From Near Death to a $20B NVIDIA Deal | Jonathan Ross, Groq

Key Takeaway

Replace open-ended requests for opinions with an intent statement on one decision this week: “I intend to do X because Y. What am I missing?” Jonathan Ross argues that asking “Should we?” invites reflexive pessimism, while declaring intent mobilizes people to solve the problem and surfaces only material objections. Pair that clarity with a simple measurable goal, then give capable people room to surprise you with how they achieve it.

Episode Overview

Jonathan Ross, founder of Groq, explains how combining Groq’s LPUs with NVIDIA GPUs led to a major partnership and why faster AI inference can improve model quality, not merely response time. He also shares hard-won lessons on autonomous leadership, managing organizational politics, hiring, capital constraints, and using AI to turn question-asking into a core human advantage.

Full Transcript

Transcript of From Near Death to a $20B NVIDIA Deal | Jonathan Ross, Groq from David Senra. Auto-generated from episode audio; may contain minor errors.

Let's start with this alleged $20 billion partnership you have with Nvidia. Can you tell us about the structure of the agreement and how it came about? Well, the most interesting thing is that the idea first came about 3 weeks before the money was in the bank. Well, Johnson moves fast. Of course, that's how you stay ahead of the curve. Yes. How did it come about? Well, we had been working on GPU and LPU integration. The best way to describe why this helps is if you were building a logistics network for the United States and I told you that you could have 18-wheelers or vans for last-mile delivery, which would you choose?

And the answer is both, right? And so , the combined GPUs and LPUs ended up delivering better performance across all performance curves. We had implemented it and had gone to Jensen to ask if we could buy about 100,000 GPUs because we were going to implement them ourselves. Jensen saw what we had done and thought it might be best to make them available to all his customers. You and I had this conversation at NVIDIA GTC and you were talking about the fact that these technologies are very complementary.

Can you explain that a little more? When you process an LLM token , what happens is that you are doing all these different matrix multiplications . Some of them are more restricted in terms of computing power, and others in terms of memory performance. Those that are more restricted in terms of computing power we put on the GPU, and those that are more restricted in terms of memory performance we put on the LPU. And bottlenecks are everywhere. There are all sorts of different bottlenecks. There is no single strategy.

There is no such thing as perfect architecture. So, the conclusion was that by combining these two things, the bottlenecks in all the different maples are overcome. There's another thing I love that you said when we had this conversation: when AI communicates with other AI, speed is becoming increasingly important. Yes, I mean that a human can expect to wait one or two seconds to receive a response when typing a command into the computer. The AI ​​is simply there waiting because it produces these tokens so frequently that it thinks much faster.

Now, if you incorporate LPU, the speed is much greater, so it all comes down to how fast you can move. AI is really good at using AI. That's really what Agentic is, right? Humans benefit from AI, and so does AI. Just as you would research me before I appeared on your show, the AI will begin to research different tools it will use while using this other tool. And so it is passed on to another AI. And that's how you get this exponential growth. I'm going to digress from the topic for just a moment and we'll come back to it shortly .

I spoke with many founders about this recently, but the issue is that when agents make payments, the number of payments that are going to be made is going to skyrocket. Do you have any ideas about this ? Yes, I think it's still early days and one of the limitations is that payments aren't yet designed for this, but if you can make micropayments, the number of payments will skyrocket. I did a small hobby project and for the hobby project I needed a couple of different phone numbers so I could have a bunch of different agents with me on Signal and WhatsApp.

And I had to go to Twilio, prove that I was a human being in order to get the number and all that process, and it turned out to be a huge headache to be able to finish it all. On the other hand, if I could have simply allocated a budget to the AI ​​and it had been able to use it, it would have spent it without me ever noticing, and it would simply have stayed within the limits of the established budget. Tell me more about your hobbies and side projects, because this has come up in a couple of conversations we 've had.

I like doing cutting-edge things on my personal computer. So I don't have access to the work's codebase , which involves risks. For example, I set up a server on GCP or AWS and start creating things. I've created everything from apps that tell you how long it takes to travel by plane, which routes have the best seats and all that, to apps that do certain mathematical operations, like the daily report and all that. Very simple things. I use it a lot at work, but I always start with a hobby project before bringing it to work.

What is the daily report? Every morning I receive an email that tells me what's happening in the world based on what interests me and how I interact with it, and a lot of research. It's very similar to the daily presidential report , but customized for me. And is this in text format? You read it. It's a text with links that I click on to find out more. However, the big change I made to the daily report was that I started receiving a lot of text and reading it, and then I realized it was AI.

He has done a lot of research. It has the context. Why don't I have him summarize it all? Just give me a bunch of headlines and I can ask follow-up questions. I just spent some time and recorded an episode with Gustaf, who is the big boss at Spotify right now. And he did something very similar because he tries to avoid any kind of feeds. The entire thesis underlying Spotify's organizing principle is that time is well spent. And he says, "Well, you know, me just surfing and getting angry about X isn't helpful, but there's some information I want to know." And then, there are a couple of different places their agents go to and basically summarize things that might interest you.

He says, "Hey, don't include anger bait. Don't include politics." But he also emphasizes that these are the people who interest me . Who are the people I'm interested in talking to? And if the same people I'm interested in are talking about the same topic, I want to know about it. And then he can read it, but now he's turned it into a kind of personal podcast and actually listens to it on Spotify. It's like his daily briefing, but he listens to it in a matter of 5, 10, or 15 minutes.

Yes. I mean, for me, interactivity is important because it's kind of like... Did you ever play the 20 Questions game when you were a kid? Yes. So, with 20 questions you can find out what someone is thinking. Questions allow you to distill what truly matters to you. If you have a podcast, it's static. You can hear it, but if you can interact with it, you can get the information you want most. So, since I'm learning through AI, I 'm not reading static content. I am interacting with him.

I'm glad you mentioned the questions because I have a number of your quotes that I saved, and by the way, I love your tweets. You said: " Success in the information age was about being able to answer questions. Success in the AI ​​age will be about being able to ask the right questions." Can you elaborate on that? Yes. And this also points to a shift in how people move from being individual contributors or collaborators to being all leaders, but AI leaders. And what good leaders do is not do the work themselves .

They don't have the answer themselves. They just ask the question. They're just con artists, you know, who take everything they hear into account and then ask the question that no one else asked or that everyone is thinking about but is afraid to ask. So with AI, because it can go out and solve all these problems for you. You can do the research report. Because the question you ask determines what you get, and that determines the outcome. Since the information age, we have all been trained to answer questions.

That's what school is all about. Remember this, remember that, remember this. With AI, just ask AI. She knows. You just have to think of the right question. It's a fundamental change. I want to tell you about the main sponsor of this podcast, RAMP. I've been reading a lot about SpaceX lately. SpaceX is one of the most valuable private companies in the world. And one of the main themes in SpaceX's history is constantly attacking and questioning its costs. Ramp helps many of the world's most innovative companies do exactly that.

