This is an edited transcript of “The Ezra Klein Show.” You can listen to the episode wherever you get your podcasts.
I’ve covered the economy for a long time — including a financial crisis and a pandemic — and I can’t remember a stranger and more chaotic year than this one, where it was so unclear what the story was and how everything would turn out.
Other years, the economy has been either bad or good. But what is the economy right now? From “Liberation Day” to the deals and the pauses and the carveouts, it’s unclear what our tariff policy is.
Then there’s the giant A.I. build-out that is keeping the economy afloat. But is that a bubble? Does it herald massive labor-market disruption? Is it good for us? Is it bad for us?
The economic data has also become completely divorced from how people actually perceive the economy. If you look at consumer sentiment, people feel like the economy is as bad as it was at the depths of previous recessions. And yet the economic data — the job market, wages, inflation — all kind of look OK.
So as 2025 comes to an end, I wanted to do a show wrapping up the strangest year in the economy that I have ever covered.
Tracy Alloway and Joe Weisenthal, who are at Bloomberg and co-host the excellent economic podcast “Odd Lots,” are here to talk it all through with me.
Ezra Klein: Tracy Alloway, Joe Weisenthal, welcome to the show.
Tracy Alloway: Thanks so much for having us.
Joe Weisenthal: Thrilled to be here.
We’re here almost at the end of 2025. Tracy, let’s start with you. How would you describe how the economy is doing right now?
Alloway: That’s actually a really tough one, which probably says something about this moment in economic history, which is that no one really knows anything.
We had these tariffs that came in, and many people were expecting those to have an inflationary effect. We haven’t necessarily seen that, even on things like unemployment. But a lot of people have been concerned about a recession for ages now, and that just hasn’t materialized.
I think a lot of traditional economic thought that should dictate how things develop and unfold isn’t bearing out. So I would describe the economy as unexpectedly chaotic, perhaps. It’s just not acting the way a lot of people thought it would in the current situation.
Is it chaotic, or is it unexpectedly normal?
Alloway: Maybe normal. [Chuckles.]
I kept thinking as I was looking at this data that, if you just showed me the macrodata of the year, if you showed me the jobs numbers month on month, if you showed me gross domestic product and inflation, and I didn’t know any story line, I would say: Hey, pretty normal year in the economy.
Alloway: Yes, that’s fair. There’s a layer of policy chaos built on top of an economy that seems surprisingly resilient to that chaos.
Joe, what’s your state of the economy gloss?
Weisenthal: I want to echo basically the same thing that Tracy said. Look, there is clearly a labor-market deceleration happening — I don’t think anyone would dispute that. Unemployment is, as of November 2025, 4.6 percent.
All that being said, the temptation is to assume that, OK, you have this creaking labor market, and then it snowballs, and then you have a proper recession. But people really have been talking about the imminent recession for three years.
Yes, I remember Joe Biden’s imminent recession.
Weisenthal: There were so many imminent recessions, and I think everyone is just very gun-shy right now about calling anything.
We also have this weird thing, where understanding the economy is murky in the best of times. Then layer on to that fact that, prior to the government shutdown, data collection — response rates to government surveys — have been getting worse and worse. Then you figure: OK, during the shutdown, that impaired the data collection process on top of that.
So you have multiple, I don’t know, error bands is maybe how you would present it. Then you have this very strange economy where we know there is this one sector of the economy that’s doing absolutely phenomenally well, which is A.I. and other tech-adjacent things, and then the other areas that probably are sort of stagnating — maybe a little “stagflationary” vibes.
So even if we had a very clear picture — let’s say we had great response rates in all the surveys, and they had been running, and we hadn’t had a government shutdown — this is just a very strange underlying condition. So levels and levels and levels of uncertainty.
The ontology of the economy is unclear. [Chuckles.]
Weisenthal: Yes, absolutely.
So I want to go through some of these stories that you have covered very closely, what happened at the start and where they’ve settled.
And I want to start on tariffs. Tracy, you mentioned the tariffs. So take me back to the week of “Liberation Day.” There were a lot of emergency episodes of your podcast that were great.
What happened then? And what has happened at a high level since then?
Alloway: So that was a crazy week for obvious reasons — capital-H history being made.
I think the surprising thing to everyone was the actual rollout of the tariff announcement and just how unstructured it seemed to be in many ways. This idea that we were going to impose tariffs on small islands out in Oceania —
Weisenthal: Tariff the penguins!
Alloway: Right. Whose only export is bat guano or something like that. It just didn’t make any sense.
Markets don’t deal with uncertainty at the best of times, and this was like a boatload of uncertainty being dumped onto the market. So you saw this huge reaction.
What was even more surprising was that the market recovered so quickly. We spoke to a lot of businesses back in April and May and asked them: How are you dealing with the tariffs?
And we heard, for instance, from a women’s clothing company saying: It’s been absolute chaos. We don’t know what’s going to happen this year. I’m supposed to be putting in orders for the winter Christmas season. I don’t know if I can actually do that with my suppliers in China. I don’t know how much to order.
And yet, fast-forward to now, and things seem to be ticking on relatively well. The big question is going to be whether or not that’s an overhang from earlier periods in the economy. People have already bought a lot of stuff. They stockpiled inventory coming into tariffs. So maybe at some point, all that uncertainty, which you would presume would cause businesses to invest less in their own companies, maybe eventually will hit.
One of the reasons the tariffs were hard to cover is they kept going up and down. Then there would be these bilateral deals with other countries.
Joe, I want to show you a chart from the Budget Lab at Yale that tracks effective tariff rates since the beginning of the year.
Can you talk through what you see on the chart? What happened? And what do you make of it?
Weisenthal: So this is a chart of the U.S. average effective tariff rate since the beginning of the year until now. At the start of the year, the U.S. was an open economy — we had very few trade barriers. It was less than 5 percent on average. Probably looks like it was somewhere close to 2 percent.
Then obviously, when we started getting those initial tariffs on Mexico and Canada — and then, of course, there was “Liberation Day” in early April — nearly 30 percent effective tariffs across the board. Then we started getting the deals and the carveouts and the bilateral arrangements, and we’ve settled in this area that’s somewhere between 15 and 20 percent.
So we are now a very high tariff country. And I think if you had taken those first numbers seriously, people would have looked at them as like: No, we just can’t trade at these levels. But with the modifications they became, I think, tolerable.
The way I’ve been thinking about the tariffs is this: On the one hand, you might say: OK, tariffs are inflationary. They raised the price of goods. That’s inflation.
Another person might say: Tariffs are disinflationary. Tariffs are a tax, historically. And most economists would say, when you raise the tax on things, you’re taking money out of the economy. That’s disinflationary.
So to add to this uncertainty, you can make both arguments. The way I conceive it —
It’s disinflationary because it slows down economic growth?
Weisenthal: Yes, because you’re taking money out of the economy —
You’re taking money out, and people don’t have that money to spend.
Weisenthal: The way I resolve the tension in my head is not to think about inflation versus disinflation itself, per se, but just to think about this idea that we have raised the cost of doing business in the United States. That, I think, we could safely say.
Tracy mentioned that we talked to a women’s clothing retailer. We were also in Alaska this summer, and we talked to this guy who owns the biggest furniture chain in Alaska, which was really fun. And he was talking about how they were going to find a company in India that manufactures this couch or that chair, instead of in China. So they find workarounds, which is why trade hasn’t come to a halt.
But that company in India, maybe they don’t take as many orders as previously because they’re worried that by the time that couch gets to the port, maybe the tariff schedule is going to be different, and then the importer doesn’t want to take delivery of it at the new tariff rate, etc.
