Below is the full dialogue:

Shen Xiangyang: Last night, I tossed and turned, with one crucial reason being my eager desire to introduce to you the most outstanding CEO in the universe. But I also secretly worried about your company, as Apple's stock price rose last night, while NVIDIA's performance was slightly overshadowed. I can't wait to know the closing results of the stock market! Waking up this morning, my first question to my wife was whether NVIDIA managed to hold up. You have long been at the helm in the field of artificial intelligence. Could you share your thoughts on artificial intelligence again, and the possible impacts that this technology, or AGI (General Artificial Intelligence), may bring?

Huang Renxun: As you are aware, when artificial intelligence networks can learn and master the understanding of various data such as bytes, language, images, and protein sequences, a revolutionary and pioneering ability emerges. Suddenly, we have computers that can understand the meanings of words. Thanks to generative AI, information can freely convert between different modes, such as from text to images, from proteins to text, from text to proteins, and even from text to chemicals. This tool, which originally existed as a Function Approximator and a language translator, now faces the question of how we can fully utilize it. You have witnessed the emergence of venture companies worldwide, combining these different models and abilities, showing infinite possibilities.

Therefore, I believe the truly amazing breakthrough is that we can now understand the true meaning of information. This means that as a digital biologist, you can understand the meaning of the observed data, thereby accurately capturing key information from a myriad of data; as a chip designer, system designer at NVIDIA, or agricultural technologist, climate scientist, researcher in the energy field, in the exploration of new materials, this is undoubtedly a pioneering feat.

Shen Xiangyang: Nowadays, the concept of a universal translator has taken shape, giving us the ability to understand everything in the world. Many have heard you describe the astonishing impact of artificial intelligence on society. Those viewpoints deeply touched me, even in some aspects, it made me feel shocked. Looking back at history, the agricultural revolution allowed us to produce more food, the industrial revolution significantly increased our steel production. Entering the information technology era, the quantity of information has exploded. Today, in this era of intelligence, NVIDIA and artificial intelligence are jointly 'manufacturing' intelligence. Can you further elaborate on why this work is so important?

Huang Renxun: From the perspective of computer science, we have reinvented the entire stack. This means that the way we developed software in the past has undergone fundamental changes. When it comes to computer science, software development is indispensable, how it is achieved is crucial.

In the past, we relied on manually writing software, conceptualizing functions, designing algorithms with imagination and creativity, then transforming them into code, and inputting them into the computer. From Fortran to Pascal, and then to C language and C++, these programming languages allowed us to express creativity with code. The code ran well on the CPU, we input data to the computer, asking what functions it discovered from it, observing the data we provided, the computer could recognize patterns and relationships within.

However, the current situation is different. We no longer rely on traditional code writing methods, but have transitioned to machine learning and machine generation. This is no longer a simple software problem, but involves machine learning, which generates neural networks and processes them on GPUs. This shift from coding to machine learning, from CPU to GPU, marks the arrival of a new era.

Moreover, due to the extremely powerful functions of GPUs, the types of software we can now develop are extraordinary. Built upon this strong foundation is the vigorous development of artificial intelligence. This is the transformation brought about by its emergence, leading to significant changes in computer science. Now, what we need to consider is how such changes will affect our industry? We are all rushing to utilize machine learning to explore new realms of artificial intelligence. So, what exactly is artificial intelligence? It is actually a familiar concept to many, involving cognitive automation and problem-solving automation. Problem-solving automation can be summarized into three core concepts: observing and perceiving the environment, understanding and reasoning the environment, and then proposing and executing a plan.

For example, in autonomous driving cars, vehicles can perceive the surrounding environment, reason about their own and other vehicles' positions, and finally plan the driving route. This is actually a form of a digital driver. Similarly, in the medical field, we can observe CT scan images, understand and reason the information in the images. If abnormalities are detected, it may indicate the presence of a tumor. Then, we can mark it and inform the radiologist. At this point, we are playing the role of a digital radiologist. Almost everything we do can be associated with artificial intelligence applications, which can excellently perform specific tasks.

If we have a sufficient number of digital intelligences, and these intelligences can interact with the computers generating this digital information, then this constitutes digital artificial intelligence. However, currently, despite the seemingly large overall consumption of data centers we all have, data centers are primarily producing something called 'Tokens,' rather than true digital intelligence.

I can explain the difference between the two. 300 years ago, General Electric and Westinghouse Electric Companies invented a new instrument - the generator, which eventually evolved into an AC generator. They were very wise to create 'consumers' to utilize the electricity they produced, including light bulbs, toasters, and other electrical devices. Of course, they also created various digital devices or appliances that required power.

