Transcription
Good morning, friends. It is a great pleasure to be here again, in the congress hall. I hope everyone had a good day yesterday, and you are enjoying today. It is a great honor for me to introduce Jensen Huang, a person whom I admire, whom I have followed for a long time, and who has become a true mentor to me on my journey of learning about technology and artificial intelligence. It is incredible to witness how he has led Nvidia to what it is today. I am not accustomed to measuring myself through comparison, but there is one comparison I particularly like. Since Nvidia went public in ninety-nine, the same year as BlackRock. >> [laughter] >> Nvidia's cumulative shareholder return has averaged 37% annually on a compounded basis. Just think about that. What would that have meant for every pension fund if they had invested in NVIDIA at the IPO stage? How much success would that have brought in securing people's future pensions? Meanwhile, BlackRock's average annual compounded return was 21%. For a financial company, that is not bad at all, but it still falls noticeably short. And this, in my opinion, is a clear indicator of Jensen's leadership, Nvidia's strategic positioning, and also reflects what the world believes when talking about the company's future. Jensen, congratulations on this incredible journey. I am sure there are many more such years ahead of us. Thank you. I really appreciate it. The only thing I regret is that after the IPO, I decided to buy something nice for my parents. I sold Nvidia shares when the company was valued at $300 million and bought them an S-Class Mercedes. The most expensive car in the world. They still regret it. Do they still have it? Of course, yes, they still have it. Excellent. Then let's move on to the main topic. But first, I want to say, the discussion about artificial intelligence today largely boils down to how it will change the world and the global economy. And today, I want to talk about how AI can add value to the global economy and how it can become a foundational technology that everyone in this room can use to improve our lives and the lives of people around the world. We need to discuss how AI will change productivity, the labor market, infrastructure in virtually all sectors. But what is particularly important, how it will change the world itself and how more and more segments of the economy can benefit from artificial intelligence. And how we can ensure not a shrinking, but an expanding global economy. Frankly, I cannot imagine a person who would have a clearer and deeper understanding not only of what AI is, but also of the entire infrastructure around it. The infrastructure that needs to be built for its development. Since many of the largest hyperscalers use Nvidia, and considering the entire scale of interaction around infrastructure and AI potential, I believe we have a truly important voice to listen to today. Jensen, thank you again. This is your first visit to the World Economic Forum in Davos. I know how busy your schedule is, so thank you for taking the time. I really appreciate it. Let's get straight to the question: why do you believe artificial intelligence has the potential to be such a powerful engine of growth? And how does this technological moment differ from previous technological cycles? First of all, when we think about AI, we interact with it in every possible way, whether it's ChatGPT, Anthropic, Claude, and we see the almost magical things it can do. It's useful to go back to basic principles and understand what is fundamentally happening with the computing stack. We are observing a platform shift, and a platform is the foundation on which applications are built. This is a platform shift similar to the transition to personal computers, when applications for a new type of computing began to be developed; to the transition to the internet, a new computing platform on which entirely new classes of applications emerged; to the transition to mobile and cloud. In each of these transitions, the computing stack was rethought, and new applications were created on its basis. This is a new platform shift in the sense that today you use ChatGPT. It's important to understand that ChatGPT itself is an application. But what is fundamentally important is that new applications will be built on top of GPT. New applications will be built, for example, on top of Anthropic's Claude. This is precisely where the platform shift lies. Artificial intelligence becomes understandable if we realize that it can now do things that were previously impossible. Previously, software was essentially pre-recorded. People manually described an algorithm or a recipe that the computer had to execute. It could only work with structured information: name, address, account number, age, place of residence. Tables with a clear structure were created, from which the software extracted data. We call these SQL queries. SQL is the most important database mechanism the world has ever known. Practically everything used to run on SQL, and now we have a computer that can understand unstructured information. It can look at an image and understand it. It can read text and understand it without a rigid structure. It can listen to sound and understand its meaning, its structure, and reason about what to do with it. Thus, for the first time, we have a computer