn't want the hierarchy, He doesn't want this telephone where you're losing the signal the more people it touches. He wants to hear directly from the individual that's coming up with these ideas and helping to push that individual, reinforcing visibility, approachability, and the cross-pollination of ideas. I think it's so cool. Did you know Sam's Club isn't a store? It's actually a club with cool finds and like a whole community.

39:47It's a club. Of course, Jason. It's in the name, Sam's Club. Oh, yeah. Come join us. Sam's Club.

40:24though, Mint's story is better. Every plan,$15 a month, even unlimited. That's it. Happy ending. Zero tears. Give it a try at mintmobile.com slash switch. Upfront payment of$45 for three months, $90 for six months, or$180 for 12 month plan required. $15 per month equivalent. Taxes and fees extra. Initial plan term only greater than 50 gigabytes may slow when network is busy. C terms. I find it interesting, and this is just more of a comedic comment than anything else. People, please don't read into this. When Elon was there with Doge working for the government, He required all government employees to send the top five things or whatever.

40:57Exactly what Jensen, what they talk about in this book that he does at NVIDIA. I guess Elon did this with government employees and who knows what was sent back or what was even read. I think it was actually Elon just kind of poking everybody, just letting them know who was in charge. But anyway, as a side comedic note. Okay. Let's talk about this other thing that we talked about earlier in the show with CUDA and how this was really important. What I think we failed to capture as we were just kind of quickly going through the summary was this was not popular when he first started doing it. A lot of people were like, why are you doing this?

41:31This has like five customers. It was like four academic people and one person who needed it for industry. So there was no money to be made by him creating this software CUDA interface for the GPUs. There was no demand. And one of the things that they talked about in the book, which I really liked, was he would create things based on what he thought was going to be needed because he understood the engineering. And a lot of the times you hear in the VC world or just entrepreneurs in general, they say, well, what does the customer want? What's the customer demand? What's the proof that there's something here?

42:09And when he started doing this CUDA thing, there literally was not any type of market demand. It was a total distraction compared to where they were making revenues. And it was almost like this leap of faith that he was just kind of looking at the sheer horsepower of computation that he was producing. And he just kind of had this gut feeling or intuition that there was something way bigger here and leaned into it. And he even talked in the book about how the shareholders, and he would take so much heat from people as to the R &D that was being dumped into this, especially when they're looking at the revenues that it was generating.

42:46So I guess my question to you is where in the world, because I'm left reading that and saying, well, I just don't know if I was in that position in time, whether I would have made the same decision that he made, which was obviously the right decision. So is there some skill in there that we're supposed to extract out? Or is the takeaway that you just got to be lucky, which I hate saying that's what it is, but maybe it is what it is. What was your take? Or maybe you read something in the book that kind of illustrated it better than what I picked up on. Well, I think it's a really interesting point.

43:18And I think building on what we were just talking about, this flat system, I think this flat non-hierarchical system allows them to pivot. Because if you look at most S &P 500 companies, when you have this huge employee bloat with 10 different layers of bureaucracy, it becomes incredibly hard for you to pivot away from your main product. Whereas I think when you've created this flat system with you at the top, ultimately, you're dictating you can turn on a dime. And I think that throughout Nvidia's history, it's shown a few times he's willing to turn on a dime. He starts going down an avenue and then immediately cuts it.

43:52So I would like to think that it was more than just luck because first off with like the NV1 chip, the first chip you kind of mentioned at the start of the book, he realized pretty quickly he was trying to do something called quadratic processing with the way to visualize gaming environments. And no one else was doing this. And it really, it was crashing all the Windows computers. It really was not doing what it was meant to be doing. And he basically was 30 days from going out of business and they got like a$5 million injection of capital from Sega. And they immediately were like, rather than continue to try and make it better, they dropped it onto the next thing.

