00:00David, there is something that I've wanted to talk to you about for a long time, and we will get to it, the debate around should the United States or should the United States not export some form of AI chip to China, and we will get to it.
00:11But I'm sure you can appreciate, you know, the AMD and OpenAI agreement is very interesting. And so as a place to start, could I just ask your sort of interpretation of what it signals when you have a US chip maker like AMD after NVIDIA and one of the frontier model shops like OpenAI working so closely together on infrastructure?
00:32Well, and I think the main thing here is that there's an AI boom going on, and the market is highly competitive. We have a number of leading model companies, we have a number of leading chip companies, and they're all competing with each other, and they're all booming.
00:45And what we see here is that this AI boom just keeps going and going and keeps driving businesses to new highs. And this is all a result of President Trump's pro-innovation, pro-export, pro-AI policy.
00:59We saw this help drive US GDP to 3.8% growth rate in the second quarter. So I think this seems, this boom just seems like it's going to keep going and going.
01:09And we just heard, I think it's Jensen Wang was the latest on a podcast, shouting you out, shouting out those that are in the team at the White House at the moment, guiding when it comes to technology, David.
01:17But I'm interested as to whether you as an investor have great comfort in these relationships being built, financial relationships between chip designers and the chip users.
01:27Well, I think it's up to them. You know, I don't really take sides in these deals. We want all of our American AI companies to be successful, and they're all competing with each other and cooperating with each other.
01:40I guess it's called coopetition. And that's a great thing to see. We just want, you know, putting my AI hat on, we just want these markets to be competitive, and we want American companies to be successful.
01:51And that's what's happening right now.
01:53I'm looking, therefore, at the future rollout, the buildout, anything that gives you pause in terms of actually the infrastructure needs here at the moment, David.
02:03Well, anytime you're working in the world of atoms, it's going to be more complicated than when you're working in the world of bits.
02:09And so it's very important for us to scale up the amount of power that's available to AI companies.
02:15And you're seeing that, thanks to President Trump's policies that are pro-energy.
02:19The president, you know, backed the idea of drill, baby drill, going back many years.
02:23I think he was very farsighted in this regard. He understood that energy is a basis for everything.
02:28It's certainly the basis for this AI boom.
02:31And we're in the process of allowing a lot more power generation in the U.S.
02:35President Trump is allowing so-called behind-the-meter power generation.
02:40In other words, these AI companies can stand up their own power generation.
02:43We need to squeeze more out of the grid.
02:45And we're enabling new oil and gas and nuclear, so all of the above.
02:50So I think that energy is the main thing, and President Trump is supporting that.
02:55David, final question on AMD and OpenAI.
02:58Did the parties consult the White House about moving forward with this arrangement and their plan to work together?
03:06No, no. I mean, not with me, and I wouldn't expect them to.
03:09We just don't get involved in the deal-making between private companies.
03:13Again, we're here to create a policy environment that is supportive for all of our AI companies.
03:19I appreciate that.
03:20Okay, so David, I asked you to come on the program because we spoke at the end of last week for a story that was done out of D.C.
03:26on the idea of, well, what is a China hawk, but also within that broad description, should the United States or should the United States not export deprecated AI chips to China?
03:42You've been very generous with your time, but I just invite you again for our Bloomberg Tech audience, explain how you view yourself in this debate and what you think is strategically most important for this country.
03:53Well, first of all, I would say that I consider myself to be a China hawk.
03:59I want the U.S. to win this AI race.
04:02We understand that China is our main competition globally in this AI race, and we want to do everything we can to win.
04:09And President Trump, in his AI speech on July 23rd, laid out some of the tentpoles of that strategy.
04:15We need to be pro-innovation, we need to be pro-infrastructure and pro-energy, and we need to be pro-export so that the American technology stack dominates the world.
04:23So this is all in the service of the United States winning this AI race.
04:27We understand that it's going to have major economic and national security ramifications, and we're in it to win it.
04:34There is a lot of focus on the direct-to-China part, but the other way that some look at it is there is a marketplace outside of America and outside of China, the Middle East being an example.
