684. He Helped Clean Up the Last Crash. Does He See Another One Coming?

US Economy Sectors

0:00The U.S. economy is made up of 20 different sectors, as the government calls them. There is the agriculture sector, manufacturing and construction, health care, transportation and energy. One of these 20 sectors is fundamentally different from the rest. Why? Because it intersects with every other sector with an intensity and at a scale that we have never seen in human history.

Gary Gensler Interview

0:31I am talking about the finance sector. Listen to Gary Gensler. Finance is a critical piece of American exceptionalism. We're about 4% or so of the world's population. We're 25% of the world's economy, but we're 50% of the world's capital markets. It's easy to argue, and some people have, that we are living through the greatest period of wealth creation in history. The financial markets that support this growth are exceedingly complex, and few people understand them as well as Gary Gensler.

Gensler's Career

1:06He spent the first 18 years of his career at the elite investment bank Goldman Sachs. He first got into government work in 1999 when his Goldman elder, Bob Rubin, recruited Gensler to the Treasury Department. After the 2008 financial meltdown, Gensler was asked to help clean up the markets that drove the crash. Barack Obama installed Gensler as chair of the CFTC, the Commodity Futures Trading Commission. And later, during the Biden administration, Gensler chaired the SEC, the Securities and Exchange Commission.

1:38Now he is a professor at MIT with a specialty in financial booms and busts.

Financial Booms and Busts

1:44The debate really is, when do you have these booms that lead to just recessions, and when do you have these booms that lead to real washouts? I wanted to speak with Gensler about the AI boom in particular. That's the big financial risk in the economy right now. And why is AI such a big risk? We have a parlay bet right now. It's that the model companies like OpenAI and Anthropic and so forth, and the hyperscalers like Microsoft and Google, that their capital expenditures will lead to enough revenues.

2:19The second part of the parlay bet is that it will also lead to sufficient productivity gains in the economy in the near term. I'm not talking about in 10 and 20 years. In the near term, enough productivity gains to make up for what is going to happen. So what is going to happen? Today on Freakonomics Radio, we get Gary Gensler to help us sort that out. This is Freakonomics Radio, the podcast that explores the hidden side of everything, with your host, Stephen Dubner.

Gensler's Background

3:02All right, let's start with Gary Gensler's birth. October of 1957, the same month that the Soviet Union put Sputnik in the air. And the same year, 1957, that you start with artificial intelligence. Gensler is off by a year on what people consider the birth of modern AI. That was 1956, but that's not his point.

3:35Few people really think AI is as old as I am or that I'm that young. And now Gensler is working at one of the epicenters of the AI revolution. I'm not a card-carrying PhD, but MIT saw fit to take this old math guy and make me a professor of the practice. And I guess you don't view me as a competitor, but Simon Johnson and I started this podcast, Power and Consequences. Oh, I know. I've been listening to it. I like it a great deal. Well, thank you. I mean, we just try to do explainers and we're broadening out and inspired by what we do with students.

4:08I think there's an overlap of what we're trying to do. We use data, narratives. We use historical narratives a bunch. To me, history helps inform thinking about the present and mostly what it means about the future.

Navigating Systems

4:22I was wondering if you could just compare how it is for you to navigate three different systems, academia now, government service for a long time, and before that, private sector Goldman Sachs for a long time. First, a similarity. I've been blessed by working with people of very high caliber professionalism. The Goldman Sachs merger department in its day was like the Green Beret of M&A bankers working on various campaigns and political efforts, even when we lost, but working with really talented people.

4:55And now, my God, the finance group and my co-podcaster, Simon Johnson, just won the Nobel Prize in economics two years ago. The similarity. Very high level of professionalism and gray matter. There's real differences, though, at Goldman Sachs. I learned about trying to discern value. What's the value proposition of a company? How do you sell it? Helped me in politics. Also, what was the value in any transactional situation, but also broader strategy?

5:25The second thing I learned at Goldman Sachs that has helped me throughout is negotiating skills. They were amongst the best skillful, courteous negotiators. I took that into the political world. I worked on a lot of legislative initiatives with Paul Sarbanes, that thing called Sarbanes-Oxley. I find my colleagues at MIT are really collaborative. They're always curious if I reach out to them and say, hey, could I stop in? In academia, though, it's not the same level of teamwork that I learned at Goldman Sachs.

5:58In politics, you have to always find consensus. There's a hierarchy that messaging helps drive the political consensus, and the policy analysis is part of it, but it's kind of third. Whereas in academia, it's in reverse. The analysis seems to come first, and I'm constantly saying, well, what about the political economy? What about can you get the consensus to move that forward? Importantly, what's your message that you can boil down to six seconds?

6:29In politics, I'm usually surprised and frustrated by the lack of empirical attention paid to even significant things like the way that our government looks at the financial markets. To me, it feels as though the political temperature drives too much decision-making versus empirical thinking driving decision-making. Can you just talk about that for a moment? The beauty of the political world is the American public gets to decide.

6:59If you can't describe something to the American public with a narrative arc, with a clean, why does it matter to me at my kitchen table, then it's hard to break through. An example is our unsustainable debt right now. The United States has gotten very comfortable and relaxed in this idea that every year we run federal budget deficits running around 6% of our gross domestic product. When you add up all of that debt and deficits, we're now at about 100% of our gross domestic product, to put it in dollars, $31 trillion.

7:36And that's going to keep growing. That's something we haven't had the political consensus, the political will to address. And I think we lost something our first 200 or so years, the capital markets and the political markets were aligned where we couldn't borrow too much. And then something shifted about 40, 50 years ago. The broad bond markets of the world said, well, no, we'll keep buying these dollar assets and keep basically supporting the U.S. government twin deficits.

