Transcription
Good morning everyone. It's really nice to be back here in Congress Hall. Hopefully everybody had a good day yesterday and are enjoying it today. It is my real pleasure to introduce Jensen Wong, who is somebody I admire, somebody I've watched, and somebody who has been a teacher to me on the journey of learning about technology and AI. It is amazing watching how he led Nvidia. And um I don't measure myself on on comparisons, but I like this one comparison. So since Nvidia has been public, which was in 1999, same year as Black Rock. [laughter] Oh boy. Okay. No. Uh, Nvidia's total return for its shareholders, it's been a compounded 30 37%. Just think about that. What what would that mean to every pension fund if they invested in Nvidia as an IPO? The the amount of successes we have with everybody's retirement. At the same time, uh, Black Rock's annualized uh, total return has been 21%. Not so bad for a a financial services company, but uh uh it certainly pales and uh and so that but that is just a really great indication of Jensen's leadership, the positioning of Nvidia, and also it is a great statement about what the the world believes in the future of Nvidia. So Jensen, congratulations on that journey and I know we have many more years of that journey ahead of us.
>> Thank you. I appreciate that. My only regret was at the IPO, after the IPO, I wanted to buy my parents something nice and so I sold Nvidia stock at a valuation of $300 million. Uh the company was at a valuation $300 million and I bought them a Mercedes S-Class. It It is the most expensive car in the world. [laughter] [snorts]
>> They regret it.
>> They still have it.
>> Oh, sure. [laughter]
>> Yeah, they still have it. Yeah.
>> Good. Let me go into the subject matter now, but I just want to say uh you know, the debate on um on AI is about how it's going to change the world and the global economy. Today I want to talk about how AI can add to world to the world economy and how AI can increasingly become a foundational technology that everyone in this room can be utilizing, enhancing our lives, enhancing the lives of everyone in the world. And we need to talk about how it's going to reshape productivity, labor, infrastructure across virtually every other sector, but importantly, how it's going to reshape the the the world and how can more segments of the world benefit from AI and how can we ensure that we have a broadening of the global economy, not a narrowing of the global economy. And I can't think of another person who has a clearer view on not just what AI is, but the infrastructure around it. The infrastructure that is necessary to build around it. And because so many of the major hyperscalers are utilizers of what Nvidia creates and the whole engagement around the infrastructure around AI, the potential of AI, I think we have a great voice to listen to this afternoon or this morning. So Jensen, once again, thank you. This is his first time here at the World Academic Forum in Davos and I know uh you have a really busy schedule so it's uh thank you for taking that time.
>> I appreciate that.
>> So [clears throat] let me go right into it. Why do you believe that AI has the potential to be that significant engine of growth and what makes this moment, this technology different than past technology cycles?
Yeah, this is u uh first of all, when you when you think about AI and you're interacting with AI in all these different ways, chat of course, using Gemini of course, using Anthropic Claude of course um and the magical things that that it could do. Uh it's helpful to reason back to the first principles of of fundamentally what is happening to the computing stack. Uh this is a platform shift. A platform is something where applications are built on top of. And this is a platform shift like the platform shift to PCs. New applications were developed to run on a new type of computer. A platform shift to the internet, a new type of computing uh platform uh hosted all kinds of new applications. A platform shift to mobile cloud. In each and every one of these platform shifts, uh the computing stack was reinvented and new applications were created. Uh this is a new platform shift in the sense that today you're using ChatGPT. It's important to understand that itself is an application, but very importantly, new applications will be built on top of ChatGPT. New applications will be built on top of Anthropic uh cloud, for example. And so so it's a it's a platform shift in that way. Um AI is really easy to understand if you realize what it can do that you could ever do before. Uh software in the past was effectively pre-recorded. Humans would type and describe the the algorithm or the recipe for the uh computer to execute. Uh it was able to process structured information, meaning you've got to put the name, the address, you know, their account number, uh their age, where they where they live. You create these structured tables that software would then go and retrieve information from. We call it SQL, SQL queries. SQL is the single most important database engine the world's ever known. Almost everything ran on SQL before. Now, now we have a a computer that can understand unstructured information, meaning it could look at an image and understand it. It could look at text and understand it. It's completely unstructured. Uh it could listen to sound and understand it, understand the meaning of it, understand the structure of it, and reason about what to do about it. And so for the first time uh we now have a computer that is not pre-recorded but it's processed in real time, meaning that it's able to take the context of the circumstance, whatever the the the environmental information, the contextual information, and whatever information you give it. [clears throat] Um, it could reason about what is the meaning of that information and reason about uh your intent, which could be described in a really unstructured way. Uh, you describe it however you want to describe it. We call it prompts. Um, but you describe it however you like to describe it and to the extent that it can understand your intention, it could uh perform a task for you. Now, the important thing about this is that because we're reinventing that entire computing stack, the question is what is AI? You're asked when you think