The Times Opinion editor, Kathleen Kingsbury, spoke with Peter R. Orszag, who leads the global financial firm Lazard, in front of an audience at the Economic Club of New York on Thursday. Their conversation centered on a recent paper Mr. Orszag coauthored, “The New Supply-Side Era: The O-Ring Meets Geopolitics,” which argues that geopolitically driven supply shocks, rather than demand fluctuations, best explain economic volatility.
Mr. Orszag’s background includes serving as a director of the Congressional Budget Office and in the Obama White House as director of the Office of Management and Budget.
During Thursday’s talk, he said the debt-financed artificial intelligence boom brought potentially serious economic risk and predicted that the Federal Reserve chairman, Kevin Warsh, will defy expectations by acting independently of the White House. Mr. Orszag signaled that more rate increases are likely, warned of a sharp jump in food prices next summer and said TSMC’s Arizona chip plant barely dents U.S. exposure to a Taiwan disruption.
What follows is a transcript of their conversation edited for length and clarity.
Kathleen Kingsbury: Peter, you’ve compared today’s global economy to the Challenger disaster. The space shuttle was a brilliantly engineered system undone by one cheap part, the O-ring. Describing the part’s role in the disaster, Richard Feynman’s point wasn’t really about the O-ring — it was that NASA’s managers had talked themselves into a belief that it leaked a little last time and we were fine, so it’s probably fine. Where in the economy — in markets, in companies, in Washington — do you see that same reasoning right now?
Peter Orszag: For decades, we had expanding globalization, and effectively infinitely elastic supply. If you had the equivalent of an O-ring, if you had a component that failed, or you had a problem sourcing it, easy just to substitute it and get another one. Just wait for the next container ship, and you’re all set.
That has started to change because of the fracturing of the global economy. Globalization metrics have not declined, but they’ve kind of gone sideways, and there is this fracturing. We’ve built supply chains that have these multiple single sources of failure; we’re living in a global economy that has a wide series of O-rings, not just one, built into many production processes.
The key point is that we’re living in a new supply-side era, in which if you look empirically at the variation in both economic activity and business activity, it’s disproportionately now coming from the supply side, not the demand side. But this is not your grandfather’s supply side. It’s not about marginal tax rates and savings rates. It’s about often geopolitically motivated things that happen in the global economy that then have really significant consequences for the economy, for business and for inflation.
And you could say, “Well, that’s interesting, but so what?” To ground the discussion in a practical process, this trend is at the heart of what the Fed and other central banks have to account for today, which was not the case 20 or 30 years ago. At the beginning of this year, Adam Posen, who runs the Peterson Institute, and I suggested that inflation in the United States would not come down; it would stay high, stay elevated, and we were almost laughed at because that was not the conventional wisdom. The overwhelming majority of forecasts said that inflation would come down. The point here is not to claim credit for the prediction, but to point out why we said what we said. And we said that if you want to know what’s happening today, you have to look at the supply shocks that were happening a year or two ago, and you could clearly see that the full impact of the tariff shock had not played through, that there was the potential for additional geopolitical risk and so on. Our prediction was more based on a methodological change or a philosophical change than just a one-time thing. Now, we all have to be armchair geopolitical analysts in order to evaluate what’s happening in the business.
Kingsbury: The paper argues that these supply shocks actually behave differently — slow to build, slow to unwind, with long lags through margins and inventories. But you also note the economic damage from any single shock — Hormuz, the chip shortage — has actually been smaller than their 1970s equivalents. Which is it? Are we more fragile, or has the system actually gotten better at absorbing hits, just more often? And why?
Orszag: The impact of any individual shock, I think, is smaller today than in the past. Look at the energy shock coming out of the conflict in the Middle East. The U.S. economy, not only from a net production perspective, but also from the perspective of energy use per unit of G.D.P., is less exposed to energy price shocks today than it was in the 1970s. So when something happens, the impact on the economy is smaller today.
But even though the impact of any given shock is smaller, we are subject to an ongoing series of these shocks. The thought that, oh, we just need to get past this one thing and then we’re good, is very likely to be wrong. The U.S. economy has survived the incoming shocks that have been happening for two reasons: a massive degree of fiscal support and the A.I. buildout. The fiscal deficit is around 6 percent of G.D.P., which is providing a substantial cushion of support to the economy. Artificial intelligence is generating activity in the economy, but that may have its own risks associated with it.
