Help us bring debate to communities and classrooms across the nation.
DonateWe are excited to announce the second debate of The Hopkins Forum, a partnership between Open to Debate and Johns Hopkins University’s Stavros Niarchos Foundation (SNF) Agora Institute.
The AI revolution is underway, and the U.S. and China are racing to the top. At the heart of this competition are semiconductors—especially advanced GPUs that power everything from natural language processing to autonomous weapons. The U.S. is betting that export controls can help check China’s technological ambitions. But will this containment strategy work—or could it inadvertently accelerate China’s drive for self-sufficiency? Those who think chip controls will work argue that restricting China’s access gives the U.S. critical breathing room to advance AI safely, set global norms, and maintain dominance. Those who believe chip controls are inadequate, or could backfire, warn that domestic chipmakers, like Nvidia and Intel, also rely on sales from China. Cutting off access could harm U.S. competitiveness in the long run, especially if other countries don’t fully align with U.S. policy.
As the race for AI supremacy intensifies, we debate the question: Can the U.S. Outpace China in AI Through Chip Controls?
John Donvan
This is Open to Debate. I’m John Donvan. Hi, everybody from Baltimore, for an episode we are pleased to be producing in partnership with the SNF Agora Institute at Johns Hopkins University. That’s where we are, we’re on this beautiful Hopkins campus. And we’re also in the middle of a still-unfolding global and tech-centered situation centered on the growing rivalry between China and the US where one of the front lines in that rivalry, where the stakes appear to be among the highest, is the contest to be first in artificial intelligence.
John Donvan
Not long ago, we saw a wild card thrown into the mix with DeepSeek, the Chinese AI model that they are claiming they produced at a fraction of the cost of ChatGPT and the like. That rattled a lot of people here who are looking for ways to make sure that China doesn’t ever catch up. One idea is to keep China from getting its hands on the most advanced semiconductor chips upon which the currently most advanced AI is built. Indeed, in one of his last acts in office, Joe Biden greatly expanded a regimen of export controls. Now, president Trump is announcing that he’s rescinding some of those controls and replacing them with different rules, also aimed at limiting China’s access.
John Donvan
So this idea of stopping exports, does that make sense? Is there a downside? Is there a better way to stay ahead in the AI race? That’s what we’re gonna be getting at in this debate. Framed around this question, can the US outpace China in AI through chip controls?
John Donvan
Let’s meet our debaters. Answering yes to that question, I first wanna welcome Lindsay Gorman. Lindsay is managing director and senior fellow at the German Marshall Fund’s Technology Program, and venture scientist at Deep Science Ventures, Lindsay Gorman, thanks so much for joining us.
John Donvan
Your partner, Lindsay, is a former member of the House of Representatives, also a former CIA officer. He also served on OpenAI’s board before he left two run for president, the only candidate in that race with an extensive AI platform and is now Chief Strategy Officer for Chaos Industries, Will Hurd, you are also first time new father. Congratulations to that and welcome to Open to Debate.
John Donvan
On the other side of the stage and arguing on the other side of the argument saying that chip controls will not help the US outpace China, I wanna welcome Susan Thornton. Susan was a diplomat for 28 years, including serving as Assistant Secretary of State for East Asian and Pacific Affairs. Now senior fellow at the Yale Law School Paul Tsai China Center, also a former debater with us. So thank you for returning and it’s great to have you back on stage, Susan.
John Donvan
And at the far end of the table, I wanna welcome Paul Triolo, senior Vice President and partner at DGA Albright Stonebridge Group, that’s a global business strategy firm where he focuses on all things China and all things technology. Pleasure to have you join us, Paul. Thanks for joining us on this stage.
John Donvan
So we’re gonna get right to opening statements, and our first speaker is going to be Lindsay Gorman. Lindsay, you are answering yes to the question. The floor is yours.
Lindsay Gorman
Who’s heard of DeepSeek AI? Who has asked it for information about China’s leader Xi Jinping, or about Tiananmen Square? The answer you get is let’s talk about something else. We cannot let an autocratic regime build and define the most consequential technologies of our era.
Lindsay Gorman
The United States has significant hard computing power advantages. That is the ability to produce high-end chips, designed specifically for training AI models. With our allies we control a strategic choke point on compute, a bottleneck without which training these models is impossible to do efficiently. And China knows this. China has long sought to displace the United States as the global technology power. And in fact, early thinking about export controls on semiconductors was in fact informed by Chinese scholars who identified a handful of critical technology choke points that would stop China from achieving its own indigenization goals, i.e., building everything at home. Semiconductors were on that list.
Lindsay Gorman
The US has gambled that this is the decisive decade for AI competition. That whoever takes the lead now will have outsized advantages. Properly implemented controls can have an effect and also have an increasing and compounding effect over time in retarding China’s I- AI advantages and giving the United States a head start.
Lindsay Gorman
Don’t take my word for it. The CEO of DeepSeek himself has said, “Money has never been the problem for us. Bans on shipments of advanced chips are the problem. And they have to consume twice the power to achieve the same results.” Despite headlines about Chinese advances, the reality is that chip controls have been effective. The United States has tools and it should play its cards, now that the stakes are high and the time is now.
John Donvan
Thank you, Lindsay. Uh, now to make an opening statement on the other side of the argument, Paul Triolo. Paul, you are arguing that the US cannot outpace China in the AI, AI race using chip controls. Your chance to make your argument.
Paul Triolo
Thank you. We believe that the export controls are not working and it will not lead to US, uh, dominance in AI. And there are several reasons for this.
Paul Triolo
First of all, the goal of the controls has never really been stated very clearly. What I hear is through… The, the, the goal of the controls is to throw sand in the, in the machine in China, to slow China down, um, from developing AI. Now, we know that, uh, Secretary Gina Raimondo, who oversaw a lot of the controls over the last two years called this effort a “fool’s errand,” quote, unquote, most experts believe that China is only a… Chinese companies are only three months behind US leaders, uh, in developing advanced AI models.
Paul Triolo
The other issue is that, that both the Biden and the Trump administration have changed the goalposts in terms of the, the controls, expanding the controls over the last two and a half years, which has been really confusing to industry. It’s disrupted global supply chains and it’s disrupted critical technology relationships. And this comes with steep costs.
Paul Triolo
The other thing is that proponents of the, of the controls argue that, er, initially argued that they would be effective within a few years. Lately, I’ve heard people talk about a decade of, of that… Before the controls will be effective. That’s, that’s a pretty big difference. So we don’t think that controls will be effective in the future either, and that’s because of several reasons. First of all, there is no agreed upon clear goal and no clear multilateral agreement on how to control these technologies. We have neither of those.
Paul Triolo
Uh, secondly, um, we have no real, uh, uh, uh, airtight enforcement mechanisms to make the controls work. And that’s because the expert controls were not designed to maintain US technology hegemony.
Paul Triolo
And third, and maybe most important, the Chinese view AI as a critical, uh, and essential to their economic development. So they’re very capable. They have lots of STEM engineers, they’re very capable of innovating, and in fact, the controls have forced China to move r- rapidly up the innovation ladder and to develop innovations around both semiconductors and AI technology. Witness DeepSeek and Huawei.
Paul Triolo
And then finally, I think most importantly is that, that the, the controls have really, uh, highlighted two various sensitive geopolitical issues. One is they’ve precluded so far, uh, any real collaboration between the US and China on the critical issue of AI safety. And this is happening just as we’re on the verge of a key inflection point, the development of artificial general intelligence.
Paul Triolo
And finally, the hardware base for AI resides in Taiwan. And the risks around conflict over Taiwan are heightened by the controls and the focus on AI dominance.
John Donvan
Thanks very much. Next up, giving his opening statement back to the center stage is will Hurd. Will, you are arguing that we can outpace China with chip controls. Please tell us why.
Will Hurd
Ladies and gentlemen clap once if you’ve heard this name before, Baidu. Okay. What about DJI? What about TikTok? Okay, everybody’s heard of TikTok. These are three companies that… Chinese companies that operate in the United States of America, freely without problems.
Will Hurd
But their competitors, or their American counterparts, let’s say Twitter, Google, Shield AI are not able to operate in China. If there was a level of reciprocity between our two countries, we wouldn’t be here having this debate about chip controls. I want the US and China, and when I say China, I actually mean the Chinese government. I want us to be frenemies. Let’s work on things we can work together on, compete on things that we compete on. But that means you’re doing it based on the rules that you’ve already accepted, instead you’re gonna agree to. And if you violate those rules, guess what? You can’t trust people.
Will Hurd
I’ve always talked about artificial intelligence, the equivalent of nuclear fission. Nuclear fission controlled gives you nuclear power, unlimitless clean power, nuclear fission uncontrolled, nuclear weapons can kill everybody. I don’t think that’s hyperbole to explain that and, and to use that frame when it comes to, to artificial intelligence.
Will Hurd
The first mover advantage of when we get to AGI, and I know we’re gonna have some debates on what does AGI mean, artificial general intelligence. There’s different names for it. Really powerful AI that’s gonna be hard for a human to control. Let’s, let’s… That’s kind of my working definition. The first mover advantage of that is gonna be so important.
Will Hurd
And lemme be frank, I don’t think American might… We may not get it right, but I know for a fact that Chinese government won’t. ‘Cause we have seen them use technology in order to surveil their population, to try to predict what their population is gonna do, to ultimately control this. And that’s why when it comes to this, this powerful technology, one tool in our toolkit, and it’s just one tool, is controlling chips going to the Chinese government.
John Donvan
And our fourth opening statement comes from Susan Thornton. Susan, again, you are arguing we cannot outpace China with AI, in AI with chip controls. The floor is yours.
Susan Thornton
This is not a new Cold War. The US and China are in a competition, but it’s not an ideological competition. It’s not a military competition. It’s an economic competition. And our two countries are the two most powerful countries in the world by a lot. And we are going to remain interconnected for the foreseeable future. And we just saw an example of the realization of this, uh, with the trade war that was launched on April 2nd and over the weekend was substantially walked back, in a recognition that the two countries are inextricably intertwined.
Susan Thornton
I think the main thing we should be asking ourselves about this question relating to chip controls and how it’s gonna affect the competition in AI is what is the cost benefit of us policy actions? We have to face the reality that China is already building AI. It’s going to be building AI. It’s already within three months catch up to top US companies. Um, you know, out of the top thousand AI scientists out there in the world, a third of them are Chinese. China is one third of the entire global technology market. So it’s clearly a player.
Susan Thornton
So the question is, what will we have to pay for trying to slow them down through the chip controls? And what are we gaining through the chip controls in slowing them down? The costs are obvious. Costs to US companies by giving up one third of the technology market. There are costs to our relations with our allies who don’t wanna join us in these policies of chip controls because they don’t see China as an enemy in the same way that we tend to. We have a lot at stake in our relationship with China, not just on this issue, on many, many, many issues.
