Professor Woo organized these lectures. Please join me to welcome Dr. Lisa Su to give the speech and also speak on the case of the AI and the AMD transformation to AI leaders. Thank you.
And then we have a quite different speech or speaking on her topics, because when I check with the audience and pre-registered audience, there are questions. There are too many. Well, actually when we made an announcement on her speech at NTU, we limited to 500 audience in this lecture room, and it was all registered in about one minute. Therefore, too many questions, so I cannot ask her to answer all the questions. As you just walked into a lecture hall, you can see that there is another room in the third row. They are also using the video to listen to what she will talk about today. So we will use a different kind of fireside chat for her speech. She will also accept some questions from the students during the break after about 10 questions when I chat with her. And finally, she will accept some other questions. Also, I think she's very easygoing, as you can see when you walk into the room, so probably after the lecture she could speak with some students and take photos. So that's today's lecture. Okay, thank you.
Wonderful, thank you. How is everyone this morning? Yeah, so first of all, President Chen, thank you so much for inviting me. You invited me several times to come to NTU, and I was honored every time to come here, but sometimes when I come to Taiwan it's very, very busy, especially around Computex timing. So this is actually a great time that I can visit. Thank you for the warm welcome, and Professor Woo as well. Thank you so much. It's really my pleasure to be here to meet all of you very talented students. You make me feel very old, but hopefully I can give you a few stories for your career. Okay, please.
I will have some questions for you first about the 10 questions initially, and after that we have open to the audience for the students' questions. Okay, the first question is that AMD has evolved from a CPU-focused company into a powerhouse across CPUs, GPUs, and adaptive computing. What were the key turning points that drove this transformation? Yeah, please.
Yeah, so maybe let me ask the students: how many students here are electrical engineering and computer science? Please raise your hands. Okay, and then how about medical field? Not too many. So maybe I should say my background was in electrical engineering and computer science. That's where I went to school at MIT for bachelor's, master's, and PhD. My specialty was semiconductor devices. So when I was a young student, that is what I studied. Now for AMD, I've been in the semiconductor industry for over 30 years. What you find is that the technology moves so fast. For AMD, I've been with the company since 2012 and CEO since 2014. Probably the most important thing with any company is to decide what do you want to be when you grow up. For AMD, my belief is that our specialty is in high performance computing, and high performance computing is really at the center of all of our lives today. I'd like to say that AMD's technology touches billions of people every day, whether you're talking about cloud data centers, the edge of the network, PCs, automotive, or industrial. You probably can touch some of AMD's technology. But from a technology company, the most important thing is you have to have a long-term strategy and roadmap. Our focus was really on three things: first is high performance computing; second is to have really excellent partnerships, so TSMC is at the center of our partnership overall; and the third piece is that the semiconductor industry is changing. Many of you have heard that Moore's law is slowing down and changing, and we made a big bet on chiplet technology as a very key area going forward. These are some of the things that were important for AMD. Today, I'm very happy to say that when you look at the top supercomputers in the world, we have five out of the top 10 supercomputers based on AMD's technology. From that standpoint, I really feel that high performance computing is at the center of what's most important going forward.
Okay, very good. Yeah, well, the second question is: where do you see the most exciting frontiers in chip design today? I heard that many students still view chip design as the most exciting thing.
I would say that there's always a movement. Sometimes hardware is more exciting, sometimes software is more exciting. From my standpoint, the most exciting part of design is actually holistic design. You have to really design hardware, software, and systems together, especially when we're talking about AI going forward. You cannot design each piece separately. You actually have to think about what is the overall system application, especially thinking about the algorithms and the applications, and then you can think about how you want to design your chips.
Okay, yeah. Next question, I will ask a different thing, not the question here. Okay, please ask very interesting questions. Well, as you know, DeepSeek, okay, we know DeepSeek. They made an announcement in January, as I can remember. It looks like they have some impact on the AI industry. What's your comment or opinion about DeepSeek? How do they impact AI?
The one thing I can say about the AI industry is that every week there is new technology and innovation coming out. I think DeepSeek was one of those areas. When it came out in January, I think there was some surprise. When you look at what they've done, it's actually pretty clever technology in terms of really gathering all of the different innovations using the mixture of expert models and so on. But the main thing that we saw with DeepSeek is that when you can find a way to make AI more accessible and actually cheaper, you actually will have more usage and application of AI. So the main thing that DeepSeek has done is it really has accelerated the application of AI. Of course, that's not a market statement, that's a technology statement.
Okay, yeah. I'm quite curious: after this DeepSeek, well, the United States still wants to invest about 500 billion for AI infrastructure. Those kinds of investments, will they still want to do that after DeepSeek?
From everything that we see, the key point about AI is that there is no one-size-fits-all for the computing ecosystem. You have the very large frontier models where the few people who can invest at that level, hundreds of billions of dollars, are going to go as fast as possible because they want to build the best model. But you also find that that is just one application of AI. You also see there will be many different medium-sized models, smaller models, and you will see that people will use different algorithms to get there. So my expectation is that there will continue to be very significant investment, including very large investments as well as more affordable investments going forward.
Okay, very good. Yeah, yeah. Next question is: what differentiates AMD's AI strategies from others in the space, and what role do you see for the open ecosystem and partnership in accelerating AI adoption? Actually, I would like to ask how you were going to compete with Nvidia or others. I think this question is not very exciting. I would like to hear how are you going to beat Nvidia in your opinion.
The most important thing I can say about AI is that the market is huge. We expect that the market over the next three or four years will be a 500 billion area. In that type of market, you need all different kinds of computing. There's no one company that has all of the solutions. Actually, my view is that you need every kind of computing: you need CPUs, you need GPUs, you need ASICs, you need FPGAs. In that framework, I think our AI strategy is number one to have a very complete solution. Complete solution includes hardware, software, and systems. And the second piece is to have an open ecosystem. One of the most important things is to foster collaboration and to have an open ecosystem. That way, people can learn from each other. That was also another important part about DeepSeek: it was an open model, and as a result, many people innovated on top of that. So my answer to your question is the main thing to say is the market is huge, and our AI is throughout every aspect of our business, from the largest data centers to even at the PC level. We believe that there will be a lot of AI, so there's a lot of market for all of us.
Okay, yeah. Well, the next question you already answered: the core principle or priority that guides your company through the AI transformation. So I would like to ask you another question: what is your suggestion to Taiwan to compete in the future AI technology? Because we are very good in semiconductor manufacturing and also AI servers or others, but in the future, AI will transform many things or change many things. What is your suggestion to Taiwan on how we are going to compete with the world to maintain the advantage right now?
So I think the Taiwan ecosystem is very special. When you look at the Taiwan ecosystem, starting from the TSMC ecosystem, which really has built this incredible manufacturing capability, and then including all of the ODM, OEM, manufacturing, and design ecosystem. I think more investment on the software and system side will come in. From an AMD standpoint, we're very proud of our team here in Taiwan; it's a very important piece of our overall R&D. But the other thing I would say is that in this world today, the ecosystems are very interconnected. From a United States standpoint, we do rely on a lot of the Taiwan manufacturing ecosystem, and trying to build resilience into that ecosystem is important.
Okay, I see. Yeah, yeah. I think the next question is that your family and education were very important to your early life and also your career. Could you talk a little bit about that, and also is that connected to the very difficult period for you in turning AMD into a world-leading company? So do you think President Chen should go from being university president to talk show host?
I hope so. In that case, I can make more money. So I'm very happy to say that I was born in Tainan, and my parents immigrated to the United States. As you can imagine, that was a very big move for them. When we went to the US as a family, the key part that my parents always instilled upon me is that education is so important. For all of you in school, I remember when I was in school at MIT, I very much wanted to finish school to become a business person and then become a semiconductor leader. But it was not easy, especially in those early years. My parents and family have always been very supportive of that, and I'm very lucky in that way.
I see. Yeah, I'm just curious. The problem is, you know, my father, right? That's a problem. Yeah, well, when we see the industrial leaders in the earlier stage, for example, Bill Gates, Zuckerberg, and Steve Jobs, as I can remember, they did not finish university education, but they became leaders. And now the new leaders, for example, Jensen Huang or you, you finished your degree. What's your comment on that? I'm just curious why the previous leaders did not even finish university education, but now the new leaders finish education. What's the difference?
My thinking is that there are many different kinds of leaders. We have a lot of respect for all kinds of leaders, but I truly believe that university education is so helpful. These years that you're in the university is really the best time to learn and to decide what you enjoy the most, so that you decide your next step in your career.
Okay, yeah, I see. Well, next is maybe I should ask you what do you think, President Chen? Well, I think in Taiwan you have to finish the education. In the United States, maybe not. Because in Taiwan, the education system and also job system, you have to finish the degree. And also for the parents, the Chinese-based parents usually ask the kids to finish the degree. Okay, my another question is a little personal. I know you and Jensen are relatives. Do you exchange ideas on running the AI company, or do you run individually? Have you ever exchanged ideas with him?
First of all, I have tremendous respect for Jensen. I think as a founder of Nvidia, he has really built an amazing company. I would say that we run independent companies though. Probably as a semiconductor industry, we do have a semiconductor industry association in the US where all semiconductor companies are there, and together in that forum we would exchange some ideas.
Okay, yeah. Well, anyway. Well, next I would like to ask this for my students. For students who want to make an impact in technology, what advice would you give about taking risk, choosing mentors, or building resilience?
Yeah, so I can say in my experience, I had one piece of probably the best advice that I received from a mentor. They told me, 'Lisa, whatever you do, I would suggest that you pick the hardest problems.' The advice they gave me is to run towards problems. Actually, I think that's the best advice I can give to students, because there are so many interesting things to do in the world. Choosing a project, a problem, or a company where you can make a big impact, you will learn so much and be able to contribute so much. I can say for sure that I learned the most from the biggest mistakes I made. So mistakes are not a bad thing; mistakes are actually a good thing because you can learn from those lessons and be better in the future.
Oh yeah, that's very good advice. Yeah, I remember when I submitted my first paper at NTU, it was rejected. I took a look at the comments and found out I really did not see the experiment data carefully. Then I revised my manuscript and submitted to a better journal, and it was accepted. Well, sometimes we learn from mistakes. So now we open the question to the students. Only the student can raise a hand. Okay, student, yeah, please. Is there a microphone?
Good morning, Dr. Su and President Chen. My name is Jian Ling. I come from the Department of Political Science. In September 2023, there was a book from Columbia University Law School, 'Digital Empires,' describing the three basic models of the internet world: the rights-based model from the EU, the market-based model in the United States, and the state-based model from China. So throughout 2023 to 2025, we're seeing a lot of development in AI. My question is two questions. The first one is regarding the EU model, rights-based with a lot of regulation. Do you think the regulation published will hinder innovation in Europe? And the second question is: what is your evaluation of the United States political environment right now, especially the AI policy between the Biden era and the Trump era? Thank you.
Okay, thank you very much. Yeah, please. Political science is not the easiest question. Technology is actually a little easier. I think I would maybe not talk about the individual countries, but I would talk to you a little bit about my thinking on AI. Because AI is such a new technology. If you think about the last few years, we're all learning together. Countries are learning, companies are learning, individuals are learning. The initial thought process is that AI could be quite dangerous if it's not fully guarded with good guardrails. However, on the other side, you don't want to stop innovation. From my standpoint, I think AI is the type of technology where we have to innovate and go as fast as possible, and understand that there does have to be regulation, but not too much regulation so that you slow down progress. In general, that's my philosophy. I think when you look across the world, people are recognizing that it's a very dynamic time. Every week, sometimes every day, something changes. But generally speaking, I think countries that embrace technology are going to really benefit from it, and many countries around the world are all building their AI policies.
Okay, yeah, thank you. Any more questions? Yeah, please. Yeah, we need a female to ask the questions. Yeah, here, here is the mic. So please.
Oh, hello. Yes, good morning, Dr. Su and President Chen. Thank you for your very marvelous speech. One of my relatives says that Taiwanese people are most likely to achieve in the field of semiconductors and artificial intelligence. So my questions are two. First one is: how can students who are not from the school of engineering play a role in this? For example, I myself am a freshman from the Department of Languages and Literatures. And the second one is: now that Taiwanese Americans are among the most highly educated ethnic groups with the highest income in the States, how do you think that will affect the status of Taiwan in the future? Thank you very much.
Yeah, please. Sure. So on your first question, I think everybody should be educated in AI. It doesn't matter which department you're in. When you say you're in the literature and linguistics department, I think AI can also be very beneficial to understand how to apply AI in your field. Most of what we see today is that people are still learning very much what you can use AI for in each field. So we talk a lot about the development of AI, but there's also a whole field on the application of AI going forward. So I would certainly encourage all students to do that. And then your second question about Taiwanese Americans... can you say again?
Taiwanese Americans are among the most highly educated ethnic groups with the highest income in the States. How do you think that will affect the status of Taiwan in the future?
Oh, I'm not so sure that I can say that. But I will say that I think Taiwanese Americans and Taiwanese in general are extremely hardworking and really thinking about how to contribute to their fields and their professions. I think that's something that we can be very proud of. Although some people say that I'm more American than Taiwanese, I'd like to think I'm Taiwanese American.
