Bloomberg Surveillance Sep 21, 2026
With Alexis Crow, Partner and Chief Economist at PwC
Members of the Federal Reserve's own committee flagged AI spending as a source of strong demand and inflation, in minutes released in July.
The argument about AI capital spending usually runs as a question about whether the companies doing it earn a return. Alexis Crow's version is different: it is about whether the equipment being bought will still be useful, and about what the building does to prices while it happens.
"They're worried about an unwinding of the AI trade. That's definitely on the table."
Crow is a partner and chief economist at PwC and teaches the macro section of a course at MIT's Sloan School alongside Simon Johnson, who she says still uses chalk and asks students to keep their laptops shut.
The full segment is covered here so you can skip it.
Here are the 6 arguments that matter.
Key Takeaways
The Fed's own June minutes name AI spending as a source of demand and inflation, and say it is spilling into other sectors
Clients are worried about an unwinding of the AI trade, not about whether the technology works
Europe's equity market is financials and pharmaceuticals, with energy added since the Iran conflict
It also runs a large trade surplus, which the US does not
The risk in data centers is obsolescence, not demand — she is not sure the equipment will still be useful, given Moore's law
She sees no moat around AI, agreeing with Steve Eisman's version of the argument
The Bank of England's governor has described a bubble that bursts while the technology keeps paying, and she thinks both can happen at once
1. Chalk At MIT Sloan
Tom Keene introduced Crow through her teaching rather than her day job, asking what it is like to pick up chalk at MIT's Sloan School with a Nobel laureate watching.
Simon Johnson in a classroom is unlike anyone she has seen
There's nobody like Simon Johnson in the classroom. I'll tell you that much, really. And I've seen some greats in my time. And there's nobody like him.
Alexis Crow
Her own part of the course is the macro context for case studies students then do in emerging markets and in Europe. The detail she found remarkable was the teaching method.
Chalk, and no laptops
As you say, he actually still uses chalk, and that's pretty remarkable, requesting students not to use their laptops.
Alexis Crow
2. Old World, New World
Keene put his standing question to her: the United States lets innovation fail and lets markets clear, Europe does not, so how does Europe become more Anglo-American against an older culture. Crow answered with organizational shape.
The difference is hierarchy, and it is older than any policy
One world was Catholic and very hierarchical. The other world was Anglo-Saxon.
Alexis Crow
Keene accused her of blaming the Reformation. She replied that this was spoken like a true Catholic.
3. What Europe Actually Has
Paul Sweeney said the European market trades at a discount because it carries no technology weight, that Europe missed the last thirty or forty years of technology, and that he does not see this cycle going differently — so is Europe simply a net taker of AI, and is it comfortable with that? Crow told him to look at what the market actually holds.
Europe's index is financials and pharmaceuticals, and always has been
Equity market performance in Europe has historically been financials and pharma. Pretty one-sided, to be honest.
Alexis Crow
Two things have been added to that. Energy has gained since the Iran conflict, and industrial production is still there, which produces the surplus she pointed at next.
A large trade surplus, which the US does not have
Hey, guess what? We don't have that in the U.S., despite all the efforts to the contrary.
Alexis Crow
She expects the infrastructure spending tied to defense and aerospace to be the interesting thing to watch in Europe. On AI itself, her point was about appetite rather than capability.
Europe is ambivalent about general-purpose technology
I also, by the way, think that there's an ambivalence, there's a cultural ambivalence toward some of these general purpose technologies in Europe and whether or not these are going to benefit citizens overall.
Alexis Crow
4. The US Cocktail
Sweeney asked for a six- to twelve-month economic call given higher rates worldwide, higher energy costs worldwide and still-strong growth. Crow listed what is mixed together in the US.
Record debt issuance, a changed buyer base, bank deregulation and hedge funds
So here in the US, we definitely have this interesting conflagration of record high levels of debt issuance, major changes in the composition of ownership in the bond market and shifts there.
Alexis Crow
She added deregulation in traditional banking and a deepening relationship between banks and hedge funds to the same list. Growth is still driven by AI and the capital spending behind it, but in a higher interest rate environment that puts pressure on some corporate borrowers.
The share of that build-out funded with debt has jumped
And you look at this share of debt that's being used to fund some of these hyperscaler plays and some of the data center plays, and that's dramatically increased in the last year as well.
Alexis Crow
5. Unwinding The AI Trade
Keene asked whether the people she talks to at PwC have settled into the strong nominal GDP story or are worried about the list she had just read out. Her answer was one sentence.
What they fear is the trade unwinding
They're worried about an unwinding of the AI trade. That's definitely on the table.
Alexis Crow
She then set current spending against previous technology booms using Federal Reserve data: software and hardware spending as a share of GDP looks like past cycles, and data center spending is the line that does not. Keene interrupted to say he had not heard any of this in the Fed meeting or in the economics around it. Crow said it is in the record.
The committee has already said it in writing
In the June meeting minutes that were released in July, you did see some of the committee members concerned that spending on AI was contributing to very strong demand and inflation here in the US and spilling over from the AI sector into other related sectors as well.
Alexis Crow
Sweeney said he was starting to get nervous about what AI means for the economy, corporate earnings and equity valuations, and asked how it could end badly. Crow's answer was about the buildings.
The question is whether the equipment is still useful later
Are we developing things that will be kit for the future and equipment for the future? I'm not entirely sure, given Moore's law.
Alexis Crow
And there is a political leg to it as well
You've obviously also seen the political risk against this and potential for a snapback there as well.
Alexis Crow
6. A Bubble And A Payoff
Keene brought up Steve Eisman's interview the day before, which turned on the absence of moats around AI businesses, and said he does not see one. Crow said she does not see one either.
He then asked how the market clears, invoking Friedrich Hayek: at some point it has to, and there will be losers. Crow's answer was that the loss and the payoff are not mutually exclusive.
A bursting bubble and continued investment can happen together
Well, it's interesting. We've chatted about this in the past, but Governor Bailey at the Bank of England has said he sees a scenario where you could have a bubble bursting, but companies continuing to use AI and to invest in AI.
Alexis Crow
Her own view is that both unfold at once, with the investment continuing because it raises productivity.
Bonus Insights
The rest of the hour
Crow took the third segment of the program. The same hour carried Torsten Slok of Apollo Global Management on the AI-driven economy, Meghan Robson of BNP Paribas on credit, Dana Telsey of Telsey Advisory Group on retail and Bruce Wolfe of Alight on retirement income.
The moats question came from another network
Keene's reference was to Steve Eisman's appearance on a rival network the previous day, which he flagged as the source of the no-moat framing rather than presenting it as his own. Keene also said Eisman has been unwell recently and that it was good to see him back.
Crow's bottom line is that the risk in the AI build-out is not whether the technology works but whether the physical equipment holds its value and how much inflation the building creates on the way, and that the cleanest description of the outcome she has heard is the Bank of England's: the bubble bursts and the investment carries on.
Products, Companies & Tools Mentioned
PwC (Her firm. She is a partner and its chief economist, and the client conversations she reported are about an unwinding of the AI trade)
MIT Sloan School of Management (Where she teaches the macro context for the course's case studies, alongside Simon Johnson)
The Federal Reserve (Its June meeting minutes, released in July, record committee members worried that AI spending was adding to demand and inflation)
The Bank of England (Governor Bailey's scenario, which she endorses: a bubble bursting while companies keep using and investing in AI)
Listen to the full episode
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: