THE NVIDIA WAY SHOWS HOW RELENTLESS EXECUTION BUILDS AN ADVANTAGE

Dalton Anderson: Welcome to Venture Step Podcast where he's ghost entrepreneurship, industry trends, and the occasional book of you. I'm your host, Dalton Anderson. In this episode, we're gonna be reviewing a book that originally finished by Tae Kim. The book is titled The Nvidia Way: The Making of a Tech Giant. In this book, it breaks down Nvidia and extensively reviews competitors and the interesting stories about executives, co-founders, and different people in various levels of the organization at Nvidia. throughout its time of earning its right to survive and fight another day. The other co founders involved in this book are Chris Malikowski and then Curtis Priem But majority of the book is centered around Nvidia and it leadership style of Jensen. Well, in the book, Jensen, Chris, and Curtis meet at a Denny's. You've all heard the story. They pitched Jensen, and Jensen was like, Cool idea, man. Where's the money? Show us the money. And they're like, Well, we're here. Talking to you because that's not our thing. We're chip people, we're manufacturer people. We don't know anything about the business development side of things. That's where you come in. And he's like, Okay. And it was this tug of war thing where Jensen wasn't convinced that they were convinced that this could work, because it's really tough to create a chip company. And then the other part of the problem was Everybody had wives and kids. And so none of the wives really wanted their man to leave a really good job. And then maybe the other guy didn't leave. So it was this little tug of war match. And they winded up leaving all at the same time. Or Jensen and I think Curtis. And then Chris left first. He's like, I'm doing it with with or without you guys. And then Jensen and Curtis left at the same time, same day. So that was the founding of Nvidia, but one of the big things that Nvidia did first was they had a hit chip and then they didn't One of the things that they did and the way that they approached it was They approached it all wrong. Chris and Curtis were focused on not what the market wanted, but what they wanted. They wanted to embrace the best technology. They wanted to build A architecture of the future. They wanted to do all sorts of things that they were capable of that other people weren't. But it wasn't that other people weren't. It's just that the market wasn't ready for it. And so they were building something that the market wasn't willing to accept. This came with many, many issues with the NV1 V1 and NV2, which had the wrong ecosystem, which was a drain on resources, almost bankrupt. the company and it prevented them from having a smooth integration with major manufacturers and OS systems, especially Microsoft, which was a popular OS at the time and still is to this day. And what happened was they invented this new architecture, which wasn't compatible with a lot of the operating systems at the time. So then they had to run these workarounds or in adapters. So whenever an update came out from Microsoft, their chip would break, the audio wouldn't work, the visuals would be all jacked up. So overall was a very bad experience for everybody involved. And so people didn't buy Nvidia. So it was a huge deal. And one of the ones was Microsoft Direct X. It was incompatible with Nvidia's architecture, which then put them in a really big bind on like just overall compatibility in general because of this issue. So those missteps nearly bankrupt the company and they had to fight to survive. I think they had eight months of survival and they had to get a chip out within six. And it normally took them a lot longer. And so they just skip a whole bunch of steps. They had to recycle, build out a whole new architecture because they realized that their current architecture approach just didn't work. It just didn't work. So with the limited amount of money they had and recognizing that it wasn't gonna work, Jensen proposed purchasing an emulator, a virtual simulation for chip architecture design to try to hash out the bugs before they went to manufacture the chip and then test the chip. So to prevent additional issues before the chip was manufactured. Because the way that they did it before was they manually created the chip architecture, then they sent the architecture designs to a semiconductor manufacturer like that has a foundry and the clean rooms. That would manufacture the chip, They would give it back to them, they would test the chip, and then To fix the bugs, they would send it back to them, fix the the bugs with the new architecture, they would send it back a new chip. So this is like the second round, and you do that back and forth. And so This emulator prevented the majority of that by testing out everything virtually and fixing the chip architecture virtually. It was a painstaking process how they described it. And once you did that, It allowed them to move much faster and it saved the company, although The machine that they were purchasing was a massive expense with the limited money they had, but they didn't have a choice. It was either purchase the machine, risk it all, or don't purchase the machine and most likely fail. So it was good that they purchased the machine because their first bug that they had, they couldn't they couldn't boot up the chip. It had massive architectural issues. They thought that the machine wasn't working. And it really was that the chip was so broken that it wouldn't work. and so they had to redo everything again. And then they finally got it to boot up and run. And they talked about how slow the emulator was. They would have to wait like eight seconds for the mouse to react to where they wanted to go to click on the computer, this the virtual setup. And so some of the employees to become more efficient, they taped. like icons on their desk on where it would be on the machine if they moved their mouse from the center point to like the left or the right or up or down to so they wouldn't have to wait eight seconds for it to to move over. Which I thought was very interesting. Earned their right to stay alive. just willing your way into existence is how I would describe Nvidia for a while. They had some missteps that made them who they are today. But one of those was creating chips that the market wasn't ready