Well, good afternoon everyone. I'm delighted to see such a full house and delighted to have the opportunity to kick off what I hope will be an absolutely fabulous discussion. So, we are gathering right now at a moment of extraordinary transformation in artificial intelligence. AI of course is not just reshaping business education, but it's raising important questions about US leadership, economic and national security, and the future of work itself. At the GSB, it is our responsibility to not only help our students to understand this technology, but to understand the broader context in which AI leaders, policymakers, and practitioners are making decisions. This is exactly why we are launching the Stanford Leadership Institute. Now the institute brings together academics, policy makers, practitioners and our students to help develop an understanding of the very context in which leadership takes place. My hope is that today's discussion will surface practical insights, raise important questions, and deepen all of our understanding of how we can responsibly harness the potential of AI while thoughtfully addressing its risks. We are thrilled to be joined today by our speakers. We have Jensen Huang, the founder and CEO of Nvidia, Congressman Ro Khanna, who represents California's 17th district, and General H.R. McMaster, former National Security Adviser and a senior fellow at the Hoover Institution and Freeman Spogli Institute for International Studies, who's going to moderate the discussion. Please join me in welcoming these wonderful speakers.
Thank you so much. Thank you. Sorry, I'd shake your hand except Congressman. You can be in the center. Okay. This is Huang. They came here to see you. I think this might be the hot seat. I think I'll sit over here. I would shake your hand. I put some honey in my tea and now everything's sticky. And my first thought was, what are you, a child? Hey, well it is a great honor to be with both of you gentlemen and to be with all of you. Thanks for coming out for this panel discussion. It seems to me this is kind of like a setup for a joke. You know, a washed-up general, tech innovator and CEO and congressman from Silicon Valley walk into a bar, you know, what happens next. So, hey, why don't we just jump right into it. There's so much to talk about, obviously, and I really want to hear what's on your minds in each of these general areas. But first of all, I think we can all agree that artificial intelligence related capabilities are extremely powerful. They have huge impact across society. They hold great promise. They're affecting the way that we wage warfare and they determine I think whether or not we have a competitive advantage economically or militarily and from a national security perspective. Could I hear just both of your thoughts on how do we, if you agree, which I know you both do, that we should maintain our competitive advantage. How do we do it?
Jonathan, why don't you start? First of all I think it's helpful to take a step back and ask ourselves what is it that we did. We have reinvented computing as we know it. How software is developed, how software is written, what software can do and how software is processed at its most fundamental level. In a lot of ways that's all we've done. Now of course the nature of computing also changed in the sense that the way we did computing in the past is what was called retrieval-based computing. All of the content was pre-recorded. You know, you wrote a story, you designed something or you recorded a video, you recorded a speech and you store it on cloud databases and data centers and based on how you clicked something or whatever the recommendation systems are and whatever the algorithms are, it would present that pre-recorded content to you. The way that computing is done today is called generative. It takes all the context, the prompt, your intentions and it understands because it now understands, it can perceive, it understands, it reasons and it could, you know, do something, write your story, summarize, write software and so the new type of computing is generative. It is therefore seems intelligent. But when you look under the hood and you open the data center and you open up your computer what you see is software running on top of computers. Okay. And so this is a new type of software. It is incredible in that sense but it's no more incredible in that sense. It's not an alien. It didn't come from outer space. You know, it's not things like that. Okay. And so I think that number one now because of the nature of this new computer, everything about the computer industry has changed because it's so capable and it could do such amazing things. Everything from the nature of companies, the position of companies and the nature of data centers went from storage of files to now generation of tokens and I call them factories. You turn electricity into tokens. It's manufacturing something. The classical data centers used to be a file server. Now you have basically token generators. Well, that takes a lot of computers and so now the question is what can it do? Well, I think that it's fairly clear to all of us that the latest rev of AI which allowed us to go from perception to generative to now agentic systems, this next click of AI has proven to be incredibly capable and we expected it to be and it's doing really great things. Now the implications to all the different industries we can go into a little bit later. So the first thing is to understand number one what is the technology from an industrial perspective. What is it? Well, from an industrial perspective because of the way I described computing, AI is essentially a five layer stack of an infrastructure. It's energy on the bottom, chips next, infrastructure like a cloud, AI factory, AI models and then most importantly AI applications and those applications could be enterprise software or consumer software or drug discovery or robotics or manufacturing so on and so forth. These five layers each one of them have industries and markets and lots of different companies. And I think the most important thing to take away is that if United States wants to, and of course we do want to stay in the lead, it is vital that we win in every single one of the five layers. And each one of the layers has its own issues. Each one of layers has its own dynamics. Each one of its layers has different companies in it. But it's completely vital that we enable every one of those layers to succeed. And then finally, the single most important layer to succeed. And if this layer doesn't succeed, the flywheel will never happen. And if the flywheel doesn't happen, the technology will never scale. The industry will never scale. The most important thing is that the application layer is diffused into society, into our industries and that AI is actually being used. Right. If we cause ourselves because of anything that we decided to say, that we decided to do and it caused our country to be so fearful of AI that we resisted it, that we regulated it out of society, we regulated out of industry and we slowed ourselves down, it would be really quite unfortunate that this industrial revolution that we invented, that we started, that we're in the across the board leadership position at, that somehow we didn't take advantage of.