The average company that operates with RAMP reduces its expenses by 5%. And one thing SpaceX has demonstrated is that a religious dedication to cost control can help increase revenue, because you can take advantage of opportunities you otherwise couldn't. And we see that in the RAMP data as well. The average company that uses RAMP also increases its revenue by 16% . So when you manage your business with RAMP and your competitors don't, you have a huge competitive advantage that builds up over time. Ramp is the only platform designed to make your finance team faster and happier.

Many of the best founders and CEOs I know operate their businesses with RAMP. I run my business with RAMP, and you should too. Visit ramp.com to learn how they can help your business save time, save money, and increase revenue. That's ramp.com. I want you to write to me about your views on leadership. Yes. What is your description of leadership? I got it from a John Levy book, but it didn't go into much detail. The first principle of leadership is to have followers. TRUE? It's that simple.

You are not a leader unless you have followers. But when you think of leadership as having followers, it's also like thinking of investing as making money. There are many ways to be an investor. You can be a venture capitalist. You can give debt where you can take equity. You may be in the initial stage. You can be a Series A. You can be growth. You can be a crossover. You can make convertible promissory notes. As if there were private capital. There are many different ways to be an investor.

Public market, right? The same applies to leadership. And with leadership, the mistake I often see in new founders is that they say, " How can I be a leader?" And the problem is that they don't realize there are an infinite number of ways to be a leader. And then they go and listen to someone and receive all that advice and try to put it into practice, but it's not true for them. That's why one of the things that struck me as very different as a leader is that most of the people I listen to on your podcast are control freaks.

They want things done their way, and that's true whether it's on the Founders Podcast or this one, or both , really. That's why, if I had tried to be a control freak, it wouldn't have worked for me. I'm one of those rare people who can go to a restaurant and tell the waiter, "Bring me what you think is best." Or I don't even have a driver's license. I haven't had a driver's license since I was 18 . Why don't you have a driver's license? Because I don't feel the need to drive and I prefer to think.

I don't need to control the driving. I want to control my thoughts. I want to concentrate. I want to be with my phone. I want to do something useful. And it has been like that since I was 19. That's why I love delegating things in a way that others don't. One of the things that makes my leadership different is that when I hire people, I hire very autonomous people who tend to act on their own and who would be terrible in most corporate environments. But I can't hire the same kind of people either.

And that's what happened to me. If I had done the opposite, I wouldn't have been very successful. If other people do the same, they won't succeed. On the other hand, there are many things that are very similar, such as the fact that most of the best leaders in Silicon Valley lead from a place of inspiration for their people rather than trying to make people afraid. But that is one form of leadership. There are many leaders out there who are successful because they manage to make people feel very afraid.

So you just have to choose the leadership style that best suits you. But also, when you choose where you're going to work at the beginning of your career, you should probably work in a place where you'll learn lessons that are good for you. If you're more the type of person who can show appreciation and gratitude, you probably shouldn't work for someone who inspires fear because you're not going to learn any lessons you can use yourself. Wow, this is very important. This excites me so much because, tell me, have you ever spent any time with Toby Luke?

No. Okay. Well, the conversation I had with him, I don't know, it was on the show about five months ago. I still think about her every few days, and in fact, we go over the previous episodes and mine, and I constantly come up with new ideas. And that's why you see all these clips we're posting on X, which could be from things we did 6 months ago. And Toby, one of the things I love that he said in the conversation, was something like: " Dude, there's no one right way to do things." There are probably a hundred ways to achieve your goal.

You have to choose the one that's based on you. This is why you see all these technology companies that are so different from each other. Like Apple, which is completely isolated. Google , everyone has access to the codebase. They are completely different forms taken to the extreme. We just had Dana White on the show and all he said was that the first thing you have to do is know yourself. So I'm going to ask you a question once you've discovered your leadership style. Wait a second.

And then you really need to know who you are, okay? And what suits you. And the second thing is what you really want to do in life. So, once you have those two most important questions, you simply wake up once you figure them out and focus on achieving whatever it is. Once you've solved step one, you wake up and tackle step two. So, when you realized you could n't work, your leadership style wasn't typical and you needed autonomous people who, by the way, if you ever started another company, I imagine it would just be you and a group of agents.

It would probably be AI. Therefore, I believe this will be a shortcut for founders in the future, and we can delve deeper into it. But the first thing you have to do as a founder is to move from the technical things you know how to do and with which you can add value to learning how to manage people, and that probably took Grock 3 to 4 years. Tell us more about this. I was a terrible leader. He was one of the worst leaders in the world when I started.

Well, I gave people too much freedom because I tend to delegate, but I trusted people who probably shouldn't have been given that level of autonomy during that time. I didn't hire people who could operate autonomously, but I was someone who delegated and gave autonomy by nature. So what ended up happening was that things came to a standstill because they didn't know what to do, and I wasn't telling them what to do, and they were used to being told what to do. And in the end I got so frustrated that I would go in and tell them what to do, but it felt so unnatural to me that they wouldn't accept it.

Why do you find it so difficult to tell others what to do? I work through questions. I like to set a clear direction. So , for me, my leadership style consisted of setting a goal that would take me a while to achieve, but that was so simple that I could put it on a challenge coin and give it to everyone. So at Grock everyone had a coin with 25 million tokens per second and an upward trend, and everyone knew that was what we needed to achieve.

And it's kind of like when you ask an agent, an AI agent, what you know how to do to do something. The fewer restrictions you place on him, the more freedom he will have to solve your problem. That's why I loved working with incredibly creative people who came back with surprises. Repeat that part . Repeat that part. This is important. The fewer restrictions you place on someone, the more freedom they will have to solve the problem and the more freedom they will have to surprise you with the solution.

Therefore, if you want to run a highly creative and innovative organization, what you really want to do is minimize the number of restrictions, but you also have to give them the things that matter. If you are unable to clearly summarize what you are trying to achieve, then you will either restrict people too much or too little. When you try to do something as a founder, you are inherently trying to disrupt an established industry. There is a moat. You're trying to do something differently. If you don't do anything different, what's the point?

They are already well-established and well- funded companies. If you can assemble a team that is also disruptive and innovative, and not just you, then you go from being Superman to being the Avengers. And that was my natural instinct for leadership. If I had been more of a command and control person, I should have doubled down, but you have to do what comes naturally . Have you spent any time studying Kelly Johnson, the guy who did Skunk Works at Lockheed? Not too much. Okay. He has a great quote where he says it reminded him of what you just said when you did the co-coin challenge for Grock.

He says that extreme performance often stems from a brutally clear priority. Yeah . Yes. And you see it because if you don't give people a goal that's clear enough but limited enough so that they do n't pleasantly surprise you with the result, then you're not giving them the ability to innovate. I want to double down on that for a second. The only way for your team to innovate, without you being the innovator, is if they are able to surprise you in a good way, which means you shouldn't limit the objective too much.