When it all shakes out — inflationary versus disinflationary — we don’t really know. But if you add up all of these factors, it’s going to make uncertainty about sourcing decisions, it’s going to change your pricing, and you don’t know how consistently you’re going to access your goods.
In the end, it raises the cost of doing business. It throws sand into the gears, so to speak, of the economy in ways we may not feel for a long time but that might over time degrade the economy or degrade our standard of living.
It settles high, right?
Weisenthal: Yes.
It’s more than half as high as it was right after “Liberation Day.”
And Tracy, I listened to a lot of “Odd Lots” in that period, as in every period.
Alloway: Well, thank you.
And I was listening to the Flexport C.E.O. tell me that global shipping was going to collapse. I was hearing people say that kids were not going to have Christmas toys. And then Donald Trump said: Well, what do these kids need all these toys for anyway? [Chuckles.]
Weisenthal: Which he’s right about.
It’s the antimaterialist turn in Bloomberg here.
Weisenthal: Yeah. [Chuckles.]
And I was hearing that things were just going to break down if tariffs held at high levels.
And then they held at high levels, and things did not break down. Why?
Alloway: It’s an excellent question. First of all, a lot of that hyperbole that we heard about empty shelves and things like that was right after the announcement. And so there’s still a debate: Had you stuck with those levels, maybe we would have seen that.
But even where they settled at the higher rate — actually, if you zoom out on that chart and go back to the 1930s, we’re at the highest effective tariff rate since the Great Depression, basically. So you’re right: This is really surprising.
I think one of the things that’s happening here is there’s a tendency to oversimplify the business environment. People think there’s a company in the U.S. that imports stuff from a supplier in China, and that’s all that happens — that’s the way things actually get into my house.
But of course, there are all these middleman entities in between that process. What that means is you actually have a pretty diverse cushion to absorb some of the tariff costs and maybe even some of the operational issues.
So maybe the shipping company lowers some of its rates in order to encourage business. You’ll have an actual importer who’s bringing that stuff and then selling it wholesale, and maybe they’ll start reducing their prices to offset some of the tariff rate.
So if everyone is giving up a tiny slice of that value chain, then the impact on prices can end up being a lot less than you expected.
Weisenthal: One thing I appreciated, or really realized, after the Covid shock is that American businesses are really good. They’re really well run. And I think this is an important point.
If you think back to March 2020, so many people had no idea what was going on, and some people sort of retreated into their homes. I will admit that I was one of them. I was like: All right, I’m just going to stay on my computer.
And then you look at the creativity of American corporations, which were like: We’re going to figure out a way to turn this restaurant into an overnight commercial kitchen for delivery.
We saw this incredible amount of resilience during that period. We saw a tremendous amount of companies figuring out: OK, what do we have? And how can we keep operating during these extreme conditions?
I have come out of the last five years with a greater admiration for the creativity and resilience of corporate America to withstand these shocks, and management and executive teams essentially finding a way to very quickly pivot and figure out: How are we going to keep running our business under this new uncertainty?
But not just us, right? When we were hearing from all kinds of people who had a line of sight on global shipping data and port data, one reason the predictions of collapse felt so vivid was that it was so inhumanly complex. You would think something that intricate cannot possibly be as flexible.
Weisenthal: To withstand a shock of that magnitude. Yes.
This many shocks. Like, shock after shock after shock. I mean, there have been wars in this period.
Weisenthal: Yes, that was my realization after the Covid shock. That had such an extraordinary impact: Every corner of the globe restructuring life almost overnight, or maybe in the span of a couple weeks.
And of course, there were tremendous shortages everywhere and all kinds of disruptions. But somehow, the machine kept ticking in a way that, looking back, I think would have surprised a lot of people.
Alloway: Not to be supernegative, but I do think we shouldn’t downplay the impact on productivity.
There are a lot of man-hours being devoted to figuring out the tariff schedule — maybe figuring out how to game it a little bit — and then filling out paperwork at the docks and stuff like that.
I don’t know about you, but I shop a lot. I bought something from the Netherlands in September, and I never got it because the shipper said they couldn’t figure out how to mail it to the U.S. and who would actually have to pay the tariffs.
I ended up getting into a credit dispute with them. It took up a lot of my time. I still don’t have the item.
Weisenthal: We get so much content from Tracy.
Alloway: Yes — from my daily life. But that’s one thing, right? So imagine this multiplied by millions of things across various companies. It’s a lot of hours, a lot of manpower.
But let me ask you about the other side of this. We’re talking about the catastrophes that either didn’t or only sort of happened. And I take your point that there was real disruption here, even if it wasn’t the economy shattering.
But obviously, the point of the tariffs, which were a chosen policy in the way that the global pandemic was not, was to create benefits.
And as I would listen to the Trump administration, I would hear that it was going to bring a ton of manufacturing jobs and capacity back to the U.S. I heard a lot about how much revenue it would bring in. Maybe we wouldn’t even need income taxes anymore. I would hear a lot about the security benefits of this.
So the tariffs are a policy meant to create a gain for America. Did they?
Weisenthal: This is the funny thing. We’ve been talking about tariffs for so long, as if it were just something that happened. But to your point, there was ostensibly the idea that it was going to make the economy better.
We chose this! [Laughs.]
Weisenthal: Yes, we chose to do it, which makes it very different than the pandemic, obviously.
I have seen no evidence that we have seen some gain from it — none. There has not been some great boon for the American work force. We know that unemployment has been ticking up.
It is true that they’re collecting a decent amount of revenue from the tariffs — that is real. But the idea that has somehow obviously redounded to the benefit of the American consumer or home buyers, or made our deficits more sustainable, per se — I haven’t seen any evidence of it.
So in terms of the good? I don’t know.
Alloway: There’s also a major policy tension where Trump was claiming: This isn’t going to have an impact on prices. Your prices are going to stay the same. It’s not going to slow down the American economy. But then he was also arguing that this was going to be a huge revenue generator for the U.S. government.
You can’t have both. You can’t charge people a bunch of money and raise a bunch of money, and expect not to be taking the people’s money. It has to come from somewhere.
I didn’t mention one other policy aim here. So we had “Liberation Day.” We tariffed a bunch of penguins for a while. They begin to remove some of the tariffs. And then there is a really big policy pivot that makes a lot of people on the right happier: They settle down the tariffs on our more normal trading partners, and they jack them up on China.
And the new policy rationale we are given is that this is a trade war with China. We are going to isolate China. We are going to take our manufacturing capacity back from them. We are going to reverse their manipulation of world markets.
Now we’re here at the end of the year, and we finally have a deal with China.
Tracy, what is that deal? And how does that fit or not fit the China theory phase of the trade war?
Alloway: So it is true that competing with China is one of the few areas of bipartisan agreement at the moment. I think everyone feels this sense of competition.
But what China has actually been really good at is creating ecosystems for certain products and industries. For instance, on the rare earth side of things, we hear all the time that rare earths are a major choke point for the U.S., and there are worries that China is going to cut off supplies.
The way China has approached that industry is they have mining extraction — again, it probably benefits from a lack of environmental regulation — and they’ve also built manufacturers. And then they have a very vibrant E.V. industry and computer industry that’s built around that rare earth supply, and is there to off-take the supply — to actually consume it.
It’s very hard to replicate those types of ecosystems in a short time frame.
I buy that. But we went from a 100 percent-plus tariff on China as an effort to make them less competitive, to, I believe, now it’s going to be a 20 percent tariff on China? So it seems like we’re not —
Weisenthal: That’s the thing: We don’t really know where the White House stands. The question you posed: Is the entire trade war about isolating China? — that was one view.
The U.S. is not the only country where people have very serious concerns about manipulation or the effect that Chinese manufacturing has on their own national champions. All the anxieties that we have in the U.S. are shared by plenty of other countries — and not just in Europe, but elsewhere in Asia, South America and so forth.