Susan Thornton
And that certainly we want… The one thing we need to do is avoid going to war. Uh, it’s been mentioned that Taiwan, the most sensitive issue in US-China relations has now been dragged right into the middle of this AI issue because they’re the place that produces all the cutting-edge chips that we’re trying to control. If we don’t get our hands on, uh, AI controls between the two biggest AI powers, you know, who knows who’s going to be controlling and stopping AI from inflicting major damage on humanity. It’s an area where the US and China need to cooperate.
John Donvan
We’re gonna take a quick break now, and when we come back, we’ll get into more discussion on our question, can the US outpace China in AI through chip controls? I’m John Donvan. This is Open to Debate. We’ll be right back.
John Donvan
Welcome back to Open to Debate. I’m John Donvan and I’m joined by our debaters Will Hurd, Susan Thornton, Lindsay Gorman, Paul Triolo. Before the break we heard opening statements on our question, can the US outpace China in AI through chip controls? Let’s keep going.
John Donvan
I wanna just summarize what I think I’m hearing. The side, uh, that’s arguing, uh, yes to trip chip, chip controls. Will Hurd and Lindsay Gorman are, I think fundamentally talking about, uh, that the US already has an advantage, cannot let go of that advantage. That whoever dominates in this situation dominates in numerous ways that would have long-term consequences. They also make the case that China has demonstrated that it cannot be trusted ’cause it doesn’t abide by the rules. And that the first user advantage is just one that is not to be fritted away because the stakes are just too high.
John Donvan
Team arguing against the proposition, Paul Triolo and Susan Thornton, uh, they’re talking about a broader context. First of all, making the case that, um, we’re not talking about a military rivalry they say. We’re not talking about an ideological rivalry. We’re talking about an economic rivalry, asserting these kinds of controls would have consequences that would be negative for the United States and have to be brought into the calculation. They raised the specter of war.
John Donvan
And fundamentally, they’re just saying that these things don’t work. That China’s already almost there, that they’re only three months behind the US and that these controls would take up to 10 years actually to have an impact. So we’re hearing both different philosophical points of view and different, uh, differences on understanding of what these controls would actually do.
John Donvan
I wanna go first to you, Lindsay, as you had an opening statement in which you presented the situation as very much zero-sum and the stakes being enormously high and your opponent’s making the case that it’s, it’s, it’s not i- ideological and it’s not military, it’s economic, and therefore, you know, a somewhat kind of contained thing, not as broadly dire as you’re suggesting. If you could take a minute to respond to that.
Lindsay Gorman
Yeah, thanks, John. I would definitely say that it is ideological when we have, uh, a government trying to remake the world in its own autocratic image. It’s hard not to think of these things through an ideological light. And I did wanna take up, uh, some of the points that the, the other side made.
John Donvan
Before you do that, let me see what their response is to, to what you had to say. Who would like to respond to, to that point?
Susan Thornton
Yeah, I mean, I think during the Biden administration there was this framing about democracy versus autocracy. I mean, clearly the Chinese has a different governing system than ours, but I don’t think that this AI race necessarily extends to Chinese government imposing their ideology on the United States or anyone else.
Susan Thornton
This was an issue in the Cold War with the Soviets saying that they could have a global sort of, um, you know, communist, um, sort of dominoes would fall. Uh, but that’s not the situation here, I don’t think. And so I think that that’s just a mistaken way of thinking about this issue.
Will Hurd
But Susan, I, I think we gotta, we gotta take what they say about themselves, right? The Chinese government has made it very clear in their own documents, oftentimes in English, that they are trying to use their, um, a, a, a, a mastery in a number of technologies in order to surpass the United States, um, as the global global superpower, right? Now, whether we want to call that ideological, economic, um, they are trying to displace us there.
Will Hurd
And, and, and how are they gonna do that? Um, if these, if these tools are designed with their, um, uh, their belief system entwined in it, and we have to start using that, it will have an impact on us, right? Um, we can… I know this is not a debate about TikTok, I’m sure y’all probably done one before. Um, but-
Will Hurd
But, but the, the use of those tools to try to influence, uh, our society and, and that is, that is, to me, is ideological because they are trying to imprint how they do business on… And, and we know how they’ve adopted technology before.
Paul Triolo
I, I might just throw out a… The…. A, a comment by, um, the historian, Rana Mitter on China, who, who says that, um, he’s a historian of China. He says that the, the, the, the most… The least reliable way to predict where China will be in 20 years has always been to extrapolate in a straight line from where it is now.
Paul Triolo
So when we talk about this idea that China wants to overthrow the, the so-called rules-based international order, which seems… Which, which the other side seems to be implying here, we have to be very careful about, about what that means. China, for example, participates in many parts of that global order now, and has been successful, for example, in the global financial system. So there are many parts of that global order. Some of them China accepts and has benefited from some of them. China i- i- i- i- is uncomfortable with and would like to change and some of them it rejects.
Paul Triolo
Um, but, but, but China’s… There’s the evidence, uh, that, that China wants to overthrow and replace the US, uh, for example, in the international global order, I think is, is, is, is, is misplaced.
John Donvan
Uh, Lindsay, I, I cut you off before when you said you had some points that you wanted to respond to, and I wanted to get back to those points. But as the conductor of the orchestra here, I would like you to choose one of them that you want to rebut and then let your opponents rebut to your rebuttal.
Lindsay Gorman
Um, I think I’d pick up on this critique that the goal is not clear and that, um, mo- quote, most experts think China is three months behind. Um, and specifically in regard to clarify what, what most experts mean by that timeline, um, I think it’s the model itself.