So if I were to sum up what you said is to really work on removing barriers to adoption, to AI adoption and think about how you can accelerate the development of the technology but then the application of the technology. And Congressman Khanna, what do you think are the keys to maintaining our competitive advantage? What are you most concerned about in terms of how we stay ahead?
Well, first of all, let me just say it's an honor to be here. An honor to be here with Jensen. When I first met Jensen, his first question was, 'Congressman, how do you understand Moore's Law?' And gave me a pause and then he proceeded to explain it very simply. And so he is someone who's an economic patriot and I respect even where we disagree, I've found him to be very thoughtful. And General, I've appreciated you putting our country first and thank you to Stanford GSB for hosting us. The first thing in my view that gives America a comparative advantage is that we have people from around the world who want to come here, study here, innovate here, be part of the research universities here. If you look at the AI startups, 60% of them have been founded by immigrants. If you look at the AI researchers, 72% of them did not have an undergrad degree from the United States. They had an undergrad degree from other parts of the world. 38% of them are Chinese nationals who came to the United States. You're making me think of Jensen, you know, the founder of Nvidia. Yeah. And so I believe there's something not just about getting the world's talent, but having the world's talent interact with each other from diverse perspectives that creates magic and innovation. The second thing is our research universities are where we are. We have 14 of the world's top 20 research universities. You have Tsinghua and Peking University. Two in China. We have 14. Now, Nature Magazine says that they have nine out of the 10 top quantity, but we still have the top quality in research. That's not by accident. That's because we've funded research universities. Of course, people remember it was in Stanford that because of NSF funding in 1969, you had the ARPANET internet from UCLA to Stanford. We need to continue to fund our research universities. Third, I would say it's our academic freedom. You know, in our country, you could be, as I know as a politician, you could have any job and your view on war and peace or economics matters as much as a PhD, a congressman or a president. And people are not afraid to say you're wrong. They're not afraid to question authority. They're not afraid to speak out. They're not afraid to challenge convention. That is our comparative advantage. We need to keep that. And finally, which is here at Stanford, the tech transfer program. The idea that universities can collaborate with the private sector. We have a magic formula with the government, universities, private sectors working together. Those in my view are the fundamental principles that have allowed and will allow America to continue to lead whether it's AI or other technology.
Great. Well, to maintain competitive advantage then we're going to remove barriers to application which can be emotional, attitudinal, and you're emphasizing human capital, research funding, academic freedom. I think the power of our free market is part of that. Got to say that here at the home of Milton Friedman here, you know, at the Hoover Institution. But you of course there are certain actions that we have to take beyond even just funding research to maintain our competitive advantage that involve not only really incredible innovators like Jensen and your team but government policy. And so how do you view economic statecraft? What is the role of the tools of economic statecraft? I'm thinking in terms of an issue that you've had to confront directly, export controls, inbound and outbound investment screening, creating the right incentives for investment by maybe countering elements of economic aggression, you know, dumping of certain materials to maintain a grip on critical supply chains for example. But also, you know, deregulation and removing barriers that are really impeding us maybe from maintaining that competitive advantage at all the levels you described from energy all the way through to application. But how do you view economic statecraft, really the role of government and government policy in maintaining this competitive advantage?