But we don’t really know. There are China hawks in the administration who clearly feel like this is an existential threat.
Trump himself — who arguably, more than any American of the last decade or whatever, is responsible for the sort of massive national turn on China — may be one of the least hawkish members of the administration when it comes to China. He clearly has a lot of admiration for Xi Jinping — I think he likes him. He may think it’s unfair, but he clearly does not hold it against China. He clearly admires or appreciates the fact that the various strategies that the country has undertaken were done in pursuit of the national interest.
So I don’t think we actually know. And again, if you’re going to do this “isolate China” strategy, then you really — or I would say intuitively — want to have as much trade as possible with the non-China world. Really destroy any barriers. But we haven’t done that. So I don’t think we really know where this administration views these sorts of tensions with China, in part because I think it’s divided.
This gets to a reality that I think a lot of us suspect but is inconvenient to talk about: We are used to covering White House policies like they are highly connected to White House goals — that there is a legibility to the connection between means and ends.
But the Trump administration often seems to me to have different levels operating completely separately. There are policies that come out of highly ideological members of the administration, sometimes in conflict with one another. And then there’s Trump, who sees the world through deals and relationships.
What I see happening in the tariffs, if you track the year, is a series of policies that are then overtaken by a series of deals and relationships.
There will be a standard policy: We’re going to tariff everybody. We’re going to tariff China 100 percent, 110 percent, 165 percent. And then slowly, Trump will get worked on by the person. Some kind of tribute — maybe one that we can see, or I suspect, more often that we can’t see — is going to get paid. And all of a sudden, the tariffs are down.
And there’s no policy that comes through clearly, because there never is a policy. There are only, in the end, deals.
Tell me if that’s wrong or if you have a better narrative than that.
Alloway: I think there’s definitely an element of erraticism — is that a word? — when it comes to the Trump administration’s policy. And it gets back to that tension between alleged policy goals.
Again, if you’re really worried about the U.S. deficit, are tariffs the best way to actually generate money for the U.S. government? Probably not. Which then poses the question: Well, why are we doing this?
I guess if you were going to be very cynical about it, you could argue that Trump really likes deals because it gives him those short-term wins and those short-term headlines. People forget things fairly quickly when it comes to the news flow, and so if all they see is “The U.S. Strikes a Deal With China” or “The U.S. Is Bringing China to the Negotiating Table,” people forget all the chaos that it took to actually bring us to that moment.
I do think with China, in particular, the rare earths thing was really important. I think there was a sort of —
Do you want to describe what happened there?
Alloway: Sure. So basically, China said: OK, you’re going to tariff us at 50 percent, or whatever. We’re going to cut off the supply of magnets that are used in lots of batteries, computers, things like that.
And I think that sparked an element of panic among the type of people Trump listens to. Business executives are thinking: Well, this is an incredibly important component for my particular business. I can maybe get some of it from elsewhere in the world, but certainly not at the cost and scale that I’ve been getting it from China.
So China has done a phenomenal job of basically putting itself right in the middle of a crucial choke point for the entire global economy, and they were able to use that to their benefit to reduce the tariffs. They didn’t get them all to zero, but they brought them down a lot.
So did we fight the trade war with China and lose?
Weisenthal: I do think Tracy is absolutely right that the rare earths, specifically, to some extent because of our vulnerability there, may have undermined the entire prosecution of the trade war, so to speak, because it’s so specific. So arguably, yes.
I think to your question, though: I think that offers such an important insight into how he thinks. The idea of a strategy or a policy is abstract, whereas a deal is him. A deal is something where he could shake the hand of someone else. That is real. That is tangible to him.
Alloway: It’s good TV.
Weisenthal: Yes. Whereas all this other stuff around: What is our long-term strategy? It’s abstract. It’s depersonal. I do not think this is the way Trump conceives of government.
But he’s got people around him. I actually find what just happened with the A.I. chips shocking.
Some background on this for people who have not been following it as much: We have had, since the Biden administration, pretty tough export controls on forms of chips that are very useful for creating frontier A.I. systems. And Trump just cut a deal, at the urging of the C.E.O. of Nvidia, to ship some of the more advanced Nvidia chips to China.
And it is, on some level, impossible to me. Having covered a number of White Houses, you normally have a bunch of advisers around, being like: Sir, that’s not our policy. Everything you have said, everything we are trying to do, is trying to maintain dominance on this specific frontier — A.I. — against this specific competitor — China — so you can’t give them the chips just because one of the C.E.O.s, with whom you’ve had the country take a cut in his company, wants to.
Their giving China the Nvidia chips just struck me as the final collapse of the China policy, at least at any level of intelligibility, in the Trump White House. I just didn’t have a way of reading it aside from that.
Alloway: If you’re going to ask us to explain it, we’re going to struggle.
[All chuckle.]
Weisenthal: The David Sacks argument is that the important thing is that Nvidia, the American chip company, remains the dominant infrastructure for the development of A.I.
I don’t find that to be completely unreasonable. I don’t find that to be a completely absurd argument.
So there’s, on the one hand, the dominance of A.I. The question is: Well, who is at the edge of developing models? That’s certainly one way of measuring who is at the frontier of A.I.
But it does not strike me as crazy, per se, to redefine the question of the A.I. race as: On whose chips and on whose software architecture will all A.I. models be built in the future?
I don’t have a view on which is right or wrong. I’m just saying it does not strike me as necessarily absurd — that view that it is a win if everyone is using American-designed chips.
I disagree with you, but I don’t think it’s absurd. I guess what you could say is the Trump administration spent a year evolving on the question of China.
I would find it genuinely interesting if a member of the Trump national security team came out and gave a speech about how they rethought everything on China. But nobody did that. We just went from one policy to the other without anybody really explaining how the theory of the policy changed.
Alloway: I think that’s right, and I doubt anyone in the Trump administration was like: OK, we need to change our approach right now. We’ve had this big realization.
What I would say is: If you think about how China views the sort of technology competition with the U.S., the thing that comes up quite a lot domestically in China is — have you guys read “The Three-Body Problem”?
Yes.
Weisenthal: I never read it. I should. I started it.
There’s a pretty good Netflix series.
Weisenthal: Yeah, I’ve watched the Netflix series.
Alloway: Yes, the Netflix series is good.
A very brief synopsis: Aliens are threatening Earth, and basically, all of humanity comes together at some point and develops technology to get rid of the aliens in a domestic context.
[Laughs.] I would say that does some violence to the plot, is my take.
Alloway: It definitely does, but we don’t have that much time, so I’m shortening it.
Yeah, they’re long.
Alloway: But in China, there’s this idea that if the U.S. completely cuts off the country from global technology, then China is going to accelerate its own technological development and basically do everything that the rest of the world does, probably more cheaply and at better scale.
That was a concern that we saw starting to bubble in the Biden administration when they toughened up the initial restrictions on chips. We saw China start to allegedly produce some pretty advanced things.
So I think there is a concern there that if you press too hard, China is just going to double down on its own development.