8:11We have trade deficits and we have the federal government deficits. That would be an example where the analysis says this is unsustainable. Where does it tip over? That's the debate. But the political consensus just isn't there.

Financial Markets Questions

8:27We asked some of our former guests on this show for questions for you. This is from Jessica Riedel, who's at Brookings now, used to be at Manhattan Institute, so both sides of the divide. She says, do you think that financial markets can absorb $200 trillion in projected budget deficit borrowing over the next 30 years under the current policy baseline without interest rates significantly rising? I would just add, I don't know if interest rates significantly rising is the biggest threat or, as you noted, people are still buying our debt, but that could stop too.

8:59The answer to Jessica would be no. I look at something in the capital markets that spread between two-year borrowing, treasury, and 10-year borrowing. It's called the yield curve. And something happened over the last 20 or so years that narrowed, meaning the cost of borrowing for 10 years is only 40 to 50 basis points or a half a percent greater than for two-year debt on average. For some time, it was even less. And when I grew up at Goldman Sachs, it usually was in the 80 to 100 basis points.

9:34You'd say, well, Gensler, what's 30 or 50 basis points between and amongst friends? But the central banks around the globe plus China's excess savings pulled that down, and it allowed the United States to lean in. I don't think we'll be allowed to lean in and borrow from the world that $200 trillion without the yield curve steepening, interest rates, real interest rates adjusted for inflation moving up.

10:08At some point, there'll be a reckoning. The Democrats and Republicans have done a wonderful job of accusing each other over the past couple decades of being the ones to inflate the debt and deficit. But they both pretty much do it. There are a lot of members of Congress who really see this as a significant threat and would like to address it and absolutely are hamstrung by their parties. If they raise a public word about it, politically, they're toast. Do you see any solution to that stalemate?

10:38It's really hard. And I think the challenge is really the American public itself. We're basically borrowing from the future and we're borrowing from foreigners to spend in the present. If you look at the federal budget in aggregate, it takes in about 17% of our economy, gross domestic product, in revenues, that's the taxes, and the tariffs. And we spend about 23% of the economy, six percentage points, mismatch. Then you look at that 23% of the economy spending at the federal level and you say, well, what can we adjust?

11:14The bulk of it, well over a majority of it, is what in Washington is called entitlements, what we all know is Social Security and Medicare and Medicaid. Only about 4% of our overall economy is the discretionary spending, half of which is the Defense Department and half the non-Defense Department. And so you say, well, can we take this on? Can we fill the deficits that are running at $1.7 trillion a year right now?

11:46Can we fill that with just adjusting the non-defense discretionary? And the answer is no. The arithmetic of it all really runs into the political side. I was hoping you'd have some magic solution though. We just lost a remarkable public servant, Alan Greenspan. We can debate the pluses and minuses of his career, but one thing he did in the 1980s, he was asked by then-President Ronald Reagan to chair a commission on Social Security reform.

12:19Somehow there was a political consensus at that moment to make some reforms on Social Security, which effectively cut benefits for you and me. They said, all right, retirement age isn't going to be fully at 65. It's going to creep up. And I think now it's closer to 68 if you want full benefits. That political consensus doesn't exist today. Some people look to the 1990s and say, what led the federal budget surpluses at that point in time?

12:50And did it have some relation to productivity gains that were coming in the economy with computerization and the internet and so forth? The optimist would say, could that happen here? I think it's far more challenging because we've had some very significant public policy changes since then on the tax side that we only bring in about 17% of our economy into federal budget revenues. And going into those 1990s, we were closer to 20% of our economy.

13:24The Bush tax cuts, the Trump tax cuts, both in his first term and his second term, really have changed the whole fiscal picture.

SEC and CFTC Experience

13:36You chaired both the SEC and CFTC. I want to exploit your experience and wisdom in those two areas to talk about the economy now. Well, let's start with the CFTC. You chaired that from 2009 to 2014. The big headline that I'm aware of is Dodd-Frank, the swaps oversight provisions, central clearing, real-time trade reporting, bringing transparency to a market that hadn't been transparent. But if you could just walk us through what you saw as the hits and then either things you weren't able to accomplish or things that didn't work out the way you'd hoped.

14:14Let me say something about both, Cherry. The Commodity Futures Trading Commission and the Securities and Exchange Commission. And we in the United States are one of two large nations that have several market regulators. Japan followed us in the 1940s and did something similar. But one that oversees the securities markets. Securities are things that you buy and sell that issuers, companies raise money with. And then something around the derivatives markets, financial contracts that derive their value from an underlying.

14:45And in the case of the Commodity Futures Trading Commission, initially, they were financial contracts deriving their value from agricultural things. Corn, wheat, later oil, later interest rates. In both jobs, I am proud of and leaned into a field that seems a little geeky to many people, but it was the market structure or more technically micro market structure. I think you can create great public good by the rules of the game.

15:15Is a market fair? Does a market have access? How transparent is it? And how much do you promote integrity? Basically, getting rid of fraud and manipulation. Those are the key things of market structure. And if you have really good markets, it benefits somebody just filling up their car with a tank of gas, because behind that, somebody might be hedging the risk of the oil price going up or down due to war in the Strait of Hormuz or just due to production cycles.

15:49To answer your question of the Commodity Futures Trading Commission, it was a remarkable time to serve. There it was, the 08 financial crisis, 10 million people either losing their jobs, losing their homes, and many, many more people than that having financial uncertainty at the time. President Obama decided to put together a team and said, we not only have to stabilize the economy, but we also have to put in some reforms to make such failures less likely.

16:21Never going to get rid of them completely, but less likely. We did work with the rest of the Obama administration, with Tim Geithner, with Mary Shapiro, with congressional colleagues, Barney Frank, who we lost earlier this year, Chris Dodd, Blanche Lincoln, Colin Peterson. These were the relevant chairs of the committees. And we put together provisions to lower the risks and create greater competition in these derivatives markets and what you call swaps markets.

16:51I'm proud to say that not only did we get a law passed, but then we were asked to implement various rules. We did 67 different rules. I think 85% of them bipartisan, meaning Republican and Democrats voting together. Almost two-thirds were unanimous. And remarkably, 15 or so years later, they haven't been overturned. The big problem in the middle of that crisis, the credit default swap market, the interest swap markets are functioning at lower cost.

17:26The interest swap market, a lot lower spreads, they're called. They're institutional markets. Many of your listeners would go, why does Stephen have this guy on talking about interest rate markets? But it matters to your mortgage. It matters to your employer. It matters to your auto loan, your student loan, the interest rate markets. I'm very proud of that. I'm proud of one other thing in the enforcement area. We found a bunch of cheats. I don't know what else to call it. We found a bunch of people at big banks that were fraudulently setting the interest rates, and they would say on a daily basis, this is what we will borrow and lend in the interest rate market called the London Interbank Offer Rate.

18:10Now, why did that matter? Many people's mortgages, many people's loans were set in relation to that market. That was called floating rates. Every day, 16, I think it was, banks put in a rate, and we found that a number of them were just lying and others were colluding, and I think we helped clean that market up. What you're talking about, some people may remember that being called the LIBOR scandal, yes? That is the acronym LIBOR stands for London Interbank Offer Rate.

18:42When you were talking about how derivatives went from being on corn futures to derivatives on interest rates, the big picture of the last maybe 30 or 40 years, to me as a layperson, has been the amount of, let's call it, financial engineering in our economy. Can you just take a step back and describe that development pros and cons generally, how much of the economy itself but also the downstream effects are influenced by really a relatively small handful of people at a relatively small handful of firms who have found new ways to make money that are vastly different from old ways to make money?

19:22If I could tease apart two pieces of it, one is financial engineering, or you might say financial innovation or even technology, and the other is market concentration. Finance, since antiquity, has been coming up with new technologies. In fact, long ago, we humans came up with this thing called money. That's a technology we invented. It's a good one. Yeah, it's a really good one. Very valuable. We went on to do a bunch of stuff on math, and I'm not talking about high-end math like AI math.

19:58I'm talking about like double-entry bookkeeping. In the 1980s, a firm named Solomon Brothers invented the first interest rate swap, and others invented how you can take loans and sell them in the market called securitizations in the mortgage market and then in the asset-backed securities markets. And now we have, of course, credit default swaps, and more recently, I'm guessing we'll talk about prediction markets as well. Well, one former public servant who I had the honor to work with closely, Paul Volcker, said at one point in time after the financial crisis that the only technology he thought that really benefited the public was the automatic tell-it machine, ATMs.

20:39I thought Paul was a little too harsh. We should also say that in the U.S., especially, corporate governance is a big deal in making markets function the way we want to. I think a gold standard in the United States, we are the big market for now. There's a lot of reasons, but part of it is because we set up institutions, the Federal Reserve in 1913, the SEC in the 1930s, the Commodity Futures Trading Commission, on and on.

21:10We set up these institutions, and they've, on the main, function pretty well, focused on public goods, what's good for the market, access, transparency, fairness, competition, and the like. Let's talk about concentration. We're doing an episode right now on the price of beef in the U.S., which is really high. And if you look at the concentration in the packers right now, there's only like four major meat packers. But if you go back 130, 150 years, it was the same thing, a different four or five.

21:44And so that seems to me, as a layperson, to be a feature of our capitalism. Maybe it's a good feature. Maybe it's a bad feature. I don't know. But I'd love you to talk about the kind of ongoing concentration of markets and in which markets maybe now you see that concentration as problematic. It's a feature of finance since antiquity. The Medicis had their day. They did. They did. They did. J.P. Morgan, the man, not the company, had his day and his son in the late 19th century and early 20th century.

22:19I think it's a feature of finance. And then you might go, well, why? It has to do with the economics of networks. Just as Google is a network for search in the United States or Beidou as the search network in China, just as Amazon has that big network effect. There is a powerful set of economics around the centrality of a handful of either commercial banks, those that collect deposits and make loans, or the market makers themselves that they get in the center of the market.

22:57One of the key things they get an advantage of when you're in the middle of the market is you get more and better information. There's what economists call an information asymmetry. I have more information than you do. I'm glad to trade with you every day, Stephen, if I have more information than you. So you're called the less informed individual. Sometimes I would hear on Wall Street that the broad public were like the slow deer and the Goldman Sachs or Morgan Stanley were the fast deer.

23:30Well, now it's changed. Maybe now it's Citadel or it's Jane Street or jump trading firms that are at the center of this capital market. But one other thing happens in finance, and it's a feature, is the borrowing of money and then trying to make money on borrowing money. That's called leverage. Banks do it, and when they get into trouble, people want their money back, and that's called a run on the bank. Market concentration and leverage and technological waves are all features of finance.

24:02When I was born in the 1950s, and you said what was finance's total aggregate part of the U.S. economy, it was maybe 3%. When I went to Wall Street in 1979, it was maybe 5%. And now it's 8%. You'd say, do we have a better economy? For sure, we have a bigger economy. But I'm saying, do we have a better economy? Yeah, what's your answer there?

24:32We have a less equal economy. It's a far higher bit of inequality, wealth and income inequality. It leads to polarization. And there's a lot of political scientists who study the relationship of inequality and polarization. I think finance is a critical piece of the American exceptionalism. I think finance is also the thing that might be right now leading to this AI boom slash bubble that we're living through right now.

AI Boom and Bubbles

25:02Coming up after the break, does every boom turn into a bubble? I'm Stephen Dubner speaking with Gary Gensler. And this is Freakonomics Radio. I spoke with Gary Gensler on July 10th. The markets have been a bit choppy since then, but the larger trends are still holding true.

25:36By whatever index or measure you want to look at, U.S. stock market valuation is at an all-time high relative to our GDP. I think roughly two and a half times GDP. It's hovering right around 235% of our economy. It's called the Warren Buffett index. He just simply takes the value of the stock market and divides by our economy. That's an all-time high depending on the day. You can look at price earnings ratios in the rearview mirror where we're about 30 times.

26:11You can look out in front of us. The predicted price earnings ratio is not quite the highest. We have been a little higher at times, or there's a famous economist named Schiller that does a Schiller index, which takes an average of 10 years. We're pretty much at the high there as well. Regardless of all that, what we're having right now in our stock market, everybody would agree, is this investment boom in artificial intelligence. Think data centers, think chips, not just computational chips like the graphic processing units of NVIDIA, but think memory, Micron, Samsung, this SK Hynix out of Korea.

26:52All of that spending has gone up nearly five-fold in three years, from about $140-ish billion to $750 billion. To scale it in terms of our economy, just this year in 2026, that $750 billion is about 2.5% of our gross domestic product. It's estimated next year to go to a little over a trillion. All of a sudden, it'll be 3% of our gross domestic product. That's more than we've done on nearly any of our general-purpose technologies.

27:27I'm curious to know the degree to which that concerns you, and for whom especially. For instance, Citadel has data showing that the bottom half of U.S. households now own more than $600 billion in equities and mutual funds, which is an all-time high. And that is, to a lot of people, really good news, because one complaint for a long time has been that the stock markets were mostly an opportunity for middle and upper income people. But when new investors get into a market, whether it's the stock markets or crypto or whatever it may be, that is often one of the preconditions for a bubble.

28:07Talk to me about your assessment of full absolute risk of the markets and for whom. It's a risk for the entire economy. Right now, our economy is driven, by and large, by two things, both related to AI. One is the actual capital expenditure. When you move from the numbers I mentioned, $140 billion to $750 billion, that's $600 billion of more spending in just three years for capital expenditures on these data centers and the like.

28:39Secondly, the stock market itself creates a wealth effect, generally for the upper income Americans. I like the figure that Citadel puts out there, but I'd note, in aggregate, the U.S. stock market is hovering around $80 trillion. So if the bottom 50% of households hold $600 billion, it's less than 1% with all respect. So when someone like Citadel puts out a number like that, is it marketing, essentially?

29:09As we say in politics, it's messaging. You're going to find the political consensus, and then you figure out the analysis. Look, I would say this. The economy is at risk because if the capital spending just plateaus, that takes off growth. If this capital spending goes to a trillion next year and then declines just to a modest $500 billion, that's a shrinking construction, electricity provisions, all the data chips.

29:42All of that shrinkage starts to go the other way. And so right now, we have this ebullient valuation in the stock market. We have it pushing price pressure up from every laptop you want to buy, any iPhone you want to buy. Even your automobile has in it memory chips. There's only three big memory companies, one of whom just tapped the U.S. market, SK Hynix. And it's pushing pricing up. Pick your favorite general purpose technology.

30:14Stephen, there's a question back to you. For 200 years, you could start with canals in the 1830s or the internet just 20-some years ago. I have mine. I'm ready to go. All right. All right. Electricity. Ah, good one. I think it's wildly overlooked generally. Electrification, it's a good and wild story. And it's one of those things that we totally take for granted now. Yeah, it was 1880 when Thomas Edison created Edison Electric and built that first electric plant here in Manhattan, where I'm recording.

30:44And what happened to the stock market? And did we have a recession in the 1890s? I would guess yes. Yeah. And what happened in the 1920s when we really had big utility and all the utility companies were starting to electrify rural America as well? I'll say yes again. And I'll say yes to every single one. There's booms and busts. Right, right. I'm glad you picked electricity because in electricity's case, we weren't spending 3% of gross domestic product on the capital spend. Railroads, a lot of people pick because it spent so much.

31:17After the Civil War, we were just building those railroads and government policy had made a difference to giving away land for building the railroads. That big boom, we were peaking at 6% or 7% of our gross domestic product spending on railroads. And then the 1870s, we just washed out the economy. Sometimes we peak at 2%, 2.5% to 3% of gross domestic product like we did in the late 1990s on the internet. I think a lot of it has to do with leverage.

31:48How much borrowing is in the system? There's a book I'm guessing you know, Ken Rogoff and Carmen Reinhardt, just after the global financial crisis where you were heavily involved. On the cleanup crew. The cleanup crew, right. And their book was called This Time is Different, Eight Centuries of Financial Folly. They were talking about that financial crash. Let's use that term now for AI because there are, as with any big new technology that's getting a lot of investment, there are people who claim this time is different. This technology is fundamentally different.

32:20The investment around it is different. The consequences, if or when parts of it don't go well, are also different. I'm guessing you don't feel that way about AI, but I'd love to know. Look, I'm an identical twin. I would say Rob and I are different, but we're also a lot similar. So let's just give you context. What's similar is that we have a set of technologies that come along every 20 to 40 years, the last being the internet, maybe before that you would say the computer and general electronics and

32:54so forth, before that automobiles and the highways and the great highway system, all the way back to canals in the 1830s, what you usually have, history tells us, is this investment phase, big capital expenditure phase. We're having that right now. It's in the data centers and the chips and the memory and the like. And revenues do not match it. Right now we have, give or take, $750 billion of spend, and we might have native revenues, if

33:27we're being generous this year, $150 billion, maybe $200 billion. So we know something. It's not in equilibrium right now. And we knew that when the internet was building all those cables in the late 1990s as well. It's never been the case with a big new technology that it is in equilibrium only because a new technology needs an awful lot of startup investment, or am I wrong there? You're correct, but it's also the nature of financial markets, and it's a nature of human

33:59incentives. Think about the chief executive officers in any one of these eras. They said, oh, my God, they're building a canal. I got to build a canal. They're building a railroad. I got to build a railroad. And now Mark Zuckerberg can't help himself, right? I mean, he spent $80 plus billion on the metaverse, those goggles. And so Google rightly is saying, if I don't do this spend, I could lose my ad revenues to OpenAI. So there's a fear of missing out, FOMO, but there's also defensiveness.

34:32I've got to protect my high profit margins I have currently. And so you get an overinvestment that comes from the actual incentive systems right in the moment. Financial incentives and FOMO incentives. And you get this enthusiasm, and the capital markets support it over and over again. Decade after decade, we support these booms. And then at some point in time, it plateaus. And all of a sudden, it's a little bit like that cartoon.

35:04You're watching the character run, and all of a sudden, they ran off the cliff, and their feet are still moving, and there's not the revenues to support where they are. The difference that a lot of people point to is, two, one that I think is right, and the second one I worry about. One is they say, listen, there's not as much debt borrowing because a lot of this bill is from these big tech companies, the so-called hyperscalers, the Microsofts, the Googles,

35:35and they're using their own cash flow. They're not using capital. They're already public. They've been public a long time, and they generate a lot of cash, yeah? And they don't have a lot of debt yet, though they're now tapping the debt markets, right? All of them are doing things that are off-balance sheet financing. They're also having a lot of other companies called neocloud companies like CoreWeave and others lease or buy the chips, and then they lease from them. There's a lot of interconnected finance that if we have a crisis, people will look back and

36:07say, where's that Freakonomics podcast with Gensler? He mentioned that. But wait a minute. Sorry to interrupt, but let me ask you this. How do you like that style of ramping up? I'm just trying to gauge your level of concern. I think that we have a stock market that is highly valued by any measure. We have a boom in the capital expenditures that will plateau and maybe even decline in the next few years. So when that comes, that's a reversal that you see will happen.

36:40And for all of these chip manufacturers, construction, and so forth, something has to give. If AI is successful, it's also going to lead to a lot of disruption. It's going to be significant disruption. The S&P 500, a couple hundred of those companies, their valuations kind of need to go down because they've been disrupted by somebody. I'm not predicting which one, but it could be software as a service or other service companies and so forth. So somewhere AI has got to really work out.

37:11And I'm an optimist that it will over time work out, but not in the next several years. It will be much slower, I think, in terms of really changing productivity growth. So let's talk about who loses out in the AI transition, especially when it comes to employment. We saw this with the so-called China shock when U.S. manufacturing jobs were sent overseas. Some predictions say that the net job loss from AI will be much larger, but it strikes me that nobody really knows yet.

37:41AI needs to be thought of. And I say this to these great students at MIT, I say, you've got to look at the task level, not the job level. What are the tasks of the producers of Freakonomics that they can offload to AI? There's still going to be a human asking the questions. I think for a while, Stephen, I think you're not an AI agent, are you? I am not an AI agent just yet. I do find AI helpful in my research, but I'm just using it essentially as a better version

38:12of a search engine or maybe a talented and smart assistant who also makes a lot of mistakes. Correct, correct. You have to be dubious. You have to really challenge it. You can't offload too much because they'll tell you something that's just not correct. But what tasks will be automated and then later subsequently what whole processes will be transformed, like Henry Ford transformed the factory floor and did that whole assembly line. And then extending that metaphor, when you create the car, you add mobility and that leads

38:45to billions of opportunities that didn't exist. So that's the creative destruction argument that the people that I hang out with, the economists like. But when I raise that argument to AI skeptics, younger and older, they see it more in the this time is different way. There's no way that that's going to happen. So what do you say to that skepticism? I feel it. I understand it. There's been skepticism about a lot of different technologies for centuries. There's that old story of the Luddites in the United Kingdom about weaving and millwork.

39:18This is more challenging in some ways because to the extent that you concentrate on a winner-take-all or winner-take-most model, if the scaling goes that way, then a lot of the profits and revenues go to that one or two main central models. Then if companies, law firms, consulting firms, software firms are offloading and automating tasks,

39:49then the worry is, well, does everybody else get paid lower for what they're doing? There's going to be a lot of disruption in the 2030s. I think there'll be a lot of political angst and changes of coalitions around this throughout the 30s and maybe the 40s. I was hoping you were going somewhere positive with this, but you're saying that what we've got now in terms of uncertainty, let's say political partisanship, let's say angst, as you put it, you're predicting at least 10, maybe 20 more years.

40:22I think that we're in a transitional period of time, and we've seen this at other times. In the industrialization of the late 19th century, it was a little bit about electrification. It was a lot about the railroad, a lot about industrialization, moving from the farms to the cities, but that was a period of time we had the progressive era. It was one of the most dynamic periods of time that we actually broadened the franchise to vote to women by 1920. We did some heinous things on race, though. We went backwards, and we even stalled the Jim Crow era.

40:55But the progressive era was one that we did antitrust laws. We set up the Federal Reserve and the Federal Trade Commission. I think the 2030s and 2040s will be a challenging political time. I think that the American public is going to rise to the challenge and say, we need to change some of the imbalances that are in society. And I hope you asked me about China.

41:21Coming up after the break, what about China?

China and AI

41:24I'm Stephen Dubner in conversation with Gary Gensler, and this is Freakonomics Radio. We will be right back.

41:31Gary Gensler is former chair of both the Securities and Exchange Commission and the Commodity Futures Trading Commission. Before that, he was a partner at Goldman Sachs, and now he is a professor at MIT. We're talking about how the AI boom may play out, and that depends on who wins the AI race. China and the U.S. are in the lead right now.

42:04China and the U.S. have different approaches. We're spending really significant. We're spending probably seven times as much as China this year on building those data centers. And yet, their models are only maybe four to nine months behind us. There's a lot of price pressure in the U.S. Our companies in the U.S. are saying, well, if you want to use OpenAI or Anthropic, we have to charge you. And at some point in time, we have to pay our bondholders and our stockholders. China might be four or nine months behind, but it's good enough.

42:38Like, when you do your searches for Freakonomics, I don't know if you're looking at the budget for how much it's costing you, but if you thought, I can do almost as good a search for half the price, you might say, I'll use a DeepSeq model from China or a Moonshot model or Alibaba-Quinn model. We're in an interesting time where we may have put a lot of cards on this thing as a nation, but we'll have the Maseratis and the Ferraris of AI, and they're going to be building the Volkswagens, and the Volkswagens will sell a lot.

43:16Well, just think about it. In China, they have a pretty darn good electronic vehicle. They do. It's taken over most of the world except for the U.S., yes? Yes, BYD. And so, is that where we're going to be on this AI front in two or four or five years, where we get some commoditization of good enough AI for most tasks? China's also ahead on industrial robotics. So, I think that's going to play out significantly as well.

43:51Have you ever heard of an exercise called a premortem, Gary? This was invented by a guy named Gary Klein, who's a psychologist. He's got a great first name.

44:02I agree. So, the premortem asks, let's say you work at a firm and you're launching a new product. You're the CEO and you say, I think we're doing great. I want to make sure it succeeds, but I also want to take out an insurance policy. Let's get together, everybody who knows really what's going on here and imagine that our product came out and failed. And let's try to identify the causes right now. And if you can successfully do that, then maybe you can amend them and the launch can be successful.

44:32So, if I were to ask you, Gary Gensler, to conduct a premortem, imagining that there's a massive market correction in the U.S., let's just call it 20%, what would be the causes? We've got private credit looking wobbly. It could be an AI collapse. I'll let you fill in the rest. But if you had to face that dark scenario, what would be the causes? And then let's say what would be the consequences, too? My friend, 20% is not the dark consequence.

45:05There's a lot of room between us right now at a stock market that's 236% of gross domestic product, and the average of the last 25 years is maybe 110% or 120%. Okay, you want to take it to 50% then? I'm just saying that what did you want to do a premortem on? And look, the consequential underpinning of the economy and the capital markets right now is this AI investment.

45:35I don't think many people will disagree with that. Where there's debate is how that's going to play out. On your premortem, if the market had a correction, so to speak, and it was off, whether it's your 20% or more, it's likely to be that folks look at this and say, the revenues and the profits aren't there. The companies are still there, but the revenues and the profits, when you think some of these tech companies have profit margins, that's the gross profits divided by their revenues, but profit margins in the 50% to 80% range.

46:16I mean, the memory companies and NVIDIA are in that 70% to 80% range, and a number of them have moved their pricing for their chips, particularly the memory companies, up four and five fold over the last year. This is not like four or 5% inflation. It's like 300%. There's remarkable innovation going on. There's remarkable technological advances going on. Your premortem, it would be, well, gross margins collapse at these companies.

46:50Their pricing pressures right now are going on because there's this huge demand factor. And then you see the chip companies, the data centers all rushing to build new factories, new plants, new laboratories. Your premortem would say, well, there's not enough revenues to support all that. There's an adjustment, and there's an adjustment, and how does it spill through to the private credit field? How does it spill through to these neocloud companies?

47:22There would be some breakage. And I go back to Warren Buffett. He says, you really find out who's been swimming with their trunks off when the tide goes out. We'd find out who was swimming with their trunks off, but it would be where is the leverage, which companies not just have to retrench, but actually go into some bankruptcy. There would be some somewhere. I think for the economy writ large, the spending, the consumption would come off a bit. The high net worth individual consumption would come off.

47:55One last thing I think is if AI continues to be successful, and if the anthropics and the open AIs can charge more, the Googles and Metas can charge more, more and more U.S. companies are going to turn to the Chinese models and say, I'm going to go to the good enough model. I'm going to go to the Volkswagen of AI, not the Maserati of AI. Most of our economy does not buy the frontier desk chair for their office.

48:29The automobile, whatever you name it. You live in a place like New York City, you think everybody drives a BMW. But that's not what all the Uber drivers are driving, right? You need the production model of AI mentality. Hundreds of companies in 2026 are already turning to their technology buy, and they're going, look, we've got to lean into AI, but we've also got to manage the costs right now. Let's do what the technologies call orchestration. I don't want vendor lock-in.

49:00I don't want to be dominated by somebody that can charge a lot more. Before AI was all the rage, especially in the markets, crypto was not all the rage, but there was a lot of rage, positive and negative rage, let's call it. And you, when you were running the SEC, were very involved in crypto regulation. You declared the industry the Wild West, and you launched enforcement actions against a bunch of firms, Coinbase, Binance, and others. I'd love you to talk about the work that you did toward crypto regulation during the Biden administration.

49:35And then what, to me, as a layperson, would appear to be a pretty total washout by the Trump administration, overruling and undoing. I've also got a question here from someone who's been on the show before, Brendan Ballew. Brendan wrote a book about private equity a couple years ago, former DOJ. He said, many of Gensler's regulations were enjoined by courts, and that will only become more common as more of the federal judiciary is appointed by President Trump. Even if a Democrat gets elected president in a world where Congress remains dysfunctional and averse to regulating finance and where courts are eager to stop new regulations, what are the institutions that will rein in either private equity or crypto, etc.?