Our situation is a little different than most industries. I would say that the computer industry, the computing technology industry is one of America's national treasures. Unlike all of the other industries, this industry leads the world. Between computing technology, of course, I'm at the epicenter of computer technology education. This is one of our nation's national treasures, no doubt. The other one of course is our financial services industry, the backbone of the world's economy. Almost every other industry we need subsidies. Almost every other industry we need protection. These two we don't. These two industries we thrive. We lead at a level that is hard for most people to understand. There's no car company in history that has ever had 90, 95% of the market. Nvidia's position in China was 95%. And so in our particular case, regulation is not about helping us succeed. This industry was very successful. And so now the question is how does regulation affect us? First of all, we want United States of course to continue to be the world leader. We're an American company. We want America to win. The question is how to do so in the nature of technology and this is where we have to think about what is the essence of the technology when we talk about AI what are we really talking about. AI is not a model, that's not what AI is. Computing is not an operating system, you know, it's not exactly what it is. And so it's really important to understand what is the AI industry, how do we continue to nurture through this industry so that it enhances our national security, so it enhances our economic security, so that we have a thriving industry. If that's where policy goes to enhance those pillars, it's really important to then take a step back and understand what is it that we're regulating, in what way do we want to regulate in such a way that we retain and maybe even enhance our global competitive advantage. And so I think the mistakes that people have about thinking about AI is that AI is a thing. It's an industry that's five layers deep and we have interdependencies with many of our adversaries. In order for United States to advance our AI, we need our energy industry to grow. If we want our energy industry to grow, we need China. If we want our infrastructure industry to grow, we need China. And the reason for that is because the supply chain is so deep. We have so much dependency on them on so much of the core industrial technologies that build up our industries. We just have to be thoughtful about all these different things. Think about the big picture and ultimately optimize towards United States of America winning but not a particular industry winning or losing.
Ro, I'd like to ask you the same question, but maybe emphasize what you've been a big proponent of, which is to reduce our dependence on so much of the world's manufacturing on the southeastern coast of China. And you mentioned that we need China to really maintain our competitive advantage, which is kind of by design, I think, by the Chinese Communist Party and their policies to create the dual circulation economy to get an exclusive grip on critical supply chains and materials. You mentioned energy, of course, the reliance on batteries and turbines and so forth. Congressman Khanna, what are your thoughts about what tools of economic statecraft could we apply to maybe, in the short term do what Jensen is saying, which is maintain a competitive advantage in any way we can if it's importing certain components from China that so be it if it has to be that. But is there also something we can do longer term to reduce what I would call the coercive power of the Chinese Communist Party over our economy?
Well, I agree with you. We can't have China be the world's monopoly when it comes to rare earths, when it comes to key starting materials for drugs, when it comes to active pharmaceutical ingredients. I mean, Jensen is right that the tech industry here and the financial industry here has been extraordinary in America's comparative advantage and one of the reasons we're the world's greatest economy. But we made a colossal mistake in this country, hollowing out places where I grew up in Bucks County, Pennsylvania, where I saw steel shut down, where we saw the entire Midwest and so many places lose factory towns, lose industry. This idea that we could just be a financial nation, an innovation nation without having an industrial base was a mistake. It was a mistake for our national security and it was a mistake for our social cohesion and I would argue that we are still facing some of the consequences of that with the type of politics and anger we see in our politics because a lot of people lost their pride. People who had served in the wars, whose grandparents served in the wars, who built America. Suddenly we said, 'Okay, go move.' And if you can't be in finance or tech, well, tough. And they said, 'No, we built this country.' So my view is we need a Marshall Plan, 21st century Marshall Plan for America, new economic patriotism. Now, that means though, not just strategic tariffs. I'm for strategic tariffs for things not being dumped. But if you put tariffs on for example active pharmaceutical ingredients and you don't have any industry here, what do the tariffs do? It just means the price is higher. So we also need policies. An industrial development bank or Raghuram Rajan has this great piece in Foreign Affairs. He was a former MIT professor saying let's have investment in emerging critical technology to help it scale to help build the new industry here. Whether it's rare earths, whether it's critical minerals, key starting materials, robotics, whether it is in places where advanced steel where we want to make sure we have some self-reliance and then I would enlist people like Jensen and others to say help us, business leaders help us re-industrialize Ohio, Pennsylvania, Michigan, parts of this country, but not with the stuff of the past, the things that we're going to need for the future. And how do we create those jobs and industry? And I believe that can be a mission that brings this country together across party, across geography and gets labor, business, technology, government all working in the same direction.