.op-aside { display: none; border-top: 1px solid var(–color-stroke-tertiary,#C7C7C7); border-bottom: 1px solid var(–color-stroke-tertiary,#C7C7C7); font-family: nyt-franklin, helvetica, sans-serif; flex-direction: row; justify-content: space-between; padding-top: 1.25rem; padding-bottom: 1.25rem; position: relative; max-width: 600px; margin: 2rem 20px; }
.op-aside p { margin: 0; font-family: nyt-franklin, helvetica, sans-serif; font-size: 1rem; line-height: 1.3rem; margin-top: 0.4rem; margin-right: 2rem; font-weight: 600; flex-grow: 1; }
.SHA_opinionPrompt_0325_1_Prompt .op-aside { display: flex; }
@media (min-width: 640px) { .op-aside { margin: 2rem auto; } }
.op-buttonWrap { visibility: hidden; display: flex; right: 42px; position: absolute; background: var(–color-background-inverseSecondary, hsla(0,0%,21.18%,1)); border-radius: 3px; height: 25px; padding: 0 10px; align-items: center; justify-content: center; top: calc((100% – 25px) / 2); }
.op-copiedText { font-size: 0.75rem; line-height: 0.75rem; color: var(–color-content-inversePrimary, #fff); white-space: pre; margin-top: 1px; }
.op-button { display: flex; border: 1px solid var(–color-stroke-tertiary, #C7C7C7); height: 2rem; width: 2rem; background: transparent; border-radius: 50%; cursor: pointer; margin: auto; padding-inline: 6px; flex-direction: column; justify-content: center; flex-shrink: 0; }
.op-button:hover { background-color: var(–color-background-tertiary, #EBEBEB); }
.op-button path { fill: var(–color-content-primary,#121212); }
Know someone who would want to read this? Share the column.
I think we should move to talking about another major economics story, which is artificial intelligence and the huge A.I. build-out.
I want to start here, Tracy, by showing this chart from JPMorgan.
Walk me through what you’re seeing here.
Alloway: This is U.S. real G.D.P. growth contribution from tech capital expenditures.
As you might expect, technology related, A.I.-related capex has become a much more important driver of G.D.P. growth. I can’t tell on this particular chart, but I heard 40 percent of U.S. growth in 2025 is estimated to be coming from A.I., which is more than the growth we’re getting from consumer spending, which is pretty phenomenal in the context of the U.S. economy.
I would also just add, right before we came in here to record, I saw a number from Standard Chartered. They were saying two-thirds of U.S. growth this year is coming from tech and A.I.
You never want to be too dependent on one particular industry. You especially don’t want to be too dependent on an as-yet-unproven technology in a highly cyclical industry for your growth.
What is all this money buying? You’ve had a bunch of episodes this year on the A.I. build-out. What is being built out?
Weisenthal: Gigantic computers, basically, in huge complexes that are as large as Central Park, in somewhere like Abilene, Texas, or whatever. They’re these mammoth data centers that are these assets that are a combination of real estate, so they’re real estate plays; they’re high-tech plays because they’re filled with those Nvidia chips or other chips; they’re huge energy consumers, or increasingly energy producers in their own right, because it takes a while to get anything access to the grid, and they’ll have a natural gas facility producing energy on the campus directly.
So it’s this extraordinary build-out all over the country of these data centers, which, of course, becomes very political and very important to the economic growth.
The really important thing is: Since basically 2008, 2009, 2010, this handful of tech companies has been where a lot of the action is, particularly for growth and income and making money in the U.S. economy.
These companies used to mostly just spend on labor. Their main cost was their software engineers. And everything else was pretty cheap, because cloud compute was pretty cheap. But they’re famously asset light, as they say.
So the other big story of this chart, setting aside this U.S. G.D.P. component, is the degree to which these very important companies for the U.S. economy have suddenly become big spenders on stuff in a way that they never had any history of before.
So you see these companies taking out more debt, or they’re coming up with these special financing vehicles where it’s off-book, and suddenly there are these off-balance-sheet arrangements to finance these things.
So there is this fundamental restructuring, I would say, of the profits and losses and the balance sheets of these big companies that is novel in their corporate history. And that’s really a story of the last two or three years.
I was watching this interview with Sam Altman.
Archival clip of Sam Altman interview:
Brad Gerstner: So I think the single biggest question I’ve heard all week, and hanging over the market, is: How can the company with $13 billion in revenues make $1.4 trillion of spend commitments?
Sam Altman: First of all, we’re doing well more revenue than that. Second of all, Brad, if you want to sell your shares, I’ll find you a buyer.
[Brad laughs.]
Altman: I just — enough.
The argument he’s making there is: There are so many investors of every different level piling onto this. OpenAI remains a private company, but if you could buy OpenAI stock, people very much would.
But these are banks and private equity players. And there are all kinds of highly sophisticated financial players and companies coming in behind these investments to make them.
That means there are all these board meetings and shareholder meetings where an argument is being made about how these huge build-outs of these giant computers will lead to profits that justify the build-outs.
When Meta and Microsoft are making these arguments to their shareholders and to the other investors they might want to work with, what is the argument? What does this look like if it pays off economically?
Weisenthal: There’s a lot about the A.I. boom — the A.I. race — that probably intuitively reminds me in many dimensions of the race to build a nuclear bomb.
[Chuckles.] I wasn’t sure where that was going, but I didn’t quite see that.
Weisenthal: So someone was describing to me OpenAI as being like the Manhattan Project, except the goal is not to build the bomb.
Because the big fear is that — as many people in this space fear — you’re going to have this runaway A.I. that kills us all. So how do we build something that doesn’t kill everyone?
Alloway: It’s very existential for those companies.
Weisenthal: Right, that’s exactly it. So there’s that. There’s this fear of who is going to get there first. There’s the U.S. versus China element, etc.
I think there is something similar going on with the way companies feel about being part of buying into A.I. as products — that this technology is going to be so powerful and so important that even if you can’t articulate why you need to be adopting it for your company, you better have an A.I. project. You better have an A.I. experiment going on. Because the stakes are so high for whoever figures it out — a way to plug A.I. into their business, reduce labor costs, get more productivity and so forth — that the gains are going to be so great that you literally just can’t afford not to be investing in it from a consumer experience.
If you listen to someone like Mark Zuckerberg, he clearly thinks that. If you look at every other company, it is this fear.
I do think it’s worth pointing out. There are examples. Clearly, engineers use A.I. coding all the time. Other companies are figuring out a way. We did an episode with the C.E.O. of Affirm, and he was talking about how A.I. allows them to see companies that are misstating what they do in their terms of service.
So it is true that already companies are figuring out ways to deploy these tools, but I really do think there’s a tremendous amount of fear driving this, at both the hyperscaler level and the customer level, that someone else is going to figure something out before them.
Let me zoom in on that. I’ve covered these A.I. companies for a long time. I covered the Anthropic guys when they still worked at OpenAI. [Laughs.] And that is how they all used to talk about it — that it is a race to build superintelligence, the one A.I. that will rule them all.
Alloway: To build God.
Yes, to build the machine God.
But I think if you believe that version of it — that there is a race, and there’s a piece of tape at the end of the race, and one of the companies is going to pass first, and then second place is the first loser — that actually implies something very dangerous about this build-out. Which is that it only matters for one of them. And the assets are going to be, not useless but not that useful for the others. And I believe they believe it — or certainly used to believe it.
It has been strange watching these people turn into, like, SaaS businesses. It seems to me more like what you see is: Anthropic is trying to own coding. Meta is going to try to own social relationships and A.I. and use it to kind of manipulate you into buying things. OpenAI is going to be a layer in enterprise software like Microsoft. Google has everything in Google.
What you’re getting at is actually my fear about it: That there are two stories — that they’re acting as if it’s a race to the finish line, and so all that matters is being in front. But if that’s not true, then actually there’s way overinvestment.
Weisenthal: And then there’s this phenomenon where it’s like: While we’re on the race to create superintelligence, we’re going to create these slop apps that everyone has.
So Meta has Meta AI, which is sort of like Instagram, except it’s all A.I.-generated garbage. And ChatGPT has Sora, which is a little bit better, I think, but whatever — it does not feel like a way station on the path to superintelligence when you see them rolling out these things. As you mentioned, they’re starting to look more and more like traditional software businesses that just plug into various layers.