50:18So I want to know what it's like for you to see how crypto has been obviously embraced by the Trump administration and Trump personally benefiting from his own crypto adventures. That's kind of one part of it, whether you want to talk about that as a grift or not. But I'm also curious to hear you talk about what happens when well-constructed regulation gets undone in a heartbeat when there's a new administration. I think there's a mix of questions in there, my friend. I'm sorry, I have a wandering mind. One is we live in a functioning democracy, and I think it's a really positive thing that when the American people speak and they elect a new lead, a president, a new Congress, that policy can shift.

51:02That's the American public's right, and that's a good thing. You also asked about the courts and how does it feel. I think the courts have shifted, and that's a market in itself. I used to say to my colleagues, we have to think about not where the puck is, but where the puck is going, that famous Wayne Gretzky quote. In the courts, the puck was moving, very much so during the Biden administration, really for the last 20 years, but it's shifted a fair amount in terms of constraining what the administrative state and the executive branch can do in terms of using the laws to promote public good.

51:42I think Brendan is a little bit unfair, if I might say this. I think that of the 40 or 45 or so regulations that we did while I was proud to be chair of the SEC, while a handful of them were challenged in court, and we did lose a number down in the Fifth Circuit. It's a handful, and I would not have wished to have lost them, particularly as it relates to transparency and private equity, which I think you mentioned.

52:13The private capital markets, we were trying to get more transparency to those pension funds and endowments that were investing in private alternative investments like private equity. And the Fifth Circuit down in Mississippi said, no, they didn't think that we had had that authority under the Dodd-Frank Act to just provide transparency to the limited partners. There were two or three other cases that unrelated to those crypto fields. The last part that you asked about, look, I say this, all the capital markets, since antiquity, now, and into the future, trade on a mix of fundamentals and sentiment.

52:55But now you get to some markets that feel nearly entirely sentiment and not much fundamentals. Those are usually those markets that are not sustained in valuation. You might think, I was going to go to tulip bulbs. I didn't need to go to tulip bulbs. I could go to meme stocks of some sort. So what is it that supports the valuation of thousands upon thousands of crypto assets? If you're an investor, you really have to think long and hard, well, what supports the valuation?

53:30Maybe there's one or two of these things that persists for some time because there's nearly 8 billion people in the world. And Satoshi Nakamoto, who wrote that seminal paper, the Bitcoin white paper, whomever Satoshi Nakamoto was, came up with an interesting ledger system. That's what it is, a database system called blockchain technology. But I would note, I don't know of any financial firm, I don't know of any crypto firm that uses the blockchain technology to keep their books and records.

54:03It is used to move this asset in a permissionless way around the globe. And in some cases, that creates an economic benefit of avoiding sanctions. It creates an economic benefit to some who want to avoid any money laundering laws and gun running. Its advocates say it also prevents friction and profiteering from the banking system. I assume you're not sympathetic to that argument. I think that it hasn't proved out to be that beneficial.

54:36Are there frictions in the banking system? Yes. All you have to do is look at the stock market and say, what's the valuation of Visa and MasterCard, and why are they trading at somewhere between $400 and $700 billion of market cap each? So there's economic rents. There is market concentration in the U.S. payment system, and the Chinese payment system, the Indian payment system, are cheaper to do commercial payments than in the United States. But we have a very efficient payment system when it comes to paying our mortgages, being paid.

55:11I'm guessing, Stephen, you're not paid in bags of gold, are you? Only in December. The rest of the year, I'm getting regular crypto payments. Oh. No, no, no, no, no. Yes, I'm not paid. I've never been paid in a bag of gold. I'm open to the possibility. I assume that you get paid in digital dollars, bank account dollars. I do. I love them. They spend really well. They're clean and easy, and they go everywhere, yeah. They go everywhere. It's a very efficient system.

55:42They're the type O of currencies. We have seen this debate, and Jamie Dimon has gotten some publicity recently on it, and I think he's right to raise these points. If you have something that is purportedly backed by U.S. dollars, stable coins, if they're well-regulated, maybe will be backed by U.S. dollars. In the case of Tether, I'm not sure. I think close to 20% of its backing is not dollars, but it's Bitcoin and investments and alternative investments and things.

56:13But this debate, what happens if you have a U.S. dollar-backed digital dollar that doesn't have the same rules and doesn't comply with any money laundering laws and the like? Will that destabilize the U.S. banking system? And will it disintermediate the banking system? That's playing out right now, and it's a fascinating political debate, but I think as an economic point of view, I would say if that three or so hundred billion dollars of stable coins grew, as Secretary of the Treasury Scott Besson said, to two trillion, it's going to take something away from the U.S. banking system.

56:54And the person who would be signing that legislation has his own financial stake in crypto. When you've got that kind of behavior coming out of the White House itself, what does that do to public trust? Well, you have challenges in terms of public confidence in their elected leaders, Congress, the executive branch, the president themselves, and the president and elected leaders' families and colleagues. We have serious enough differences on policy. We have serious enough differences on how to take our great nation forward and what we were talking about earlier about how do we do that in a world that has so much inequality and polarization.

57:35And then you layer onto it an additional sense that some individuals, and in this case, the president himself and his family, are taking economic gains out of it. Whether it's insider trading or not, you just think about the disclosures that you referenced earlier about very large gains. As the numbers are publicly reported, $1.4 billion of profit to the president has disclosed on his financial disclosure forms in crypto.

58:08Crypto alone. And we should say there's someone on the other side of those bets. Look, I mean, most of the field doesn't have fundamentals. Let me ask you two questions, though, about insider trading. First of all, I don't understand why elected officials in D.C. and elsewhere who may have access to private information based on their job and who may have the ability to influence market moves, why they are allowed to buy and sell stocks the way they are.