Lindsay Gorman
But no experts think that China is three months away from manufacturing the top-of-the-line semiconductors. And that’s crucial to this debate because we’re asking can these chips slow down, and I think that was a very clearly articulated goal, slow down China’s progress. And what we’ve seen with DeepSeek AI is that in fact it has benefited from advantage… From advances in the US ecosystem, and it has also trained its model on generations of con- of chips before they work controlled.
Susan Thornton
But I would say that if you take it by the measure of producing a semiconductor, I mean, China’s been working… I think the main critique of this would be that what we’ve done is we’ve massively incentivized a very dynamic Chinese innovation ecosystem to come up with all kinds of ways to get around this. And they’re working very hard on indigenous innovation.
Susan Thornton
I mean, CEOs like Bill Gates and Jensen Huang of NVIDIA have both said that what we’re doing is the Chinese will go on without us. We’ve massively incentivized their innovation ecosystem to develop all kinds of technologies. And we see that with DeepSeek, we see that in other areas. It’s not keeping them from catching up on AI, which is the, which is the question that we’re supposed to be discussing today.
Paul Triolo
Lemme just add that on the semiconductor side. Uh, right now Chinese companies have, uh, stockpiles of existing US, uh, GPUs, for example. They have their own indigenous sort of in-house GPUs they’re using. And they have some Huawei GPUs that Huawei is beginning to produce that are pretty advanced.
Paul Triolo
And so for the next two to three generations of model development, they have plenty of, of, of compute, for example, juice. And in the meantime, then, as Susan notes, um, we’ve incentivized the domestic industry to get much better. So companies are collaborating amongst each other. They’re trying to overcome some of these bottlenecks and choke points.
John Donvan
Okay. I wanna, I wanna let the other side respond to that. What, what I’m extracting from that is they’re saying that this… All- what’s already happened is already teaching the Chinese that they have to do it on their own, and they’re starting to do it on their own, and they don’t need us to do it on their own. Will?
Will Hurd
I, I, I, I think that’s some of the flawed thinking that has been going on about how we’ve looked at China for, for, for, for quite some time. Um, with chip control specifically. Again, I say it’s a tool in the toolkit, is not gonna prevent the Chinese government from developing or continue to develop this.
Will Hurd
We know this is on their, their, their their, um, uh, ten-year plan in order to… W- they are making, there are making investments, but saying that this is going to, this is going to increase that timeline. They’re already trying to do it. So the goal is to slow them down and not let them use our technology in order to build upon.
Will Hurd
And we also know that if the Chinese government invades Taiwan and tries to take Taiwan, one of the strategic objectives is to corner 70% of the semi- advanced semiconductor manufacturing, right? So, so saying that’s-
John Donvan
But I think your opponents are saying they’re more likely to do that, you know, they, they raised the specter of war and that continuing to heat up the atmosphere would make that sort of scenario more likely. Am I getting that right?
Susan Thornton
Yes, that’s right. You’re preventing Taiwan from giving the chips to China and creating a, a dynamic there.
Will Hurd
Why give our adversaries? And, and look, I, I think part of the, part of the reason that there’s, there’s debate about this is are they an adversary or not, right? Are they trying to surpass us or not? Like, I think that’s a whole-
John Donvan
Very good question. Do we, do we know the clear answer to that? I just want a yes or no, or not sure.
Susan Thornton
Because that’s the economic competition, right? Shouldn’t doing that if you were Chinese?
Will Hurd
But to, to corner global markets that they can decide the rules of the road. We are already seeing them in, in, in multilateral forums, uh, trying to change rules that benefit them and their, their style of doing business. So yes, it, it… They’re… They are using, um, a per- a economic power to achieve these broader objectives. And whether you want to call that economic or ideological, it is ultimately-
Susan Thornton
Well, China’s growing more powerful. That’s true. And so with that, it gets more influence in the global system, that’s true. And with that, maybe they want to make adjustments to certain global institutions. I mean, at the… This point, the US is withdrawing from all those global institutions, so we’re making it very easy for them to do that. But economic-
Susan Thornton
Economic power is what, you know, most countries aspire to. They wanna grow their economies.
John Donvan
I wanna draw Paul and Lindsay back into the conversation. And Lindsay, what’s another point that you wanted to rebut?
Lindsay Gorman
Yeah, if I could also respond to that, to that quickly. China’s indige- indigenization efforts far predate US export controls. Um, these are longstanding goals. Um, but I did also wanna pick up on this point that Paul made initially on AI safety. I think no one wants these technologies in the hands of terrorists. No one wants, um, AI enabled bioweapons to be developed. Who, who succeeds well write the rules of the future.
Lindsay Gorman
Uh, we don’t wanna compromise that for a shared objective, which should be a shared objective on AI safety, and we can walk and chew gum at the same time. We have a history of arms control negotiations with some of our fiercest adversaries. Um, that doesn’t mean we shouldn’t also try to control technologies from going to them, even as we have diplomacy there.
Paul Triolo
Well, let me just respond to the, um, the idea that China always, um, intended to, um, to go it alone, for example, in semiconductors, this is really not the case. As long as Chinese companies could buy the best equipment from US, Japanese and Dutch makers, for example, they were happy to do that. But once we put restrictions on that, that forced, uh, Chinese companies, indigenous companies to work together to improve the capabilities of those domestic companies who then can compete by the way, against US outside of China.
Will Hurd
I’d have to push back a little bit to my, my friend Paul’s, uh, uh, statement there. Uh, the Chinese have always had about 10% of, of global supply, of advanced chips. Uh, Taiwan has about 70, uh, then the, the mix in South Korea, uh, some in Japan. And so saying that, you know, this was a, a, a nascent industry is not true because, ’cause 10% of the global market in something this advanced is pretty significant. And that predated any of these, any of these, um, uh, uh, uh, kinds of debates.