about AI, you think about the AI models, but it's really important to understand industrially, AI is actually essentially a five-layer cake. At the bottom is energy. AI because it's processed in real time and it generates intelligence in real time, it needs energy to do so. Energy is the first layer. The second layer is the layer that I live in. It's chips. Chips and computing infrastructure. The next layer above it is the cloud infrastructure, the cloud, the cloud services. The layer above that is the AI models. This is where most people think AI is. But don't forget that in order for those models to happen, you have to have all of the layers underneath it. But the most important layer, and this is the layer that's happening right now, the reason why last year was an incredible year, frankly, for AI is that the AI models made so much progress that the layer above it, which is ultimately the the layer that we all need to ex to succeed, the application layer above that. And so this application layer could be in financial services, it could be in healthcare, it could be in manufacturing. This layer on top ultimately is where economic benefit will happen. But the important thing though, because this computing platform requires all of the layers underneath it, it has started, and you guys are everybody's seeing it right now, has started the largest infrastructure buildout in human history. [clears throat] We're now a few hundred billion dollars into it.
>> That's it.
>> We're a few hundred billion dollars into it. Uh Larry and I, we get the opportunity to work on many projects together. There are trillions of dollars of infrastructure that needs to be built out. And it's sensible. It's sensible because all of these contexts have to be processed so that the AI so that the models can generate the intelligence necessary to power the applications that ultimately sit on top. And so when you go back and when you reason about it layer by layer by layer, and you you realize that the energy sector is now seeing extraordinary growth. The chip sector, TSMC just announced they're going to manufact they're going to build 20 new chip plants. Foxconn working with us and Wistron and Quanta building 30 new computer plants, which then go into these AI factories. So we have chip factories, computer factories, and AI factories all being built around the world.
>> And memory.
>> And memory. Right. Exactly. Those chip fabs, Micron has started investing $200 billion in the United States. SK Hynix is doing incredibly. Samsung is doing incredibly. You could see that entire chip layer growing incredibly today. And now, of course, we pay a lot of attention to the model layer, but it's really exciting that the layer above them is really doing fantastically. And now, one indicator is where the VC funding is going into. Last year, 2025 was one of the largest years in VC funding ever. And last year, most of the funding went to what is called AI [clears throat] native companies. These are companies in healthcare, companies in robotics, companies in manufacturing, financial services, all of the large industries in the world. You're seeing huge investments going into those AI natives because for the first time, the models are good enough to build on top of.
So, let's just dive a little further. [clears throat] Obviously, everybody, I'm sure, uses their own chatbot and getting information. But you're you're you're talking about the dispersion of AI is going to be the key. Let's talk about it like go into a little more upside ideas related to the dispersion of it in the physical world. You mentioned obviously healthcare is a great example of that. But where do you see the transformational opportunities in areas like transportation or science?
Well, last year [clears throat] uh I would say say um last year I would say three major things happened in AI in the AI technology layer, the model layer. The first one is that the the models themselves started out uh being curious and interesting, but they hallucinated a great deal. And last year uh you could you could we can all reasonably accept that these models are better grounded. They could do research. um they can reason about about uh circumstances that maybe they weren't trained on. Break it down into step-by-step reasoning steps and come up with a plan to and to answer your your uh question. Um do your research or perform the task. So last year we saw the evolution of language models becoming AI systems that we call agentic systems. Agentic AI. The second, the second major breakthrough is the the breakthrough of open models. Um, several years ago, is it was it a year ago, DeepSeek came out and DeepSeek was a lot of people were quite concerned about it. Frankly, DeepSeek was a huge event for most of the industries, most of the companies around the world because it's the world's first open reasoning model. Uh since then, a whole bunch of open reasoning models have have emerged and and uh open models uh has enabled companies and industries, researchers, educators, re you know universities, startups uh to be able to use these uh open models to start something and create something that's domain-specific or specialized for their needs. The third area that had enormous progress last year was the concept of physical intelligence of physical AI. AI that understands not just language, but AI understands, you know, if you will, nature. And it could be AI that understands the physical world here. AI that understands proteins, chemicals, um, uh, natural n physics of physics, for example, fluid dynamics, particle physics, u quantum physics. AI that are now now learning all these different structures and different uh different languages, if you will. Proteins is essentially a language. And so all of these AI are now making such enormous progress that these industries, industrial companies, uh whether it's you know manufacturing or drug discovery, are really making great progress. And one of the one of the great indicators is a partnership that we had with Lily, uh that they realize now that AI has made such extraordinary progress in understanding the structure of proteins and the structure of chemicals. Um essentially be able to interact and talk to the proteins like we talk to ChatGPT. Uh we're going to see some really break big breakthroughs.