44:28And then you saw, again, I mentioned it very briefly, with they saw the rise of the mobile phone market. So they were like, well, maybe we want to go into mobile chips. But there was already a lot of competition, Nokia, BlackBerry and stuff. And as a result, he was just like, you know what, this is not a zero to one market, it already exists, we don't want to take from them. And so he pivoted. And so I would like to think there was a bit of foresight in seeing this is where the world is heading as opposed to trying to compete in the prevalent markets that already exist. But it's definitely, it's a fascinating one.

44:58I think there's a point, I'm trying to remember the point. I've lost it. Anyway, I find it really, really interesting. I think there's definitely a skill to a certain extent, but I think that his ability to pivot, his flat non-hierarchical system enabled him to follow this instinct. Yeah. We often talk about, I know in the Bitcoin community, show me the incentives and I can show you the most probable path of what's going to come next or who's going to try to attack the network. And for somebody that's just so deep in this particular space, parallel processing, the manufacturing process of all this hardware, the software that goes on top of it from an optimization standpoint so that it can be used for other things.

45:42I think when you just, because he's so deep, another thing in the book that they talked about is his work ethic is mind-blowing, like waking up at 4 a.m. for decades on and just communicating with everybody. I think at that level, maybe he can just understand the incentives of what he's building, the incentives of the market, all the people that are out there using the hardware. And I think maybe when you're just sitting in that spot and you have humility at your core of who you are, you're not doing things for the wrong reasons. You're just doing it because you're deeply curious and you have just operational excellence.

46:18it allows you to see the incentives and the vectors to where they're pointing of where things are going next. And you probably won't get that out of him because I think he always errs on the side of when he's speaking in public to be ultra humble to the point where if he does have any secrets, he's guarding them and not telling anybody. He's just kind of like making you think that he really doesn't know where things are going, but maybe inside, he knows very deeply where he thinks things are going to go. And I think it's probably more of that than anything else. I think he puts on this perception in the public eye that he is just maybe lucky, but under the hood, he's deeply skilled and deeply knowledgeable on so many different domains that the person looking at it from the outside just cannot possibly comprehend.

47:08When I think we're also, we're trying to look at this industry to the average individual, to the layperson, most people have no idea what Nvidia is. I had a few friends ask me, hey, I'm going to go record a press and we're going to talk about this book. And they're like, what's Nvidia? Yeah. And I'm talking about people that have PlayStations and gaming consoles and computers, and the average individual just has no idea. And so, I think that to that point, when you're so deep in an industry and then all of a sudden, you start to see the rise of neural networks, and then you start to see the rise of like you create GPUs, and you can do parallel processing.

47:44And all of a sudden, you can ask a question to an AI, and it responds back to you. I think it becomes very clear, this is the path we need to go down, because you're seeing this stuff before anyone else. It's kind of like, when you're in Bitcoin, it makes sense. Why are we going down this path? And Bitcoin keeps going up because of printing money. But people that have never spent five minutes looking into Bitcoin, it doesn't make any sense. And so I think it's just because he's in that world. He's in that world so immersed in it. And I think the analogy that comes to mind is, I'd say, Apple with the iPhone versus BlackBerry.

48:16At the time, if you'd asked customers, hey, what do you want? Everyone would have said, I want a BlackBerry. I want to have a full-size keyboard on my phone. And then the moment the iPhone came out, it was something like within three quarters, BlackBerry had pretty much gone bust. And so it really shows that people don't know what they want because they can't envision things that they've never used before. They can only envision a future of things which they've used before. And to that point, people are using this and they don't even realize they're using it. It's completely masked out of their purview.

48:45They understand the iPhone because they are literally holding it and they hear it in the news and they use it literally on an hourly basis. But what they don't see is this computation, this parallel computation that's happening in the background that's like completing the word that you're writing. Apple does a terrible job at this, by the way, but it's completing your sentence for you. And how is that happening? It's happening because you have AI that's assisting in some of these computations in the background that are being run on NVIDIA chips. For all these people that have no idea what NVIDIA is, and I agree with you, I think most people are clueless to this company.