Now, let's see what we are doing. We are creating smart tools like Copilots, ChatGPT, etc., which are different types of intelligent 'consumers' we have created. They are essentially devices that consume energy, just like light bulbs and toasters. But imagine those amazing intelligent devices that everyone will use, they will be connected to a new factory. This factory used to be an AC power plant, but now, the new factory will be a digital smart factory.

From an industrial perspective, we are actually creating a new industry that absorbs energy and generates digital intelligence. This digital intelligence can be applied in various different scenarios. We believe that the consumption level of this digital intelligence industry will be enormous, and this industry did not exist before, just like the AC power generation industry did not exist before.

Shen Xiangyang: You have outlined a hopeful and bright future for us, which is largely due to your and Nvidia's outstanding contributions to the field over the past decade. Moore's Law has always been highly regarded in the industry, and in recent years, the 'Huang's Law' has gradually become familiar to people. In the early days of the computer industry, Intel's Moore's Law predicted a doubling of computing power every 18 months. However, in the past 10 to 12 years, especially under your leadership, the growth rate of computing power has even exceeded this prediction, achieving a doubling or even higher rate of growth each year.

From the consumer side, the computing demand of large language models has increased at a rate of over four times each year over the past 12 years. If this rate continues for 10 years, the growth in computing demand will be an astonishing figure - as high as 1 million times. This is also an important argument when I explain the 300-fold increase in the price of NVIDIA stock over the past 10 years to others. Considering this huge increase in computing demand, NVIDIA's stock price may not seem high. So, when you use your 'crystal ball' to predict the future, do you think we will witness another 1 million-fold increase in computing demand over the next 10 years?

Huang Renxun: Moore's Law relies on two core concepts: one is the design principle of Very Large Scale Integrated Circuits (VLSI), which is inspired by the works of me, Professor Carver Mead of Caltech, and Professor Lynn Conway, inspiring an entire generation; the second is that as transistor sizes continue to shrink, we are able to double semiconductor performance every period of time, about every one and a half years, achieving a performance doubling, and therefore a tenfold performance increase every five years, and even a hundredfold increase every ten years.

We are in a trend where the larger the scale of neural networks and the more data used for training, the more intelligent AI seems to perform. This empirical rule has similarities with Moore's Law, let's call it the 'Scaling Law', and this law seems to still be in effect. However, we are also aware that relying solely on pre-training - using massive data globally to automatically mine knowledge - is far from enough. Just as graduating from university is a crucial milestone but by no means the end. Next, there is the post-training phase, which involves in-depth research into a specific skill, requiring the comprehensive use of various techniques such as reinforcement learning, human feedback, AI feedback, synthetic data generation, and multi-path learning. In summary, post-training is about selecting a specific field and dedicating yourself to delving deep into it. This is like when we enter our professional careers, where there is a lot of specialized learning and practice to be done.

After this, we will eventually reach the so-called 'thinking' stage, also known as the computational thinking stage. Some things you can see the answer at a glance, while others require us to break them down into multiple steps and start from first principles, solving them one by one. This may require multiple iterations, simulating various possible results, as not all answers are predictable. Therefore, we call it thinking, and the longer the thinking time, the higher the quality of the answers tends to be. And a large amount of computing resources will help us produce higher-quality answers.

Although today's answers are the best results we can provide, we are still looking for a critical point, where the answers obtained are no longer limited to the best level we can currently provide. At this point, you need to determine whether the answers are truly reliable, meaningful, and wise. We must reach a point where the obtained answers are largely trustworthy. I believe this will take several years to achieve.

At the same time, we still need to continue to improve computing power. As you mentioned before, over the past ten years, we have increased computational performance by 1 million times. NVIDIA's contribution is that we have reduced the marginal cost of computation by the same magnitude. Imagine if there are things in life that you rely on, such as electricity or any other option, when its cost is reduced by 1 million times, your behavior habits will undergo fundamental changes.

In terms of computing, our views have undergone a revolutionary change, and this is one of NVIDIA's greatest achievements ever. We use machines to learn from vast amounts of data, a task that researchers cannot accomplish alone, and this is the key to the success of machine learning.