So I agree: I think it’s very muddled. I’m sure there’s a mix within the companies. I’m sure there are people who are like: No, we are here to build superintelligence — and there are probably others who are like: We are here to build enterprise software and build a subscription.
What you are watching happen inside the companies, since I feel more confident talking about it, is the very normal thing that happens in politics and in companies, which is that you talk yourself into the story that your bottom line needs you to believe.
So the way to build superintelligence is through building SaaS software. Because that’s going to get you the scaling, it’s going to get you the investment.
The most striking thing to me in covering this for years now is: The things I have heard from people building A.I. are just wild, and they were really wild in 2021 and 2022. Like, truly, the wildest things I’ve ever heard in my reporting, in terms of what people believed to be true in, like, 10 years.
I’m not even sure they’re wrong about what will eventually be true, but then watching them end up running these totally normal-looking businesses except for the scale of the investment — “how much of your Slack should be written by A.I.” kind of thing — it is amazing.
I guess it’s true for religion, too. You’re trying to tap into transcendence, but you also need to fund the real estate investments for the church.
[Weisenthal and Alloway laugh.]
But there is this incredible mixture of the sci-fi and the mundane, and watching the companies have to bring those two things into alignment has been sociologically very interesting, and a reminder of the incredible power of capitalism to persuade people of things.
Alloway: I was just about to say: First of all, if you couch everything in existential terms, then you know the limit on your capital expenditure is basically infinity, so that’s part of what’s happening here.
[Klein chuckles.]
Alloway: There are two approaches to building out A.I. at the moment, and I call this “the coffee-pod theory” of A.I.
America is building really expensive cappuccino machines that it thinks are going to produce the most amazing cup of coffee that the world has ever known. And because of that, everyone in the world is going to want to buy one of these cappuccino machines.
That is not the only way to approach A.I. development or the A.I. business model. China has taken a very different approach. Again, China is doing the Nespresso coffee-pod version of A.I. technology. They’re producing something that’s relatively cheap, something that’s pretty standardized and something, again, that it sees the entire world being a market for.
I don’t think we yet know the answer to: Which particular business model is going to win out? But then — you mentioned capitalism — on the customer side, we’re also seeing this narrative dynamic where, in late-stage capitalism, it’s kind of hard to boost returns forever. So now this new lever has appeared. It’s called A.I., and all you have to do is pull it — or at least put out a press release saying that you are pulling the A.I. lever and cutting workers and saving a bunch of money — and you’ll see your share price go up.
And I think that’s pretty important. Investors are still responding very well to any utterance of A.I. in a business press release. There may come a moment where people are actually like: Wait a second, we want to see the cost savings. But it’s not happening yet.
In terms of questions I would like to ask the “Odd Lots” hosts, high on my list are: What is late-stage capitalism? And do you believe in it as a useful conceptual tool?
Weisenthal: [Chuckles.] Tracy said it! So I’m going to make her define it. I have no idea. It’s all on Tracy, not the “Odd Lots” hosts. Tracy is the one who mentioned it.
Alloway: I mean, OK, it’s a little bit of an intellectual crutch — I will give you that. Because we’re always living in late-stage capitalism, right? Late-stage capitalism is now.
But if you look at our existing situation, it’s a relentless search for growth. And a relentless search for growth is also something that is, in many ways, very unique to America.
I hate to keep talking about China, but other countries take a different approach. We’ve seen China take the lead in a number of strategic industries, which counts as growth, but that growth hasn’t translated into huge returns for investors.
So if you look at a line of the Shanghai Composite, it’s been going sideways for many, many years. China is willing to make that trade-off: We’re going to develop important industries and give people jobs, and maybe shareholders just aren’t going to make that much money off it.
In the U.S., it’s pretty much all about shareholder return and getting that line going up forever. And our political economy is basically built on that entire system, where if you put your money in the S&P 500, you’ll probably have a decent retirement, and so everyone will be fine.
So the way you understand it is not so much that there’s something specifically late about this stage of capitalism, so much as this is financialized growth capitalism.
Alloway: Yes. And also late-stage capitalism implies that there’s an end at some point, and I’m not sure there is.
That actually is a better segue to this next chart I was going to show you than I thought it would be. There are some suspicions going around that this whole thing has become a circular money machine — that the hunt for growth, the hunt for justifying share prices and investment and valuations, is leading to money just constantly passing hands to create almost the appearance of activity.
So Joe, I’m going to show you this chart, which is a Bloomberg chart.
Weisenthal: I recognize this chart from a distance. Yes.
And it’s extremely hard to parse, including for me.
Weisenthal: There’s no need to parse it. The point is almost not to parse anything—
The chart is a vibe. It’s a visual more than a chart, too —
Weisenthal: The point is just to gaze upon this incredible level of interlinkages—
This, to me, is almost the most interesting chart to look at in A.I., so why don’t you look at it and tell me what you see in it?
Weisenthal: Here we have a chart with Nvidia at the center, and basically everyone is invested in everyone else. So Nvidia invests in OpenAI, then has an investment in CoreWeave. And CoreWeave buys chips from Nvidia, so the revenue gets recycled.
Basically, everyone is linked to everyone else.
And bidirectionally. It’s not like you pay someone and they pay someone else. Like, you pay them, and they pay you. You invest in them, and they invest in you.
Weisenthal: Yes. You’re exactly right. So I’m going to invest in you, and then not only are you going to buy chips for me, you’re going to make an equity investment.
So obviously, there is the web of complexity, which I think we associate with 2007, 2008 — which is the sheer volume of the web of relationships and so forth, and just how hard that is to decipher.
But then there’s the other element that goes back to the dot-com bubble. If you looked at a lot of the companies that were riding high on the dot-com bubble, they had real revenue.
The poster child for this was Yahoo or Cisco. So you have these companies where you can say: OK, maybe they are a little rich on the stock market, but look, we know they’re real businesses.
The issue is that underneath these real businesses, there was a lot of financialization going on. By that, I mean specifically: There was a host of start-ups, and they were raising money in I.P.O.s, and then that I.P.O. money that they raised would immediately be put into either ads on Yahoo or purchases of Cisco equipment. And when the I.P.O. market closed down when there was a little bit of risk-off appetite in the stock market, then suddenly, the revenue collapsed at those giants. So yes, what looked like sustainable, healthy businesses were really being funded by financial markets.
And I think that the concern is, when you look at the A.I. boom, you have all these companies doing very well. Nvidia is absolutely a real business. It absolutely has real revenue, it absolutely has real profits — no one is denying it. Is there some richness in the valuation? Sure — maybe — I don’t know. But very plausibly, they’re real businesses.
So I think that the issue when we talk about a bubble in A.I. is: Sure, there may be rich valuations, but the fear would be that these are not sustainable revenues and therefore not sustainable profits.
So let’s talk about the question of a bubble. Tracy, you all have done a bunch of episodes talking to different people about this. Make for me the best case you can both against the idea of a bubble and then for it.
Alloway: Oh, man. So against the idea of the bubble is very simple. It’s this idea that we were talking about earlier, which is that this is basically a winner-takes-all strategy, and if everyone develops the products that they say they’re going to develop — if they develop A.I. models or systems that magically solve every business’s or person’s problems in the entire world — then perhaps you can justify some of those valuations.
Weisenthal: It’s not a bubble if magic occurs. [Chuckles.]
Alloway: That’s right. And that’s what a lot of these companies are promising. They’re promising magic, right? That’s the way they talk about it.