That's right. As Congressman knows, the AI industry's growth is the engine that's enabling the United States to re-industrialize chip manufacturing, computer manufacturing, and building all these AI factories. We are re-industrializing United States. We're creating so many manufacturing jobs and plumbing and construction, electricians and now fine tool outfitters. Their salaries are doubling, tripling. It's fantastic. But we need a thriving economic engine, a really really strong thriving economic engine so that American companies can afford to invest in the United States. We're going to invest half a trillion dollars in setting up manufacturing of chip plants and computer plants here in the United States. It's not possible if we don't have a thriving business. Sure. And so one of the ways is to lean into this and help diversify our reliance of manufacturing and bring it on shore, create a much more balanced economy. Now we can't just be all cars. I mean, our country needs to have a great labor force in the information area, but also in the manufacturing area, in the labor area, in the crafts. And so we have the opportunity now with a thriving industry to do that. And so everything that we can do to keep the industry thriving is really a good thing to do. And maybe what government can do is incentivize the kind of investments you're talking about and obviously try to create the thriving economic environment that generates the capital that allows you to make those investments.
I think, Congressman Khanna, what you're talking about is a transition in the global economy that occurred really largely after China's entry into WTO, in which many Americans benefited from that transition because of cheap goods that were available from China, but many Americans were left behind. You've talked a lot about democratizing AI, Jensen. You've talked a lot about encouraging adoption across the whole economy, which I think is a very similar theme to this idea of democratizing AI. In your recent paper, you talk about not having AI concentrate in the hands of a few billionaires, present company accepted. But getting into everybody. So what I'd like to ask both of you is what are your ideas about how can everybody come along for God's sake. But this gets to your point too about the doom scenario of people losing jobs and so forth. How do you, to use your term, democratize AI? How do you ensure that this massive transition you're talking about and opportunity doesn't leave Americans behind and allows Americans all to benefit from the technology and its impact across so many different sectors?
Well, you know, I start with the premise that if America has been good to you, you need to do good for America. America has been very good to the three of us on this stage. We have been able to in our own ways live the American dream. One of the things that I respect about Jensen is that he has talked about having a sense of social contract, of an obligation to contribute back to the country. And there has to be a recognition of how a lot of folks in East San Jose or where I grew up in Bucks County see the country. They see 19 billionaires, I don't know if Jensen's one of the 19, who have literally $3 trillion, 12.5% of the GDP. It's triple the wealth concentration of the Gilded Age. And 70% of Americans don't have a view of the American dream. We've got massive economic inequality. You know what? They don't trust us. Even though we invented AI, the highest skepticism of AI is in America. And why is that? Why is it that other countries are more trusting of it? Because they don't trust the elite. They don't trust the people in Congress. They don't trust the president. They don't trust the business leaders. They don't trust media. They feel like we have not delivered for them. So, we have an obligation to figure out how we're actually going to get this AI revolution to work for everyone. And I can go into things in detail, but two places in my view that are so important is to have a jobs program, a commitment to jobs. Now, I was at Brown University the other day, and I said, 'How many people are concerned about jobs?' and 80% of the hands went up. When this was an issue that William Julius Wilson was writing about in black inner cities, few people cared in America. When J.D. Vance started writing about it in white working-class areas, more people cared, but not enough. Now you got Brown graduate kids having this issue. So there is an opportunity. Better than Stanford, I'd say. I mean, you know, Stanford won't admit it. But we have an opportunity to have the most patriotic affirmative jobs agenda in this country where we could say with the federal government that for young people coming out, we're going to hire you. You can work to rebuild your community. You can rebuild it. I was just at a park. You can help do that. You can do counseling. You can help in the care economy. You can make local government more effective or you can come to the federal government do a moonshot bill. You could join the army. Or join the army. I mean that's, you know, as you know it's 1% right. But I want to have some sense of giving back even if you don't do that and we work with the business leaders right. I mean, Jensen has worked, Nvidia has been creating these partnerships with HBCUs. You could join something like that where you get these skills and let's figure out how we take this moment of anxiety. Yeah. Where the reality is no one knows what the disruption will be, where the new jobs will be. I mean, Jensen talks about the radiologists having more demand. I would have never guessed that. We need to have humility. We don't know. But we could take this moment to say we are going to have an affirmative jobs agenda that's going to give people a sense of rebuilding their communities in the country, working for America, bringing this country and maybe giving some new national purpose to this country. And that to me would be one constructive response to this fear of AI and technology.