Shen Xiangyang: I am eager to hear your views on how Hong Kong should act in the current opportunities. Now, something particularly exciting is 'AI for Science', which you have always been very passionate about. Hong Kong University of Science and Technology has invested a large amount of computing infrastructure and GPU resources, with a special emphasis on promoting cooperation between departments, such as the interdisciplinary fusion of physics and computer science, materials science and computer science, biology and computer science, and so on. You have also explored the future of biology before. In addition, it is worth mentioning that the Hong Kong government has decided to establish a third medical school, and the Hong Kong University of Science and Technology is the first university to submit this proposal. So, what advice do you have for the university president, myself, and the entire university?

Huang Renxun: First of all, I introduced artificial intelligence at the supercomputing conference in 2018, but at that time it faced many doubts. The reason was that the artificial intelligence at that time was more like a 'black box.' Admittedly, to this day, it still maintains the 'black box' characteristics to some extent, but it is now more transparent than before.

For example, both you and I are 'black boxes,' but now we can ask AI questions like 'Why do you propose this suggestion?' or 'Please explain step by step the process you used to reach this conclusion.' Through such questions, AI is becoming more transparent and easy to explain. Because we can explore its thought process through questions, just like professors would understand students' thinking through questioning. It's not just about getting the answers, but also about the reasonableness of the answers and whether they are based on first principles. This was not possible in 2018.

Furthermore, AI currently cannot directly derive the answers from first principles; it learns and draws conclusions by observing data. Therefore, it is not a solver that simulates first principles, but rather mimics intelligence, mimics physics. Then, does this imitation have value for science? I believe its value is immeasurable. Because in many scientific fields, although we understand first principles such as Schrödinger's equation, Maxwell's equations, etc., when faced with large systems, we find it challenging to simulate and comprehend. Therefore, we cannot solve based solely on first principles, there are limitations computationally, even impossibilities. However, we can use AI, train it to understand these physical principles, and use it to simulate large systems to help us comprehend these systems.

So, in which specific areas can this application be effective? First, the scale of human biology ranges from nanometers to years in time. It is simply impossible to achieve with traditional solvers over such a broad scale and time span. Now the question is, can we use AI to simulate human biology to gain a deeper understanding of these extremely complex multiscale systems?

In this way, we may call it creating a digital twin of human biology. This is where our hopes lie. Nowadays, we may already have the computer science technology that enables digital biologists, climate scientists, and scientists dealing with exceptionally huge complex problems to truly comprehend physical systems. This is my expectation, hoping that this vision can be realized in this interdisciplinary field.

Mentioning your medical school project, for The Hong Kong University of Science and Technology, a unique medical school is about to be born here, even though the traditional professional fields of this university are technology, computer science, and artificial intelligence. This is quite different from most medical schools in the world, which mostly introduce artificial intelligence and technology after becoming medical schools, which often face skepticism and distrust from people about their technology. However, you have the opportunity to start from scratch, creating an institution closely intertwined with technology from the beginning and driving the continuous development of technology here. The people here are well aware of the limitations and potential of technology. I believe this is a once-in-a-lifetime opportunity, and I hope you can grasp it firmly.

Shen Xiangyang: Of course, we will take your advice. The Hong Kong University of Science and Technology has always excelled in technology and innovation, continuously advancing the forefront of fields like computer science, engineering, and biology. Therefore, as the third medical school in Hong Kong, we firmly believe that we can take a unique path, combining traditional medical training with our strengths in technical research. I am confident that in the future we will seek more advice from you. However, I would like to slightly change the topic and talk about leadership. You are one of the longest-serving CEOs in Silicon Valley, probably far beyond others. You have been the CEO of Nvidia for about 30 or 31 years, right?

Huang Renxun: It's been nearly 32 years!

Shen Xiangyang: But you never seem to get tired.

Huang Renxun: No, I actually feel very tired. When I got here this morning, I said I was super tired.

Shen Xiangyang: But you continue to move forward. Therefore, of course we want to learn some experience of leading large organizations from you. How do you lead an institution with tens of thousands of employees, amazing revenue, and a large number of customers with such a wide coverage? How do you lead such a large organization with such amazing efficiency?

Huang Renxun: Today I want to say, I am very surprised. Usually, you only see computational biologists or business students, but today we see a computational biologist who is also a business student, which is fantastic. I have never taken any business courses, never written a business plan, I have no idea how to start. I rely on all of you to help me.

What I want to tell you is, first of all, you should learn as much as possible, and I have been constantly learning. Secondly, about anything you want to dedicate yourself to and consider as a lifelong career, the most important thing is passion. Consider everything you do as your life's work, not just a job. I think this mindset will make a big difference in your heart. nvidia is my career.

