(Photo: Michael M. Santiago/Getty Images)
My mind’s been stuck on one question lately: What’s the biggest risk to the economy? There seem to be a ton of issues at the moment.
Inflation is eating away at pay raises; persistently high energy prices are fueling price growth; the Iran war is driving much of that increase; consumer sentiment has been weak; and the labor market is holding up, mostly.
But when I asked Morgan Stanley chief U.S. economist Michael Gapen this question on Monday, he didn’t pick any of them.
I think the biggest risk to the economy would be a sudden, sharp slowdown in the rate of AI spending, or the market calling into question the sustainability of AI spending. Both would probably happen simultaneously. That would lead to likely a decline in equity valuations and spilling over into the consumer.
Obviously, if we were having this conversation in March, it would have been geopolitics.
Geopolitics is still a factor, but I think most of us are now looking at how much the U.S. and global growth backdrop is anchored by AI spending.
So, if that's what's giving us the resiliency to geopolitics and everything else, then getting rid of that resiliency is probably the bigger risk.
Translation: With so much of the economy on shaky ground, AI spending has become its main source of support — and its biggest risk.
Warning signs
The trigger for a sudden, sharp slowdown could be anything:
An earnings miss that calls profitability into question. Politics. Bottlenecks. Or “an AI bot gone rogue,” Gapen said. “You pick your scenario.”
And if you’ve been watching the stock market or paying attention to Q3 tech earnings, you know that investors are already concerned about the sustainability of AI spending.
Three of the biggest tech companies in the world — Amazon, Google parent Alphabet and Meta — all reported spending more cash on data centers and other capex than they made from running their businesses.
That news sparked a tech stock selloff last week that dragged down even shares of Microsoft, which reported positive free cash flow of nearly $19 billion.
Then came Thursday’s selloff, which erased $2.5 trillion of market value from the Nasdaq. (Some investors appeared to be locking in profits, while others fear that a Democratic Congress could slow data center builds.)
“It just confirms my view that [AI spending] is the bigger risk to the economy,” Gapen said in a follow-up email on Friday.
Lastly, I’ve been particularly curious about the potential risks of recent AI financing.
2026 is the first year Big Tech has had to finance much of its AI spending with outside money rather than free cash flow, Gapen pointed out.
The shift into equity, debt and private credit markets is “the initial signs of potential financial contagion down the road,” he said.
And what about the off-balance-sheet deals, like the special purpose vehicles Broadcom set up with Apollo and Blackstone to finance Anthropic’s chips?
Having worked at the Federal Reserve during the global financial crisis, references to off-balance-sheet special purpose vehicles raises concern.
That said, such financing vehicles during the financial crisis had links directly back into the financial system and that linkage was responsible for creating adverse feedback loops in the economy.
One difference today is these off balance sheet financing vehicles are in private credit markets, with fewer direct linkages to the traditional banking system.
We’ll see if that matters for economic stability, but the pattern appears different.
How it spreads
Any shock would hit stocks first, then the wealthiest households, then consumer spending — the top driver of the economy.
The first two ripples may already be starting:
Sentiment among consumers with the largest stock portfolios fell 5% in early November, the University of Michigan reported on Friday.
For context: The richest 10% of households own close to 90% of stocks; stocks also made up a record 48% of U.S. financial assets as of Q2.
“That's why I argue you get both of these things: AI spending and upper income household spending, or you get neither,” said Gapen.
The more probable timeline
While Big Tech will keep spending more on AI in the coming months, the boost that spending adds to economic growth may be peaking now, according to Gapen.
And while the growth rate of hyperscaler capex is expected to slow over time, that’s not a “bad thing … if it reduces the risk of an AI boom followed by an AI bust.
“Elongating the timeline for capex spending could mean it is executed more efficiently,” he added.
Actions to consider
For everyone: Be prepared for AI spending to remain the main story and the biggest risk in 2027 and 2028.
For business leaders: Watch cost, risk, and compliance. The question is “how to utilize it as an input without excessively spending on it.”
For workers: “Be prepared for a very different labor market.” Skills and occupations will change, and so will how we work.
Hope’s macro view
In the 1991 Steven Spielberg film “Hook,” Tinker Bell (Julia Roberts) tells a grown-up Peter Pan (Robin Williams) that to fly, he has to hold on to a happy thought. His was the birth of his son.
For the economy right now, that happy thought is AI spending: It doesn’t have to disappear for us to fall — just fade.
More Rising Tides:
Subscribe to Macro Talk to read the rest.
Become a paying subscriber of Macro Talk to get access to this post and other subscriber-only content.
Upgrade