Well, I say to everybody, move to California. Don't leave. It's the highest taxes in the world, but it's okay. And the weather is great. And a great member of Congress, great congressman. First of all, I think the narratives of AI destroying jobs is not going to help America. Yeah. First of all it's just false. Of course with every technology and every single day that goes by, jobs of the past are changed. Could, would you mind, Ro mentioned this. Would you mention the example you've quoted before about radiologists. Do you mind sharing that with the audience? It's a great example.
At the beginning of the AI revolution, one of the smartest and most influential computer scientists and one of the fathers of modern AI said that in 10 years time, the one job you don't want is radiology. And the reason for that is in a decade AI is going to completely revolutionize radiology, it will permeate every aspect of it, will automate radiology and reading scans, a radiologist will be obsolete. This is the one job you shouldn't go after. Well, a decade later, he was completely right. AI has completely permeated through every aspect of radiology. Every single radiology scan is now assisted by AI and the number of scans that are being studied by AI has gone through the roof. He's completely right. The part that was exactly opposite is the number of radiologists increased. And so the question is why is that? It makes absolutely no sense whatsoever that the task that the radiologist does, radiology was completely automated.
And
Why would they need more radiologists? And the reason for that is obviously very clear. Most mature people, when you think through it, your job, the purpose of your job, and the tasks that you do in your job are related but not the same. Using myself as an example, if they were the same, then somebody would observe that what Jensen does really for a living is typing and talking. And typing and talking have both been automated to a superhuman level by AI. And yet I'm busier than ever. I'm busier than ever. And so I think the first thing is to separate those two ideas. Now, what's amazing is this. Then you say, why did I say we actually did harm? Well, telling people who want to go into radiology that the future of radiology is dead caused the number of people who are developing a career in that field to decline. And so now look what happened. We need more radiologists than ever and we don't have enough. The purpose of a radiologist is to help diagnose disease, work with patients, work with doctors, diagnose disease. They're able to admit more patients, study, scan, do more scans, do a better job with healthcare. The hospitals are making more money. They notice that the radiology department is doing incredibly well. They hire more radiologists so they can increase their revenues, take care of more patients. That flywheel is only sustainable if we have radiologists. Software engineers. Somebody said that AI is going to destroy all of the software engineering jobs. Well, as it turns out, we now have Agentic AI inside NVIDIA. It's everywhere. Every single software engineer is using it. And the one thing that you will observe, two things you'll observe. Number one, the software engineers that know how to work with AI are the most popular software engineers. The software engineers that know how to use AI, know how to use agent tech systems, working with agentic systems are the most popular and the most successful. Number two, the software engineers are busier than ever. And the reason for that is because back then they used to have an idea and they would code it. It would take time to code. Now we have an idea, it takes no time to code. Now all of a sudden the company is waiting for you for the next idea. So you're in the critical path all the time. And so what we see is agents are contacting software engineers perpetually in text. What's the answer for my next thing? What's the answer? What do you want to do now? I just fixed that. What's next? Your agents are harassing you, micromanaging you, and you're busier than ever. And yet our company is able to do more. We're doing things faster. We're doing at larger scale. We're thinking about doing things that we never imagined. And so here's the flaw. The fundamental flaw, and it drives me nuts that it's not obvious. The fundamental flaw is that there are people who think that NVIDIA has to write, we have to code, pick our favorite number, a billion lines of code a year. That if we just finish coding 1 billion lines of code, job done. That's the definition of a year. And so if we have AI automate that 1 billion lines of code and instead of having 10,000 people do it, it only takes 1,000 people to do it. All of a sudden 9,000 people are unnecessary. Well, turns out a billion lines of code was all we could do with those many people in the time that we had. I have dreams to write a trillion lines of code. And so the fact that we now have AI assistance help us, we could explore more space, do better work, do things at a greater scale, do things more cost-effectively, do things better. And so the jobs didn't disappear. The task was automated. Now, of course, there are some jobs where the task is exactly the same. And those jobs where the context doesn't matter, you know, the world's always exactly the same, then I think those jobs will be affected. But it is very likely that overall, this is not even, you know, with all technology evolutions it's not even, but overall my sense is, my belief is we're going to create more jobs in the end. There'll be more people working at the end of this industrial revolution than at the beginning of it, just like at the end of the last one at the beginning of this one.
You know, I had a question later on workforce adaptation. I know you've done a lot of thinking about this as well. Do you want to just follow up on that, Ro?
