GO TO SPOTIFY

QUOTES:

At the time we started the company in 1993, we were the only consumer 3D graphics company ever created. We were focused on transforming the PC into an accelerated PC because, at the time, Windows was a software-rendered system. Revo 128 was a reset of our company because, by the time we realized we had gone down the wrong road, Microsoft had already rolled out DirectX, which was fundamentally incompatible with NVIDIA's architecture. Thirty competitors had already shown up, even though we were the first company when we were founded. The world was a completely different place.

The question about what to do as a company strategy at that point—I would have said we made a whole bunch of wrong decisions. But on the day that mattered, we made a sequence of extraordinarily good decisions. That time, 1997, was probably NVIDIA's best moment. The reason for that was our backs were against the wall. We were running out of time, money, and, for many employees, hope. The question was: what do we do? The first thing we decided was that DirectX was now here. We weren’t going to fight it. Let’s figure out a way to build the best thing in the world for it. Revo 128 was the world’s first fully hardware-accelerated pipeline for rendering 3D. At the time, the PC industry was still developing, and it wasn’t good enough. Everybody was clamoring for the next fastest thing. If your performance was ten times higher than what was available, there was a large market of enthusiasts who would go after it. We were absolutely right. The PC industry had a substantially large enthusiast market that would buy the best of everything, and that remains true today.

For certain segments of the market, where technology is never good enough—like 3D graphics—we chose the right technology. 3D graphics is never good enough. Back then, we called it a sustainable technology opportunity because it’s never good enough, and technology can keep improving. We chose that.

The typical plan companies go through—building the chip, writing the software, fixing bugs, taping out the new chip, and so forth—wasn’t going to work. If we only had six months and could tape out just once, then obviously, we had to tape out a perfect chip. I remember having a conversation with our leaders. They said, “But Jensen, how do you know it’s going to be perfect?” I said, “I know it’s going to be perfect because if it’s not, we’ll be out of business. So, let’s make it perfect. We get one shot.” Everybody understood that we had no choice. Not doing it would be crazy because we’d be out of business anyway. Anything else was crazy. It seemed like a logical thing. Quite frankly, as I’m describing it now, you’re probably thinking it’s sensible. Well, it worked. We taped it out and went directly to production.

When you push your chips in, you have to believe it’s going to work. We assumed we taped out a perfect chip because we emulated the whole chip before taping it out. We developed the entire software stack, ran QA on all the drivers and software, and tested all the games and VGA applications we had. When you bet the farm, you’re saying, “I’m going to take everything risky and pull it in advance.” That’s probably the lesson. To this day, everything we can prefetch, everything in the future we can simulate today, we prefetch.

A lot of us gain common sense and reasoning ability by reading, so why wouldn’t a machine learning model also learn reasoning capabilities from that? From reasoning capabilities, you could have emergent abilities. Emergent abilities are consistent with reasoning and intuitive. Some of it could be predictable, but it’s still amazing. The fact that it’s sensible doesn’t make it any less amazing.

I don’t think there’s anything special about NVIDIA. We had the courage to build a system like this. NVIDIA is not built like the military or armed forces, with generals and colonels. We’re not set up like that. We’re not a command-and-control system with top-down information distribution. We’re built much more like a computing stack. The lowest layer is our architecture, then there’s our chip, then our software. On top of that, there are all these different modules. Each of these layers and modules are people. The architecture of the company is like a computer with a computing stack. Your title is unrelated to your place in the stack. It’s about who is best at running that module or function in that layer. That person is in charge and is the pilot in command. That’s one characteristic.

Your organization should be the architecture of the machinery for building the product, right? That’s what a company is. And yet, every company looks exactly the same, but they all build different things. How does that make any sense? Do you see what I’m saying? You know how you make fried chicken versus how you flip burgers versus how you make Chinese fried rice—it’s different. So why would the machinery or process be exactly the same? It’s not sensible to me that, if you look at the org charts of most companies, they all kind of look the same. You have one group for one business, another for another business, and another for another business, and they’re all supposedly autonomous. None of that makes sense to me. It just depends on what we’re trying to build and what the architecture of the company best suits to achieve that.

In terms of the information system and enabling collaboration, we kind of wire it up like a neural network. There’s a phrase in the company: “Mission is the boss.” We figure out what the mission is, then wire up the best skills, teams, and resources to achieve it. It cuts across the entire organization in a way that may not make sense but looks a bit like a neural network.

The downside is the pressure on leaders is fairly high. In a command-and-control system, the person you report to has more power than you because they’re closer to the source of information. In our company, information is disseminated fairly quickly to many people, usually at a team level. For example, in a recent robotics meeting, we discussed certain things, made decisions, and everyone heard it at the same time—new college grads, three vice presidents, two executive staff members. Nobody has more power than anyone else. Does that make sense? The new college grad learned the decision at the same time as the executive staff. Leaders earn their jobs based on their ability to reason and help others succeed, not because they have privileged information.

Over the last 30 years, I’ve read my fair share of business books. As with everything you read, you’re supposed to enjoy it, be inspired by it, but not adopt it outright. The point of these books is to share experiences. You’re supposed to ask, “What does it mean to me, my world, my context, and what I’m trying to achieve?” From that, being informed by what we learn, we’re supposed to develop our own strategies.

You can’t wait until an opportunity is right in front of you to reach out for it. You have to anticipate. Our job as CEOs is to look around corners and anticipate where opportunities will be someday. Even if you’re unsure when or what exactly, you position the company to be near it. It’s better to be approximately right than exactly wrong.