Kingsbury: You’ve called the U.S. economy “a levered bet on A.I.” — infrastructure mostly financed through debt, not equity. If that bet is wrong, however, isn’t that a classic demand-side financial fragility story, cutting against the thesis that demand shocks have been tamed and supply shocks are now the dominant risk?
Orszag: I want to make it clear that we’re saying the supply side is the more important. It doesn’t mean there are no demand-side effects. But let me talk for a second about the levered bet on A.I.
I think it’s really important for us to distinguish the technology from the business model. I am a huge fan of the technology. I am a massive user myself, and we can go through all the different ways in which I think it’s an amazing time to be alive because of this technology.
But separate that from the business model. History is replete with episodes in which debt-financed processes run into trouble. And I think the risk that we face at this moment in time has less to do with the technology, more to do with the fact that the investment levels are massive.
A recent Brookings paper estimated the investment between now and 2032 at roughly 3½ percent of G.D.P. And just to contextualize that, the buildout of highways was more like one. Railways, railroads might have been two. Electricity infrastructure was about a half a percent of G.D.P. So this is a massive, massive bet, and to your point, it is increasingly debt-financed.
And here’s, I think, the business model concern: If it is debt-financed and token spend doesn’t increase as rapidly as the underwriting behind the debt, you can have severe financial distress open up at one of these firms. That is a real significant concern. Token spend is going to come under pressure not just from open-source models, but also from local and edge compute, which doesn’t draw on the cloud architecture and tokens in the same way. If it were equity finance, less-than-expected token spending doesn’t matter. But when it’s debt finance, it really does. And I think this is a pretty significant concern that we should be focused on because I think the odds that the debt financing plays out as expected are certainly far from 100 percent.
Kingsbury: You compare what’s happening in today’s A.I. economy to the fiber-optic cable glut of the 1990s. Can you tell us a little bit about that example?
Orszag: Again, separate technology from the business model, because there are lots of things where there’s a significant amount of investment, and the people who have been making that investment don’t reap the full or even most of the benefits of it. But then there are long-term benefits to the economy as a whole, and I think that’s what happened with the dark fiber period in the telecom buildup, where you had all these fiber-optic cables going unused.
The telecom buildup, at one percent of GDP, was a fraction of the projected A.I. spend, which is estimated to cost more than $10 trillion from 2025 to 2032. It’s a dramatically different situation, but the point is that there can be long-term benefits from the investments that we’re making from a technology perspective, but it doesn’t mean that the firms that are making the investment will benefit or even survive. That distinction is relevant to any new technology financed by debt.
Kingsbury: So I think it’s natural then to turn to the question of Taiwan and TSMC’s $265 billion Arizona buildout. Taiwan now produces over 90 percent of the world’s most advanced chips, and modeling of a blockade scenario puts U.S. G.D.P. down 3 percent in Year 1. Does the Arizona build de-risk anything?
Orszag: Barely.
I think it’s healthy for the U.S. to be building out domestic facilities and capability, but even in the best-case scenario, a very small share of overall chip production — and an even smaller share of the frontier chips — will be produced in the United States. When you’re facing a problem, it’s good to start creating workarounds. But this is not an easy or short-term fix, and it’s not a panacea. So there still will be a significant amount of exposure to the production that happens on Taiwan itself, and from that perspective, Taiwan is effectively the O-ring Island at this point.
Chips are such a key component. In the paper, we go through an example of relatively inexpensive chips during the pandemic; auto manufacturers thought that they wouldn’t need as many of them. The chip production went elsewhere. When the auto manufacturers realized that demand was not declining as much as they thought and wanted to ramp up production, they had a problem because they couldn’t get back to the front of the queue for chips. The inability to access these relatively cheap chips caused auto manufacturing lines to shut down. So the point about the O-ring is not actually a direct comparison to the space shuttle. It’s instead that, looking at something from a share-of-cost perspective is much different than a consequence-of-failure perspective. We have lots of components from that era of ever-expanding supply that are relatively cheap and, from an accounting or cost perspective, look like they can’t be that important, but whose absence or failure causes the whole thing to shut down.
Kingsbury: For those of you who are interested in today’s realities of that, my colleague Farah Stockman wrote an excellent piece about a single screw that has slowed down production of U.S. ammunition in recent years.
We’ve been talking about hardware, but China has bet large on the idea of cheap, open-weight models over frontier ones. If reasoning becomes a near-free commodity, that would be a pretty massive supply shock in the way that you’re defining it. But it also would mean that the U.S. would lose dominance in something it’s currently paying trillions of dollars to build. What outcome do you think is most likely in the next three years?