If you want to be a CEO of a company, you have a lot to learn, you must constantly reinvent yourself. The world is always changing, your company and technology are always changing. Everything you know today will be useful in the future, but far from enough, so I basically learn every day. I watch YouTube on the plane on my way over, chat with my ai. I found an ai to be my mentor, ask many questions. AI will give me an answer, I will ask why it gave this answer, let it explain to me step by step, apply this reasoning to other things, give me some analogies. There are many different ways to learn, I utilize ai. So, there are many ways to learn, but what I want to emphasize is, you must constantly learn.

Regarding the insights on being a CEO and a leader, I have summarized the following points:

First, as a CEO and leader, you do not need to play the omnipotent role. You must firmly believe in the goals you pursue, but this does not mean you have to know every detail. Confidence and certainty are two completely different concepts. In the process of pursuing goals, you can move forward with confidence, while maintaining an open mind, gladly accepting and embracing uncertainties. This uncertainty actually provides you with a space for continuous learning and growth. Therefore, learn to draw strength from uncertainty, see it as a friend that propels you forward rather than an enemy.

Furthermore, as a leader, your decisions should always revolve around the mission, taking into account the well-being and success of others. Only when your decisions truly benefit others can you earn their trust and respect. Whether it's internal company employees, partners, or the entire ecosystem we serve, I am always thinking about how to promote their success and safeguard their interests. In the decision-making process, I always take others' best interests as the starting point, using it as a guide for our actions. I believe these may be very helpful.

Shen Xiangyang: Regarding teamwork, I have a question that I am very interested in asking. You have 60 direct subordinates reporting to you. How do you conduct your employee meetings? How do you effectively manage so many senior executives? This seems to reflect your unique leadership style. Huang Renxun: The key is transparency. I will clearly explain our reasons, goals, and actions that need to be taken in front of everyone, and we work together to formulate strategies. Regardless of the type of strategy, everyone will hear it at the same time. Because they all participate in making the plans, when the company needs to make a decision, it is something everyone discusses together, not just me making the call, nor me telling them what to do.

Shen Xiangyang: About teamwork, I have a very interesting question to ask. You have 60 direct subordinates reporting to you; how do you conduct your employee meetings? How do you effectively manage so many senior executives? This seems to reflect your unique leadership style. Huang Renxun: The key is to maintain transparency. I will clearly articulate our reasons, goals, and actions that need to be taken in front of everyone. We collaborate to develop strategies. Regardless of the strategy, everyone will hear it at the same time. Because they all participate in planning, when the company needs to decide on something, it is something we all discuss together, not just my decision or me telling them what to do.

We discuss together and come to a conclusion together. My responsibility is to ensure that everyone receives the same information. I am usually the last one to speak, based on our discussion results, to clarify the direction and priorities. If there are any uncertainties, I will address them. Once we reach a consensus, understand the strategy, I will push the work forward based on the fact that everyone is an adult. Concerning my code of conduct - continual learning, confidence but embracing uncertainty - if I am unclear, or they are unsure about something, I hope they can speak up proactively. If they need help, I hope they can seek support from us. Here, no one will face failure alone.

Later, when others see my behavior pattern - as a CEO, as a leader, I can show vulnerability, seek help, admit uncertainty, make mistakes - they will understand they can do the same. What I expect is if they need help, they speak up bravely. But besides that, my team consists of 60 individuals, each a top talent in their respective fields. In most cases, they do not need my help.

Huang Renxun: That was indeed a little trick I used as an icebreaker, not with any real meaning. I went to university at 16, met my wife when I was 17, and she was 19 at the time. Being the youngest student in the class, with only three girls out of 250 classmates, I had to learn some attention-grabbing skills as I looked like a kid. I approached her, told her that even though I looked young, her first impression of me must be that I am smart, so I mustered up the courage and said, 'Do you want to see my homework?'

Shen Xiangyang: I must say, your management approach has indeed achieved significant results. Your speech at the degree award ceremony is still fresh in my memory; you mentioned many data points about The Hong Kong University of Science and Technology, especially the number of alumni-founded startups and unicorn and listed companies nurtured by our school. This university is truly known for nurturing new entrepreneurs and companies. However, even in such an environment, we still have many master's students studying here today. You and your team established your company at a very young age and have achieved the remarkable success you see today. So, what advice do you have for our students and faculty? When and why should they start their own businesses? Besides the promise you made to your wife to start a company before you turned 30, do you have any other advice?

