Full Interview: Nvidia CEO Jensen Huang on the Future of AI [Summary + Transcript]

01:03
Jensen Huang
Well, Apparently all the GPUs.

01:06
Omar Al Olama
I think this is one thing I'm waiting to ask Sam about because it's a really big number. Talk about ambition. We have a lot of ambition here in the UAE. We don't lack ambition, But. Is there a. View that you can give the government leaders today with regards to compute capabilities, Artificial intelligence, How can they plan well? Where do you think the deployment is going to make sense and what advice do you have?

01:33
Jensen Huang
Well, First of all, These are amazing times. These are amazing times because we're at the beginning of a new industrial revolution. Production of. Energy through steam,production of electricity. IT and Information Revolution with PC and Internet. The now Artificial Intelligence. We are experiencing two simultaneous. Transitions and this has never happened before the first transition is the end of general purpose computing and the beginning of accelerated computing. It's like specialized computing. Using. CPUs for computation as the foundation of everything we do is no longer possible. And the reason for that is because it's been 60 years. We invented. Central Processing Units in 1964, The announcement of the IBM. System 360. We've been riding that wave for literally. 60 years now. And this is now the beginning of accelerated computing. If you want sustainable computing, Energy efficient computing, High performance computing, Cost effective computing,

02:43
Jensen Huang
You can no longer do it with general purpose computing. You need specialized domain specific acceleration. And that's what's driving at the foundation our growth, Accelerated computing. It's the most sustainable way of doing computing going forward. It's the most energy efficient. It is so energy efficient, It's so cost effective, It's so performant that it enabled a new type of application called. AI. The question is, What's the cart and the horse? You know, First is accelerated computing and enabled a new application. There's a whole bunch of applications that are accelerated today. And so now we're in the beginning of this new era. And what's gonna happen is.

03:27
Jensen Huang
There's a about a trillion dollars worth of installed base of data centers around the world and over the course of the next four or five years we'll have two trillion dollars worth of data centers that it will be powering software around the world and all of it's going to be accelerated. And this architecture for accelerated computing is ideal. For this next generation of software called generative ai. And so that's really at the core of what is happening. While we're. Replacing the install base of general purpose computing. Remember that the performance of. The architecture is going to be improving at the same time. So you can't assume just that you will buy more computers. You have to also assume that the computers are going to become faster. And therefore, The total amount that you need is not going to be as much. Otherwise, The mathematics,

04:25
Jensen Huang
If you just assume. That computers never get any faster, You might come to the conclusion we need. 14 different planets and three different galaxies and four more suns to fuel all this. But obviously, Computer architecture continues to advance. In the last 10 years, One of the greatest contributions, And I really appreciate you mentioning that, The rate of innovation. One of the greatest contributions we made was advancing computing and advancing AI. By one million times. In the last 10 years. And so whatever demand that you think is going to power the world, You have to consider the fact that it is also going to do it 1 million times larger, Faster, You know, More efficiently.

05:16
Omar Al Olama
Don't you think that creates a risk of having a world of haves and have nots? Since we need to constantly invest to ensure that we have the cutting edge and to ensure that we are able to create the applications that are going to reshape the world and governments as we know them. Do you think that there's going to be an issue of countries that can afford these GPUs and countries that can't? And if not, Because it would be surprising if you said the answer is no, If not, What are going to be the drivers of equity?

05:46
Jensen Huang
Excellent question. First of all, When something improves by a million times and the cost or the space or the energy that it consumed did not grow up by a million times, In fact, You've democratized the technology. Researchers all over the world would tell you that Nvidia single-handedly democratized high performance computing. We put it in the hands of every researcher. It is the reason why. AI researchers. Jeff Hinton in University of Toronto, Jan LeCun, I think Jan's going to be here, University of. New York, Andrew Eng in Stanford, Simultaneously discovered us. They didn't discover us because of supercomputers. They discovered us because of. Gaming GPUs that they used for deep learning. We put. Accelerated computing or high performance computing in the hands of every single researcher in the world. And so when we accelerate the rate of innovation, We're democratizing the technology. The cost of.