Orszag: One of the reasons it’s worth focusing on token-spend trajectory is the pressure from open source. It doesn’t necessarily mean that the market share goes to those open-source models. But it could mean that the pricing of frontier or U.S. closed models comes down to maintain some market share, with the origin of that competitive pressure being, at least in part, the open-source models.
The Chinese are playing for foundational control of key parts of the global economy. Since the end of World War II, the U.S. benefited dramatically from controlling technological infrastructure, controlling the financial stack, controlling the tech stack. It gave us an ability to exert a lot of influence on what happened in the rest of the world. And I think if you look at what’s happening in the digital financial stack, if you look at what’s happening in A.I., if you look at what’s happening in lots of parts of the physical world, including in renewables and so on, the Chinese are manufacturing at scale, whether it’s pure manufacturing or virtual, the infrastructure for the next several decades. Near-term profitability is not necessarily the only objective, especially from the Chinese side.
Kingsbury: And that doesn’t even include the rogue, unexpected choke points in the geopolitical system that have become normal at this point. In writing about your London School of Economics speech in Bloomberg, John Authers raised Goodhart’s Law, the idea that a choke point stops working the moment it’s weaponized because everyone exposed to it then has a reason to route around it. You address Goodhart’s Law in your paper. Let’s dig into that. Rare-earth elements feel like the obvious example with the one-year truce that was just extended, the initial Nexperia standoff was resolved in just a few weeks. If blocked choke points keep resolving that fast, is that a new permanent era, or is it just a bad couple of years?
Orszag: First, just to clarify, Goodhart’s Law is from monetary policy, and it basically says as soon as you start targeting some metric from a policy perspective, it immediately distorts that metric and its useful life diminishes. The analogy here is: As soon as you start or even threaten to exert pressure on one of these single sources of failure, you’re creating an incentive for people to have a workaround. We’re seeing this play out on the Strait, where there’s lots of discussion of not only the East-West pipeline, but building new pipelines, finding new ways of working around the Strait itself, et cetera.
So the question then becomes: How easy is that to do? How cautious should you be in exerting any pressure on the choke point? Rare earths, the chips, and what have you, a temporary peace is not the same thing as the vulnerability having been addressed. We are still extremely vulnerable to a cutoff of rare-earth supply from the Chinese, despite the fact that there’s a bit of a détente. And I think the way to actually interpret the stability in the U.S.-Chinese relationship right now, which is something that we should welcome, is that both sides are playing for time. The U.S. is playing for time to build out more rare-earth capability, especially on the processing side. The rare-earth materials themselves are more easily obtainable than the building of the facilities to process them. And the Chinese are playing for time to build more parts of the chip production ecosystem. Both sides have an incentive to play nice for a while, while there’s this, effectively, mutual assured destruction.
Kingsbury: Your paper is pretty candid that the reorganizing of supply chains around the geopolitical blocks as you’re describing will cost trillions of dollars. Who pays that bill? Is it the consumers, the shareholders, the taxpayers, and has anyone really been honest with the public about that?
Orszag: This is the inevitable consequence of the ever-expanding globalization period ending. Globalization worked really well from a cost perspective. Didn’t work so well from a resilience perspective. It will be more expensive to create more resilience amid this fracturing of the global economy. How much more expensive depends on the sector. And we don’t necessarily need that resilience in every area. And so the extra cost may not be that substantial depending on the sector, but it will be more expensive. And that’s inevitable. It’s very rare to be able to have a more resilient structure that doesn’t also have some additional expense associated with it. There will be a cost to moving away from that ever-expanding globalization era.
Kingsbury: What you’re actually talking about in the paper is a shift from a rules-based economic system to discretionary state capitalism. You raise the question of golden shares, and government equity stakes in companies like Intel — firm-specific deals replacing the predictable regulation that we’re used to. Can you talk a little bit about whether you think that’s good for American capitalism ultimately?
Orszag: First, a little backdrop, and then I’ll answer the question directly. Over the past 20 years, the spread in corporate performance has exploded. I think a lot of people are used to the inequality chart. If you rank companies by one common metric of performance, which is return on invested capital, it’s the same chart. The median’s basically flat, and the 10th percentile is actually down a little bit, so the spread has widened dramatically, and so I think the question is: As we enter this new supply-side era, is that likely to continue?