49:20And what's so crazy to me is this company in market cap value is a trillion dollars more than Apple, a trillion dollars for today when we're recording this. It's like$4.2 trillion in video. Apple's like 3.1 trillion. Just to kind of give people an idea of the sheer size and just think, how many of these things are there on the planet right now, Seb? And I'm holding up an iPhone for people that are listening. How many of these things exist in the world? And when you think through that and you're saying, wow, just take New York City alone. How many iPhones are there floating around New York City?

49:59Think about how many NVIDIA chips there are. And it's freaking mind-blowing, man. It's crazy how big this company and what they're making must be behind the scenes. And we don't see any of it because it's not something the normal person sees ever. Totally. And I think the other thing is when you read this book and even the way that we're talking about it, sometimes it may sound like it is a one directional thing. He's thinking up, this is where the market is going, and I'm putting these products out into the market. But I was listening to a talk by him and a lady called Cleo Abrams. If you just type in Jensen Huang on YouTube, it's the first one that comes up.

50:37It's got like 3.7 million views. So I highly recommend anyone going and listening to it. And one of the things he says is this is a reciprocal relationship. Like with the graphics processing units and parallel processing, They set the stage for this neural net and being able to process large amounts of data so AI could start to emerge. But they're not the ones that enabled AI to emerge. They just gave the tools to enable AI to emerge. Then you saw things like there was this contest called ImageNet. And ImageNet was essentially, the whole goal was we want teams to be able to take pictures and categorize pictures based on what's in the picture.

51:14And so if you've got your Google Photos and you look through your photos and you want to go and search, find a photo with a cat. Rather than having someone going and tagging cat, how do we get AI to go and do all of this categorization and tagging? And so there's a team called AlexNet and they used NVIDIA GPUs and they trained them through a neural net AI to start to recognize photos. And they went into this contest in 2012, blew away the competition. Very low error rates, completely blew away the competition. And so this was someone external to NVIDIA, seeing the benefits of parallel processing.

51:48So then NVIDIA then takes this technology, this advancement in AI, and then looks back, okay, how can we start using this with our GPUs? And so in this podcast, he talks about how the GeForce GPU, which is kind of their top of the line kind of gaming GPU, today, when they're rendering a 4K, let's say a gaming screen, or a realistic world, that 4K screen, there's 8 million pixels on the screen. Well, traditionally, you would have had to have rendered all 8 million of those pixels using the GPU. Well, today, they only render 500 ,000 of the pixels. The rest are all rendered by AI. And so what that means is because their focus is now only on 500, they can put way more effort into that 500, make far more detail in the 500 ,000, and then AI is able to kind take that and create a phenomenally realistic screen, but it makes it far more efficient.

52:40And so there's kind of the symbiosis where they're creating the technology, the technology is being used for AI. AI is then being used back on their technology. And so it is reciprocal in this advancement as well. Another just kind of ad hoc comment on what you're bringing up there is this is just the idea of compression. So when you take a WAV file, an audio file that is really big and has like all the raw data in there and you compress it into an MP3 and you play it on a device, it sounds exactly the same, but it's just compression. It's a compression algorithm that you use to take the WAV file and make it much smaller without our ears really being able to notice the difference.

53:16And so what's AI doing? AI is compressing data. If it's taking something that's a 4K image that has all these megapixels like Seb was laying out, then you're able to compress that into a process and procedure to render it in a way that puts it up there, you're effectively doing the same thing. You're just using different means of compression that can be applied across almost any type of file type. And I think that that's really beyond fascinating. I can only imagine where some of this compression AI is going to take us. It's going to be nuts. Again, this is one that blew me away. I started digging into, in the book, it talks about this called the DGX1.

53:56And at the time, the DGX1 was kind of, this was in 2016, it was top of the line GPU processing, and correct me if I'm wrong, in its function. But basically, it was being used for AI to basically train these neural nets. And it was$250 ,000. And the first one was sold to OpenAI. Elon Musk received it into the office. And it was like absolute top of the line at the time. And what's fascinating is on this podcast with this lady, Cleo, he brings in, and this is eight Eight years later, this is 2024, the podcast, he brings in a mini version, which is one tenth of the size. It's got six times the processing power, and it uses one ten thousandth of the energy expenditure.