And what you're saying took you 3 to 4 years from the founding of Grock to realize that in basic management, dealing with people is like doing small things. One of the great things I learned at NVIDIA... In fact, we're going to learn a lesson from NVIDIA because I'm there now. Jensen is top- notch. And one of the things I've noticed because I've worked at other tech companies is that there's no politics. There is no policy; It's the least political large organization you'll ever see. And I learned this lesson in Grock, but I didn't take it to the extreme.

So seeing it at the extreme shows me how valuable it is. There is no circumstance at NVIDIA where Jensen has one-on-one meetings with people and tells them one thing. I learned this at Grock because what would happen was that I would have a conversation with one person and then I would have a conversation with another person and then what would happen was that they would both hear very different things and talk to each other and come to very different conclusions. But when I had a room full of people and I said something to them, it was amazing how everyone heard the same thing.

That's why, when you're managing groups of people, if you want to reduce the amount of politics and people getting out of control and forming clichés and all that, stop having one-on-one meetings. Hold large meetings with all the people you want to tell something to and tell them everything at once. And don't let anyone email you. For example, send a copy to everyone in the email. Like when someone sends you something. For example, if someone says, "Hey, this person is making a mistake here." Copy that person to the email.

Let him intervene. Otherwise, you would be allowing politics to take place. What else have you learned from Jensen? I wanted to be too clever playing chess in 3D. While Jensen is very similar to "what does the customer need?" I want to build trust with the client. I always want to tell her things that are true, that I believe in, and that I can stand behind . And if I have something that isn't what the customer wants, I'm not going to sell it to them. I'm going to sell customers things they really need and that I think they need.

I'm not going to think about how to create notes. How do I do all this? It's simply what the customer needs. Simply create that for them and everything else will fall into place. I did a few Founders Podcast episodes about Jensen. One of them was How Jensen Works, which basically removed all the biographical information that was in that book, NVIDIA-style. And I think there are about 19 main ideas that I cover in that podcast. But one point we mentioned recently is, well, if you're a founder today, what will it be like to found a company in the future?

Perhaps it's simply you, a co-founder, and the assistance of about 10,000 artificial intelligence agents. And Jensen has this great line about how people are afraid of managing AI agents that might be smarter than them. He says, "I'm already doing it." He has, I do n't know, like 60 direct subordinates. He says: "Each of them is smarter than me in their field, and I have no problem organizing and managing them." And I thought it was a great metaphor. One of the things good founders do is ask questions again.

And what a good founder is able to do is, even if it's not their area of ​​expertise, someone approaches them, says something, and asks a question, and that person says, "Oh, [ __ ], I hadn't thought of that." TRUE? Over and over and over again. It's a skill you can perfect. It's in every book about Bezos. Yes. And I think it's universal. I believe any good founder is capable of doing it. And again, going back to those early stages of going from non-founder to founder, you'll end up hiring people who will tell you, "No, no, I'm the expert in this field." In other words, trust me.

You're a child, trust me. And you have to learn to be confident. Well, one of the things that was very useful to me in the early days of being a founder was that, when we were about 35 people, I was able to closely follow someone who was running an organization of 2,000 people. And it was funny because there was no confidentiality agreement or anything, but at every meeting we went to, he'd say, "Oh, this guy has a confidentiality agreement. Don't worry, you can say whatever you want in front of me." And I thought, "Okay." But we went to all the meetings and I would just sit there in silence thinking, " What would I do?" And in the end, every time he made a decision, it was exactly what I would have done.

What I hadn't realized until that moment was that I lacked confidence. Part of the problem with being a leader was that you needed to have confidence in a decision so that other people would have confidence to execute it. Many people are overconfident. Some people have very little confidence. The thing is, if you have very little confidence, you're probably the type of person who overthinks things, but you still need to reach a point where you act with confidence. And when I realized that I was making the same decisions as this very experienced founder or CEO , I started to act with confidence and people started to follow my instructions much more.

I didn't change my decisions, but I did change my leadership. How many employees did you have at Grock when you partnered with Nvidia ? About 450, but it was more difficult to manage than a typical group of 450. So, in the military, depending on your rank, you are allowed to have a certain number of people reporting to you. And the higher your rank, the more you can have as an officer. But when you have scientists reporting to you, the number is actually much lower, drastically lower.

Because they had such a creative organization, they were much more difficult to manage. The problems became apparent. So it was probably more like managing a group of 5000 people than 450 in many ways. In other ways, it was like managing an even smaller group because innovations simply arose on their own. But the better the people were, the harder they were to manage. Can you explain what state Grock was in? Yes. I think at one point, you know, he was almost out of money. Isn't this accurate?

Very early on at Grock, and this is a lesson for founders because if you're doing a business that requires a lot of capital, you're going to need to raise money. And one of the things we went through is that we raised money from certain investors who fell out of favor with others, and since they didn't want to co-invest, every time we tried to raise new funds, we just ran into difficulties . And there's also a bit of biodality in how. Can you tell us more about ...?

Well, the way I like to put it is that typical West Coast venture capitalists are more like lemmings. And the typical East Coast venture capitalists think they're smarter than everyone else. So, when looking for capital on the West Coast, if one fund invests, everyone wants to. In New York, a single venture capitalist investing means nothing. They are going to conduct their own analysis. They don't care at all what other investors do. The other side of the coin is that, on the west coast, if one venture capital investor declines, they will warn the others and , like lemmings, they will all reject it as well.

We had this problem because no West Coast venture capital investor wanted to invest in us. In the end, we had few of the typical venture capitalists on us. We had like a bunch of East Coast investment funds investing . Wait. That's really funny. The biggest deal Nvidia has ever made, by almost 3 times, and the West Coast VCs lost it. Everyone chose other things that were safer. That's why there's something called the Keynesian beauty pageant. Have you heard of him? No, I don't think so. Okay.

John Maynard Kees, the economist. Well, the idea is that it works in parallel with venture capital, and when you see this, you can start to understand some of the behavior, some of the behavior of lemmings, because it's actually a good idea to follow other investors. For example, in the Keynesian beauty contest, imagine I give you a magazine with a bunch of models. And your job is to choose the models and say which one is the prettiest. But what determines who is the most beautiful model is not who is the most beautiful.

It's who has the most money invested. And that model, depending on how much money you invest, you get back all the money that has been bet on all the models. And if your model doesn't win, you get nothing. You lose all your money and it goes to the winner. Well, if that's the case , if someone has a lot of money, they can bet on any model they want, whether she's pretty or not. And a lot of people made other bets. It doesn't matter. They simply walk in and make a big bet.