Will Hurd
But like… But to, to my friend Lindsay’s, Lindsay’s point, uh, we have these kinds of tools in many technologies, um, that has existed for, for a long time. And the reason we have them is to make sure that the people that are using and are using for the intended purpose and not gonna ultimately erode our national security.
Susan Thornton
Yeah, I, I don’t think our side is saying that we shouldn’t have any export controls. And we’ve had export controls in many technologies on China for a long time, and, um, since, um, the ’80s really. And, um, in 2019, we imposed, um, a ban on China importing the highest level of lithography machines from the Netherlands, which effectively prevents them from being able to manufacture advanced chips.
Susan Thornton
So… And we did it through the Wassenaar mechanism, which is a joint multilateral export control. All of our partners agreed. The Dutch agreed. It was a very well thought through and tight control. Doing that makes sense because it’s, it’s bought into by everyone in the world. It’s understood by the Chinese, they don’t like it, but they, they understand why we wouldn’t, you know, give them the latest cutting edge technology.
Susan Thornton
Uh, what we’re arguing against is all these latest controls and all these changing of the goal posts and all of these other things are really disruptive, are making an enemy of China, are not favored by our partners and our major costs with very little benefit.
John Donvan
Could, could you explain for people who don’t know, when you’re referring to lithography machines.
Lindsay Gorman
These, these are manufactured by the Dutch co- company, ASML, the highest in machines that go into the chip manufacturing process.
Lindsay Gorman
They’re necessary for the chips. If China were content to just buy chips made by NVIDIA all the time and had no designs on its own indigenous process, why would it need to buy the DUV and EUV machines from, from the Netherlands?
Lindsay Gorman
And I actually agree with Susan that a multilateral export control regime, an update to Wassenaar maybe that doesn’t have Russia in it, would be a better mechanism for us to multilaterally, especially among democratic allies and partners, implement these controls in an effective way.
Lindsay Gorman
The issue is we don’t have the time, this three month window on the models, a, a, a longer window on the chip manufacturing, these are critical days, weeks, months, and years where the AI playing field is being set and the market dynamics are being set. So we can’t wait a decade to set up the perfect export control regime.
Paul Triolo
Yeah. I just wanna respond to, to Will’s comment and, and I agree that there are some technologies that, that of course have long been controlled for very specific military end use purposes, right? The problem with AI though, is that, for example, in China, the vast majority of this technology is being used in civilian applications. Healthcare, uh, scientific discovery, materials, science with, with no military application.
John Donvan
Alright, so that’s actually take… Goes to my question. I would like to ask you to think through, for people who are not very, very entrenched in this topic, what happens to us if China dominates it?
Will Hurd
Uh, let’s start on the battlefield. Um, use of autonomy in, in the battlefield. We, we’ve seen this play out in, in, um, in Ukraine and how, uh, the asymmetry, asymmetry of warfare, um, has, has been enabled because of the use of AI where you can now have a $20,000 drone that blow up a, a $60,000 radar that’s protecting, that’s protecting a country.
Will Hurd
I can make the argument that what we see in Ukraine is gonna be just a fraction of the kind of activity you would potentially see, um, in the, in the First Island Chain, in the level of sophistication.
Will Hurd
Let’s take, let’s take cybersecurity. Um, we have seen in 2020- uh, 2025, a four-thousand-percent increase, 4,100 and some change, increase of AI created, um, spear-phishing campaigns. Half of, of cyber attacks have been generated, have been generated by AI. How do you make that a military application? That’s difficult because guess what? That same AI is gonna create that email that helps me buy, uh, that table I just bought, um, for, for the renovation of, of my, of my house.
Lindsay Gorman
Yeah, I, I mean, I think those are brilliant points. Um, in terms of the picture, this is about who controls the fundamental means of production of the next century. And in China, we have a coercive power, one that issues cyber attacks, that uses economic leverage when countries vote for a dissident to win a Nobel Peace, Peace Prize, or when they criticize the Chinese government.
Lindsay Gorman
So the world that I’m imagining is one where someone speaks out against something the Chinese government has done, and their access to the new personalized medicine they’re developing goes away, the lights go off. Uh, not to mention a sophisticated surveillance regime that they’re already piloting, um, in, in the Xinjiang province.
John Donvan
Lindsay, why, why… What about… What is it about the AI, weaponization of AI that would be different, say, for the, you know, the 65 years that both the Soviet Union and the United States had a nuclear weapon, neither attacked the other, they were both nuclear superpowers. They both held, held this danger in their hands. You’re sort of… I think you’re suggesting that in AI it’s different, there can only be one dominant power?
Lindsay Gorman
Uh, yes and no. I think the key difference there is that it’s true that AI is a general purpose technology. It’s not just military technology. Uh, yes, the US and the Soviet Union had mutual… Mutually assured destruction and adv- and incen- incentives not to unleash those weapons in a, in a very hotly contested situation. But with AI, this is also an economic argument.
John Donvan
Okay. I’ve let your opponents talk for some time, so you get the last word before questions to respond to what you’re… What the, the, the kind of dire situation that they’re talking about now in detail.
Paul Triolo
Yeah, I think what I’ve heard in talking to people about this, uh, I think, um, the former White House czar was on Ezra Klein podcast, and he was asked, what does the world look like when China gets to AI first, AGI first, again, advanced artificial general intelligence. And he immediately went to the cyber, the cyber issue, that somehow if China had an advantage in AI they would take down US critical infrastructure.