And we think the answer is yes. If you’re at the frontier, if you’re the top firm, you want to make sure that your entire supply chain is very high quality because the cost to you from an O-ring moment, from having one component that’s not quite as good, is actually larger than it is for the laggard firm, because you’re otherwise at the frontier and one thing messing up really collapses. Your excellence depends on every part being excellent.
A second thing that’s happening, at the same time, is: Because these choke points or single sources of failure are more geopolitical in nature, it’s changing the interaction between business and government, and this is where you get to the discretionary state capitalism point. That can happen in different ways. It can happen in the sense of: Here are the rules of the road. We’re going to define them, and there is a different relationship between business and government. Or it can be: We’re trying new things. Come see me if you want to. We’ll reach some special deals.
We are doing a lot of that, and I think what I would say is the evidence suggests that over time, that practice tends not to produce economic outcomes or business outcomes that are as beneficial as a rules-based system, because it creates a lot of unwarranted variation across firms that doesn’t have to do with their underlying performance. That creates an inefficiency, and then it does also tend to almost inevitably lead to either the appearance or reality of corruption because the payoff to being one of those favored firms is so high.
Two years ago the Nobel Prize in economics was actually awarded to a set of three economists who basically made the argument that a rules-based system and that kind of governance lead to better outcomes. The thing is, it’s not immediate. It takes time for those effects to play through.
Kingsbury: Lazard advises government and companies on exactly these kinds of choke points and geopolitical risks, what you call “contextual alpha.” Does your own business benefit from the changes in the world that you’re describing?
Orszag: It’s not that one environment is better for us than another environment. Our business is dependent on helping our clients understand the environment they’re operating in. And so Lazard — to the contextual alpha point — has always had a comparative advantage going beyond the narrow business question to a broader frame of, “Here are all the things that could affect whether this is a good decision or not,” that goes back to Felix Rohatyn and throughout Lazard’s history. I would say in today’s world that’s more important than ever because whatever the operating environment is, this operating environment has a larger impact on narrow business questions than the one from 20 or 30 years ago.
Kingsbury: You’ve talked about monetary policy a couple of times here, and you’ve written elsewhere that the A.I. capital expenditure is funded through cash flow and private credit rather than bank lending. But if monetary policy has weaker traction now, as you’re arguing, what actually stops an inflation surprise once it starts, if not the Federal Reserve?
Orszag: I think it does have to be the Fed. It’s a new paradigm in that historically, if you go back to Econ 101, the argument was that if a supply shock like the 1970 oil shock comes along, the central bank shouldn’t respond to it. You often hear some administration officials still making this kind of argument: It’s a one-time price level change. It doesn’t affect ongoing inflation. It just raises the price level, and therefore, since the Fed is focused on ongoing inflation and not one-time changes in the price level, they should just not do anything about it and let it play through.
But if you’re seeing a series of these things and they’re hitting one after another after another, then that sure looks a lot like ongoing inflation. It’s not just one thing.
We’ve had six of these shocks in six years. I think the argument that the Fed or the E.C.B. or any other central bank should just passively say, “Oh, that supply shock. Don’t worry about it,” is much weaker today, given this new environment that we’re in, than it was 20 or 30 years ago.
I think the Fed needs to get pretty good at looking at inflation today, looking at the supply shocks that we were hitting a year or two ago, and making sure you’re playing through the consequences.
Let me give you one specific example. The Fed focuses on core inflation, which excludes energy and food. But those things do seep into underlying inflation. I think it’s very likely that we’re going to see significantly higher food prices next summer in an almost predictable way because of the constraints on fertilizer and other inputs into the agricultural cycle that we already know are happening, and you can almost inevitably see that impact.
And that’s the point: If you want to be evaluating monetary policy and inflation, you have to get really granular at a sectoral level on all these shocks hitting the economy, because they have lags in terms of when they show up in prices, and you’ve got to be playing that game even as a central bank.
Kingsbury: Do you think that the current Fed is acting aggressively enough right now?
Orszag: Inflation has remained elevated in the way that we expected, and I think it was appropriate for the Fed to respond to that by raising rates at the last meeting.
Kingsbury: Do you think they should raise rates again?
Orszag: I think from here, a lot will depend on the degree to which these ongoing supply shocks hit, but that the weight of the evidence or the risk balance tilts toward probably needing higher rates in the future.