for. And they were going against the grain as a new player. For what? I think that's a big takeaway and call out for you if you're building anything, or if you are in charge of designing stuff at a company. What does the market want? Where do you think the market's going? And what are the current offerings? If you try to build something so advanced and outside of people's box without earning the right to have the trust, it's gonna be a really difficult conversation. you got to earn the right, one to survive, but you also got to earn the right to innovate. And trying to do that as like a brand new company and rethinking. How you're going to persuade the market into just jumping into this brand new architecture or thing is most likely not gonna work. You can see that play out very clearly in this approach. And so they switched to the Riva. They made a new product line, the Riva one two eight, and they designed the Riva through the emulator and they were able to have a successful launch, which saved the company. The Riva one two eight came out in nineteen ninety-seven. When they had their launch. And they weren't doing well. Their competitors were crushing it. And they were looking at the talent that Nvidia had. Nvidia had great talent. At Nvidia, their competitors knew that they were on the ropes. Problem was they were not ruthless enough. And they said that if it was on the other side, Nvidia would have killed him off. Because Jensen's ruthless. And he would have recognized it and he would have acted on it. But their competitors, they just thought that we'll wait until we get a really, really good deal. Or we we don't need to press them out of the market. They're just gonna fall out. And then they came roaring back alive. Then they dominated for years. So now that we're a market leader, how do we stay ahead and become a consistent dominator in this space? And the way that they did that was recognizing that computer manufacturers, they release on six month cycles. Chip manufacturers like Jensen's company, they released on 12 to 18 month cycles. So the computer companies were working with whoever was gonna have a chip coming out near the release. And they'd work in tandem together. The problem was somebody could come out with a really good chip and then it would be put on that machine. The next part of this is Jensen recognizing that chip companies, they lead for one to two years. Because if they have one bust, they're a year away out of the marketplace and somebody else takes their spot. So what if we changed our speed and moving at the speed of light is what he would always say. Moving at the speed of light as fast as humanly possible. What if we went from 12 to 18 months for our manufacturing a chip to six. And we match the computer manufacturers. We just launch a new chip every six months. We're always in the conversation with these manufacturers now. And no one can really take our spot as long as we offer a better product and consistently update it. And we might not have the best product, but then we'll have consistency and we'll constantly be able to add new features every six months. And so we could tell them, like, hey, The features that you wanted in this one, you got three of the five. The other two are gonna have to wait till the next release, but it's gonna be there in six months. Whereas another manufacturer might come out twelve months or eighteen months later. The gist of it was computer manufacturers appreciated the consistency of not having to integrate with a new GPU and just having Nvidia as their sole provider. This gave them a huge advantage when you think about competitors. Competitors were not doing this. And the time that they restructured their whole organization to to enable this process, it's too late. They were already integrated with the manufacturers, had great relationships. Everybody loved them. It was a good product. Customers love Nvidia. And time and time again, Nvidia was crushing it. That time it was recognizing that hey, we just need to put a chip out there that people want and then make it easy for people to use it. And that goes back to the principle of removing friction. Build something marginally better that removes friction, make it backwards compatible, make it easy to use. That's what they did. So they went from being very high friction. to low friction and then also added backwards compatibility. So if you put in one chip, you upgraded your computer, it would just work, which is great. The takeaway there is Having more releases gives you more chances to encounter market feedback. This is a mechanism that will allow you to grow and discuss the product much faster than if you had very few releases. They focused on small, meaningful changes versus massive changes. it's almost a direct rejection of their first couple products, whereas the first couple products was, hey, we're gonna rethink the market. This one's like we're gonna be the market and we're gonna evolve the marketplace by just relentlessly launching chips, which allows you to have a closer relationship with your customers, where the product was always available and there was always something new coming out. A good example would be Garmin. So Garmin, they release something every five years or something like that. And so if you buy a watch and it's close to that five year cycle, people might just wait to buy a watch. But if you had a watch it come out every year, it doesn't matter that you're I don't know, seven months in to the cycle. You'll you'll buy a new phone because it's like, like It's releasing every year, like the difference between this year and the next year is not that much. But the difference between a a five years and the five years prior is massive. So when you think about if you are a chip fan and you're trying to buy a new chip, and then all the other chips on the market, they release once a year or maybe close to every two years. And then you have another chip that you're hearing good things about. And they release every six months. Every six months they're releasing. It's like, wow. What are they doing? And what's stands out is willingness to defer features that you can just push to the next release and then staying close to your timelines and cutting out scope. Instead of scope creep, it's being very clear on the due date. The due date is X and we're delivering Y and