Paul Triolo
And frankly, I, I, I, I just don’t buy that as an argument because I think that, that, that would not likely be the first response of, of the Chinese government. First of all, as I think the moderator noted, the definition of AGI is very complicated. We’re not gonna wake up one morning and know we have AGI, so it’s gonna be a continuum. And so whoever gets to an advantage will have time to d- to figure out what to do with it. But I don’t think the first reaction is gonna be to launch a cyber attack.
Susan Thornton
I think to see China as the biggest danger that’s headed our way, stemming from A- AI and AGI is really misplacing our focus on what we should be doing to make sure that we’re all gonna be able to use this technology, use it for the benefit of humanity, and have it be safe. And China is a huge part of that equation, and we need to be working with them.
John Donvan
We’re gonna take another quick break, and when we come back more on our question, can the US outpace China in AI through chip controls? I’m John Donvan and we’ll be right back.
John Donvan
Welcome back to Open to Debate, I’m John Donvan and I’m joined by our debaters, Susan Thornton, Lindsay Gorman, Will Hurd and Paul Triolo. We’re debating this question, can the US outpace China in AI through chip controls? Now it’s time to take questions from the audience.
John Donvan
Right down in front. If you could stand up and if you would also tell us your name and your… If you’re connected to Hopkins, we would love to hear that.
Coco Yao
So, hi, I’m Coco Yao. I’m a senior student at Hopkins. I’m majoring in biomedical engineering and minoring in CS. I’m here today because I wanna learn how the decisions are gonna shape the future of AI and how that affect global dynamics. So my question is, is the containment gonna be sustainable because the AI is just gonna keep developing and, uh, with the two countries getting increasingly strained, are they gonna, uh, like sit down and then make, like. A checkpoint for, is the AI development, like, safe or, like, like, kind of like a standard-
Will Hurd
Look, look, I wish we could do that. I wish you could have track two conversations, um, on, uh, safety around AI, right? Because here’s the thing that we are learning.
Will Hurd
Uh, you’re learning that the, the, the, the more powerful AI gets, um, the less likely it is to follow human intention, right? And, and for the ability to have the conversation to pull in that third of, of some of the great AI brains in, in the world together to have that conversation be great. But guess what? The Chinese government is not allowing those, those track two kinds of conversations, right? So yeah, I, I wish that was the case, but unfortunately I don’t think it’s gonna happen any time soon.
Paul Triolo
Look, can I, can I also respond, China is very interested in discussions on, on AI. I was in the Paris AI Action Summit in February when China in fact launched its AI Safety Institute. Uh, so China’s very eager to have those discussions.
Paul Triolo
But it’s a complicated issue, ’cause industry is also split. Some of the leading model developers like Anthropic and OpenAI don’t support China’s participation. And so right now, uh, we’re in the middle of a, a big international discussion between these AI summits that are held every year to determine whether China will be allowed to participate.
Sam Carliles
Okay, I- I’m Sam Carliles, uh, Johns Hopkins Institute for Data Intensive Engineering and Science. People have talked about, um, the, the problem of running out of internet, right? And, and sort of the out from that is the conversations that users have with, with the AIs, right?
Sam Carliles
But not everybody’s conversation is as informative as everyone else’s. At some point, assuming your hardware keeps up, um, that becomes what, what the competition is about. When do we get there? Are we already there? And when we’re there, what does that competition look like?
Will Hurd
If you’re asking a data question, right? The, the data that’s going in to, to these systems, and if you have garbage in, you have garbage come out. And is the competition really about who is putting the best data in? Heck yeah. That’s what everybody’s trying to figure out is, is you have to have data, you have to have compute, and you have to have power to inform that compute.
Will Hurd
Um, I also think that the US always has to have… We’re always gonna have less data because we have a little thing called civil liberties. Um, and that, um, our, our, our, our competitors in China are able to take whatever data they ultimately want. So I think they’re always gonna have an advantage, um, when it comes to the data, to the data piece. But there’s no way that you can potentially control that, um, in this export-controlled fashion.
Paul Triolo
I, I, I… I’m not sure that that’s, that’s, that’s really the, the nub of the issue. I think right now, for example, we’re in the age of agentic AI. So if they- they’re trained on a certain set of data, but then they’re going out on the internet and they’re grabbing new data up-to-date data, and they’re doing deep research on this. DeepSeek has… Does this, uh, Manus AI, which is a Chinese company, does this, uh, Claude Sonnet is doing this very well.
Paul Triolo
So we’re really on the cusp of dif- a different type of AI being, being deployed, where the model is doing much more after it’s… Based on its train- its training data, but also using and, and leveraging new data. So that issue, I think, becomes less important.
Paul Triolo
Um, but China, you know, China’s training… There- there’s only a small percentage of data that China’s not using that’s different from the, from the US. So I think the idea that, like, China’s AI is gonna be authoritarian or, you know, is gonna have some, is gonna have some censorship, of course it’s gonna, because that’s abiding by Chinese law. But a basic data sets that, that every… All the major model users are using are pretty similar, at one level, there’s only a small delta.
Susan Thornton
I just wanted to make one more point because we’re talking about data and we’re talking about compute. But the other thing you need to do… Be an AI leader is you need talent. And the US used to be number one by far at g- gathering all of the international talent together in the US and our various AI labs and, uh, honing that tech talent and getting the most kind of, uh, most innovative research out of that. But I have to say, lately we’ve been bleeding talent back to China.
Susan Thornton
Um, I think we still have, um, the most, um, you know, top tier AI researchers from multiple countries, but a lot of them are Chinese and a lot of the Chinese, uh, talent is going back to China now because they see that as a more dynamic ecosystem for research.
