The AI party has a tab
Paul Krugman’s latest warning is basically: the AI boom isn’t just a shiny growth story, it may be muscling other investments out of the room. His point is that when hyperscalers pour a mountain of borrowed money into data centers, credit demand rises, rates can climb, and somebody else’s project gets shoved to the back of the line.
That’s a lot less fun than the usual “AI will change everything” pitch deck. But it’s also the kind of thing investors should care about because higher rates don’t just annoy homebuyers — they can make new factories, office builds, and other capital projects more expensive too.
Not your 1990s internet boom
Krugman’s twist is that this time may not come with the same foreign cash cushion the dot-com era had. Back then, America sucked in a wave of overseas investment. Now, he says, the balance-of-payments picture doesn’t show the same magical money hose.
The numbers in the piece help explain why the warning is getting louder:
- JPMorgan’s Jamie Dimon expects hyperscalers to spend nearly $1 trillion next year
- About $220 billion of debt has already been issued by five of the biggest hyperscalers this year
- Goldman Sachs says lower-rated AI-related borrowing has reached $88 billion
That’s not just “big spending.” That’s “the credit market is working overtime” spending.
What it means for you
This doesn’t mean AI is a bubble tomorrow or that the sector is doomed. It does mean the boom may have spillovers that show up far outside Silicon Valley — in rates, in construction, and in the opportunity cost of all that capital.
Big picture: even if AI keeps winning, the rest of the economy may be paying part of the bill.