How does a company like Nvidia get built? More importantly, how does a nearly-25-year-old company – a publicly listed one, no less – keep pushing the boundaries year in and year out and chase new frontiers?
Nvidia is the acknowledged market leader in the gaming industry and a darling of the stock market. Not only has its stock price zoomed 170% over the past year (according to a report by The Motley Fool), it has also rewarded its shareholders handsomely via dividend payouts and share buybacks.
Nvidia’s top line for fiscal year 2017 came in at $6.9 billion while net profit more than doubled to $1.7 billion. Few would argue that driving the hunger towards new frontiers is Jensen Huang, who founded Nvidia back in 1993. He remains an entrepreneur first and foremost and driven to push the boundaries for himself, his company and in the process, the world.
In Munich earlier this week, Huang launched the NVIDIA Drive PX Pegasus, which can support Level 5 autonomous cars and promises to take the self-autonomous driving space to a whole new level. With the chip slated to be available in 2018, the automotive industry is the new pearl in the Nvidia oyster. And yes, gaming and powerful GPU remain at the core of it all. Not surprisingly, the announcement sent its stock price to new high of $192.95 – after all, the Drive PX Pegasus promises to speed up AI programmes and make self-driving cars fully autonomous.
Over a tasty breakfast at a five-star hotel in Munich, I got the chance with a few others to interact with Jensen Huang. And here I bring to you some morsels of our conversation – something that all entrepreneurs can learn a lot from. I certainly did. Here’s Jensen’s take on what has made Nvidia not just a very profitable company but an acknowledged leader in the space.
“If you have a sense of purpose and you know what you are about then these kinds of questions are fairly easy to answer. And the reason for that is that it kind of pumps through your veins, the company’s veins. It's always been the case that our company has been…doing work that wasn’t otherwise possible without us. I know it’s kind of strange but if you frame your life’s work and you frame your sense of purpose around work that only you can do and doing work that’s really hard to do and doing work that if you were successful somehow, you will make a greater impact.
“(There are the) only three questions we ever ask…about anything. We ask these three questions consistently, every single time, (in) every single project, and it never fails. The first question is simply, is it hard to do, and if we succeeded in doing it, would it make an impact? Number two, is it work that only we can do? Meaning nobody else has ever done it before, maybe it’s not even a technological way maybe it’s the way you went about solving the problem, or the way that you go to the market with the problem though ideally its a technical problem because Nvidia is a technology company. And the third: is it work we would derive great joy from?
“Ideally, if someone told me we are not going to make any money for 10 years, then you have got my attention. If you told me that we won’t make any money for about three months but boy, it’s going to be a quick return on investment, then I am not going to be involved in that conversation. And the reason for that is if it takes only three months to do, then somebody else can probably do it, and it probably has been done. If it takes only three months to do, then it can’t be that hard. And if it takes three months to do and we are going to make such an extraordinary impact. Really? That’s kind of surprising.
“Nvidia is an accelerated computing company. What that basically means is that we have to go and add something to a computer that was already invented by some of the world’s largest and most important companies. By adding something we can solve a problem that otherwise could not be solved. You have to accelerate something. And the choice of problems we chose: computer graphics. Virtual reality from the very beginning, was a problem. (In our) original presentation…I wrote (that) this is a problem that is sustainable. Meaning, it is going to take us an enormous amount of time to solve this problem. We would be solving for 25 years and still not be solved.
“I have just explained very simply the selection of a business that followed the three rules.
“The question is: how does it sustain you and if you make it your life’s purpose, then you are going to seek out these incredibly hard problems to solve, that only you can solve, that brings you enormous joy when you absolutely make no money. It sustains you. Along the way you have to be successful, otherwise, it discourages you. And so, of course, we select many of these types of challenges over time and we succeed sometimes, we fail often.