Now, when you see some of these bets happening in Silicon Valley, where people are making big bets, it's because that used to be what won. However, what has changed is that, unlike the Keynesian beauty pageant , where the winner is the one with the most money, in reality, there is a point where you get enough money and you don't need any more. And for the first time in history, startups are not hungry for money. They have everything they need and more. So now everyone is getting the funding they need.

And investing more money is not an advantage. But people continue to act as if investing more money gives the startup an advantage. Okay. Can you tell us the story you told when we invaded GTC together about the drastic change in Grock's fortunes and how quickly it happened? Yes. Going back to what you asked about us almost running out of money. We were about 3 weeks away from running out of money at one point. But that was many years ago. That was many years ago. And you weren't, you weren't even close to running out of money this time.

No, no, no, no, no. We were doing well in that regard. You were doing well, but the value of the last valuation you raised compared to this deal was drastically different. Actually, it was only a little more than double. Well, it wasn't a big leap. And we were also able to raise the amount for which we made the license. Okay, let's go back to what we were talking about on stage, about how quickly they came up with this idea and how quickly it changed Grock's entire trajectory from that moment on.

Well, it was that 3-week period from the time we submitted and requested to buy GPUs until not only was the deal closed, but the money was transferred. But how many months before were you working on this? There were probably three or four, maybe a little more. But what happened was that, at first, I didn't think it was going to turn into something so serious. Well, I didn't suggest doing it. In fact, it was Sunny, my operations manager. This brings us back to the whole issue of autonomy.

This is the story I want. Yes. So he had the idea to try and merge our chips, and he didn't have to explain what you mean by merging our chips. The LPU and GPU, as mentioned, are better in different parts of what is called the decoding layer of an LLM. The GPU is better at attention processing, and the LPU is better at applying weights, which is what gets trained instead of memory. And what we realized was that this is what most people do wrong when they try to do it on their own.

They will be responsible for processing the reading of tokens, also called prefilling, on one piece of hardware, and then they will generate the same tokens on a different piece of hardware. But token generation is the difficult part. That's the idea. You know, reading is easier than writing, right? And that's also true for AI. And then what we discovered was that, again, this was all a group of people. It wasn't just one person, it was a group that innovated. Once he brought up the idea of ​​why we didn't put them together because there were different bottles, then the team thought, "Oh, this part goes here , this part goes there." We implemented it and it worked.

And we weren't afraid to show it to Nvidia either, because we wanted to become Nvidia customers. We wanted to buy a GPU. And what ended up happening was that we went, we presented it, it made a lot of sense, and the deal was closed. Okay. From Jensen's point of view , once he sees this... Yes. Of course. Why does he decide to act so quickly? This is the nature of a successful entrepreneur. You move fast. You don't wait. Clear. Waiting has an opportunity cost. Technology is not a business where you can wait a year.

My question is, why is this so important for your business? Right now , when you start using AI, it might seem fast, but that's because you 're not used to using it much faster. It's like when you first used the internet, it seemed faster than sending things by mail. But when you have broadband, you realize it's much better. I'm never going back. The difference is that broadband required people to optimize their websites to make them more usable. If the servers are slow, you don't get any benefit.

Therefore, it took some time to implement it, improve it, and obtain videos that could be streamed and all that. The difference is that you put these LPUs in a system and suddenly token generation becomes faster. It's like getting instant broadband on these existing models. And now, instead of having to wait a minute for an answer, you can get an answer in 10 seconds, and that really starts to add up. Well, let me give you an example of why it's not just about speed, but also about quality.

Another thing I did was create Google TPU , and at Google there was a point where I had already moved to Google X. I wasn't working on TPU at that time. I had already done that, and someone else from the TPU team who was at Google X came and showed me an email from DeepMind that said, "Hey, we're having a contest. We think we're going to lose. There's a prize pool. Is your chip as fast as you told us?" And we said, "Yes." It's like in Ghostbusters, when someone suddenly asks you directly, " Are you a god?" And you say, "Yes." And someone asks you, "Is your chip as fast as they told me?" And you say, "Yes." So we answer yes and they tell us: "In 30 days there will be a big contest." We were going to play against the world champion, and we played our test matches and lost.

We needed to win. So they had no choice but to carry that TPU chip. So we did it and a lot of interesting things happened . I think you know what an ELO rating is? Yes. Okay. For those who don't know what an ELO rating is, it's like your ranking in chess or go, and something like a 200-point advantage is insurmountable. The probability of you winning is basically zero. Alpha Go running on GPU had an ELO score of around 3200. The total score was around 35,3550.

And I might be wrong about the first digit. It may be 2000 instead of 3000, but as you know, it was over 200 and then when it went into L in TPU, it actually went up to 3900 or something ridiculous or 2900, whatever the first digit was. He made a spectacular leap, so I think he didn't expect to lose, but he lost spectacularly, but it was exactly the same model. What changed was that the ability to calculate more made the results smarter. The way these models work, thinking fast, thinking slow, by Daniel Conaman.

Yes, I read that book. What the AI ​​does is, if I have 270 possible moves, which is what you have on a Go board, the AI ​​will rank those moves and say this is the best, this is the next best move, and so on. What happens is that you virtually play that best move, then you virtually play the countermove, then you virtually play the next one, and then you see how the game unfolds. Another option you have is to try that second-best move. And sometimes, that second-best move , when you play it, turns out to be the right move in this context.

It's not the one you'd normally make, but in context it's better and you can see that as you play it. In the second game, there was a famous move called move 37, which was creative . It was original. Actually, it wasn't completely original. It was a 1 in 10,000 game move. It had been in the canon of games he trained on. But when we went back and played it on GPU, it never found that move because it was too deep in the chain. So over time, these GPUs have become as good as, and even better than, TPUs .

As the creator of the TPU, I must admit that GPUs are now better. This is the advantage of having a whole industry behind you, an ecosystem and everything, but at that time TPUs had some novel innovations, right? Now it incorporates the LPU and you can dig deeper faster, you can search faster, and in fact, you can make a model smarter by making it faster. And so the conclusion is that now that you have this ability to reflect, think deeply, and change the outcome based on your thinking, being able to think faster makes you think smarter.

And that's the advantage of pairing the LPU and the GPU. I came across one of my all-time favorite quotes while reading the book 0 to 1. The quote says, "The most powerful pattern I've noticed is that successful people find value in unexpected places. And they do it by thinking about business from the ground up, rather than formulas. That's exactly what Apploven has done with its advertising platform. Apploven connects you with over a billion potential new customers within mobile games. Apploven lets you capture all the attention.

Apploven ads are full-screen video ads that are watched for an average of 35 seconds. That 's retention that surpasses other advertising platforms. And you can launch on Apploven in minutes. You set the goal, and Apploven achieves it. There's no complex setup or specialized knowledge required , and Apploven scales rapidly. They can show your ads to over a billion potential customers. Other companies have seen immediate results, scaled up to spending hundreds of thousands of dollars a day, and increased their revenue by millions. So, you need to get started quickly before all your competitors do." They're on Apploven.