Huang Renxun: That was indeed a little trick I used to break the ice, not with any real meaning. I went to university at 16, met my wife when I was 17, and she was 19 at the time. Being the youngest student in the class, with only three girls out of 250 classmates, I had to learn some attention-grabbing skills as I looked like a kid. I approached her, told her that even though I looked young, her first impression of me must be that I am smart, so I mustered up the courage and said, 'Do you want to see my homework?'

Then, I made a promise to her, I said: "If you do homework with me every Sunday, I promise you will get full marks." That way, we could go on a date every Sunday and spend the whole day learning together. In order to eventually make her willing to marry me, I also told her that by the time I turned 30 - when I was only 20 - I would become a CEO. I had no idea what I was talking about at the time. Later, we indeed got married. So, that's my entire advice, with a bit of humor and sincerity.

Shen Xiangyang: I collected a question from a student who wanted to know: he excels in school but needs to focus entirely on studying. After reading your love story, he's worried that if he also spends time on romance, will it affect his academic performance?

Huang Renxun: My advice is, absolutely not. But the premise is that you must maintain excellent grades. She (my wife) has never found out about this little secret, but I always wanted her to think I was very smart. So, before she arrived, I already finished my homework. By the time she came, I already knew all the answers. She may have always thought I was a genius, and for a whole four years, she believed that.

Shen Xiangyang: A professor from the University of Washington expressed an opinion a few years ago. He believed that in the deep learning revolution, top American universities like the Massachusetts Institute of Technology (MIT) did not make too many groundbreaking contributions. Of course, he was not only referring to MIT, but pointing out that the contributions of all top American universities in the past decade have been relatively limited. In contrast, we see top companies like Microsoft, OpenAI, and Google's DeepMind achieve astonishing results, partly due to their powerful computing capabilities. So, faced with this situation, how should we respond? Should we consider joining Nvidia or collaborating with Nvidia? As our new ally, can you provide us with some advice or assistance?

Huang Renxun: The issue you mentioned indeed touches upon a serious structural challenge that universities currently face. We all know that without machine learning, we cannot drive scientific research as rapidly as we do today. And machine learning depends on strong computational support. This is similar to researching the universe without radio telescopes or studying fundamental particles without particle accelerators. Without these tools, we cannot delve deep into unknown areas. And today's 'scientific instrument' is AI supercomputers.

One structural problem that universities face is that researchers usually have to raise funds on their own, and once they receive funding, they are reluctant to share resources with others. But machine learning has a characteristic that requires these high-performance computers to be fully utilized during certain time periods, rather than being idle all the time. No one will occupy all resources all the time, but each person will need immense computing power at some point. So, how should universities deal with this challenge? I think universities should become leaders in infrastructure development, driving the research development of the entire school by centralizing resources. However, implementing this in top universities like Stanford or Harvard is very difficult because computer science researchers in these universities can usually raise a lot of funds compared to researchers in other fields.

So, what is the solution now? I believe that if universities can build infrastructure for the entire school, it will effectively lead the transformation in this field and have a profound impact. However, this is indeed a structural challenge that universities currently face. It is for this reason that many researchers choose to intern or conduct research at our company, Google, Microsoft, and other enterprises because we can provide opportunities to access advanced infrastructure. Subsequently, when they return to their universities, they hope we can maintain the vitality of their research so they can continue their work. In addition, many professors, including visiting professors, combine teaching with research. Our company has hired several professors like this. Therefore, while there are many diverse ways to solve the problem, the most fundamental thing is for universities to reassess and optimize their research funding system.

Shen Xiangyang: I have a challenging question to ask you. On the one hand, we are delighted to see a significant increase in computing power and a decrease in prices, which is undoubtedly good news. But on the other hand, your GPUs consume a lot of energy, and there are predictions that by 2030, global energy consumption will increase significantly. Are you concerned that because of your GPUs, the world is actually consuming more energy?

Huang Renxun: I will answer you in this way, I will use reverse thinking. First of all, I want to emphasize that if the world consumes more energy to power the global AI factories, then when all this happens, our world will become better. Now, let me elaborate on a few points for you.

First, the goal of AI is not just to train models, but to apply these models. Of course, going to school to learn, purely for the sake of learning, is noble and wise in itself. However, most students come here, investing a lot of money and time, with the goal of succeeding in the future and applying the knowledge they have learned. Therefore, the true goal of AI is not training, but reasoning. The reasoning process is highly efficient, it can discover new ways to store carbon dioxide, such as in reservoirs; it may be able to design new types of wind turbines; it may be able to discover new materials for storing electricity, or more efficient materials for solar panels, etc. So, our goal is to eventually create AI that can be applied, not just trained.