“So you have to believe in the work you are doing, and you have to find great joy in doing it, no matter what. They say the journey is the reward and it’s absolutely true. You have to enjoy the journey, find joy in it and that’s kind of it. And if you could share this sensibility with a large number of people and genuinely find joy in the suffering which all inventors, innovators, craftsmen, artists have to (go through)… then the culture sustains.
“I always feel like we are 30 days from going out of business. I always believe that we will succeed, but the insecurity (is) about whether you are doing the right thing. Most entrepreneurs face that insecurity: whether you are doing the right thing, how you are approaching it and your opportunity for success. I started Nvidia a long time ago and I was well known for closing every company meeting saying, “Remember, 30 days from going out of business.” And I still feel that way. And so in the case of self-driving cars, I am always insecure about whether we would succeed or not. But I am always deeply rooted and deeply committed in the moment.
“In terms of self-driving cars, my confidence in our contribution has been growing exponentially over the course of the seven years. With every day that goes by, there is more reason to believe that you were right. That the investment is going to be high, the problem is going to be hard and that no one has solved it yet. And it comes back to the reinforced core values of our company: that it is important to do and it is hard to do. This positive reinforcement turns out to be the most important thing.
“With respect to the technology, at this point I would say that it remains incredibly hard to put a Level 4 self-driving car on the road. This (Level 4) means that you can literally not pay any attention: you could take a nap, read a book, watch TV, turn around and enjoy the guests in the car, and the car will safely drive you from Point A to Point B. That specification is a technological extraordinaire. We are talking about colossal work and some things will fail. Computers will fail, software will fail. and software with bugs will fail. There are conditions that the cars never experienced before.
“So the question that we have with Level 4 is if things were to fail, how do we guarantee the car continues to operate? I just said something that does not make any sense: it fails but it continues to operate. So how do we create redundancy and diversity? But redundancy is not enough, it turns out. With two computers that are the same, why would the first one fail and the second one not? It could be a systematic failure or a bug. So the same software running on the same computer could fail in the exact same way. So you need redundancy and diversity as well. And here comes the interesting question: if I knew how to solve the problem the first way, why would I not solve the problem the same time the second way? We have thought through these problems with deep learning approaches, computing hardware approaches, tools-based approaches. But I feel like we still have a lot to learn in the next 2-3 years, but the large things have been answered. So from a technology and engineering perspective, I feel a great amount of confidence to put the car on the road.
“We are the only company in the self-driving cars (segment) today that is starting to build the problem all the way from the foundational architecture of the processors to the computers, the operating system, the middleware, the applications that do the driving, put it in a car, collect the data, create software to train the data, build supercomputers to train those data, simulate the world, and re-simulate the world. We are the only company that is working top-to-bottom, edge-to-edge. We have been working on building this vision for seven years.
“Nvidia's investment in autonomous driving is absolutely the largest of any company in the world. Now let me explain the difference in our strategy. Rather than building a self-driving car, we decided early on, because we wanted to make a huge impact, and do something incredibly hard, something only we can do, that instead of building a self-driving car, we will create the infrastructure, the architecture, and a platform by which every single company on the planet can build a self-driving car. It doesn’t matter who they are. You are a startup, a large OEM, a trucking company, a shuttle company, you want to deliver pizza? Fantastic, we are happy for you! We think it’s absolutely the way forward. And we will create the platform for you to do that. And you want to use all that we created, so that it is easy for you, because your focus is really about industrial styling, and luxury and things like that, and not about technology, it is fantastic.
“But if you are a software developer and it turns out you have expertise in some parts of autonomous driving like control systems, then it is fantastic; use parts of it, use or all of it or none of it. Our mission is to enable the entire transportation industry to achieve self-driving capabilities.
“We have been investing for about seven years. It has to continue for another two or three, at quite large scales. But our observation that the world needs to have fundamental computing architectures necessary for this autonomous machine processing, and that the entire software stack is going to have to be re-engineered for concerns on safety, functionality and large fleet deployment, scalability -- because Level 2, 3, 4 and 5 have radically different problems.