Kingsbury: I’m asking these questions because we are also experiencing a Fed that has had more pressure from the White House and legislators than perhaps ever before, so what I’m asking is more of a step-back question. Your prescription for all of these questions is a kind of superabundance, as you describe it — more capacity spread across more suppliers and geographies rather than self-reliance. But you also talk about how difficult it was for Germany to adapt after Russian gas was cut off, and permitting reform is stuck in the U.S. Senate right now until a vote after the midterms. So is superabundance actually achievable? Or is it something that today’s politics simply can’t act on?
Orszag: I may be an outlier here. I think Kevin Warsh is going to surprise us in terms of the degree of independence that he exerts on behalf of the Fed.
When I became head of the Congressional Budget Office, Bob Reischauer, who was one of the icons of former C.B.O. directors, pulled me to the side and said, “Peter, just remember, don’t leave the dance with the people who brought you.” In other words, you’re in the job now. You have a series of responsibilities to fulfill that job, and whoever appointed you is no longer really that relevant because you’re in the job.
And once you’re in the seat of being the Federal Reserve chair, you have these two icons in internal mythology: Paul Volcker and Arthur Burns. Paul Volcker, who withstood political pressure, brought inflation down, and Arthur Burns, who caved to political pressure and is seen as not the best Fed chair in history. So internally, it’s very clear which model you want to be reaching for, and we’ll see how this plays out. On the German experience, I think the lesson actually is, don’t prematurely cut off sources of resilience that could benefit you. Moving away from nuclear under the German government over the last few decades is exactly what you don’t want to do. And if I were then to bring that to the United States, I would say if I had a concern about our energy policy, it’s actually not that we’ve moved more toward, at least relative to the prior administration, fossil fuels. It’s instead being anti- renewables, because I think the right approach is actually the one the Chinese have, which is you need a kind of all-sources approach, and the mix can vary over time. But to be adamantly for or against different sources of energy, almost to the point of exclusion, is a problem, and what’s interesting about that, in turn, is despite policy shifting away from renewables, capacity buildout remains very heavily tilted toward renewables, including in states that people may find surprising, like in Texas and elsewhere. Fundamentally, from a cost and business perspective, it’s still attractive.
Kingsbury: What you’re describing is a long-term vision, but I am wondering, having served in the Obama White House in 2009, what you see as the best supply-side crisis playbook, and who in the government today do you think you would trust to run that playbook?
Orszag: The thing about crises is they’re very hard to know how they play out until you’re in them, and so it depends what we’re talking about. But let’s hypothetically say that we do face some financial distress around the A.I. buildout business model. I think the interesting question is what happens then. There’s a lot of opacity in where the ultimate liabilities rest. Does someone in the federal government decide that the A.I. labs are too big to fail? The politics of that, I think, are really challenging. Do one of the major players try to play J.P. Morgan from the early 20th century and step in with a private-sector solution? I mean, there are lots of different ways you can see that play through, but it also depends on where the shock initially hits and lots of contours of it that are really hard to game out ahead of time.
Kingsbury: That’s fair. If you’re right, and we are now “geopolitical supply-siders,” what is the one assumption that investors are still making that is wrong?
Orszag: That these are one-off events. In fact, I read an analyst report from yesterday afternoon from one of the well-known financial firms in New York that basically made the argument: Don’t worry, these are all one-time effects on inflation, and so it’ll fade. And that may be right with regard to the shocks that were happening a year ago, but our point is there are new shocks. There will be new things that come up because there are so many O-rings or single sources of failure that governments or businesses or others are going to try to exploit. That you shouldn’t be shocked that this keeps happening, and therefore you need to have a worldview in which you can continue to operate despite the fact that these things happen.
The second thing is to not become overly negative, because it’s easy to look at all these risks and go to the Eeyore side of the spectrum and think that everything is dismal. And I’m always reminded, when people get too apocalyptic, of the saying that anyone who doesn’t believe in miracles is not a realist, because it’s not just the right kind of operating model, but also empirically, the fact of the matter is, things have worked out despite the fact that at any point in time in the past, you can look at all these risks and do the hair-on-fire thing. So I think we just need to also be cognizant of those risks, operate within them, but have some underlying sense of optimism, because it’s easy to get stultified otherwise.
The Times is committed to publishing a diversity of letters to the editor. We’d like to hear what you think about this or any of our articles. Here are some tips. And here’s our email: [email protected].
Follow the New York Times Opinion section on Facebook, Instagram, TikTok, Bluesky, WhatsApp and Threads.
Peter R. Orszag is the chief executive and chairman of Lazard. He is a former director of the Office of Management and Budget and the Congressional Budget Office.
The post Why Your Grocery Bill Will Get Worse appeared first on New York Times.