06:48
Jensen Huang
Building, Purchasing a Supercomputer Today is really negligible. And the reason for that is because we're making it faster and faster. Whatever performance you need costs a lot less today than it used to. It is absolutely true we have to democratize this technology and the reason why is very clear. There's an awakening. Of every single country. In probably the last six months. That artificial intelligence. Is a technology you can't be mystified by. You cannot be terrified by it. You have to find a way to activate yourself to take advantage of it. And the reason for that is because this is the beginning of a new industrial revolution and this industrial revolution is about the production, Not of energy. Not a food. But the production of intelligence and every country needs to own the production of their own intelligence,

07:41
Jensen Huang
Which is the reason why there's this idea called sovereign AI. You own your own data, Nobody owns it. Your country owns the data, It codifies your culture, Your society's intelligence, Your common sense, Your history. You own your own data. You therefore must take that data, Refine that data and own your own national intelligence. You cannot allow that to be done by other people. And that is a realization. Now that we've democratized the computation of AI, The infrastructure of AI. The rest of it is really up to you to take initiative, Activate your industry, Build the infrastructure as fast as you can so that the researchers, The companies, Your governments can take advantage of this infrastructure to go and create your own AI.

08:41
Omar Al Olama
I think we completely subscribe to that vision. That's why the UAE is moving aggressively on creating large language models, Mobilizing compute, And maybe work with other partners on this. Let's try to flip the paradigm a little bit. Let's today assume that Jensen Huang is the president of a developing nation that has a relatively small GDP. And you can focus one AI application. What would it be? Let's call it a hypothetical nation. And say that you have so many problems that you need to deal with, What is the first thing that you're going to approach if you're going to mobilize artificial intelligence in that scenario?

09:19
Jensen Huang
The first thing you have to do is you have to build infrastructure. If you want to mobilize the production of food, You have to build farms. If you want to mobilize the production of energy, You have to build. AC Generators. If you want to operationalize, If you want to digitalize your economy, You have to build the internet. If you want to automate the creation of artificial intelligence, You have to build the infrastructure. It is not that costly. It is also not that hard. Companies all around the world, Of course, Wants to mystify. Terrify glorify you know all of those those ideas but the fact that matters their computers you can buy them off the shelf. You can install it. Every country needs, Already has the expertise to do this. And you have to, You surely need to have the imperative to go activate that.

10:14
Jensen Huang
The first thing that I would do, Of course, Is I would codify the language, The data of your culture into your own large language model. And you're doing that here. Core42,saudi aramco, a, a, sadai, really doing important work to codify the Arabic language and creating your own large language model. But simultaneously remembering that AI is not just about language AI for Language. We see sevaral AI revolutions happening at the same time. AI for language. AI for. Biology, Learning the Language of Proteins and Chemicals. AI for. Physical Sciences, Learning to AI of Climate, Materials, Energy Discovery. AI of IoT, The language of keeping places safe, Computer vision and such. AI for IoT, AI for robotics and autonomous systems, Manufacturing and such. There's AI revolutions happening, AI breakthroughs happening in all of these different domains. And if you build the infrastructure,

11:28
Jensen Huang
You will activate the researchers in every one of these domains. Without the internet, How can you be digital? Without farms, How can you produce food? Without an AI infrastructure, How can you activate all of the researchers that are in your region to go and create the AI models?

11:49
Omar Al Olama
You touched upon the issue. Of, I would say, Authentic ignorance, The fear mongering, AI taking over the world. And I think there is a requirement for us to clarify where the hype is real and where artificial intelligence really has the power to create a lot of disruption and to harm us and where AI is going to be good. What do you think is the biggest issue when it comes to artificial intelligence right now? Because I think the problem of regulating AI is like trying to say we want to regulate a field of computer science or regulate electricity. You don't regulate electricity as a. Invention or Discovery you regulate a specific use case. What is one use case that you think we need to regulate against and that government should mobilize towards?