And you can do that by going to apploven.com. It's apploven.com. Let's go over a couple of ideas you've come up with in the 10 years you've been running Grock. What's your take on the reality quotient? Well, one of the things we hired at Grock was the reality quotient , which is different from IQ. And the way we look at it is that there are a lot of very intelligent people who wouldn't recognize reality even if you tapped them on the shoulder. You have to tell us more about that.

Yes. Yes. Well, you know, these people construct these very elaborate stories in their minds that are completely disconnected from reality, right? And then there are some people who are incredibly intelligent, but who couldn't do basic arithmetic or anything like that . The reality quotient often starts as the ability to recognize reality, but in its most extreme form, it's the ability to choose the dominant game being played . So what most successful founders and entrepreneurs do is make everyone else play this game. game, and they realize that if you play this higher-level game, you win.

A simple example: MySpace focused on the number of registered accounts. Facebook focused on monthly active users. It was the dominant game, right? If you have monthly activity, that's more important than registered accounts. And if you maximize monthly activity, you're going to beat someone who maximizes registered accounts. You're playing a better game. And so, as a founder, you can often do better than other people. And your job in leading those people is to try to help them connect their activities to that dominant game. So, when I led Grock, I said our goal was to reach 25 million tokens per second of capacity in our data centers.

That meant everyone had a different way of contributing to that. It meant they could manufacture the chip faster. It meant they could manufacture the software faster. It meant we could deploy more. It meant reducing energy costs so we could deploy more chips with lower operating expenses. It meant having more data centers. It meant manufacturing more chips. It meant finding ways to Optimizing the supply chain so we could place orders faster and ship things faster for customers—when I got one so everyone could connect what they were doing to that dominant game.

Where being a founder is challenging is that you often see that and tell people, and everyone wants to stick with the old way of doing things or they care more about the process. So, going from being an engineer to a founder, what I finally understood was that if I was going to do something disruptive, my job would be full-time change management. And the first principle of change management is to make it look like it's not a change. Why is that so important? People don't like change.

No human being likes change. The difference between people who seem to like change and those who don't is that they often see different things. And for the person who's okay with change, nothing has really changed. So, if I'm playing this dominant strategy here and that doesn't change, I'm just trying to maximize the number of tokens per second I've implemented. When my focus shifts, nothing has really changed. But if you're down here thinking, how do I make it so If this chip is faster and you don't think about the software, then you might see something that changes the chip as a change.

So , your change management duty is to give people enough context so they see that their work hasn't really changed. What they 're trying to achieve is exactly the same, and there's no change. Tell me about the return of luck. I read a book by Jim Collins a while back, and it had this chapter that really struck me. The thesis is that the most successful companies don't have more lucky events. They simply leverage that luck better than other companies. I read this very early on as a founder and started to realize that it was true.

And there was a really good example of this when LLMs started becoming commonplace. I remember getting a phone call from the CEO of GitHub who basically said, " I need a ton of GPUs." Now we have LLMs so we can complete code, although, you know, we're part of Microsoft and all that. You know, we just can't get GPUs. Could we use your chips, your LPUs? "And I went to the team and said, 'We have a chance. We can do it.' And they all said, 'No, it's not going to work.

It can't run on these chips.' I said, 'No, it looks like it's ideal to run on our chip, it looks almost perfect.' And they said, 'No, it won't work. There are all these things that are in GPUs that we don't have.' And I said, 'Yeah, but none of those are important for LLMs.'" And they were analyzing the wrong things. I let them convince me that we should n't go ahead, even though deep down I knew we should . This happened on another occasion, when there was another opportunity to implement an LLM and I let myself be dissuaded.

The third time, they told me no, that I would do it myself. So I just did everything. I did the calculations and determined the yield, and everyone disagreed with me that it was possible . And I thought, "No, no, look at this." And in the end, we ended up achieving exactly those performance figures. The point is that they were analyzing the wrong reasons; they were analyzing all the reasons why it could n't be done, instead of the reasons why it could be done . And we convinced ourselves not to do it.

I had many lucky opportunities that I didn't take advantage of. I mean, how much better off would we have been if we had been the first to lead LLM programs at Microsoft for OpenAI? That would have been a very different outcome. So we lost a little, but we're still winning. from the curve when we realized that fast inference was going to be important. And I remember that very early on we were talking to potential clients. We even had a video where we sped up the process, and everyone who watched it said, "Why do I need an LLM to be faster than I can read?" It hadn't yet occurred to them that they were n't going to read.

They were n't going to read it, but besides, that's not how the internet works. Do you think it's okay for a website to appear? You know I hate that. Yes, because the eyes don't move that way. The eyes move all over the place . You need to have it all there. You'll look at it and even before reading it, you'll often get the feeling that this isn't what you need and you'll start typing your question without reading everything that comes out. So I realized that quick inference was going to matter.

Nobody else did it. What do you mean nobody else did it? Even within Grock there was a lot of resistance, as if we had a lot of turnover at this point. Many people were leaving because, how many years ago was this? It was probably three or four years ago. Okay. And many people were leaving saying that quick inference made no sense . It wasn't going to add any value to the ecosystem. And even if you draw simple parallels like dial-up versus broadband, nobody could connect it.

I'm going to interrupt you quickly. Can you talk more about this because now everyone is talking about fast inference? It's like everything but what, and I wasn't paying attention to this four years ago. She was doing other things as if she wasn't paying attention. Can you talk about the difference? I think this is one of the most important parts of your company's story; it's simply how contrary it is. Perhaps it's not even the right word, but I was in disgrace. Your idea, your main idea, people said no, that it wasn't important.

Well, everything was in disgrace. Everything we did was considered a bad idea, but if you don't do things differently, you don't have any advantage, right? Why do you think people just didn't understand that four years ago? When people don't understand the first principle of something and get involved in it because it's an exaggeration, they don't understand enough to grasp why what you do is different. That disoriented you. Yes. I mean, I still stand by this. Well, what finally worked was ... and this is due to a little marketing idea we came up with.

We realized that there was no way , no matter what we showed people, that they would accept that fast inference was going to be useful unless we let them try it. And I remember this example from Eric Schmidt who was involved in this, SCSP or whatever, and they showed Anthropic, an LLM from Enthropic, about 3 months before the GPT chat time. And I remember sitting there watching it, watching demonstrations of this AI answering questions from the audience, and nobody reacted. I thought, how is it possible that nobody is reacting to this?

Now compare that to the moment in the GPT chat when everyone reacted. What was the difference? The difference was when people asked their question and got an answer to their question that was specific to them. That was magical . Seeing someone else's reply text appear wasn't magical. So I realized that and thought the only way to get people to understand the value of speed is if we implement this and put it on the internet. We did that , and what happened was that we put it online.