So all these breakthroughs raise concerns about the human element. You and I have had many conversations on this, but we need to tell the whole audience, there is a huge concern that AI is going to displace jobs. Um and you've been arguing the opposite. Obviously, the buildout of AI, as you talked about, the biggest infrastructure buildout in history is going to occur, which >> energy is creating jobs, industries creating jobs, the infrastructure layers creating jobs, land, power, and shell jobs. I mean, right? It's incredible.
>> So let's get into that a little more detail. So you actually believe we're we're going to face labor shortages? Um and so how do you see that AI and robotics changing the nature of work rather than eliminating it?
There's several different ways that we could think through it. First of all, uh this is the largest infrastructure buildout in human history. That's going to that's going to create a lot of jobs. And it's it's wonderful that that um the jobs are related to uh tradecraft. Um, and uh uh we're going to have uh plumbers and electricians and construction and steel workers and uh uh network network uh uh technicians and uh people who who uh uh install and fit out uh the equipment and uh all of these jobs. We're going to in the United States, we're seeing quite a significant uh boom in this area. Salaries have gone up nearly doubled.
>> And so we're talking about six-figure uh salaries for for people who are building uh chip factories or computer factories or AI factories. And we have a great shortage in that. And and I'm really delighted to see so so many people in so many countries really recognizing this important area. You know, everybody everybody should be able to make a great living. You don't need to have a PhD in computer science to do so. And so I'm delighted to see that. Um, the second thing to realize and and so we theorize about um the automation of of uh of tasks and things like that and what is the implication to jobs. Um, you know, I'll I'll just offer some anecdotes. These are real-world anecdotes of what has actually happened. Remember 10 years ago, uh one of the first first professions that everybody thought was going to get wiped out was radiology. And the reason for that was the first AI that became superhuman in capability was computer vision. And the one of the largest applications of computer v vision is studying scans by radiologists. Well, 10 years later, it is true that AI has now completely permeated and diffused into every aspect of radiology. [clears throat] And it is true that radiologists um use uh AI to study scans. Now, it the impact is 100% and the impact is completely real. However, not surprisingly, I say not surprisingly, if you reason from first principles, not surprisingly, the number of radiologists have gone up.
Is that because of lack of trust or is that because the human interaction with the with the results of AI, exactly, is a better outcome?
>> Exactly. The reason for that is because a radiologist, their job, their purpose of their job >> is to diagnose disease, to help patients diagnose disease. That's the purpose of their job. The task of the job includes studying scans. The fact that they're able to study scans now infinitely fast allows them to spend more time with patients diagnosing their disease, interacting with the patients, interacting with other clinicians. Well, surprisingly, also not surprisingly, actually, as a result of that, the number of patients that the hospital can see has gone up because, you know, there are a lot of people waiting a long time to get to get their scans done. And so now because the the number of patients have gone up, the revenues of the hospital have gone up, they hire more radiologists. This is the same thing that's happening to nurses. We're 5 million nurses short in the United States um as a result of using AI to do the charting and the transcription of of uh the the uh the patient visits. Uh nurses spend half of their time charting >> documenting >> and now they could use AI technology in one particular company at Bridge, they're a partner of ours doing incredible work. As a result of that, the nurses could spend more time visiting patients >> human touch >> that's right. And because you could now see more patients and we're no longer bottlenecked by the number of nurses, more patients could get into the hospital sooner. As a result, hospitals do better, they hire more nurses. And so surprisingly, AI is increasing their, not surprisingly, AI is increasing their productivity. As a result, the hospitals are doing better. They want to hire more people. You have too many people waiting too long to get into hospitals. And so these are two perfect examples. Now, the easiest way to think about whether what is the impact of AI on a particular job is to understand whether the job, what is the purpose of the job and what is the task of the job. My if you if you just put a camera on the two of us and just watched us, you would probably think the two of us are typists because I spend all of my time typing and and so if AI could automate so much word prediction and help us type, then we would be out of jobs. But obviously, that's not our purpose. And so the question is, what is the purpose of your job? In the case of radiologists and nurses, it is to care for people. And that that purpose is enhanced and made more productive because the task has been ma has automated. And so to the extent that you can reason about each one of the people's purpose versus the task, I think it's a helpful framework.