12:37
Jensen Huang
Excellent question. First of all, Whatever new incredible technology is being created, You go back. To the earliest of times it is absolutely true we have to. Develop the technology safely. We have to apply the technology safely and we have to help people use the technology safely. And so whether it's the plane that I came in, Cars, Manufacturing systems, Medicine, All of these different industries are heavily regulated today. Those regulations have to be extended, Augmented to consider artificial intelligence. Artificial intelligence will come to us through products and services. It is the automation of intelligence and will be augmented on top of all of these various industries. Now it is the case that there are some interests. To. Scare people about. This. New Technology. To mystify this technology. To encourage other people to not do anything about.

13:46
Jensen Huang
That technology and rely on them to do it and I think that's a mistake we want to democratize this technology. Let's face it, The single most important thing that has happened last year, If you were to ask me the one single most important event last year and how it has activated. AI Researchers Here in this Region. It's actually Llama 2. It's an open source model.

14:10
Omar Al Olama
Or Falcon.

14:11
Jensen Huang
Or falcon,another excellent model,very true,mistral,excellent model. I just saw another one, A smog. There's so many open source models, Innovations on safety, Alignment, Guard railing, Reinforcement learning, So many different reasonings, So many different innovations that are happening on top of transparencies, Explainabilities. All of this technology that has to be built all were possible because of some of these open source languages. And so I think that democratizing. Activating every region, Activating every country to join the. AI advance is probably one of the most important thing rather than. It's convincing everybody it's too complicated, It's too dangerous, It's too mystical, And only two or three people in the world should be able to do that. That I think is a huge mistake.

15:13
Omar Al Olama
The focus I think that we have done in the UAE is to focus on open source systems. Because we do believe that anything that we develop here should be given as an opportunity for others that can't develop. Most of this is developed using GPUs, So graphic processing units that you guys are supplying the world. What do you think the next era is going to depend on? Is it going to continuously be built on. GPUs? Is there something else as a breakthrough that we're going to see in the future, You think?

15:44
Jensen Huang
Actually, You know that in just about all of the large companies in the world, There are internal developments. At Google, There's TPUs. At. AWS, there's Tranium. At Microsoft, There's Maya. Meta has chips that they're building. In China, Just about every single CSP has chips that they're building. The reason why you mentioned Nvidia GPUs is Nvidia GPU is the only. Platform that's available to everybody on any platform. That's actually the observation. It's not that we're the only platform that's being used. We're simply the only platform that's used that. Democratizes AI for everybody's platform. We're in every single cloud, We're in every single data center, We're available in the cloud, In your private data centers, All the way out to the edge, All the way out to autonomous systems, Robotics and self-driving cars. One single architecture spans all of that. That's what makes Nvidia unique.

16:45
Jensen Huang
That we can, In the beginning when CNNs were popular, Were the right architecture because were programmable. Our CUDA architecture has the ability to adapt to any architecture that comes along. So when CNN came along, RNN came along, LSTMs came along, And then eventually Transformers came along. And now Vision Transformers, Bird's Eye View Transformers, All kinds of different Transformers are being created. A next generation state space. Models, Which is probably the next generation of transformers. Live and breathe and be created on Nvidia's flexible architecture and because it's available literally everywhere. Any researcher can get access to Nvidia GPUs and invent the next generation.

17:33
Omar Al Olama
So for those of you who are non-technical and heard, You know, Foreign language there with CNNs and some of the other acronyms that are being used. The thing about artificial intelligence is it's going through a lot of evolutions over a very short period of time. So whatever the infrastructure that was used probably five years ago is very different to the infrastructure that's being used today. But what Jensen's point was, I think it's a very important point, Is Nvidia has always been relevant. Historically, We see companies that are relevant at one phase of development, And then as the infrastructure changes, They become irrelevant. But you guys were able to innovate and push through. Let's move to a non-AI related topic for a second. I want to talk about education. So today, Knowing what you know,

18:19
Omar Al Olama
Seeing what you see and being at the cutting edge of this technology, What should people focus on when it comes to education? What should they learn? How should they educate their kids and their societies?