So I think there’s a concern, as A.I. becomes an even more dominant force in the U.S. economy, that if the bubble bursts, or even if the promised revenue and savings do not materialize to the scale that people think they’re going to, then you’re going to have an economic impact that potentially feeds on itself. Which would be — again, not to be too pessimistic — but similar to what we saw back in the run-up to the great financial crisis, where housing became an incredibly important driver of U.S. economic growth. Everyone was buying houses. Houses were being built. We saw the share of housing construction in the U.S. economy go up, and eventually, it got so big that housing became the source of wider problems in the U.S. economy.
That wasn’t always the case. It used to be that there were problems in the U.S. economy, and housing would get hit. What happened was that housing got so big that housing became the proximate source of problems in the wider U.S. economy. And the concern now is that we might be on the same path with A.I.
You showed the chart of the circularity of a lot of these businesses. I always think about that “It’s Always Sunny in Philadelphia” meme of the guy standing in front of the board with all the red strings connecting everyone. It feels very much like that once you start to untangle these relationships.
The other concern is just the opacity of how A.I. is actually getting financed right now. There’s a lot of stuff going on in the private credit market. The private credit market is where businesses get loans from — sometimes banks but mostly other types of investors — and these loans and bonds are not publicly issued, not publicly traded.
So normally, if IBM or Microsoft or whoever issues a bond, it would come with a big prospectus. There would be a lot of information available about it online, you could see the terms, anyone can buy it, and people would trade it after.
Private credit is something much more bespoke. It’s sort of a customized loan between a business and an investor. It’s very hard to get much insight on that particular market for obvious reasons. The clue is in the name — it’s all private.
So I think when it comes to financing, it’s pretty difficult to get a sense of the scale of what’s happening right now — but also to get a sense of who is actually financing what. We hear stories. You hear big private credit investors, like Apollo, who will say something like: We’re really into data centers at the moment. But it’s hard to get a sense of how much.
I want to look at this then not from the market’s perspective or the financier’s perspective but from the worker’s perspective.
You were talking about how they’re promising magic, and that’s one way to put it. The other way you might put it is that they’re promising replacement — that the thing that would make these companies extraordinarily valuable is, in fact, if companies are suddenly able to replace human labor like accountants and paralegals and H.R. workers with tireless chatbots who never want to join a union.
And one thing I have wondered about a lot is not the case where they invent superintelligence, which has its own set of possibilities and problems, but the case where the more direct economic bet pays off — the bet I hear C.E.O.s talking about and investing in.
Is that scenario good for workers?
If we are not in a bubble, does it mean we are in a labor-substitution world, which in some ways is going to be much tougher on normal people than a bubble?
Weisenthal: Yes. The way I like to think about it is either if the A.I. bet fails, then we’re going to have a recession and a bunch of people are going to lose their jobs, and if the A.I. bet succeeds, then a bunch of people are going to lose their jobs because A.I. will be able to replace labor. So either way, fail or succeed, it feels like it ends in a bunch of people losing their jobs.
I have very mixed feelings about this question, though. Economists are very sort of strict on this idea that there’s always a demand for labor. That yes, of course, historically, you’re going to have an invention that from time to time puts an entire class of workers out of business. Or that there’s no longer a need for this work because we’ve developed a technology. But then that means savings from someone else, and then they spend it somewhere else, and that creates new labor demand.
Furthermore, they would say: That’s literally what economic progress is. We’re not toiling in the fields the same way because we’ve gotten so much more productivity. This is, by definition, what progress is.
What feels different about A.I., obviously, is that it’s all happening at once and the sheer range of potential vocations that A.I. could disrupt, whether we’re talking about lawyers or accountants or coders, etc.
So again, progress is labor-saving technology. The ability to get more with fewer man-hours is what economic growth is at its core. But what makes A.I. different or why it raises anxiety in a way that other labor-saving technologies might not is just the sheer range of professions —
Well, let me combine your two scenarios there into the one that I actually worry about the most on the labor market side, which is: You could have this thing where the A.I. bubble pops at some level. This creates some kind of recession, which leads to firms wiring themselves for A.I. in a way they haven’t before, and bringing in the technology in a way that’s actually much worse for labor.
And then you have a scenario where there’s been an acceleration of labor substitution. And yes, in the long run, according to the economists, the labor markets will adapt — although, again, A.I. is a bit of an unusual technology because it’s meant to mimic us. But markets adapt over time. People don’t. Not like that.
Weisenthal: No, we really only have so much time here. Our productive years are very limited.
And we know that to be fired during a recession scars workers for life.
Weisenthal: Yes. So I’ll just say two quick things. One, I think an interesting twist in the story of the last several years is that in 2021 through maybe 2023, for the first time in decades, firms realized or learned that they couldn’t just put a help-wanted sign in the window and there would be a line of labor.
So I think, actually, we’ve already seen the beginnings of this — setting aside A.I. — where there’s been this catalyst for labor-saving technology that started even before ChatGPT. Because for the first time, I think it was sort of taken for granted that there would not be an endless supply of labor.
There have been further developments since then that have driven this besides A.I. Obviously, the change in flows of immigration is one. And then you have demographics. We know that an aging population is going to put an incredible amount of strain on the productive population because we have to care for the elderly, and so forth.
So already there are these catalysts for firms to feel like we have to get more productivity out of our existing labor force, even before we get to the A.I. question, even before we get to the recession question.
But to the point: Yes, recessions are catastrophic. They’re really bad. I think economists and policymakers are too comfortable with the inevitability of recessions, as though recessions are natural, recessions are healthy, that this is what clears out the brush. But they ruin lives, they impair earnings forever. They’re terrible for workers.
Alloway: I would just add that what’s also different about A.I. is we’re not talking about industrial automation. We’re talking about automation that’s really centered in the knowledge economy. So things like writing, filling out forms —
Weisenthal: Podcasting.
Alloway: Podcasting. I would describe it as a lot of the fun stuff — like writing music and doing art and things like that. That’s really where A.I. is dominating.
Meanwhile, we’re still waiting for the robots that can fold our laundry or, I don’t know, serve us a burger or something like that.
Weisenthal: Watch our children during the day.
Alloway: Yes, something like that. So I think that’s also why there’s a lot of nervousness around this.
There’s been a lot of talk about, first, whether or not we are seeing any evidence in the labor market data that A.I. is doing anything. But then also there’s been more and more evidence that there’s something strange in the hiring and firing side of the economy, where things seem more frozen than normal.
Could you walk through both of these questions: Do you think that there’s an A.I. effect on the labor market? And then: What is the frozen labor market that people are talking about?
Alloway: It’s hard to tell whether A.I. is having an impact on the labor market. You know if people or businesses are cutting or adding workers broadly. You don’t necessarily always know the reasons.
I will say that I remember very clearly a moment — I think it was last year — when the Challenger jobs report came out. And there was a little anecdote —
The Challenger jobs report?
Weisenthal: There’s a company called Challenger that produces its own layoff tally.
Alloway: Right, so they’re counting up the number of layoffs in the U.S., and there was a tiny little bit of text at the bottom of this report that said a bunch of companies said they were laying off workers because of A.I.
That was the first time I ever really saw job losses being attributed to this new technology. But going back to our earlier point about the narrative, it’s hard to tell whether businesses are actually doing this because they’re replacing workers with A.I. or whether they’ve just figured out: If I say I am cutting people because of A.I., investors like that, and my boss really likes it, so that’s what I’m going to say.
And then, in terms of the broader employment environment —
Which is, by the way, culturally a little bit grim.
Alloway: Well, incentives matter, right? A lot of the world works because people are doing what their boss wants them to do.
Beyond that, though, broader employment — the way everyone has been characterizing it is that “low hiring, low firing” environment. We’re really seeing companies basically stick with the work force that they have.
Two things to say on that: I think it gets back to Joe’s point about the scarring from the pandemic. Everyone found themselves caught short of labor supply in 2020. No one wants to repeat that process, so they’re actually holding on to people.