54:35This is in eight years. And so we talk about this problem, which is where is all this energy going to come from for all of these AIs? Where is all this energy, these massive data centers that are crunching these numbers? But in eight years, we have reduced the energy expenditure by 10 ,000 times. That's just mind-blowing. Yeah, that's nuts. The last thing that I want to talk about, Seb, was this idea of his speed of light principle. Do you remember this in the book, the speed of light principle that he brought up? Refresh my memory. Okay. So he was trying to figure out, this is in production, and this is probably one of the reasons I like this, because guys, producing anything that's physical is super difficult, especially when you're competing against other people that might bring something else to market and that makes your product obsolete.

55:22And we talk about this a lot in Bitcoin mining and how it's so difficult to compete in that space. So you're thinking through Jensen and he's building all of this hardware and the competition is crazy fierce. Well, he had one of his employees that was looking at the entire production line of all the parts and pieces to make these really complex end items. And he asked the executive, he said, how much would it cost to have this to us at the most breakneck pace that you could produce it? And the person came back and they were like, it would be this many days and it would cost this much. And Huang was just like, there's no way.

56:00It's faster than that and it's going to cost more than that. There's no way that that's the timeline. And the person that was working for him was somewhat taken back and And they're like, no, that's what it is. I asked the suppliers and the vendors, and this is what it is. And he says, that's not right. And he's like, public lambasting, boom, you're done. Get me the right answer. So the person comes back and they said, as they talked to each one of the vendors, the vendors could do it faster, but the price was so outrageous that they didn't even quote them that price. They got Jensen the answer that he wanted, which was they could have it.

56:37And I'm exaggerating because I don't remember the exact numbers from the book, but it was something like we could have it there within a week or three days, but the cost would literally be this crazy, insane amount of money. And Jensen was like, that's the answer. That's the answer I wanted. I don't want the vendors to come up with what they think the answer is for us, because maybe we have a buyer that would want it in the three days and not the two weeks that you were telling me it would take. And he came up with this principle, which he called, I think it's called the speed of light principle or the price from physics is really kind of what he's getting at.

57:14And the reason he wants to know this number is because it's almost like in the universe, the speed of light is the one number you can't exceed. He wants to know that when he's manufacturing something because, hey, maybe he might have an Elon Musk that comes knocking at his door and say, hey, I want to buy$10 billion worth of GPUs. How many does that get me? I don't care about how many I get. I just care about the time. Or I have somebody that's very price sensitive and they don't care about the time. But knowing that number in production is so vital in program management land, they call this the critical path.

57:52But I think this idea that he's talking about in the book goes beyond the idea of critical path because a lot of people just kind of take the quotes that their vendors give them and they plug it in and they figure out what the serial and parallel tasks are. And they say, okay, this is my critical path and this is what it's going to take. But Jensen's like, no, I want to know the speed of light. I want to know absolutely the best you can possibly do and whatever the cost is, I don't care. Just tell me that number. And then he pieces that together. And what this gives him is the ability to actually figure out what pricing should be by dissecting each one of these swim lanes at each one of these things.

58:28And if you're a listener and you're a program manager, I think that this is a really important idea because it forces you to figure out what you think the cost should be versus what you're being quoted the costs are by the vendors. But yeah, no, I found that really interesting. Anything you want to add on that particular idea, Seb, or anything else in the book you want to... Previously, it was a bit of a throwaway comment, and we kind of touched on it, which is this idea that along those lines, as a result of this, he completely changed the industry. Because I think up until that point, from my understanding, chip cycles tended to be yearly, every two years.

59:05And he managed to cut it down to every six months new chips were coming out. Because this kind of gets back to that point of just like, iterate, iterate, iterate, execute, execute, execute. It's just like, we can completely change the world we live in, but we've got to constantly be pushing the limits. We've got to constantly be pushing the limits. And I think it's a phenomenal mind to try and actually figure out what are the boundaries of my ability to create change, as opposed to just taking for granted what other people are telling me my boundaries are, even when those are not really the boundaries.