that is not pre-recorded, but operates in real-time. This means it can take into account the context of the situation, any information about the environment, contextual data, as well as any information you provide it. It can determine the meaning of this information and understand your intention, even if you describe it in a completely unstructured way. You can formulate a query in any way that is convenient for you. We call these prompts, but essentially you describe the task as you see fit. And to the extent that the system understands your intentions, it can perform the task for you. It is important to understand the following. Since we are reinventing the entire computing stack, the question arises: what is AI? When you think of artificial intelligence, you typically think of AI models. But from an industrial perspective, AI is essentially a five-layer cake. At the very foundation is energy. Since AI operates in real-time and generates intelligence in real-time, it requires energy. Energy is the first layer. The second layer, where I am, is chips, semiconductors, and computing infrastructure. The next layer is cloud infrastructure and cloud services. Above it is the AI model layer. This is where most people see AI. But it is important to remember that for these models to exist at all, all the underlying layers are necessary. However, the most important layer, and this is what is being formed right now. This is why last year, frankly, was an incredible year for AI. AI models made a colossal leap, which allowed the layer above them to develop. And this is the layer that ultimately we all need for success, the application layer. This application layer can exist in financial services, in healthcare, in industry. It is at this level that economic value ultimately arises. But it is important to understand that since this computing platform requires all the underlying layers, we have witnessed the beginning of the largest infrastructure construction in human history. And you all see it right now. We have already invested hundreds of billions of dollars in this. Hundreds of billions of dollars. Larry and I often work on many projects together, and there is trillions of dollars of infrastructure to be built. And it is justified. It is justified because all these contexts must be processed so that AI models can generate the intelligence needed to run the applications that ultimately reside at the top level. And if we look at all of this layer by layer, it becomes obvious that the energy sector is now showing colossal growth. The chip sector. TSMC has just announced the construction of twenty new chip manufacturing plants. Foxconn, along with us, and also with Wistron and Quanta, are building 30 new computer factories, which then become part of so-called AI factories. Thus, we have chip factories, computer manufacturing plants, and factories that are being built all over the world now. And memory, yes, memory production, these semiconductor enterprises Micron has begun investing $200 billion in the US. SK Hynix is showing outstanding results. Samsung is showing outstanding results. We see the entire chip layer growing at incredible rates today. And, of course, we pay a lot of attention to the model layer, but what is particularly inspiring is how rapidly the layer above them is developing. And another indicator of where venture capital investments are directed today. Last year, 2025, was one of the largest in history for venture funding. And most of these investments were directed into so-called AI-native companies. These are companies in healthcare, robotics, industrial manufacturing, financial services. In virtually all key sectors of the global economy, we see large-scale investments directed precisely into AI-native businesses, because for the first time, models have become good enough to build complete solutions on them. Then let's dig a little deeper. Obviously, today almost everyone uses their own chatbots to get information, but you are saying that the key factor will be precisely dissemination. And let's talk about broader opportunities related to its penetration into the physical world. You've already mentioned healthcare as a great example, but where do you see transformative opportunities, for example, in transportation or in science? I would say that last year, three key events occurred in AI in terms of technology and models. First, the models themselves started as curious and interesting, but they quite often hallucinated. And last year, we can confidently say that these models have become much more robust. They are capable of conducting research, reasoning about situations they were not initially trained on, breaking down tasks into step-by-step logical stages, and building an action plan, whether it's answering a question, conducting research, or performing a specific task. Thus, last year we saw the evolution of language models into full-fledged AI systems, which we call agent systems or agent AI. The second major breakthrough is the emergence of open models. A few years ago, or about a year ago, Llama appeared, and many were seriously concerned then. But, frankly, Llama became a huge event for most industries and companies worldwide, because it was the world's first open reasoning model. Since then, many other open reasoning models have appeared. And it is open models that have