And I remember I was doing a little world tour trying to find clients, and I was in Norway and I was doing a presentation, and I realized because we had it working and I noticed that the presentation was, like, when I was doing, ahem, queries, ahem, using some of the open source models , I remember it seemed a little slow to me. Not so much, but a little slower than usual. I thought, what 's going on here? Because, you know, Norway is further from the servers, but I tested it before and it 's fine.

I did the check and our usage had skyrocketed. Someone had posted a video on X of an LLM running on Grock that was super fast and it went viral. Suddenly, everyone started creating apps using it, publishing them, and it was so appealing that when people saw it, they all started creating their own and we went viral. Tell us more about this experience you had. I think you're approaching it from the basic principles . You say that these people are getting involved in this because it's a trend and it's what's being spread right now.

What was that experience like during those years when you just went through this as if you were trying to explain why this was going to be important and you ran into one blank slate or one obstacle after another? Well, there is this common theme where many really good innovators are so because they experienced the problem before others. Remember my experience with Alph. It continues, you know, the TPUs and being able to beat the best Go player in the world just thanks to the hardware we changed to.

Yes, I was able to get this return on my luck more than others. I am more able to say, "Well, there's an opportunity, I'm going to take advantage of it." But I was also exposed to opportunities first. And you need both. So you have to be in a position to see the future in advance, and this is a common saying, right? The future is already here; It's just that it's not evenly distributed . Because I was in a situation where I could see the future , because I was really willing to seize the opportunity and double it down when everyone else was saying, "No, let's not chase this opportunity," and I was thinking, "Yes, this is the opportunity," those two things are what work together.

When you say yes, this is your chance. Would you describe the response you're getting as opposition or indifference? One of the biggest changes in my leadership, another book, Turn the Ship Around by David Marquette. I read that too. Okay. Well, technically I read books for a living. Do n't know. So I largely adopted it in my leadership once I read it because it worked very well with my autonomous leadership style. And the basic idea of ​​intentional leadership is that if I ask someone if I should do something, they have opinions.

Most people will be pessimistic and give you negative opinions. On the other hand, if you express intentional leadership, you say, "I intend to do this." People don't usually give their opinion, but if it's very wrong and there's a reason, they'll oppose it. And the example is that of the submarine commander who took charge, I believe it was the USS Santa Fe. I think it was the worst in nuclear readiness of the nuclear submarine fleet, and in a year or two he brought it to number one in readiness, and all he did was move from command and control to intentional leadership.

The prime example is saying, "to sink the ship." There were incidents where a submarine had submerged with the hatch open and nobody wanted to back down because the commander was the one in charge and in control, and they were used to doing what the commander said. But when people used this intentional leadership and said, "I intend to lower the ship to 500 feet," someone would suddenly say, "Wait, the hatch is open." Now they are involved. They're all involved and they're all saying, "I intend to do this.

I intend to do this." It gives everyone the opportunity to say what they know they are doing so that people can hear it. But you're not asking for an opinion. The problem had been repeated time and time again before: I received opinions from people and they prevented me from doing it. Returning to the three examples of the return of luck. Yes. Okay. And if only he had said, "I intend to do it." Is that what you did in the third example? Yes. I literally said, "I prepared a presentation." I said, "We're going to get to this particular speed per chip." "And instead of people saying, 'We can't do this,' everyone got down to work and said, 'This is how we do it.'" It's a very small change in the sentence, but it makes a difference in your ability to move forward .

It doesn't invite friction, but people would still give feedback when it was really important, when there was a real problem, they would bring it up. How does intentional leadership relate to the question? If you received opposition or indifference, where were you going to go with that? You were encouraging pessimism by asking for people's opinions; were you asking potential customers? These are teammates, teammates, teammates. Okay, almost everything that is difficult is difficult because you can't go to one extreme or the other. In fact, you have to decide in context whether you go this way or that way.

One of the difficult things is getting feedback. You hear this from leaders all the time. At the beginning of their career, they encounter too much resistance. Later in their careers, they don't receive enough feedback. How do you balance it to get real feedback instead of just unnecessary resistance? And part of that is just this subtlety in the wording of "I intend to do this" instead of asking for an opinion. The deal is how the best founders turn the world into their talent pool. I've spent a decade studying how the best founders in history operate.

And one thing they all have in common is that they understand that recruiting and hiring the best talent is their top priority . A-list players recognize other A-list players, which is why top-tier companies like Ramp, Shopify, 11 Labs, Uber, and DoorDash use Deal. Many of the best founders I know have personally invested in Deal after using their product. And what they discovered is that Deal is the best company in the world at creating infrastructure for global recruitment. Deal will help your company hire, pay, and manage workers anywhere in the world.

This way you can retain top talent anywhere and dedicate the rest of your time to focusing on what you do best: delivering value to your customers. The founder of 11 Labs has a great description of the value that Deal can bring to your company. He said, "We created 11 Labs to break down language and communication barriers. With Deal, we can recruit and support exceptional talent anywhere. We can accelerate our innovation and bring more voices, stories, and ideas to every corner of the world. More than 40,000 businesses trust Deal.

Discover how they can help your business today by visiting deal.com/enra." That's deal.com/enra. Let's go back to that time when you were three weeks away from running out of money. Yes. And you came up with the idea for Grock Bonds. Even before I arrived, something that came to mind while you were talking earlier about your leadership styles, you know, basically I'm going to tell them what we want to do. You have this organizational principle, but you're not going to tell them how to do it. You're going to let them surprise you.

Yes. This Phil Knight in Shoe Dog says that over and over and over again. And then I'm reading about Grock's bonds and it sounds very similar to some of the things Phil Knight had to do because Nike was so close to going public. They had to go public out of necessity because they were running out of money or very close to it. And in fact, he converted some of the loans he obtained from his employees into equity and ended up, you know, doing very well for them.

So explain this idea you had for Crocodile Bonds. We were going to run out of money and the management team I had at that time was starting to draw up a layoff list of who we were going to fire. And when I started reviewing the list, it became very clear to me that if we made that dismissal, we were lost. Because? We were already struggling to keep up with what we needed to implement because this was before the product even fit the market. This was before the product worked.

We had to write a very special compiler that had never been written before and that did not require humans to write what are called kernels. I had never been successful before. Nobody had ever done it before. And our architecture didn't work with cores like everything else did. We had to reach this point, a kind of critical mass point, before our product would work. We hadn't done it yet. And we were talking about firing people who were essential to that. So, when I realized that layoffs were n't going to solve the problem and that I was just a burnout.

math, like, you know, we were going to run out of money, but we just weren't going to have the talent we needed to succeed, and we were going to have these other costs. I realized that we had to reduce our spending without reducing our staff. And the only answer was to get people to accept a pay cut. So we brought everyone together and put up, you know, war bonds that looked like they were from World War II and we called them Grock bonds. Technically they weren't bonds.