59:34Yeah. It definitely speaks to his, how proactive he is, as opposed to a passive leader. You know, if you're a passive leader, this guy would just eat your lunch. He would destroy you. He'd destroy you. I'm definitely, so you know what, like the one thing that I'm curious about, and this is again, like, I don't want to psychoanalyze. I think what he has done, and I'm going to preface it by saying, what he has done is truly profound. The world we live in today would not be the world that would not have the technology we have today, would not have the AI we have today if it wasn't for NVIDIA. I had no idea to what extent they have completely shaped this world.

1:00:07But I wonder, there's an argument, not necessarily an argument, but there's an interview at the very end of the book, basically in the last two or three pages, And the author asks him about what do you think are the risks of AI? Oh, yeah. I wanted to cover this. Yeah, this is huge. Go ahead. Sorry. He gets slammed, absolutely slammed. And one of the questions he kind of says is, and I think to quote, he says, we invented agriculture and then made the marginal cost of producing food zero. It was good for society. We manufactured electricity at scale, and it caused the marginal cost of chopping down trees, lighting fires, carrying fires and torches around to approximately zero.

1:00:45And we went off to do something else. And then we made the marginal cost of doing calculations, long division, we made it zero. This company is not a manifestation of Star Trek. We are not doing those things. We are serious people doing serious work. And it's just a serious company. And I'm a serious person just doing serious work. And he kind of reiterated that. And so there's a part of me that wonders like, where does this come from? There's almost a fear to talk about what What are the repercussions? Yes. And there's another one quote I'll quickly share, which is the author goes and speaks to other people in the company as well.

1:01:17And the other people said, I recall the discipline of NVIDIA's executives I talked to. Jensen had them wound as tight as piano strings. They were confident, intelligent, and exceptionally well-prepared down to the smallest detail. I never once caught one slipping. I recall too with sudden clarity how disinclined those same executives have been to discuss the potential future implications of the technology they were building. The disinclination that sensed is spilled over from the discomfort, even fear from Jensen. And so you wonder, I wonder where this came from. And what comes to mind, and I'm curious to hear your thoughts, is it mentions at the start of the book a few times, but he came to Canada, sorry, came to the US when he was 10 years old.

1:01:59And he quotes like, you're always an immigrant, I'm always Chinese. He was younger of two brothers. And so he's kind of always looking up to his younger brother. And I have a sense that he feels he needs to prove himself. He needs to prove himself, hence the comment, I'm a serious person doing serious work, as opposed to just being able to step back without taking it personally. But I'm curious to hear your thoughts on that. I found this so interesting that anytime the implications of AI and where this is all leading came up, he went out of his way to just almost make the person asking the question feel super small, like they're really stupid for asking such a question.

1:02:37And he's just hiding. He's hiding from this question. He hates this question, like really hates this question. And I guess that hatred for the question is probably one of the most interesting things about this entire book. And I almost missed covering it. So I'm glad you brought it up. Yeah, why? It's a fear. It's definitely fear driving this because it's not a normal reaction. Everything else that he does is just very balanced and like, oh, you know, I don't know. Yeah, like I'm very successful and it was hard work, but I don't even know if that's it. It's just this very casual response to everything, but this question.

1:03:14Isn't that crazy? In psychology, there's this kind of question you ask yourself, which is anytime you get worked up, ask yourself the question, is my response in line with the stimulus? If it is not, then I'm probably responding from some past event. There's a book that kind of comes to mind that I loved way back when. It was called The Talent Code. And it was like, why are people successful? We covered this on the show. is it dan daniel coyle i think we interviewed the author on this yeah yeah yeah yeah yeah and so this book came out i don't know maybe like seven eight years ago it's a phenomenal book and it talks about how they looked at the world's fastest hundred meter sprinters and out of the world's fastest hundred meter sprinters on average they were one of 4.6 siblings and they were on average the fourth sibling so they were nearly always the youngest and so i was curious i looked up, was he the younger of his brother?