“These matters have to be considered somehow, so we decided we will create a scalable platform, create a whole problem so that we understand it ourselves, and make available the ability for people to pick and choose, and we would solve the problem from end-to-end, because it is not about building a chip for a self-driving car; it is about building a self-driving car. You have to train the network, stimulate it, deploy it, and then re-stimulate it.
“So we thought through the problem end-to-end, and I think the investment will remain high for a few more years. At this point, this is an endeavour I would characterize as a grand slam. That we have done something really, really good here.
“I tried really hard. I tried incredibly hard. Not to let anybody forget that we really love gaming. And in fact, all of the demos that you guys see are kind of rooted in gaming. They are all kind of rooted in gaming.
“In a way, solving a car driving problem isn’t unlike solving a self-driving car problem in a game. I am not shy about our gaming heritage, nor am I embarrassed about the joy we derive from gaming and I am supremely, supremely proud of the fact that the gaming industry, because of its extraordinary scale, and the technology that it drives has made it plausible for us to sustain enormous R&D.
[Note: Over the past three years, Nvidia spent $4.1 billion on R&D, according to media reports.]
“In life you don’t need that many good ideas. Just one is good. But it has to be a supremely great idea. And this one is a supremely great one. We were in our 20s. We made the observation, and at that time the video game industry had no CD-ROMs, there was no sound. The graphics of the PC were just characters. We were in our late 20s. 3D graphics were only available in military flight simulations. Electronic Arts did not exist. Atari, Sega, Nintendo were still recovering from the game called ET. Which destroyed the entire video gaming industry. That’s the condition.
“In that condition, we postulated – that someday, would it be fun if we all played games in 3D. If it was in 3D, we could be networked together because your view of the world and my view of the world, it would be the same world, the same virtual world. That’s one of the first principles of 3D graphics. Real 3D graphics, we could share the world.
“Imagine a battlefield with all of these tanks and planes and it’s persistent, and we are in it, all the time. And we are trying to recreate reality. And it’s going to take all of the technological might from the computer industry to figure out a way to make it possible. And not only that, we are going to shrink it into one chip. Because someday, we are going to have so many transistors we are not going to know what to do with it.
“And we are going to create this massive industry. And you know who is going to play it - this was the business plan that I wrote. They asked me for a forecast, a business forecast - and I told them its either going to be infinity or zero. There were very few numbers in between. Either everybody is going to play games or this thing is going to be a complete failure. It's not even technologically possible. Today, give me an example of one person who doesn’t play games.
“I believe every human will play games. Just like every human will listen to music, every human will read books, every human will watch movies, every human will play games. And I believe 3D graphics and virtual reality is still a few decades away. We have been working on it a few decades, it’s still a few decades away. And the technology necessary to sustain this medium is just utterly unimaginable. And now when you fuse AI into it, these worlds are virtual reality physics simulations.
“It operates in super real time sometimes, and the observation that we make is like is this the only medium in the world whereas the production value continues to go up, the technology that provides the entertainment has to go up with it.
“That’s unlike music. We make the observation that music, it does not matter whether it’s a concert, a symphony or one person sitting in front of the guitar, the medium stays the same. The production value and the delivery medium were independent. However, in the case of 3D graphics, the production value of the game and the medium are co-dependant - the only entertainment medium of this kind in the world.
“Three characteristics: 1. A massive market, 2. Incredible technological demand, 3. Every single year the technology has to improve. Those three observations and therefore the conclusion. I remain completely a believer in them and notice what happened. GPU computing was made possible by gamers, artificial intelligence (AI) made possible by gamers, all these clouds powering all this technology all over the world made possible by gamers, so yes, you’re right I’m pretty proud of the fact that we are a gaming company.
“Nvidia is successful, (and) may someday be the world’s most important computing technology company.
I think it already is.