18:30
Jensen Huang
Wow, Excellent question. I'm going to say something and it's going to sound completely opposite. Of what people feel. You probably recall over the course of the last 10 years, 15 years, Almost everybody who sits on a stage like this would tell you it is vital that your children learn computer science. Everybody should learn how to program. And in fact, It's almost exactly the opposite. It is our job to create computing technology such that nobody has to program. And that the programming language is human. Everybody in the world is now a programmer. This is the miracle. This is the miracle of artificial intelligence. For the very first time, We have closed the gap. The technology divide has been completely closed. And this is the reason why so many people can engage artificial intelligence. It is the reason why every single government, Every single industrial conference,

19:34
Jensen Huang
Every single company is talking about artificial intelligence today. Because for the very first time, You can imagine everybody in your company being a technologist. And so this is a tremendous time for all of you to realize. That the technology divide has been closed or another way to say it. The technology leadership of other country has now been reset. The countries, The people that understand how to solve a domain problem in digital biology or in education of young people or in manufacturing or in farming, Those people who understand domain expertise. Now, Can utilize technology that is readily available to you. You now have a computer that will do what you tell it to do, To help automate your work, To amplify your productivity, To make you more efficient. And so I think that this is just a tremendous time. The impact, Of course, Is great.

20:40
Jensen Huang
And you're imperative to activate and take advantage of the technology is absolutely immediate. And also to realize that. To engage AI is a lot easier now than at any time in the history of computing. It is vital that we upskill everyone and the upskilling process, I believe, Will be delightful, Surprising to realize that this computer can perform all these things that you're instructing it to do and doing it so easily.

21:11
Omar Al Olama
So if I was going to choose a major in university as a degree that I'm going to pursue, What would you give me as advice for something to pursue?

21:21
Jensen Huang
If I were starting all over again, I would realize one thing. That one of the most complex. Fields of science is the understanding of biology, Human biology. Not only is it complicated because it's so diverse, So complicated, So hard to understand, Living and breathing. It is also. Incredibly impactful complicated technology complicated science incredibly impactful for the very first time and remember we call this field life sciences. And we call drug discovery, Discovery, As if you wander around the universe and all of a sudden, Hey, Look what I discovered. Nobody in computer science, Nobody in computers, And nobody in. The Traditional Industries that are very large today. Nobody says car discovery. We don't say computer discovery. We don't say software discovery. We don't go home and say, Hey, Honey, Look what I found today.

22:19
Jensen Huang
This piece of software we call it engineering and every single year. Our science, Our computer science, Our software becomes better and better than the year before. Every single year our chips get better, Every single year our infrastructure gets better. However, Life sciences is sporadic. If I were to do it over again right now, I would realize that the technology to turn life engineering, Life science to life engineering is upon us. And that digital biology will be a field of engineering, Not a field of science. It will continue to have science, Of course, But not a field just of science in the future. And so. I hope that this is going to start a whole generation of people who enjoy working with proteins and chemicals and enzymes and materials. And they're engineering these amazing things that are more energy efficient. That are lighter weight,

23:17
Jensen Huang
That are stronger, That are more sustainable. All of these inventions in the future are going to be part of engineering, Not scientific discovery.

23:27
Omar Al Olama
So I think we can end with a very positive note. Hopefully we're going to enter an era of Discovery. An era of proliferating a lot of the things that unfortunately are challenges to us, Whether it's disease, Whether it's limitations or resources. Thank you so much for taking the time and being with us. And I know that we could have continued for another hour, But thank you for taking the stage. Thank you. Thank you.

23:50
Jensen Huang
Thank you, Everyone.