Marketplace®

Daily business news and economic stories

If the AI bubble were to burst, these terms will be oft heard

Here are three AI bubble terms you may want to familiarize yourself with now. You know, just in case.

Download
What might be the "mortgage-backed securities" or "credit default swaps" of a potential AI bubble burst?
What might be the "mortgage-backed securities" or "credit default swaps" of a potential AI bubble burst?
Michael M. Santiago/Getty Images

Before the great financial crisis, did you know what a mortgage-backed security was? A credit default swap? Had you ever heard of Lehman Brothers?

Every economic bubble has its own vernacular. The esoteric jargon and financial terms of art and even names of companies that, when the bubble is inflating, become the lingua franca of Wall Street.

But when bubbles pop, this language forces its way into common vocabulary.

The artificial intelligence bubble has not yet popped. In fact, there is plenty of debate about whether we’re even in a bubble. But many experts will confess to some trepidation about how much our economy is tied to AI, and the potentially disastrous consequences if this latest tech revolution were to go south.

Here are three AI bubble terms you may want to familiarize yourself with now. You know, just in case.

1. Special Purpose Vehicle

The cool kids call them SPVs.

Even very rich tech companies sometimes need to borrow money to build an AI data center. So when Meta announced earlier this year it was building a nearly $30 billion data center in Louisiana, it wasn’t surprising there was lots of debt involved.

“Now this isn't Meta borrowing the money,” said Gil Luria, who heads up technology research at investment firm D.A. Davidson.

Technically, it’s a special purpose vehicle borrowing the money. A separate legal entity Meta co-created with a private investment firm. Meta is agreeing to rent the data center.

“So that would be like you buying a house, but not taking the mortgage yourself, but having somebody else take the mortgage, which, from your perspective, is great,” Luria said.

But from a broader economic perspective, maybe not so great. The fraudulent energy company Enron was a big fan of special purpose vehicles. Now, there’s nothing inherently illegal about SPVs. But if demand to use that Louisiana data center falls short of expectations?

“It's Meta borrowing money without borrowing money, and at the end of the day, somebody's going to be left holding the bag,” said Luria.

2. Private Credit  

Private credit is basically investment firms like hedge funds and private equity lending money to businesses, like a bank would. Except they’re not regulated like banks.

“I don't even think Americans are necessarily unfamiliar with this concept,” said 

Advait Arun, capital markets and energy finance analyst at the Center for Public Enterprise. “We might have even heard of it previously as a shadow bank.”

That data center in Louisiana Meta is building, with that special purpose vehicle? A private credit firm is throwing in billions, which is the case for a lot of AI financing.

“These chips that companies like Nvidia make, they help those tech companies buy those chips, and they help the data center developers take out long term debt or construction debt to build the data centers themselves,” Arun said. 

If the AI bubble goes pop, private credit will be hit hard. That’s bad news for the pension funds whose money private credit firms invest, but it could also be bad news for actual, non-shadow banks.

“Banks are actually funding private credit and are getting some amount of returns from private credit,” Arun said. “And private credit uses banks’ funding to go and raise other debt or other capital, and goes and makes their risky bets.” 

3. Neocloud

Neocloud is a term that people might come to know,” said Paul Kedrosky, a fellow at the MIT Institute for the Digital Economy. 

They’re tech companies that basically rent out AI chips to tech firms, and often take out loans against the chips they’re renting.

“These neocloud providers have very high debt loads, much of it at high yields, much of it has to be turned over on a four year, five year basis. And so they’re a very risky bunch,” Kedrosky said.

CoreWeave is probably the most famous one. It started as a cryptocurrency company.

“The analogy I sometimes make is they're like some of the more aggressive mortgage providers during the financial crisis who showed up to do only ninja loans, no income, no job, no asset loans,” said Kedrovsky.

The price of insurance against CoreWeave defaulting on its debt went up nearly 60% in two months.

That insurance? It has a familiar name: credit default swaps.

Remember those? 

Related Topics

Collections:

Latest Episodes

View All Shows
  • Marketplace
    3 hours ago
    25:38
  • Make Me Smart
    4 hours ago
    17:42
  • Marketplace Morning Report
    10 hours ago
    6:34
  • Marketplace Tech
    15 hours ago
    10:32
  • This Is Uncomfortable
    8 days ago
    56:05
  • Million Bazillion
    a month ago
    32:45