And then, second, it goes back to this uncertainty, as well. No one really has a good handle on how the economy is going to unfold. And if you’re unclear on what’s going to happen, then you’re basically frozen in terms of your investment choices. So people are just choosing to — or having to — stay where they are.
Weisenthal: There’s a third option. We did an episode with our friend Conor Sen, who writes for Bloomberg, where he said: Look, you could say going into 2026 that every company has to make decisions about allocations, and if the view is that we are definitely going to spend more money on A.I. technology, then we’re going to just shift some of our spending plans for the year from hiring to capital investment.
So maybe there is a direct link — not so much that the models themselves are good substitutes yet for an employee, but just from a capital planning standpoint — that 2026 is the year we spend more on A.I., and therefore, we don’t post as many job openings this year.
For a couple of years now, economics types have debated this idea of the “vibecession,” which is a term from Kyla Scanlon.
In one of her recent newsletters, she had this chart that, out of every graph I’ve shown you guys, this is the one I have been thinking about the most.
I’m going to give it to you, Tracy. This is her chart of the vibecession and how you know it’s real.
Walk me through what you see here.
Alloway: OK, there are two lines on the chart. One is real disposable personal income per capita: how much people can spend individually accounting for inflation. That spiked in the years after the pandemic, and then it started to dip, and now it’s sort of flatlining.
Meanwhile, we have the University of Michigan consumer sentiment, which is very volatile, but in general, had been going up in the years from, let’s say, 2010 to 2020. And since 2020, it has been on a broadly downward plunging path, with occasionally tiny bits of recovery, but not really.
What I see on the chart is those two things used to track each other a bit — they orbit around each other. And basically, since the pandemic, a chasmic difference has opened up.
Alloway: This is the late-stage capitalism thing.
[Chuckles.] We found it!
Weisenthal: It’s in the chart. Yeah.
Alloway: That’s right. Eventually, if you want more and more growth, then it takes more and more to keep people satisfied. And I think one of the things that’s happening now is that it used to be that money was, in some sense, shameful. Being too rich was a bad thing. And if you were a billionaire, you were expected to give some of your money to charity or, I don’t know, contribute to the world in some other way.
Now we’re seeing this sort of grifting culture take over the world. Money is the point. Even on the religious side of things, we have the prosperity gospel now, which basically says: If you’re rich, it’s because God loves you, and so it’s good to be rich.
There’s no limit on how rich people want to get anymore. And I think that’s part of the reason that we’re seeing a broad dissatisfaction, let’s put it that way.
But then realistically, I think a lot of people are also just desperate about their future. They see house prices, they see insurance costs, they see retirement programs diminishing, and they think: Well, the only way for me to get out of this hole that’s been dug for me is to do something like gamble or bet on a meme stock or something like that.
So I think money itself is becoming more and more an important part of not just the way our economy functions and the businesses that get built but also our culture.
What is your explanation of the vibecession?
Weisenthal: I mean, clearly Covid kind of broke a lot of society. But I don’t know — I’m an “It’s the phones” guy.
Now granted, the problem with my theory — and that chart, specifically — is obviously, the smartphone has existed for long before Covid. But the other important thing on that chart is there has not been some major change in affordability.
There has not been some major change in the cost of living. Yes, there was an inflation spike. Yes, the cost of living has gone up. But in such a dramatic way that could massively explain why people are so pessimistic? I don’t think you’d see it on the chart. Real wage growth has generally been positive, and it’s been trending up lately. And it looks more or less along that same trajectory that was pre-Covid.
So I do think something is going on that I would say economists themselves are not equipped to answer. I think we’re at a point in the world in which economists only have some of the answers. But I think there are things going on in the way people perceive the economy that I would say logically precede economics, and they have more to do with cultural status. So they have politics or just the amount of time that people spend scrolling their phones, alternating between doomscrolling rage bait or doomscrolling slop.
I think these are real things. So I definitely put myself in the “It’s the phones” camp.
Alloway: I think economists have also historically underestimated the importance of relative relationships and relative gains.
Most economists would look at that chart and focus on personal income and say: Well, everyone has been getting better off. On an absolute basis, everyone is doing slightly better than before.
Realists would probably not look at that chart, but they’d look at the actual tails of the chart: How much has personal income been going up for the wealthiest segment of society versus the poorest of society? And they would say: Well, what actually matters here are the relative gains.
Even if you are slightly better off yourself, if you see someone who’s doing much, much, much better than you, you’re going to be annoyed and depressed, which is what that consumer sentiment line says.
Well, you could connect those two theories. What you’re talking about is: To some degree, we might be having very unequal wage gains. Although I will say that if you look, median incomes are going up, too. This is not just a factor of Bill Gates or Sam Altman getting all the money, and nobody else is.
On the other hand, to the extent people are on their phones all day, looking at viral videos, looking at Instagram, the comparison dimension of human life has really changed.
Weisenthal: There’s the famous phrase “Comparison is the thief of joy,” which is a phrase that people have known about forever. And now we have the ultimate comparison engine, and no one is happy anymore. Well, that phrase predicted it all right there.
Isn’t there also a line that “Wealth is when you have more money than your brother-in-law”?
Weisenthal: Yes, this is the other thing that chart does not capture at all, which is the effective wealth, because this is an income chart. So what we know is that it’s been incredible times for people who already have assets. And if you’re lucky to have the really special assets — if someone in your family had gotten interested in crypto at some point in the mid-2010s — you probably didn’t work harder than anyone else, but you happen to be standing on top of a gold mine.
So there is this distribution of wealth in this country that not only is unequal, it feels arbitrary in many respects. Why did that person get crazy rich such that their bloodline never has to work again?
They’re just standing in the right thing. It feels disconnected in many ways from the effort or time that someone puts into labor income.
Alloway: And also this is something that traditional economics just isn’t prepared to deal with. Traditional economics is all about those absolute gains, and we’re talking about relative differences.
I wrote about this in the “Odd Lots” newsletter — shout-out for the “Odd Lots” newsletter — and someone actually wrote to me saying: Well, if the poor owned more assets, then they would be in a much better position. You know: Have you tried not being poor?
I guess one way of thinking about what’s going on here is that consumer sentiment is a tricky thing to measure. You can word the question in different ways, but you’re getting at somebody’s story about the economy.
Something happening in people’s stories about the economy right now is, one, Trump came in, and he disappointed even his own people. His tariff policy is terribly unpopular. Things are very chaotic. Trump himself is unpopular. It doesn’t feel like there are people with their hands on the wheels of the economy who have a vision and a theory and competence, and you trust them. So your story that you’re living in a period when the line is going to go up is weakened.
The A.I. story is threatening to people. And then you have the comparison stories and precarity. And I think things feel both not good in the moment, but also there isn’t a story that people believe — because there is a leader or because there is a plan or because the thing that seems right around the corner seems good.
Alloway: There’s very little sign of things getting better.
Weisenthal: And layer into this the fact that we’ve had a few massive crises in a short period of time, and we have the added way the phones mess with our heads. So yeah: What is the thing that’s supposed to make you happy?
I don’t know. It’s supposed to be A.I. — and that is here to replace you.
Weisenthal: That is right.
Alloway: And drive up your electricity cost.
Weisenthal: Yes, right. That’s a really important part of it. People perceive that A.I. is a combination of: It’s going to make electricity more expensive, and you’re not going to have a job. That’s not great.
Alloway: It’s a tough sell, let’s put it that way.
As we turn the corner into 2026, if this chart looked much better at the end of 2026, either because personal incomes went up or because sentiment went up, why do you think it would be?