There are many ways to think about competition. We prefer to position ourselves to serve a need that hasn’t emerged yet. There’s no market yet, but we believe there will be one. Often, when you’re positioned there, people ask why you’re there. For example, when we first entered the automotive market, we believed cars would be largely software-defined in the future, requiring incredible computers. Most people didn’t understand it at the time. One CTO told me, “Cars can’t tolerate the blue screen of death.” Nobody can, but it doesn’t change the fact that software-defined cars are the future. Fifteen years later, we were largely right. By navigating our company to non-consumption markets, we position ourselves so that when the market emerges, there are fewer competitors shaped that way.

We were early in PC gaming, and today NVIDIA is very large in PC gaming. We reimagined design workstations, and now most workstations use NVIDIA technology. We reimagined supercomputing and democratized it. Today, NVIDIA’s accelerated computing is quite large. We reimagined software development, now called machine learning, and how computing would be done, which we call AI. We aim to reimagine these things a decade in advance.

If you build your product right and enable an ecosystem around you to serve the end market, you’ve essentially created a platform. It can be product-based, service-based, or technology-based. By being early and helping the ecosystem succeed, you create a network of networks with developers and customers built around you. That network becomes your moat. I don’t think of it as building stuff around a castle but enabling a network that helps others succeed.

NVIDIA’s first business strategy was to be a game console inside the PC, which needed developers. That’s why one of our first employees was a developer relations person. We knew all the game developers and 3D developers.

Starting a successful company is insanely hard. When I see amazing companies getting built, I have nothing but admiration and respect because I know how hard it is. You can make all the right decisions and still fail, or make many mistakes and still succeed. Skills are learned along the way, but at key moments, circumstances have to align, and the market must support your success.

The number of smart decisions we made, many still relevant today, shaped how we design chips. Back then, we pulled every trick in the book out of desperation because we had no choice. Now, everyone does it that way because it’s efficient and cost-effective. Why tape out a chip seven times if you can tape it out once? Speed is technology. Time to market is performance. Revo 128 exemplified this. It led to great decisions in product specification, market needs, and judging markets. Our backs were against the wall, but we had one shot, and we made it count.

With respect to automation, my feeling is that AI is more likely to create jobs in the near term. The question is, what is the definition of "near term"? The first thing that happens with productivity is prosperity. When companies become more successful, they hire more people to expand into new areas. If you improve productivity, they need fewer people only if the company has no new ideas. But that's not true for most companies. If a company becomes more productive and profitable, they usually hire more people to explore new ideas and markets. There are more ideas in drug discovery, transportation, retail, entertainment, and technology. As long as we believe in more ideas, the prosperity that comes from improved productivity leads to hiring more people. Looking back in history, today’s industries are larger than they were a thousand years ago because humans have a lot of ideas. I think there are plenty of ideas yet to be discovered, and productivity improvements can lead to more opportunities. While job changes are inevitable, the net generation of jobs doesn't guarantee that no one will lose their job. It’s more likely someone will lose their job to another human using AI, not to AI itself. Everyone should learn how to use AI to augment their productivity, and companies should do the same to foster growth and hire more people. I believe industries will become more productive, and many currently suffering from a lack of labor will use AI to regain growth and prosperity.

Productivity doesn't mean doing less; it usually results in doing more. Everything becomes easier, but we end up doing more because the world has infinite ambition.

If you prioritize properly and don’t let external tools, like Outlook, control your time, there’s plenty of time to achieve things. Prioritize, make sacrifices, and don't let schedules dictate your every action.

I’m still afraid of the same things I was at the beginning of the company: letting the employees down. People join your company because they believe in your vision and adopt it as their own. You want to be successful for them so they can build great careers and lives. I want them to enjoy the successes I’ve had the benefit of experiencing. My greatest fear is letting them down.

The art of writing a business plan should be concise. It forces you to condense what the true problem is, the unmet need you foresee, and what you will do that is sufficiently hard so competitors cannot easily replicate it. Other skills, like product positioning, pricing, and market strategies, can be learned. What is hard is identifying the essence of what you're solving. When I started, I didn’t know how to write a business plan, but I was fortunate that my work impressed others and set me up for success.

If I were magically 30 years old today, in 2023, having a conversation with my two smartest friends about starting a company, I wouldn’t do it. Building a company and bringing an idea to life is a million times harder than expected. The pain, vulnerability, challenges, and potential embarrassment are immense. If people knew how hard it was, no one in their right mind would start a company. That’s the superpower of an entrepreneur—they don’t realize how hard it will be. Even now, I trick myself into thinking, "How hard can it be?" because you have to. That’s the trick: you have to believe it’s not that hard, even when it is. Your belief system must be incredibly strong. You have to truly believe in and want it; otherwise, it’s too much to endure.

I’m surrounded by people who never gave up on the company or me. Having the unwavering support of family and colleagues is everything. They’re proud of you, and that support sustains you through thick and thin.

What Nvidia discovered is that by separating itself from being just a chip company and building on top of the chip, we became an AI company. This expanded the market opportunity by a thousand times. Don’t be surprised if technology companies grow much larger in the future. The size of your opportunity dictates how large you can be.

Zrzeczenie się Praw Własności i Klauzula Użycia Edukacyjnego

Prezentowane na tej platformie treści, w tym m.in. transkrybowane cytaty, nie są naszą własnością. Wszelkie prawa i własność do opublikowanych treści należą do oficjalnych autorów i twórców odpowiednich kanałów YouTube i Spotify, z których pochodzą te treści. Materiał ten jest udostępniany wyłącznie w celach edukacyjnych. Nie rościmy sobie żadnych praw własności ani autorstwa tych treści i uznajemy, że pozostają one własnością intelektualną ich odpowiednich właścicieli.