given companies, industries, researchers, educators, universities, and startups the opportunity to use them as a foundation to create solutions specialized for specific tasks or industries. The third area where colossal progress was made last year is the concept of physical intelligence or physical AI, which understands not only language but also, if one can say so, nature itself. This is AI that understands the physical world, AI that understands proteins, chemical processes, laws of physics, for example, hydrodynamics, particle physics, quantum physics. These are AI that are currently learning all these different structures and languages, if one can say so. Proteins, in essence, are also a language. And all these AI are now making such rapid progress that industries, whether it's industrial manufacturing or drug development, are truly starting to move forward at a very fast pace. One striking example is our partnership with Eli Lilly. They realized that AI has reached an incredible level in understanding the structure of proteins and chemical compounds. It has effectively learned to interact with proteins and "talk" to them, just as we communicate with ChatGPT. And thanks to this, truly major breakthroughs await us. All these breakthroughs raise concerns about the role of humans. You and I have discussed this many times, but it is important to convey this to the entire audience. There is a serious concern that AI will lead to job displacement. At the same time, you argue the opposite. It is obvious that the large-scale deployment of AI, that very largest infrastructure construction in history that you spoke about, is already happening. Energy creates jobs, chips, industry create jobs, the infrastructure layer creates jobs, land, energy, shale extraction, jobs, jobs, jobs. Then let's break this down further. Do you truly believe that we will face a labor shortage? How do you see AI and robotics changing the very nature of work, rather than leading to its disappearance? This can be looked at from several different angles. First of all, we are talking about the largest infrastructure construction in human history. And this creates a huge number of jobs. What is particularly important is that these are jobs related to applied professions. We will need plumbers and electricians, builders and metalworkers, network engineers, specialists in installing and configuring equipment. All these professions are now in high demand. In the United States, we are seeing a very serious boom. Salaries in this sector have almost doubled. We are talking about six-figure incomes for people who build chip factories, computer factories, or AI factories. At the same time, we are experiencing a severe shortage of such specialists. And I am very pleased to see that more and more people in different countries are beginning to realize how important this direction is. Everyone should have the opportunity to earn a decent living. You don't necessarily need a Ph.D. in computer science for this. And I am sincerely happy to see that this is becoming a reality. The second important point. We often theorize about task automation and discuss its impact on employment. I will give a few real-life examples. What actually happened 10 years ago, one of the first professions that was said to disappear was radiology. The reason was that one of the first areas where AI surpassed human capabilities was computer vision. And one of the largest applications of computer vision is the analysis of medical images, which radiologists do. Ten years have passed, and yes, it's true, and it has fully penetrated all aspects of radiology. Radiologists do use AI for image analysis. The impact is 100%, and it is absolutely real. However, and this is not surprising, if we reason from first principles. The number of radiologists has increased during this time. Is this due to distrust of AI, or because human interaction with AI's results yields a better outcome? The reason is that the radiologist's task is to diagnose diseases and help patients. This is the goal of the profession. Image analysis is just one of the tasks within this work. The fact that they can now analyze images almost instantly allows them to spend more time on diagnosis, communicating with patients, and interacting with other doctors. And, not surprisingly, although it might seem otherwise at first glance, as a result, the number of patients that a hospital can accommodate has also increased. Because previously, many people waited a long time for their turn for examination. Now there are more patients, hospital revenues have increased, and they have started hiring more radiologists. The same is happening with nurses. In the US, there is a shortage of about 5 million nurses. Previously, they spent almost half their time filling out documentation and transcribing patient visits. Now, thanks to AI, these tasks are automated. One of the companies, Abridge, our partner, is doing amazing work in this area. As a result, nurses can dedicate more time to patients, to human interaction. And because more patients can now be seen and the system is less constrained by staff shortages, people get to hospitals faster. As a result, hospitals operate more efficiently and hire even more nurses. It turns out that AI increases productivity, hospitals develop and require more staff, and the need is