1:04:04And he was. And so, there's a part like, is he trying to, he wants to prove himself. He wants to add value to this world, but it's almost kind of like clouding this question around like, what are the repercussions of AI? And I don't want to diminish the change and the profound technology that brought into this world, but I find it really fascinating. That is really fascinating. I found it so bizarre because that theme came up multiple times in the book. And that question just kept coming up. And then his response each time was just so aggressively putting the person down for asking it. And yeah, so I agree with everything you're saying there.

1:04:41I just don't know why he's so scared to answer the question, because he's clearly like, it makes him upset. So I don't know. If you're a listener and you know you work at NVIDIA or you know maybe more behind this, throw it in the comments when we post this up on X. We'd love to hear what you got. Seb, we have a lot more we could cover here, but you know what? I don't want to cover it. I want people to read this book. This book was really, really good. We'll have a link in the show notes to the book. Again, the name is The Thinking Machine, and this is by Stephen Witt. Stephen Witt, bravo. You did a phenomenal job.

1:05:14For our listeners that are tuning into more book reviews, Seb and I just having fun reading all these things that we find fascinating, we're going to try to get the authors to come on with us. And if they don't, we're going to record anyway, because we don't care. We might have more fun without the authors. I don't know. But we're going to invite the authors on the show from time to time. What else did I want to cover? Oh, I wanted to tell people about the next book that Seb and I are going to work on. The name of this book is Empire of AI. And this is about the inner story of OpenAI and Sam Altman.

1:05:51And I have a bit of a bias. I got to say this bias up front for people. I'm not a fan. I'm not a fan of him. I really don't. Everything that I've read online, and again, I haven't researched him all that much, but the little bit that I've read online, he just doesn't. And what I've seen is that he's not really the best person. But regardless of that, what they've done at OpenAI is mind-blowing, totally mind-blowing. So this book was written by Karen Oa, something like that. But anyway, that's the next book we're going to read. So So if you guys want to read it and you want to be prepared to hear our conversation, have at it, we highly encourage that.

1:06:30Seb, anything you want to say about the next book real fast before we wrap this up? Oh, you know what? You basically took the words out of my mouth, which is, don't get me wrong, I use ChatGPT. I think OpenAI have done such a phenomenal job and it really has laid the foundation for this AI revolution since they released in what? The end of 2022, early 2023? It's profoundly changed the world. but then you see some of the ways that Sam Altman acts in society in the way that he talks about what's happening and it brings up questions. And so I'm curious to see what this book talks about and whether it goes into some of these things.

1:07:03Yeah. The subtitle on the book is one of the reasons that I was sold as soon as I read it. The subtitle is Dreams and Nightmares in Sam Altman's OpenAI. There was really good reviews online so that we're plowing into that one next. With all of that said, how awesome is Seb Bunny? right? Seb, we are so excited to have you on the show. I don't know what our frequency of doing this is going to be, but regardless of what it is, I love having these conversations with you because you and I have these conversations in real life and when we get together and hang out from time to time. And I just knew you were the perfect person to kind of do these book reviews with.

1:07:40And you have your own book. It's called The Hidden Cost of Money. And if people haven't checked it out, this is a Bitcoin book, of course. And if you haven't checked out Seb's book, You got to read his book. He, as you can see on the show, he's crazy thoughtful. He has read tons. You see the books behind me. He has, I'm sure, just as big of a library somewhere in his home. But Seb, thanks for making time and coming on the show. Anything else you want to highlight or point people to before we finish this up? No, I just, again, like you're too kind for us. And then I just feel so lucky to be on the show.

1:08:10And I shared this, I think the first time I came on the podcast, which is I've been listening to Preston since probably over a decade now. And for going from listening to you and Stig talking about the books, seeing the evolution of the show, to bring it back to talking about the books, I just absolutely love it. Being able to share information, talk about these things, talk about how the world is changing. I feel incredibly grateful. I think we need to tell Stig to read one of these, and he can join us on the conversation, too. He needs to get back in the mix here. Seb, thank you so much. We're going to have all the links to this in the show notes if people want to check out anything that we talked about.

1:08:47and thanks for joining us.