we're cutting out Z. Which is big when you think about organizations and projects you manage at any level within your org. You always have scope creep. And so avoiding scope creep is very important. Yes, if there's times where you miss something and you have to rethink your project or your approach, that's different. But in a general sense, you need to ensure that the due date is what you're executing on if you were going to be releasing things often. If it's every two years, then that's a different story. You've got to you've gotta be perfect But I'd always lean towards additional feedback because if you could wait two years, and I've done this in insurance where I work for a carrier, they do things a bit slower than an MGA. A carrier could wait five years to launch a product. And the opportunity is gone. Even a year or two years is a long time to wait to launch any product. So when you're thinking about the market and where it's at, you're building your product for where the market's at and slightly where it might be, versus if you have additional releases throughout that two-year period, you can evolve with the market and you can help nudge the market to where you want it to go You have to either drastically push or you have to guess where you're going. you're building not for where you're at. You've got to build for where we're going to be. And it's very difficult to do that if you're not constantly in the marketplace. So this is food for thought. Food for thought. I think we can all be a little bit faster. Maybe in the AI everybody can release a little bit faster and move a little bit faster and get feedback. Like don't work on your startup idea for three years and then never get feedback on it. 'Cause it's probably a waste of time. The next section I thought was interesting was winning on a distribution and build committed partnerships. And the way that they would describe this with T SMC was rough justice. There's gonna be ups and downs. But on a long-term view, directionally, it's going to be fair. And so there might be some years where you push me, there might be some years when you push I push you. But directionally, it's going to be fair when you run it over like a five-year period. So these are long-term relationships that continue and will continue to be, but you've got to be okay okay with rough justice. And it it it really is like one having a transparent conversation with whoever you're trying to work with on a distribution side, which is very important, and then recognizing that not every year is gonna be your year and then not every year is gonna be their year. Okay. The next part about distribution is winning deals and understanding whether or not the market is gonna make a decision on a deal. And one of them was very influential for Nvidia was the Xbox deal. So Nvidia made the GPUs for the Xbox for a long time. That deal eventually was handed to Nvidia, but not before stealing the deal. And so they pushed Xbox really hard to try to get the deal and why they were the best person for this job and How they would make the chip much faster and better than their competitor. So the Microsoft agreed. But one of the big things was for the deal, they didn't want Microsoft to be able to to recant and then go back to somebody else. So they wanted twenty million dollars as a as a fee. And then they wanted a two hundred million dollar commitment. So they're in, baby. There's no backing out. Because they literally just stole the deal from their competitor and Microsoft had to pay like five million dollars to back out. But they did so, and then Nvidia's like, Okay, by the way, like I that's great that you backed out that deal. But now you need to commit the two hundred and then you need to commit the twenty because you're not leaving us, baby. We're here for life. We're here for life. Which I thought was pretty epic, honestly. This is probably one of the biggest parts of Nvidia is this relentless culture of execution, high independence and agency with long hours. A big part of that is Jensen and the executive group relentlessly attacking talent and recruiting folks over a long period. And acquiring talent aggressively. When talent goes on the market with companies going out of business, Nvidia was always first. And when they wanted somebody in the research department, they moved on them for years, for years. And that reminds me about some of the best executives I've met. Part of being an executive is gathering talent and having people willing to work for you. An executive knew that there's somebody that they were seeing at the dinner, they'd love for them to work at the company. And it's just over a five year period, they've just constantly been hammering them. Like you're gonna come, right? When are you coming over? And everybody's got a different strat, but it's one of the things is just relentlessly pursuing somebody until they say yeah. The way that they approach some of these long term recruitment efforts was they could just kept changing strategies, like things that didn't work, they would change strat, they would do nice dinners, they would do demos, they would do technical things, they would talk about the future. They they would try everything. They're like, we really need you here. And they made it clear. So I think that's one of the things is you should always be recruiting. It's an ongoing strategy to aggressively recruit and execute. Maybe one of the things is they acquired their competitors a hundred engineers at three D FX and then a different company they acquired when Intel was acquiring these engineers, they were asking them to move to California and they didn't want to move. They wanted to stay where their home was and where their kids were growing up When Jensen found that out, he flew an executive out there to close the deal for a couple hundred million. And then he asked them to make an office. So they were like, We'll acquire you. We don't want your employees to move and you can stay where you're at and we'll open an office for you. And so to this day or when the book was released, they still have that office in like Hunting Huntington, somewhere in the US. I don't even know where it's at. So that's pretty cool. That that's the That's what I'm talking about is just relentlessly acquiring talent to defend your moat of innovation and execution. An another