Let's let's [clears throat] move this beyond the developed economies. Help me understand how AI is it broadened the world and helped the world. I read an I read an Anthropic piece this past weekend that basically said the utilization of AI most recently is very dominant by the educated society and they're even seeing uh the educated component of each society being heavily more utilized and they're obviously they're they're using it against their own model code. So it may be it may have its own biases. Um, so [clears throat] how do we ensure that AI is a transformational technology, maybe like what Wi-Fi and 5G was for the emerging world? And when you intersect that, what does it mean for the emerging world and jobs? How do we broaden the global economy? And two, you know, getting back to the whole job situation with robotics and AI, there is going to be some substitution there and there is substitution in the United States already going on. We may be creating more plumbers and electricians, but we probably need less analysts at financial institutions. Lawyers need less anal, you know, because they're able to accumulate the data faster. So, let's just pivot on to the emerging world for a second. In the developing world, how do you see that play out?
Well, first of all, um AI is infrastructure and there's not one country in the world I can't imagine that you need to have AI as part of your infrastructure because every country has its electricity, you have your roads. You should have AI uh as part of your infrastructure. You of course, you could always import AI. Um but AI is not so incredibly hard to train these days. And because there's so many open models, these open models with with uh your your local expertise, you should be able to create models uh that are helpful to your own own country. And so I I really believe that that every country should get involved to build AI infrastructure, build your own AI, take advantage of your fundamental natural resource, which is your language and culture, develop your AI, continue to refine it, and have your national intelligence be part of your part of your ecosystem. And so I I think that's number one. And number two, remember AI is super easy to use. It is the it's the easiest software to use in history and that's the reason why it's the fastest growing and fast most rapidly adopted. I mean, in just a couple of two, three years, it's coming up to almost a billion people. Um, I think first of all, Claude is incredible. They've Anthropic has made a huge progress, huge leap in developing Claude. We use it all over our company. Uh the coding capability of Claude, its reasoning capability, it's, you know, its ability is just really incredible. And and anybody who's a software [clears throat] company really ought to get involved and and use it. Um, on the other hand, uh ChatGPT is probably the most successful consumer AI in history and its ease of use and its approachability, I think everybody should get involved. And whether whether it's um uh somebody in a developing country or uh you know, somebody, a student, uh it is very clear that it is essential to learn how to use AI, how to direct an AI, how to prompt an AI, how to manage an AI, how to guardrail the AI, evaluate the AI. These skills are no different than leading people, managing people, things that you and I do all the time. So, in the future, instead of biological, you know, carbon-based AI, in the future, we're also going to have um digital versions of AI, silicon versions of AI, and we and we'll have to manage them. They're just part of our digital digital workforce, if you will. And so I I would I would advocate that for the [clears throat] developing countries, uh build your infrastructure, get engaged in AI, and and and recognize that AI is likely to close the technology divide, >> right? >> Because it is so easy to use and so abundant and so accessible. And so, you know, I I'm I'm actually fairly optimistic about the potential of AI to lift um the countries that are that are um uh that are emerging. And um uh for many people who haven't had a computer science degree, uh all of you can be programmers now, you know, and so in the past, we had to learn how to program a computer. Now, you program a computer by saying to the computer, how do I program you? You know, and if I if you don't know how to use an AI, just go up to the AI and say, I don't know how to use an AI. How do I use an AI? And it would explain it to you. And you know, you say, I like to I like to write a program to create my own website. How do I do that? And it says it would ask you a whole bunch of questions about what kind of website you would like to build and then write you the code. And so it is that easy to use. And that's of course the the incredible, you know, power of AI, which which is exciting.