huge. Too many people are waiting too long for medical care. These are two very illustrative examples. The simplest way to understand the impact of AI on a specific profession is to divide the concept of work goals and work tasks. If you just put a camera and observe us, you might think we are typists because I am typing all the time. And then you might say, "Since AI can predict words and help with typing, we will be out of a job." But obviously, our goal is not to type. So the main question is: what is the goal of your profession? In the case of radiologists and nurses, the goal is to care for people. And this goal becomes more achievable and more productive precisely because routine tasks are automated. And if we consider any profession through the lens of goals and tasks, it provides a very useful framework for understanding the impact of AI. >> Subscribe right now to my Telegram channel via the link in the description. I have prepared for you the top three materials that, in my opinion, everyone should know. First: a map of a hundred top AI startups - this is the future in one picture. Second: a forecast from an insider at OpenAI, who, even before the appearance of ChatGPT, predicted everything that is currently happening with neural networks. And this year, he released a new forecast until 2027. And third, the most powerful: my analysis of an essay by the founder of Anthropic, who is essentially the second person in the world of artificial intelligence. He has laid out step-by-step what will happen in the world in the next 5 years, and most importantly, what the universal AI, which everyone fears or awaits, will be like. Go to the link in the description. Let's move beyond developed economies. Help me understand how AI can be truly transformative on a global scale and genuinely help the world. This weekend, I was reading a material from Anthropic, which essentially stated that today the use of AI is most concentrated among educated segments of society. Moreover, even within this group, the wealthiest and most educated people use it particularly actively. Of course, they analyze this based on their own models, so certain biases are possible here. So how do we make AI a truly transformative technology, much like Wi-Fi and 5G have been for developing countries? And if we consider this in a broader context, what does it mean for the developing world, and how do we expand the global economy? And second, returning to the topic of employment and the impact of AI and robotics, it is obvious that some job replacement will still occur. We are already seeing this in the US. Perhaps we are creating more jobs for plumbers and electricians, but we probably need fewer analysts in financial institutions. Fewer lawyers are also needed because they can process data significantly faster. Let's focus on the developing world for a moment. How do you see this situation developing there? First of all, AI is infrastructure. And I cannot imagine any country in the world that should not include AI in its infrastructure. Every country has electricity, it has roads, and AI is becoming a part of infrastructure. Of course, AI can be imported, but today training AI is no longer something prohibitively difficult. And thanks to the large number of open models, by combining them with local expertise, every country can create models that are useful specifically for it. Therefore, I sincerely believe that every country should participate in creating its own AI infrastructure, develop its own AI, use its key natural resources, language, and culture, develop national models, constantly improve them, and integrate national intelligence into its own ecosystem. This is the first point. Second, and it is incredibly easy to use. This is perhaps the simplest software in history. That is why it is growing and spreading so fast. In just 2-3 years, the number of users is approaching a billion. I believe that, first and foremost, Claude is an amazing product. Anthropic has made a huge leap in its development. We use Claude throughout the company. Its capabilities in programming and reasoning are simply outstanding. Any company working with software should really work with it. On the other hand, ChatGPT is probably the most successful consumer AI in history. Its simplicity and accessibility make it a truly mass-market tool. I believe that everyone should start using it, whether they are residents of a developing country or a student. It is absolutely clear that today it is extremely important to know how to use AI and how to direct it, formulate queries, manage it, set limitations, and evaluate its results. These skills are no different from managing people, which we do constantly. In the future, in addition to biological, carbon-based forms, we will also have digital versions, silicon-based forms of intelligence, and they will also need to be managed. This will become part of our digital workforce. Therefore, for developing countries, my advice is simple. Build infrastructure, get involved in AI, and realize that AI will likely help reduce the technological gap because it is simple to use, widely available, and scalable. Frankly, I am quite optimistic about AI's potential to uplift developing countries. And for many people who do not have a computer science education, new opportunities are now opening up. Today, anyone can become a