Am ärgsten nervt es mich, Vorträge zu halten. Jensen Huang enthüllt, dass NVIDIA 61 CEOs beschäftigt und Mitarbeiter, die Fehler machen, niemals feuert: CEOs sind die am stärksten gefährdete Gruppe.

„There will never be another CEO like me again.“ Recently, Jensen Huang, co - founder and CEO of NVIDIA, said so in a private interview.

It is reported that this in - depth conversation took thirty years to come about and thus presented the less well - known side of Jensen Huang to the public. The moderator Jodi Shelton started her professional encounter with Huang more than thirty years ago, when graphics processing units (GPUs) had not yet triggered the global AI revolution. From the origins of accelerated computing to the prospects of generative AI, this conversation was like a masterclass in foresight.

In the interview, Huang said that in a way, NVIDIA has 61 “CEOs”. In recent years, many people, including himself, have made serious mistakes, but in NVIDIA, no one will ever be fired because of a mistake. “We have created a sufficiently safe environment.” He also revealed that the position of CEO is much more fragile than people think. “In fact, we are probably the most fragile people in the company. But for me, it's no problem to admit this fragility.”

Interestingly, he mentioned that in many ways, he is an “unwilling CEO”. “Public speeches terrify me. I prefer to stay inside the company rather than present myself outside; I prefer to work quietly rather than give speeches; I even don't like giving keynote speeches at all, but for the company, I have to do these things.”

In addition, Huang said that NVIDIA's success does not rely on production. “Although NVIDIA invented the GPU, in terms of production, we are actually the smallest global GPU manufacturer. Many unknown manufacturers have a higher GPU production than we do.” And the fact that there is no “end goal” plays a crucial role in NVIDIA's development.

Regarding what the world will be like in five years, Huang claimed that NVIDIA's and the entire industry's investments in the field of AI would completely change the way computers work. Future computers would develop from “programmed by humans” to “learning independently under human guidance”. And 100% of all jobs would change, but 50% of jobs would not disappear. The future trend is not that there will be fewer jobs, but that everyone will be even busier. And those who are currently unemployed will probably have a means of livelihood thanks to AI.

After this interview, netizens commented on Huang: “I've never seen him so open and direct. It's incredible.”

Below is the detailed content of the conversation. We have translated and abbreviated it while maintaining the original meaning to provide readers with a better reading experience.

“It took 33 years for us to see results.”

Jodi Shelton: The public is really curious about people like you. After all, you define the future of technology, and the future of technology is the future of the whole world. So we want to uncover your personal experiences behind your success and the values that have brought you this far. What do you think of this approach?

Jensen Huang: To be honest, I don't like it.

Jodi Shelton: Really not? But you're famous now, and everyone wants to know the stories of famous people.

Jensen Huang: I never consider myself famous, and I'm not famous at all. I'm just accidentally the head of a very important company, the CEO of one of the most successful technology companies of all time. We made some right decisions early on. Looking back at 1993, we wanted to reshape the computer industry, and we had our own unique views on computer architecture. For a long time, this view was not well - received and was even controversial. You know, at that time, the entire industry's focus was on microprocessors and CPUs. By the way, we got to know each other around this time. We got to know each other at the end of 1993 or the beginning of 1994, right? Since then, NVIDIA has been doing what we still do today: the computer revolution.

Jodi Shelton: Yes, I remember exactly. Back then in Silicon Valley, it was the era of CPU monopoly, Moore's Law was everywhere, and the personal computer revolution was in full swing.

Jensen Huang: Yes. And all our early customers were startups in the field of PC chipsets. These companies were the founding fathers of the glorious era of the semiconductor industry, such as Cirrus Logic, S3 Graphics, Western Digital, Trident Microsystems. Do you still remember these names?

Jodi Shelton: Of course.

Jensen Huang: These companies can be regarded as the “predecessors” of NVIDIA. And now we still follow this path and strive to create a new computing model. It took us 33 years to see results. I'm just accidentally the CEO of this company, that's all.

Jodi Shelton: Maybe it was a natural development for you, but for the whole world, NVIDIA's rise was like a sudden inspiration from out of the blue. Since around November 2023, the entire global technology landscape has changed because of you. How do you evaluate this transformation?

Jensen Huang: You know, if you want to create the future, you have to be in the future before it arrives. To be honest, since we invented CUDA technology and brought the corresponding products to the market, we've been on the way to the future. What makes me most proud of NVIDIA is that we're not only good at developing technologies but also at transforming these technologies into products and bringing them to the market. There are so many companies, researchers, and inventors in the world who have actually developed advanced technologies, but in the end, they can only regret: “I've developed this technology for a long time” or “I've had this idea for a long time”. Every time I hear something like this, I feel sorry. Unfortunately, these excellent inventors didn't have equally excellent product innovators by their side.

A product innovator is someone who can transform a technological invention into a market - ready product. But that's not enough. You also have to conduct in - depth market research and develop strategies, and sometimes even build a new market so that the market accepts your products and strategies. NVIDIA is such a company that has the capabilities from technological invention to product innovation, strategy development, ecosystem building, and market development, and we've already done this successfully several times. So this “future situation” has long been a reality for me.

Jodi Shelton: That's really true.