04:46What's your position on that and whether the United States wants to kind of leave the world open to China to sell its own technology?
04:55Well, there was a view in the previous administration that we shouldn't sell chips to many countries, including the resource-rich Gulf states.
05:04And I think that was a major mistake, because every time you tell a country that they can't buy the American tech stack, what's their reaction going to be?
05:11They're going to turn to China and adopt the Chinese tech stack.
05:15I have a very simple metric for measuring whether we're winning the AI race, which is global market share.
05:22If we look around the world in, say, five years, and we see that the American technology stack has, say, 80% market share, that means that we won.
05:29But if we look around the world in five years and we see that the Chinese technology stack, and I'm talking about Huawei chips and DeepSeq models, for example, has 80% market share, then obviously we lost.
05:40By the way, that's what happened in 5G.
05:42We don't want a repeat of that.
05:43So, again, the strategy here should be for the U.S. to dominate the world and have the greatest market share.
05:49And I think this is pretty obvious to everyone in Silicon Valley, because we understand that the way to win technology races is to have the biggest ecosystem.
05:57If you're a technology platform, you want to have the most developers using your API.
06:01If you're an app store, you want to have the most apps in your app store.
06:05In a similar way, we want as many users on the American technology stack as possible.
06:10And I find it hard to understand why the previous administration would exclude these rich countries from participating on our tech stack.
06:21It certainly didn't help us in the race with China.
06:24If anything, President Trump's policy boxes out China from the Middle East, whereas the previous administration's policy forced these countries into China's arms.
06:34You've reworked, therefore, the diffusion ideals set by the previous administration.
06:39But when you're allowing only less powerful chips, I mean, I think the president even called them obsolete versions of NVIDIA's chips into China, does that mean that ship of innovation in China has already sailed?
06:51Because they need to have the most sophisticated, and they're going to have to build it themselves.
06:56Well, I think when you're talking about what we export to China, that's obviously going to be a very complicated question.
07:01And there's arguments on both sides.
07:04I think there's a pretty strong argument for not selling China our latest and greatest chips.
07:09Because that would be too beneficial for them.
07:12However, if you don't sell them anything, then like you're saying, that will accelerate their desire to be independent of the American stack.
07:21And so I do think there is a compelling argument for selling them a, let's call it deprecated American ship, or it's a less great American ship.
07:29And by the way, this is why the Biden administration approved the H-20 in unlimited quantities for export to China.
07:37Now, when President Trump came in, he said that those H-20 sales have to be licensed, and they're subject to a 15% surcharge.
07:45And nonetheless, all the people who approved the Biden policy started attacking President Trump.
07:49I think this is just a classic case of no one had a problem with it until President Trump agreed to do it.
07:54David, Jensen Wang went on Brad Gerstner's podcast, and there was a reaction to what he said.
08:01You would point out you are an official of the government.
08:04Jensen is a private citizen and CEO.
08:07But at the root of what he was saying was also the idea of talent.
08:10Many talented computer scientists and engineers come from China.
08:13I think that was the point he was making, but the personnel part of this story, where do you stand on that?
08:22Well, it's true that something like half the world's AI researchers are from China.
08:27And so I do think that we have to somehow be open to working with this talent or allowing this talent to use the American chip stack to some degree.
08:35So it's a complicated question always of what you allow China to do.
08:38But I think Jensen was making a point there about, again, the talent pipeline, and we have to be open to it to some degree.
08:46Again, I consider myself to be a China hawk, but I wasn't triggered by what Jensen said because I watched the entire two-hour podcast, not just the 30-second clip.
08:54And I understood what he was trying to say in context.
08:57And it is a nuanced question.
09:00And look, let me just say to all these people who are criticizing Jensen, what have you done to help us win the AI race?
09:05I can't think of a more strategic asset in the AI race than NVIDIA and Jensen himself, who for 30 years have been working on these GPUs.
09:14And Jensen is a source of huge American advantage in this AI race.
09:19So let's show him a little bit of grace here.
Be the first to comment