Two quick questions, then we're going to run out of time. We're sitting [clears throat] here in Europe. When we were talking about a lot of companies, we mentioned a lot of US companies and Asian companies. Um, talk to us about how AI uh and and and the success of Europe and the future of Europe can intersect and what and how do you see Nvidia play that role here in Europe?
Well, I get I have the benefit. Nvidia has the benefit of working with every AI company in the world and uh because we're low in the in the infrastructure layer and we power um AI across the board and we power AI that are languages, the you know, their biology, their physics, their um world models and related to manufacturing and robotics. And and the the thing that's really really quite exciting for Europe is remember your industrial base is so strong. The in the industrial manufacturing base in Europe is incredibly strong. This is your opportunity to now leap past the era of software. United States really led the era of software. Um AI is software that doesn't need to write software. You don't write AI, you teach AI. And so get get in early now so that you can now fuse your industrial capability, your manufacturing capability with artificial intelligence and that brings you into the world of physical AI or robotics. You know, robotics is is a once-in-a-generation opportunity for the for the European nations. And whether whether it's you know uh well all of the countries that I visit here, uh industrial base is really really strong. Um, the other thing to realize is that that so much of of uh the deep sciences are still very very strong here in Europe, right? And the deep sciences now have the benefit of applying artificial intelligence to accelerate your discovery. And so I I um I think that that it's fairly certain that you have to get serious about increasing your energy supply so that you could invest in the infrastructure layer so that you could have a rich ecosystem of artificial intelligence here in Europe.
So what so what [clears throat] I'm hearing is we're far from an AI bubble. The question is, are we investing enough? Let's turn the turn it around because there's so many people talking about a bubble, but the question is, what I'm hearing from you is, you know, are we investing enough to do what we need to do to broaden the global economy? And so one good test on the AI bubble is to recognize that Nvidia has now has now millions of Nvidia GPUs in the cloud. We're in every cloud. Um, you know, we're used everywhere. And if you try to rent an Nvidia GPU these days, it's so incredibly hard. And the spot price of GPU rentals is going up. Not just the latest generation, but two generation old GPUs. The spot price of rentals are going up. And the reason for that is because the number of AI companies that are being created, the number of companies shifting their R&D budget, Lily is a great example. Uh three years ago, most of their R&D budget, all of their R&D budget, it was probably Whit Labs. um notice the big AI supercomputer that they've invested in, the big AI lab. Increasingly, that R&D budget is going to shift towards AI. And so the AI bubble is is um uh comes about because the investments are large, and the investments are large uh because we have to build the infrastructure necessary for all of the layers of AI above it. And so I I think the um uh the opportunity is really quite extraordinary and everybody ought to get involved. Everybody ought to get engaged. Uh we need more energy. I think that we all recognize that we need more land, power, and shell. Um uh we need more uh trade scale workers. And in fact, that population of workforce is so strong here in Europe.
>> Yes. In a lot of ways, the United States lost that um in the last, you know, 20, 30 years. Um, but it's still incredibly strong here in Europe. It's it's an extraordinary opportunity you got to take advantage of. And so I I would, you know, in I know that where where Larry and I work, uh we uh we see the the investment opportunities um and the investment uh scale is going up. Uh the number of startups, as I mentioned earlier, that LA 2025, the largest investment year in VC history, over a hundred billion dollars around the world, most of it was AI natives. And so these AI companies are building basically the application layer above, and they're going to need infrastructure, they're going to need our investment. Um, you know, and go build this future.
And I actually believe it's going to be a great investment for pension funds around the world to to be a part of that, to grow with this AI world. And this is one of my messages as so many political leaders. We need to make sure that the average pensioner, the average saver is is a part of that growth. If they're just watching it from the sidelines, you know, they're they're going to feel left out. And we want to invest in infrastructure, >> right? In infrastructure is a great investment opportunity. This is the single >> largest infrastructure buildout in human history. Get involved.
>> We're out of time. Hopefully everybody in the audience and everybody on the web streaming seeing the the power of Jensen Wong as a leader, not just a leader in technology and AI, but a leader uh in business and also a leader in in heart and soul, which is really important today, having that leadership from the heart and the soul. So thank you everyone. Thank you everybody. [applause]