Jensen Huang: A long time ago, we had a strategy that we don't mention so often today, namely “CUDA everywhere”. Many have heard the story of how I used to travel the world promoting CUDA at universities, startups, and established companies. Sometimes there were only three people in the audience, but I took out my laptop, showed them CUDA, and told them how this technology would change the world. I've visited countless research institutions and laboratories, participated in countless industry conferences, and probably promoted CUDA more often than anyone else in the world. For a long time, I was immersed in this “vision of the future”, and through all the stories I told, I even felt that the future was already here.

Jodi Shelton: There really is such a feeling.

Jensen Huang: So I'm still very happy that all this has now become a reality. And in my opinion, all this was not surprising because NVIDIA's development is based on the fundamental principles of computer science, not on momentary luck or subjective preferences. In many ways, today's result was a necessity. But I want to say that if you increase the speed of something a thousand - fold, or increase or decrease its size a thousand - fold, no matter what it originally was, it will undergo a qualitative change. And the result of this qualitative change is often surprising.

We recognized early on that deep - learning technology has enormous scaling potential, and that's why we put the entire company into this field. We knew that AlexNet (Deep Convolutional Neural Network) would not be the end of deep learning. This technology architecture inherently has strong scalability, and together with the huge global data volume, the outbreak of deep learning was a logical consequence. But I also knew that there would be a technology that would hinder our progress, namely unsupervised learning or self - supervised learning, i.e., that the computer can learn independently without manual data labeling. Because the efficiency of manual data labeling would sooner or later hinder technological development. And when unsupervised learning made a breakthrough, I knew our time had come.

Just at a recent investor roadshow, someone told me that I had clearly pointed out this “qualitative change” back then. If you look at the earnings conference at that time, you'll find that every time it comes to technological topics that are crucial to the world, I explain this point in great detail. In every investor roadshow, at every place where I speak, I emphasize this point. Now, unsupervised learning has actually made a big breakthrough, and the scale effects of deep learning have been fully unleashed. Now we're finally on the fast track. But even now, the problems that this technology can solve surprise me. We had expected that there would be a qualitative change in technology and that the computer platform would change, but we didn't think that the results of the change would be so rich.

We can now understand the “language” of proteins, cells, and quantum, and we can read all forms of representation of the world. The way we described information before is now completely reshaped. From geometric figures and textures to 3D Gaussian models and 3D point clouds, information representations are constantly changing. It's as if people have suddenly become incredibly smart, and even the English language would change. We no longer use the old words, grammar rules, and sentences because our knowledge has developed to a new level, and we can communicate in a new way. Maybe future human communication will simply consist of “beeps”. That reminds me of the scene in the movie “Arrival”, where people suddenly communicate with abstract graphics and can transmit a huge amount of information with these graphics to achieve deeper and more efficient communication.

The amazing thing is that many problems that we can solve now were completely unthinkable before, and the solution methods are also much faster than before. Before, we always talked about Moore's Law, but now NVIDIA's development can be described by the “NVIDIA Law”, which is a thousand times faster than before. The next ten years will surely be exciting, and just the thought of it gets me very excited.

Jodi Shelton: To do the things you do, to predict the future and believe that it will come, you need an incredible amount of self - confidence. As you said before, we've known each other since 1994, and you've always been like this all these years.

Jensen Huang: Yes, I remember exactly.

Jodi Shelton: I was only in my twenties back then. You were probably a bit older than me?

Jensen Huang: I was about 29 years old back then, almost 30.

“It took 33 years for the results to become visible.”

Jodi Shelton: I still remember our first meeting exactly. Back then, I interviewed you for a magazine. I asked you: “Jensen Huang, in Silicon Valley, there is a high turnover of personnel, many people come and go. Are you worried about this problem?” And you were only 29 or 30 years old back then, and you answered me: “NVIDIA is neither a church nor a prison. Those who want to can come; those who want to can go.” I was totally shocked at that time and thought: “Who on earth is this man?” So young yet with so much self - confidence and wisdom. I've also heard a similar story. When Morris Chang first met you, you immediately told him: “I'll be your biggest customer, at least one of the biggest.” His reaction was: “Wow, this young man really has guts.” So I'm very curious where you got so much self - confidence at such a young age?

Jensen Huang: Ha - ha, you know, sometimes it's also painful to know everything. I'm just joking. By the way, Morris Chang would surely be very happy if he knew that NVIDIA is now TSMC's biggest customer.

Jodi Shelton: That's surely the case. He would definitely be proud of you.

Jensen Huang: I'm also very proud of him. You know, during the personal computer revolution, NVIDIA was already TSMC's biggest customer. Now we're again their biggest customer, and that makes me very happy. But back to the topic. I think a person has to believe in what they believe in. And this belief shouldn't be based on hearsay. You can't just believe what others say. You have to think about why you believe in something and break these thoughts down into solid fundamental principles. Then you have to regularly review these principles to ensure that your belief and actions are based on solid foundations.

If this foundation isn't stable enough or changes for some reason, then it's possible that they aren't real fundamental principles, maybe they aren't bound to physical laws or objective facts. As soon as this is the case, you have to re - evaluate and then adjust the course accordingly. I've always done it this way. And if you really believe in something, you should also put it into action. Since 1993, I've been firmly convinced that what we're doing is right, and this belief hasn't changed until now. Based on this firm belief, I constantly make logical considerations and analyses. I constantly review our past decisions and try to predict future trends.

Like yesterday, we had so many meetings, and in each meeting, I re - thought the logic of our decisions. You'll find that some of our earlier assumptions were right, but some were wrong. Only because we were flexible enough to adjust our course according to the actual circumstances have we come this far. So it's very meaningful to look back afterwards and...