Alloway: I think it would probably be because asset prices keep going up, and our broad consumer economy is more levered to asset prices than it ever has been, arguably.
I also think that consumer sentiment doesn’t matter that much for the overall economy. Because frankly, even though consumer sentiment has been going down, people keep spending on stuff. And that’s been another surprising aspect of why we haven’t seen a recession emerge from the vibecession.
And I think part of the spending story, ironically, is that, again, people are kind of desperate. So if you’re not going to be able to afford a house, then why not just buy that extra lipstick or phone and make yourself happy in the short term?
Weisenthal: Another interesting thing about A.I. is that, if you think about the technologies that emerged in the early 2010s or the late 2000s, they had several years of: Wow, this is really cool. The smartphone! This is amazing. What can it do? Oh, I love sharing pictures with my friends. I love being able to talk to fellow reporters all day on Twitter — etc.
So it seemed like the trajectory with past technologies was: Something new emerges. People are very excited about it for a while. It seems to make people happy. It’s sort of fun. And then only after years do we look around and are like: Ugh, God, this is creating all these headaches in my life.
A.I. is weird in that from Day 1, the three of us could sit here all day and just talk about why A.I. is going to be bad. We could talk about electricity prices. We could talk about how it’s going to put us out of a job. We could talk about how music is going to be garbage. We could just come up with a list. It’s almost a waste of time to talk about. Any person could come up with a million negative stories about A.I.
So I guess my optimistic take, which is not grounded in something specific that I could point to, is that if you assume that the first snapshot of any technology is wrong, that we’re mistaken, then maybe something emerges with A.I. that’s like: Wow, I could point to things in our lives that are better in a way we can’t articulate yet.
That basically, in many different areas of our lives, we experience the equivalent of a Waymo. Because people get into a Waymo and they’re like: Oh my God, this is insane!
Yeah, it’s genuinely cool.
Weisenthal: This is genuinely incredible, and this car is so smooth, and it is so clean, and it is so awesome — and that the promise would be that there turns out that they’re implicitly the seeds of many other Waymos. We just can’t see them yet.
But whether we’re talking about medicine or anything else, there are other things like that that A.I. will enable. We just don’t quite know what they’re going to be yet.
But if you could just roll out Waymos everywhere tomorrow — which you can’t, but if you could —that would actually put a huge number of people out of work.
Alloway: It’s very hard to navigate that trade-off.
If I were to answer my own question about why you might see a different feeling at the end of 2026, it would be if something has shifted in people’s sense of the politics. There’s a lot of uncertainty, and people want somebody to have a plan.
And right now, you look around the world, and China seems to have a plan. People didn’t trust that Biden had a plan, and he certainly was not able to articulate that, even if his economic policy was quite coherent in what it was attempting to do. And Trump is all over the place.
So I do think there’s something about how, in times of uncertainty, people want clear leadership. And they just don’t have it — and haven’t had it for some time.
Weisenthal: We need a Jed Bartlet.
[Laughs.] The Nobel Prize winning economist on “The West Wing.”
Weisenthal: Someone who has what people feel is statesmanship. It seems very hard to imagine, given the environment, but somehow, you could have someone who has some pretense of statesmanship, purpose, unification, coherence.
If that somehow emerged in this environment — though it’s very hard to see — that might change the way people view the trajectory of the country.
I think that’s a good place to end. Always the final question: What are three books you’d recommend to the audience?
Tracy, why don’t we begin with you?
Alloway: This is very pertinent to our conversation: Dan Wang’s new book, “Breakneck,” is excellent for comparing the political economy of the U.S. and China and a lot of the things we just discussed. This idea of: Why is China able to do some of this faster and seemingly better than the U.S.? That one is great.
The best new fiction I read this year is “North Woods” by Daniel Mason, which is the surreal story about an old house in New England. And Joe knows that I won’t shut up about my house in Connecticut. It’s really good, though.
And then, historical fiction — this one is actually for Joe. I just started reading it: “A Marriage at Sea” by Sophie Elmhirst, which is a true story of a couple in the 1970s who get shipwrecked by —
Weisenthal: Something whaling related?
Alloway: A whale! And survive at sea.
Weisenthal: Oh, that’s great.
Alloway: It’s really good.
Weisenthal: I really had a hard time thinking about which direction I was going to go, because I read “Moby Dick” this year, and it changed my life, and I’ve read a bunch of whaling-related books.
How did it change your life?
Weisenthal: Oh, God. I mean, all I do is think about whales.
[Klein laughs.]
Weisenthal: All I talk about is “Moby Dick” now.
Alloway: This is true.
Weisenthal: And every single story in the economy or whatever, I was like: “OK, he’s like Captain Ahab” — everything I just frame into “Moby Dick.”
But I’ll go in a different direction than the whales.
That’s a strong “Moby Dick” recommendation, though.
Weisenthal: Yes, it’s implicit. Just read “Moby Dick,” people, if you haven’t.
So we talked about how I’m an “It’s the phones” guy. I truly think that the new media environment is fundamentally restructuring and altering society. So there’s a fairly recent book from Andrey Mir, an independent journalist and writer based in Toronto who self-publishes his own books, which is usually a huge red flag, but they’re phenomenal. People should check out his book “The Digital Reversal,” which is about the way digital media flips a lot of things on its head, but also this idea that it seems to be happening at a faster and faster pace — the pace of crises. That’s great.
Then there are two books written several decades ago that I recommend to almost everyone. Walter Ong’s “Orality and Literacy,” which I’ve been talking about a lot. It’s basically the way our communication environment is such that we’re an oral society increasingly, not just by the fact that we literally talk more, as on a podcast, but that everything is back and forth in this, and, therefore, you don’t have this logical contemplation of the person sitting alone in a room actually reading text and judging text on its own merits. He anticipated a lot of changes with social media and the phones, and I think it’s a lot better than reading a lot of contemporary stuff because it doesn’t try to shoehorn contemporary events into a theory. It’s very predictive.
And then another book that I recommend on the same level — it just celebrated its 40th anniversary, so another one that’s sort of pre- this moment: Joshua Meyrowitz’s “No Sense of Place,” which anticipates the way electronic media would dissolve the walls between: This is where you work — and: This is where you live. Or: This is a type of conversation that’s appropriate for one environment but not appropriate for here.
This obliteration of norms from one place to another, I think, has a lot of explanatory power. So yes, “No Sense of Place” by Josh Meyrowitz is my last one.
Joe Weisenthal, Tracy Alloway, thank you very much.
Alloway: Thanks for having us. That was fun.
Weisenthal: Thanks for having us. That was a blast.
You can listen to this conversation by following “The Ezra Klein Show” on the NYTimes app, Apple, Spotify, Amazon Music, YouTube, iHeartRadio or wherever you get your podcasts. View a list of book recommendations from our guests here.
This episode of “The Ezra Klein Show” was produced by Rollin Hu. Fact-checking by Annika Robbins, with Kate Sinclair and Mary Marge Locker. Our senior engineer is Jeff Geld, with additional mixing by Isaac Jones. Our executive producer is Claire Gordon. The show’s production team also includes Marie Cascione, Annie Galvin, Michelle Harris, Kristin Lin, Emma Kehlbeck, Jack McCordick, Marina King and Jan Kobal. Original music by Marion Lozano, Dan Powell and Pat McCusker. Audience strategy by Kristina Samulewski and Shannon Busta. The director of New York Times Opinion Audio is Annie-Rose Strasser. And special thanks to Kimberly Clausing, Natasha Sarin and Kyla Scanlon. Transcript editing by Andrea Gutierrez and Marlaine Glicksman.
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.
The post ‘This is Something that Traditional Economics Isn’t Prepared to Deal With’ appeared first on New York Times.