58:40Does that make sense to you? And did you ever try to do anything about that? I think it would be a good thing if Congress moved ahead to build greater confidence in our governance system and democracy and said individual members and their staffs not be able to buy and sell individual stocks. I mean, to me, that sounds like a no-brainer. I would note, Stephen, that it is against the law for a member of Congress or their staff to trade on material non-public information.

59:11That's against the law. It is a challenge. Where does the blurry line go? If you have a member of Congress or their staff who's working on an investigation and they have material non-public information, they know, okay, I can't trade on that. But what if they're just meeting with executives of a company, if they're working on legislation that might influence the profits of that sector or that company? It would just be positive for the American public to say, aha, I have better confidence there.

59:45I think it would be a good thing to have a law that's all three branches, judiciary, executive, and legislative, plus their staffs, not be trading in prediction markets. I saw recently that Goldman Sachs said that their staff can't trade in prediction markets, except for sports, apparently. One argument in favor of prediction markets historically, this comes from the academic side. Robin Hanson, in particular, at George Mason University, has been making this argument for a long time, which is that stock markets typically prohibit insider trading.

1:00:21But prediction markets, if you can set it up in a way that allows and even exploits insider trading, it makes the market more efficient, that it gets better information into the markets for everybody. What do you make of that argument in defense of prediction markets? I have heard it. It's been debated in one form or another for decades. I think that it fails to recognize that there's a big cost to that. Robin's argument and those that make that argument, they are saying the markets will incentivize people to get information out there faster.

1:00:56An insider at Apple or Google will get that information out there because they will profit from it. But what I think is the cost, and a very real serious cost, is trust in the markets. There's a concept in economics around public goods. How do you create a broad public good, in this case, I'm talking about trust in the market, that if I participate in the market, it's a somewhat level playing field. I know that others might spend more money on their research, even you, Stephen, you're running a media company.

1:01:30So you might have more information than I, but you don't have an inside look inside those companies. Those that say, let's financialize information and let's say there's no prohibition on insider trading, they fail to take in that that will have a cost to our overall capital markets and our overall capital markets will have less trust in them. And then everybody trying to raise money in those capital markets will probably get a slightly lower price earnings ratio.

1:02:06Literally, the cost of capital will go up. Now, for individuals, it's also like a raw deal. You mean somebody else is going to have an inside scoop as to whether the president's going to bomb Venezuela tomorrow? That just, to me, undermines capital markets. I recently spoke with an economist who runs an institute that's focused on what he calls bringing the economy to the people. He argues, essentially, that an economy that generates wealth mostly for the upper end of the income distribution just isn't a good economy.

1:02:40Do you feel that the economy is either unbalanced or too strongly weighted against what people used to think of as accomplishing the American dream rising up? Do you think that's in danger? The United States is still one of the most innovative, dynamic economies of the world, but it's shifting. The inequality issue that you just raised is a real challenge. It influences the power dynamics in our political decision-making.

1:03:10And if you keep tipping towards the elite or the people that have more wealth and they can change the rules of the game, then that tips more towards their favor and you get more polarization in society. We've seen this as a nation. We went through the Gilded Age in the late 19th century, and then we had a progressive era. And even Teddy Roosevelt, a Republican, said, aha, I've got to be a bit of a populist. The current president, President Trump, I think, in part, got elected both terms by tapping into some of those concerns of the broad public.

1:03:50Now, I think his policies have not aligned with his rhetoric, but I think that he understands that the public feels the system isn't working for them. If in the future there's need for a big cleanup, and let's say there's maybe a Democrat in the White House, are you up for another round of cleanup duty, or have you had your turn? Oh, God. It's a big, great country of 350 million people. They can find somebody else. I would say this for anybody listening, if you have a chance in your country, your city, your state to serve, it's a remarkable thing to feel that you can do something for your community.

1:04:32It could be at your church or your synagogue or your mosque as well, your local school. I think there's just something remarkable about it. That, again, was Gary Gensler. His podcast with Nobel laureate Simon Johnson is called Power and Consequences. And Gensler just added one more title to his resume, Friend of the Court. In his first ever amicus brief, Gensler urged a court to reject the legal position of the CFTC, the agency he once ran.

1:05:04This will end up debated and discussed amongst nine individuals in a small conference room in Washington, D.C., and that's called the Supreme Court. None of us will be in the room. What is this controversial case? It involves the prediction market Kalshi, which argues that the CFTC has the authority to allow what is essentially sports betting. Gensler says it doesn't. We are working on an episode about prediction markets, so you will hear from Gary Gensler again sometime in the next month or two.

1:05:35We'll also hear from the CEO of Kalshi, Tarek Mansour. Every time there's a new financial instrument, it was like, oh, gambling in the stock market. People used to call grain futures gambling. That is coming up soon on Freakonomics Radio. Until then, take care of yourself. And if you can, someone else, too. Also, I hope you'll check out the new TV talk show we're making. It's called Better in Person, and you can find it on the Freakonomics YouTube channel, also on Apple Podcasts. Freakonomics Radio is produced by Renbud Radio.

1:06:06You can find our entire archive on any podcast app. It's also at Freakonomics.com, where we publish transcripts and show notes. This episode was produced by Tao Jacobs. It was edited by Pete Madden and mixed by Jake Loomis, with help from Jeremy Johnson. The Freakonomics Radio network staff also includes Augusta Chapman, Dalvin Abouaji, Eleanor Osborne, Ellen Frankman, Elsa Hernandez, Gabriel Roth, and Elaria Montenacourt. Our theme song is Mr. Fortune by The Hitchhikers, and our composer is Luis Guerra. As always, thank you for listening.

1:06:40I appreciate me. We'll sit down there one day. Just let me know. You just have to go through my book or that's me. The Freakonomics Radio Network, the hidden side of everything.