thing is high expectations of employees and being able to explain your thought process. When you work a technical position or positions that are technical or strategic, it's not really the outcome. I mean, the outcomes are important. It's not really a decision. It's the thought process to that decision. Because you're never gonna have all the right information when you're making these calls. And so inevitably you're gonna make mistakes and then things are not going to be the way that you thought they would be, right? And so one of the ways that Jensen prepares the company to have their thinking displayed in real time is whiteboards. And I think it's a much better way to describe and explain thought processes instead of a slide. The slides are static. You can hide your thinking and your weakness of thinking with pretty formats. But on a whiteboard, it's very difficult to do so. And the other part of this that I think is pretty interesting is it's repeated a lot in the book, is no matter how brilliant your ideas are. You will have to erase that whiteboard and start anew. That was the antithesis of innovation. No matter how great your idea is, it doesn't matter. You're gonna have to restart, anyways. That's just the life that we live at Nvidia. They make everybody present on a whiteboard. And then Jensen sits in the front row and just hounds the person for questions. So it's really cool. I quite enjoy that. Another thing that they do at Nvidia is the top five. So the top five is what you're working on and it's very action-oriented. And then the subject line is classified by whatever it is. And so it allows Jensen to get information or market information by several different individuals at different levels of of the org. And basically classify it as a database type of thing, an unstructured database, because then he can query his emails with the subject line and research all the related topics and then figure out what's going on on a week-to-week basis to be able to understand where the market is going and where we need to position ourselves. But the emails are very simple. So it's the subject line of whatever it is. very clear on this topic and about what. then it's five bullet points, an action oriented top bullet point then the other five things are working on. They send this report out on Monday versus on Friday because Jensen has an always on mentality. So he actually does more work and more emails or gives people more work when he's on vacation because he'll sit at the beach and he'll just rip out emails. Email, email, email, email. And they had an intervention and Jensen told them no. He's like, I'm having such a great time seeing my kids play in the sand at the beach and me writing emails. I feel so productive. And they're like, no, please, please stop. One of his weekend activities is on Sunday, he'll get a whiskey and he'll drink a whiskey and he'll write emails. That's what he likes to do on a Sunday evening. He has an always on mentality and they describe him as being able to reply to emails within a couple of minutes, typically. You don't want to send it on Friday because if you send it on Friday, it's gonna request different things from you. And you'll have to work on the weekend. So typically people send it on Monday. The next thing was market research. And this was something that was pushed by Jensen. So Jensen asked his salespeople to have a good relationship with whoever they're working with. And one of them was specialized with researchers and providing materials for the researchers and GPUs, of course. And the salesperson came back to Jensen, and then Jensen was like basically. You think this is acceptable? You really know your market by sending me this bullshit? So he changed his approach and got more detailed information. He created a raffle. This raffle was for a top-of-the-line GPU that researchers would have loved to have. And he asked them all sorts of questions. It took about 30 minutes to fill out. The result of that survey was CUDA. And then CUDA is this parallel computing platform and programming model language that allows people to utilize a GPU outside of pure graphics. And so it allows a GPU to become more valuable to people that can program and learn from it and build around it. And that became the foundations of researchers utilizing GPUs to play around with and train models. The neuronets and the deep learning models that were around the math was from a long time ago. But we never had the hardware to utilize this math. And so a new age of computing was born. And that was the training of AI. Nvidia heavily invested into AI. And one of those was CUDA building out of a very robust infrastructure, supporting PyTorch, which was made by Meta, supporting every integration you can, libraries for CUDA, for people to know and love CUDA and leaning into researchers and having them be all in on Nvidia allowed Nvidia to to one hire top talent and two have everybody used to utilizing Nvidia as part of their workflow. And so they're already baked in. They were getting very strong market feedback on where we we were going. And that allowed Jensen to prevent a competitor from taking their spot in this AI space. would I recommend this book, of course. It shows that people don't necessarily have it all figured out. You're seeing their victory lap while they remember the trenches. So keep that in mind when you're thinking about building something or that you are envious of somebody else. You have no idea the pain and suffering that they went through to build what they have achieved. One, recognize that all of this isn't free. And then two, recognize if you're willing to pay the price. You can do whatever you want. As long as you're willing to pay, is what I always say. Book was great. It really outlined that you've got to have technical excellence, but also business acumen. If you're really strong on the business side but don't know anything about the tech, you've got to round out your skills. Or if you know a lot about the tech, but nothing about the business, you've got to know both. If you wanna be effective, you've gotta know both, especially now. Especially now. Well, thank you for listening in to Venture Step Podcast. Of course, wherever we are in this world. Good afternoon, good evening, good morning. Thanks for listening in and listen in again next week. Goodbye.