It was an exchange of wages for capital. And we expected that we were going to have a fairly high dropout rate, and in fact that wasn't the case. Eighty percent of the employees participated, and I believe approximately half of them were paid the legal minimum wage, and remember that engineers are paid hundreds of thousands of dollars. These people reduced their salary to $50,000 or $60,000, whatever the legal minimum was; it was a real headache. And we saved more than 3 weeks of margin. I think it was more like 2 months.

We had 3 weeks' worth of money left when we raised it. If we hadn't done this, we would have closed down. The interesting thing is that I keep coming back to this because we've had many moments where we've had to keep the team together. And there's a phrase I have that is to put everyone's hands on the wheel. As passengers, people feel more nervous on a winding road or a scary path. But when they are the drivers, they feel they have more control. They are more willing to take risks.

By doing this, we put everyone's hands on the wheel. They were participating in saving our runway. And we had less than 10% dropout. It could have been closer to 5% when we announced Grock Bonds, which was actually probably better than our previous churn rate. I love this idea. Then, on the other hand, you went from looking for other ways to avoid firing people. The other side of layoffs is hiring. You also have some interesting lessons, I think, that you learned in the decade you were building Grock about hiring that are also quite contradictory, and many of the people who appear on the show have very contradictory ideas about hiring.

I was very good at hiring incredibly intelligent and talented people, but many of the people we brought in caused organizational problems. I've already mentioned it. And the reason is that I'm quite intelligent. And when I meet someone, I can think of a reason why I should hire them. I think a lot of people do this. They convince themselves that they should hire this person. They're great because of this. They have this experience. They have this attribute. I'm going to hire this person. We have what we call a people specification, and just as there is a product specification, we had a people specification.

It had version numbers. We used to exchange them. If you don't write down what you're looking for in people, you're not going to hire that person. You're not going to be consistent. And so , we frame the specification of people in positive aspects, things you seek like the return of luck . Give me a couple more examples of what the positive points were in that specification. Poetic design. Poetry is semantic density. It's when you say a lot in so few words, which is really important to you .

Extremely important. This phrase, I think it's on Grock's blog, where he said "make every word count". I think you repeat it over and over again. Every word matters. Oh, every word matters. There you have it. Yes. It's the smallest possible expression, the most minimal, of what you 're trying to achieve, the most poetic. And that's not just in words. It's also in the design, right? You know, something is poetic even if it's not words. And that's another thing. But each of these has a negative side, right?

So, the opposite of luck returning would be wasting luck. The opposite of poetic design would be, you know, maximalist design, like simply including every feature. Like you know, some of these products are like, where am I supposed to click? Yes. And it's very easy to spot people who have some of the positive aspects and not realize that they also have some of the negative aspects. And what you're really hiring for is to avoid the negative. Because if a person arrives with that negative aspect, they are transmitting it to the entire team.

The biggest change in my hiring process was when I went from looking for positive aspects, which is what you do when you're trying to develop talent, to looking for negative aspects, which is what you do when you're trying to select talent. Explain it. When I try to help someone grow and improve, I want to show them the way. TRUE? This is a famous example of how to increase the amount of money donated to a charity; it's not about making people realize how great the charity is .

It's not about making them feel good about donating to charity. It's about telling them where to send the money. If you tell them how, if you give them a skill or a technique, they can very often learn it and put it into practice. So, when you try to help people grow, show them the positive. Don't say, "Hey , don't waste your luck." " Show them what the return of luck looks like, which is that there was an opportunity that everyone else said no to, and we said yes, and that's what made us successful.

Right? When you're hiring, you're really looking to assess people and trying to say no to things. And that's a very different move. And some people are very good at growth, others are very good at hiring, but you have to separate those two things into very different mental modes. The reason I realized this at Grock was that we hired a head of HR who was very good at spotting problems with people and solving them. And watching her do that, I realized I'd been hiring badly. I think the way you described it to me was that you basically reversed it, and now you're hiring for loss bias.

I think the term you use is one of the attributes, and I think it's important. Humans have a natural loss bias, which is that people assign a mathematical number to it, which is that a loss is six times more painful than a gain." Profit. Like you. Look at this: someone will invest money, lose 20%, and that will be very painful, but they didn't invest in something that grew 100%, and that hurts them less than losing 20% ​​of the money . Even if they don't get the profit, the opportunity cost is much greater.

There's a personality trait in people that I call "booking the win early." And you'll see, for example, we'd be in an architecture meeting, and someone would say, "Well, if we do this, the chip will be twice as fast." And I looked around the room, and nobody seemed very enthusiastic about doing it. What's going on here ? And I started to realize that everyone was listening: "If we do this, the chip will be twice as fast." Let's put it on the next chip." And I heard that if we don't do that on this chip, the chip will be half as fast as it could be.

As soon as I heard that something could be done, I booked it. He immediately assumed that if he didn't do it , he had lost that thing. So, when I started hiring, I looked for other people who had the same kind of attitude of booking the win early. The moment they hear that something is possible, they reserve it and say, "I don't want to miss out on that thing." Many of the most successful entrepreneurs, in a way, manufacture their own discontent. I want to get there in a second.

Yes, but I think this hiring-for-loss bias and applying it not only to talent, but also to these meetings you're having on product, product design, and things that would make your product better is really important. You mentioned earlier that you learned from a Founders episode about Michael Jordan a way of doing things where he challenged his teammates to bets. Why was it an interesting idea? To you? When I heard your episode about Michael Jordan, I thought, "Yes, he's very intentionally throwing his keys over the fence so he has to go and get them." What he does is that Michael Jordan is a very aggressive competitor who bets on everything, like if he could toss a quarter and hit the bullseye or things like that, weird things like that, and he never stops doing it.

Most people are afraid to make fun of someone, a competitor, because if they lose, they will feel very, very bad. Remember that the losing bias is strong. For example, if I say, " Let's play basketball." "I say, I'm going to wipe the floor with you and then lose. That's humiliating, right? What I suspect Michael Jordan was doing was intentionally taunting the other players so that a loss would be humiliating and force himself to perform at superhuman levels. He just did it over and over again. Most people are so afraid of putting themselves out there and suffering a negative outcome that they don't get their hopes up.

In fact, they keep their goals much lower. But entrepreneurs start a company and say, ' Of course I'm going to succeed.'" I'm going to tell everyone: I'm going to raise money. "I'm going to put my reputation on the line and I'm going to be forced to surrender." Michael Jordan's coach is the one who wrote the book in which I did that episode. And the way he describes this is that what Michael would do is exactly what you're saying. He says, "Well, once you tell someone how badly you're going to screw them over, you have to go and do it." What Tim Grover realized, while studying Jordan's career, was that Jordan intentionally pushed himself harder because the more pressure he put on himself, the higher he wrote , the better he performed, and the higher he climbed throughout his career.