Mark O’Connor
Uh, my name is Mark O’Connor, and I am a cyber instructor down in Washington D.C. at Jackson-Reed High School. This week, the Trump administration announced to be rolling back some of the Biden administration’s chip export controls. Uh, with this new strategy, it promised to put forth an inclusive AI strategy while simultaneously keeping sensitive technology away from our adversaries. Uh, however, the details so far have been pretty sparse.
Mark O’Connor
Uh, my question to the panel is this. Uh, what ways would you recommend to the administration to achieve these ends?
Lindsay Gorman
I think we need to recognize that China is trying to make up the difference. Um, there are massive smuggling networks that have emerged. So our controls need to be effective. They need to crack down on some of these smuggling networks. They need to address the realities that some of these tech… That these models, uh, can be used in the cloud.
Lindsay Gorman
Um, so we need to tighten up these controls. We need to pour a lot more money into enforcement of the controls, uh, in a new way. So I don’t think we got it right the first time. And I don’t think with, you know, with nuclear technology, we also, we also reevaluated and refined our controls over the, over the years. These are new controls. We are controlling manufacturing equipment. We’re controlling US persons from helping China to build, uh, its own domestic semiconductor manufacturing chain. Uh, so we need to refine how to use them and how to use them smartly. And I hope that’s what the administration will do.
Paul Triolo
Yeah, I think what you’re referring to is the Trump administration has pulled back the AI diffusion framework, which was a rule passed, laid in the Biden administration and divided the world into tiers, three tiers, you know, allies, adversaries, and everybody else.
Paul Triolo
Uh, the industry didn’t like this and, uh, many countries that were relegated to tier two didn’t like this. So the administration has basically scrapped that formulation in favor of a global licensing regime and bilateral agreements, which we saw with Saudi Arabia and the UAE that was… Were, were negotiated by the Trump administration.
Paul Triolo
‘Cause the industry’s argument is that… The way to maintain US technology dominance in AI is to let other countries build, for example, AI, uh, uh, data infrastructure using AI, uh, hardware that’s… Or- originated from the US. And so that’s the, the new policy the Trump administration is pursuing.
John Donvan
Thank you for your question. We have another question coming down the aisle. I think we can take one more question.
Lulu
Hello, hi, I am Lulu. I’m from both the United States and China, so I can see this from both sides. My question is for the pro side, China has 1.5 billion people and 65% use mobile technology for everything generating an immense amount of data. And as we know, data is the lifeline of AI.
Lulu
If the US cannot match that data advantage or even come close, how can it possibly outpace China and AI, even with chip controls?
Lindsay Gorman
And I’d add, I think we’re gonna need to find a way to do privacy protective machine learning, um, so that we can, we can up our, our data edge. But in the meantime, that data needs to be processed with computing power.
John Donvan
Thank you, debaters, for this round of the debate. And now we move on to a closing statements. And this is where each of the debaters, again, takes the center stage and tries to persuade you that they’re right in arguing yes or no on our central question is, can the US outpace China in AI through chip controls.
John Donvan
And speaking first, um, Lindsey, you just had the last word and you get the first word in this round. The floor is yours as you argue the answer is yes.
Lindsay Gorman
Our opponents in this debate are seized of a defeatism that says America can’t out-compete China or slow its progress on the technology of our time. So let’s not try, let’s not bother, uh, it’s gonna hurt US companies too much.
Lindsay Gorman
Well, our view is that we can do better than that, that America must do better than that .our companies are, are doing well. Uh, NVIDIA’s market capitalization is, is just shy of the GDP of the United Kingdom. And that’s even not selling its top line chips to China. Instead, those chips got sold primarily in the United States. There isn’t a- an issue here with demand. Uh, it’s with supply. So doing better means that we have to throw what we can at this problem now with a smart application of tools.
Lindsay Gorman
And I’d like to take us back several years to the early 2000s when in 2002, Google had just 16% of the search query market. By late 2000s, it was at 90%. And today, here in 2025, it has maintained this lead for decades. Meta in 2012 had just crossed this threshold of 1 billion active users. 13 years to… Later today, it holds 59% of the social media market, only now getting a challenge from, from China’s TikTok.
Lindsay Gorman
There are advantages in market dynamics, in customer bases to coming out of the gate early, to building applications early and to securing an early lead. And that’s the purpose of these controls, to slow down China’s advancement so that US and companies from our, our allied countries and our partners can take the lead. There is an advantage to a head start.
John Donvan
Thank you, Lindsay. Paul, again, you are arguing that we cannot outpace China in the, uh, uh, in AI using chip controls. The floor is yours.
Paul Triolo
Yeah, thanks John. So, um, I think that in the, in the words of a, of a long time China specialist I was listening to last night, we imagine export controls can protect us from China draining our precious bodily fluids. And I think that it’s important to note that highlights the issue that, that these technologies we’re talking about, particularly semiconductor manufacturing, are applied technologies.
Paul Triolo
So the mistaken belief is that in controlling those technologies, we can control what China does. In fact, there are many ways to get to, to, to the different ends. So we mentioned earlier this idea of advanced lithography. So what the controls have done is to force China then to figure out a way to do a- advanced lithography. So right now they have three teams pursuing different paths to develop advanced lithography.
Paul Triolo
And at the end of the day, they’re gonna come up with new innovations for their companies, and those companies then are gonna be able to be more competitive both to domestically and globally. And so at the end of the day, um, they, they… They’ll become s- self-sufficient and also be able to sell those, that equipment to other places that we might not like.