Second, we should remember that AI does not care where it is 'learning'. We do not need to place supercomputers on campuses near the power grid. Instead, we should consider placing AI supercomputers slightly away from the power grid, allowing them to use sustainable energy, rather than placing them in densely populated areas. We must remember that all power plants were originally built to meet the electricity needs of our home appliances, such as light bulbs, dishwashers, and now, due to the popularity of electric vehicles, electric vehicles also need to be close to us. However, supercomputers do not need to be near our homes, they can learn and operate elsewhere.

Third, what I hope to see is that AI can efficiently and intelligently discover new scientific achievements, so that our existing energy waste problems - whether it is the waste of the power grid, which is often overconfigured most of the time, and underconfigured at other times - we can save energy through AI in many different fields, saving energy from our waste, and ultimately, expect to save 20% to 30% of energy. This is my expectation and dream, I hope to see that using energy for intelligent activities is the best way to utilize energy that we can imagine.

Shen Xiangyang: I completely agree that applying energy efficiently to intelligent activities is the best way to utilize it. If equipment is manufactured outside of a place like the China Greater Bay Area (including Shenzhen, Hong Kong, Guangdong, etc.), the efficiency tends to decrease because it is difficult to find all the necessary components. Take DJI as an example, this local commercial drone company has remarkable technology. My question is, as the physical aspect of intelligence becomes increasingly important, such as with robots - especially a special type of robot like self-driving cars - what is your view on the trend of these physical intelligent entities rapidly emerging in our lives? How should we grasp and utilize the huge potential of the Greater Bay Area hardware ecosystem in our professional lives?

Huang Renxun: This is an excellent opportunity for China and the entire Greater Bay Area. The reason is that this region has a high level of integration in the field of electromechanical technology, the fusion of machinery and electronic technology. However, for robots, a critical missing piece is AI that understands the physical world. Current large language models, such as ChatGPT, excel at understanding cognitive knowledge and intelligence, but know very little about physical intelligence. For example, it may not understand why a cup does not go through the table when put down. Therefore, we need to teach AI to understand physical intelligence.

In fact, what I want to tell you is that we are making significant progress in this area. You may have seen some demonstrations, through generative AI, text can be turned into videos. I can generate a video, starting with my photo, then giving the command 'Jensen, pick up the coffee cup, take a sip'. Since I can instruct AI to perform actions in videos, why can't we generate the correct instructions to control a mechanical arm to perform the same actions? Therefore, the leap from current generative AI to universal robots is not far off. I am excited about the prospects in this field.

There are three types of robots that are expected to achieve large-scale production, and almost exclusively limited to these three. Other types of robots that have appeared in history have been difficult to produce on a large scale. Large-scale production is crucial because it can drive the technology flywheel effect. High investment in research and development (R&D) can bring about technological breakthroughs, producing superior products, further driving the expansion of production scale. This R&D flywheel is key for any industry.

In fact, although only three types of robots can truly achieve large-scale production, two of them will become the highest in output. The reason is that these three types of robots can be deployed in the current world. We refer to them as the 'brown zone' (areas awaiting redevelopment). These three types of robots are: autos, because we have built a world adapted to them in the past 150 to 200 years; next is drones, because the sky has almost no limits; and of course, the largest in output will be humanoid robots, because we have built a world for ourselves. With these three types of robots, we can expand the application of robot technology to a very high output, which is the unique advantage of manufacturing ecosystems like the Bay Area.

If you think deeply, you will realize that the Greater Bay Area is the only region in the world that possesses both electromechanical technology and artificial intelligence technology simultaneously. This situation does not exist elsewhere. The other two strong electromechanical industrial nations are Japan and Germany, but unfortunately, they are far behind in terms of artificial intelligence technology and really need to catch up. Here, we have a unique opportunity, and I will seize it tightly.

Shen Xiangyang: I am very happy to hear your views on physical intelligence and robots. The Hong Kong University of Science and Technology is indeed very good at these aspects.

Huang Renxun: Artificial intelligence, robot technology, and medical care are the three areas where we truly need innovation.

Shen Xiangyang: Indeed, with the establishment of our new medical school, we will further drive the development of these areas. However, to achieve all these wonderful things, we still need your support, we need your GPU and other resources. (Tencent Technology Special Compilation Jin Lu)

Editor/Lambor