programmer. Previously, we had to learn to program a computer. Now, you simply program a computer by telling it how to do it. And if you don't know how to use AI, you can simply ask it and say, "I don't know how to use AI. Explain it to me." And it will explain everything to you. And you know, you say, "I want to write a program to create my own website. How do I do that?" And it asks you a whole series of questions about what kind of website you want to create, and then writes the code for you. That is, it is so simple to use. And, of course, this is where its incredible power lies. What is truly exciting? Then two quick questions, because we are running out of time. We are currently in Europe. When we talked about companies, we mainly mentioned American and Asian companies. Tell us, how can Europe's success and its future intersect, and what role does and will Nvidia play here in Europe? Nvidia has an advantage. We work with every AI company in the world because we are at the infrastructure level and enable AI everywhere. We enable AI for languages, biology, physics, world modeling, as well as for manufacturing and robotics. And what is really very important for Europe. Remember how strong your industrial base is. Industrial manufacturing in Europe is incredibly developed. This is your chance to leapfrog the era of pure software. The US led in the software era. AI is software for which you don't need to write code. You don't write AI, you train it. And if you enter this field now, you can combine Europe's industrial and manufacturing potential with artificial intelligence, and this will lead you into the world of physical AI or robotics. Robotics is a unique opportunity for European countries. In all the countries I visit here, the industrial base is indeed very strong. Furthermore, it is important to understand that fundamental sciences in Europe are still very, very strong. And now these sciences have the opportunity to apply AI to accelerate discovery. Therefore, in my opinion, it is quite obvious that Europe needs to seriously focus on increasing its energy resources to invest in the infrastructure layer and create a rich AI ecosystem here in Europe. So, if I understand correctly, we are far from an AI bubble? The question is rather whether we are investing enough. Many talk about a bubble, but what I hear from you is the question: are we doing enough to realize the potential, to expand the global economy? One good test of an AI bubble is to look at the following. Today, there are millions of Nvidia GPUs in the cloud. We are present in all clouds. We are used everywhere. And if you try to rent an Nvidia GPU today, it is incredibly difficult. And market prices for GPU rentals are rising, not only for the latest generation, but also for graphics cards from the previous two generations. Rental prices are rising because a huge number of AI companies are emerging, and also because companies are reallocating their R&D budgets. Eli Lilly is a great example. 3 years ago, almost their entire R&D budget went to wet labs. And now look at the huge AI supercomputer and large AI center they have invested in. More and more of this R&D budget will shift towards AI. And talk of a bubble arises because the investments are truly enormous. The investments are so large because we need to build the infrastructure that will serve as the foundation for all the layers above. And therefore, I believe that this opportunity is truly exceptional. Everyone should be involved. Everyone should be included in this process. We need more energy. I think we all understand that. We need more electricity, more industrial capacity, more skilled workers. And essentially, it is here in Europe that this workforce is still very strong. In many ways, the United States has lost this over the last 20-30 years. But in Europe, this potential is still incredibly powerful. This is an exceptional opportunity that should be seized. I know that where Larry and I work, we see investment opportunities, and the scale of these investments continues to grow. The number of startups, as I mentioned, in 2025, made this year the largest in venture investment history, over $100 billion worldwide. And most of this money went into so-called AI-native companies. These AI companies are essentially building the application layer on top of the infrastructure. And they will need infrastructure, they will need our investments to build this future together. I sincerely believe that participating in this is an excellent investment for pension funds worldwide to grow with the world. And AI is one of the key messages I convey to many political leaders. We must ensure that the ordinary pensioner, the ordinary contributor also participates in this growth. If they just watch from the sidelines, they will feel excluded from the process. We want to invest in infrastructure. Infrastructure is an excellent investment opportunity. We are talking about the largest infrastructure construction in the history of humanity. Get involved. Unfortunately, our time has come to an end. I hope everyone in the hall and everyone watching the broadcast online has seen Jensen Huang's strength as a leader. Not only a leader in AI technology, but also a leader in business, and also a leader with heart and soul, which is especially important today. Thank you all. Thank you. [applause]