Paul Triolo
So I think it’s really important to note that, um, and, and particularly when it comes to, uh, technology independence, that process that China’s now engaged in will make it even more independent from, from places like Taiwan. And again, coming back to that risk, we’ll increase the risk of conflict around Taiwan. So if we wanna protect our precious bodily fluids and preserve Taiwan status and ensure the safety of advanced AI export controls are in fact working against us at all levels. Thank you.
Will Hurd
25 years ago, um, when I graduated from university, I joined the CIA. My job was recruit spies and steal secrets, best job on the planet. Um, I did it, um, overseas in order to protect our homeland. And, and at that time, if you would’ve talked about, you know, a technology that was made in China, I would’ve said it was a knockoff, it was garbage, it wasn’t very good.
Will Hurd
We talk about made in China now, it’s high quality. But I could also make the argument nine times outta 10 it was made in America first because they stole it from us. They stole it from other places to integrate it into their systems.
Will Hurd
We have to begin with this notion that we are in a conflict. They know it, they say it in their own materials, right? And why would we give an edge to our competitor that is trying to use that, that, that tool against us? It’s not our only tool. I agree, we have to out-innovate and I’m gonna put my money on American innovation, on, on, on, on American stick-to-itiveness, on, on, on our freedoms and openness in order to, to do this. But let’s use all the tools in our toolkits in order to make sure that we win this challenge.
Will Hurd
I was on the board of OpenAI four months before we released ChatGPT. I interviewed Sam Altman and the CTO of Microsoft on stage, and there was a bunch of, of tech luminaries, names you would know. After the interview was over, one of these tech luminaries came up to me and said, “Why’d you let him talk about all this craziness that ain’t never gonna happen?” And I’m like, “What are you talking about? I’ve been using ChatGP-3, uh, GPT-3 for, for, for several months.” Four months later, ChatGPT comes out, end of the week we have a hundred million users. Nobody thought at that time that that was possible.
Will Hurd
We can’t stay in that thinking. This technology is evolving so fast, the next 35 years is gonna make the last 35 years look like we were a bunch of monkeys playing in the dirt with sticks. And at that time, are we gonna say, did we do everything we can to make sure our way of life continued and was maintained? Thank you.
John Donvan
And our final closing comes from Susan Thornton. Susan, again, you are arguing that we cannot outpace China with chip controls. Your chance, one more time, to tell us why.
Susan Thornton
Look, making the AI competition with China a zero-sum game, not only will not work, it is dangerous.
Susan Thornton
We don’t have to give China the cutting edge chips. We did that back in 2019 with restrictions on EUV lithography technology. That was a, a move that was widely supported in the international community. It’s an easy execution, easy implementation. Everyone agrees and China accepts it that they’re not gonna have the most cutting edge technology chips.
Susan Thornton
Now we say, you know, China’s just a little bit behind us in where it is in AI, which is the question that we’re debating today. So it seems like with all of the things we’ve done since then, we haven’t actually slowed China down that much and the costs have been very high. Um, you know, the cost in relations with China, who think that we’re trying to throttle their economy when they have 1.5 billion people, many of whom still live at or just above the poverty line. Um, cost in relations with our allies who think that we have, um, kind of gone way overboard in our obsession about this chip controls in China.
Susan Thornton
In my view, it’s partially a kind of a political impulse in the US driven by fears of China. And I think this is completely overblown and unwarranted. I think we should make sure that we focus on the things that are gonna matter to our children and their children, which is the long-term AI competition, which if not constrained and bounded by international, uh, agreements and by cooperation among countries who agree on what the boundaries are. Uh, it’ll be a very dangerous world.
Susan Thornton
I certainly think that the Chinese believe that this technology needs to be bounded. They would like to be in conversation with the US about this. We’re the two leading AI powers. Everyone else in the world thinks we ought to be talking about this, and I think we should focus on that instead of worrying about changing the chip controls every two weeks, because that won’t work. And it, it’s, it’s gonna cost us a lot. Thank you.
John Donvan
That is a wrap on this debate and I want to thank you and your… Uh, for being such a fantastic audience. Uh, again, I wanna thank Hopkins and the SNF Agora Institute for making this partnership happen. I wanna thank everybody who asked questions, but most of all, I wanna thank these four debaters for doing what they did, which was to demonstrate and embody the spirit that we try to communicate, that people can come here and disagree very, very deeply, but respectfully and civilly. And I’d love to see a handshake.
John Donvan
And a big thank you to you, our audience for tuning into this episode of Open To Debate. As a nonprofit working to combat extreme polarization through civil debate, our work is made possible by listeners like you and all of the Open to Debate philanthropic supporters. With special thanks to the Rosenkranz Foundation. We would also like to thank SNF Agora Institute at Johns Hopkins University for being our partner in producing this debate.
John Donvan
Robert Rosenkranz is our chairman. Our CEO is Clea Conner. Lia Matthow is our chief content officer. Elizabeth Kitzenberg is our chief advancement officer. Andrew Lipson is head of production. Michele Debreceni is our director of marketing. This episode was produced by Alexis Pancrazi and Marlette Sandoval. Editorial and research by Gabriella Mayer. Max Fulton provided production support.
FIND OPEN TO DEBATE ON YOUR LOCAL NPR STATION
A nonpartisan, nonprofit organization, Open to Debate, formerly known as Intelligence Squared U.S. addresses a fundamental problem in America: the extreme polarization of our nation and our politics.
Our mission is to restore critical thinking, facts, reason, and civility to American public discourse.
JOIN THE CONVERSATION