I think this is also related to something you and I have talked about, which you call manufactured discontent. There's a book on the counter. We were talking in the kitchen earlier, before we started recording, about David Oggovy, who is one of my heroes, and he calls it " divine discontent." You'll find that the best entrepreneurs, the best athletes, anyone can reach the top of their profession, right? They don't rest on their laurels; They don't rest on their laurels. There's another book right next to it, the new biography of Steve Jobs that just came out, and Steve Jobs demonstrated this concept perfectly when he says, "Well, you've created this great product, now what?" He says, " Well, I believe that if you make something wonderful, all you have to do is go back to doing it, don't think about it, the next day I'm going to go ahead and make another great product, I'm going to keep doing this." " Essentially, they're telling you that the journey is a reward.

Tell us about your idea of ​​manufactured discontent. I was talking with a group of entrepreneurs and people from other fields, some of whom had made a lot of money and some of whom hadn't. The entrepreneurs were the ones who were least content with their wealth, even though they had more money; everyone involved in the conversation was incredibly successful. But we started to realize that even though some of these entrepreneurs had made hundreds of millions of dollars, they compared themselves to others who never had to work again in their lives.

But because they weren't content with their wealth, they had a reason to keep going and start another company and make more. Meanwhile, the other very successful people were quite content with their wealth, but what they weren't content with was the product of their previous work, a previous piece of writing they had done or something like that. And since everyone in this room was successful, what we identified was that they all had something driving them and something that made them discontent. So I started to analyze my own life and , as You know, there were many periods where there was genuine discontent because we hadn't managed to get the product to fit the market.

But once we got it right at Grock, I wasn't happy with the scale, and with other elements as well, and I kept finding things to be unhappy with. Most people can be quite content with the status quo and aren't going to keep pushing to innovate. You have to have a personality that makes you constantly dissatisfied if you're going to keep pushing things forward. What are you unhappy with today? Right now, I'm unhappy with the lack of computing power in the world. AI is revolutionary. It's going to change everything for people.

It has pros and cons, but the advantages are enormous. You know, there are going to be medical breakthroughs, and if it takes us one more year to cure cancer because we do n't have enough computing power, that 's my fault. Every single person who dies from cancer, every single person who gets old, frail, and dies, could potentially be a There's a point, we don't know, where artificial intelligence finds ways to slow down aging, right? I feel like all of that is on my shoulders, and I have to do it.

I have to ensure the world has more computing power. I love that idea of ​​saying that every day we fail to accomplish this mission, there's a real cost associated with it. Edwin Land, the founder of Polaroid, Steve Jobs' hero—this guy, this is one of my all-time favorite entrepreneurs. I'm not going to stop talking about him, but long before he invented the Polaroid camera, he was trying to invent new ways to reduce headlight glare because in the early days of the automobile, a lot of people were dying from oncoming headlights.

And what he did was very similar to what you did; he had this organizing principle when you said, you know, we have 25, we need to get to 25 million tokens. He used to write on the whiteboard, "Today, 300 people have died ." Yes. Because of this. And if it takes us one more week, That would be 21,000 more people. I don't know the exact number, but it's something like that. I think you're in a perfect position. This program is a love letter to capitalism. I think we should end on an optimistic note.

I think the best entrepreneurs in the world are, by default, optimistic and aggressive at the same time. Could you give us an overview of what you think will come with AI and, as a result , tell us some of the more optimistic things you could mention about it? Because you see what's going on right now, everyone is very unpopular. People want to blow up data centers, they want to attack certain people who invent the technology. I don't think we've done a very good job of telling, so to speak, a more positive story.

Well, I think that's because people perceive a shift, right? I recently published a post that received a lot of negative comments, in which I said that there has indeed been a rationing of code, right? As a software engineer, by default, writing code is expensive. That's why I'm going to have a lot Be careful with the code I write. I'm not going to create a function unless I'm absolutely sure it's the right function to create. I'm not going to implement anything until I have it figured out.

And what we've seen in agile software development is that when you take the risk, implement something, and get feedback, you end up getting better results. But there's still a very natural predisposition to say no to things. And so, the concept of a kind of "no engineer"—it's someone's job to say no to things , right? They're the ones in the meeting who say, "We can't do this. We shouldn't do this." What I 'm seeing is that code is becoming almost free. The marginal cost is approaching zero, and it's changing the way things are done for professional engineers, where you simply implement what you experience and say, "Reimplement it this way differently based on my experience." The other change is accessibility.

It's very similar to literature and literacy, isn't it? There was a time when scribes were the only people who knew Reading and writing, and in a way, they controlled access to the written word. And then we got much simpler reading and writing , you know, alphabets instead of these kinds of idioms and hieroglyphs, etc., and suddenly a lot of people could learn to read. And then the education system came along, and everyone could learn to read, and everyone could read and write. Suddenly, it was about the quality of the written word, not just the written word.

But everyone had access. My EA now creates software applications. Like, when I go on a trip, she creates a little app that I can click on, and it tells me what the weather is going to be like, it updates live and pulls it from sources, and it gives me all my phone numbers and all sorts of additional information. That would have been impossible for someone who didn't know how to code before. So what I think is going to happen is that a lot of people are going to have access to the ability to create software to solve problems that they would never have had the technical capabilities for before, but they would have good taste and know what's good, and there will be a huge number of founders, Unlike in the past, when you didn't have access to capital or talent, I think you're going to see individual founders without large teams creating very valuable companies that solve real problems for people.

And I love that approach. We'll end with one of my favorite quotes of yours, where you said, "I look forward to a year of massive improvement for anyone who wants it." Anyone who wants to learn can now learn a subject. They just have to ask questions. The problem with traditional education is that it was forced upon you . It wasn't interesting. And if you're really going to learn something, it has to be something interesting. The ability to ask questions the moment you want to learn something is going to fundamentally change education.

This brings us to what I said earlier, that the AI ​​era is going to be about asking questions. I think a lot of people ask me what they're going to do for their children . And my answer is to stop teaching them how to answer questions and start teaching them how to ask questions. Curricula should be overhauled around this issue. In fact, it's important for the community. Maybe, you know, you have that fix the way permits are processed in the city. Maybe you need a way to improve how you spread the word about some kind of event that's happening.

Things get students to write real applications that are useful to the community they're in and solve real problems, and then get them to ask questions. When you create homework or a test for Kids, if they can look up the answer online or if they can ask AI to solve it, you haven't taught them what they need for the next stage. But if you give them a problem where they have to ask questions and get AI to solve them, then you have . Thanks for the time, man.

I'm glad you did . Thank you. I hope you enjoyed this episode. Remember to subscribe wherever you're listening and leave a review. And be sure to check out my other podcast, Founders. For almost a decade, I've obsessively read over 400 biographies of the greatest entrepreneurs in history, looking for ideas you can use in your work. Most of the guests you hear on this show met me through Founders.