Wall Street Races to Cut Its Risk From AI’s Borrowing Binge

Wall Street is gearing up to lend massive amounts of money to the biggest players in artificial intelligence — and simultaneously trying to figure out how to protect itself from any bubble that its financing may be helping to inflate.

The urgency at banks to shed risk is visible all over credit markets. The cost of protecting Oracle Corp. debt against default using derivatives has risen to the highest since the Global Financial Crisis. Morgan Stanley has looked at using a significant risk transfer — a form of insurance against loan losses — to diffuse some of the risk tied to its tech company borrowers.

Mega offerings from tech behemoths including Oracle, Meta Platforms Inc. and Alphabet Inc. have helped push global bond issuance to more than $6.46 trillion in 2025. These hyperscalers, along with electric utilities and other firms, are expected to spend at least $5 trillion as they race to build data centers and other infrastructure for a technology that’s promising to revolutionize the world’s economy.

BB Credit fear

The sums are so large that issuers will have to tap just about all major debt markets, according to JPMorgan Chase & Co. It could take years for these tech investments to pay off — if they pay off at all. The rush has left some lenders over-exposed, so they’re using a series of tools — credit derivatives, sophisticated bonds and some newer financial products — to shift the risk of underwriting the AI boom to other investors.

Real Risks

“The technology is impressive. But that doesn’t mean you’re going to profit from it,” said Steven Grey, chief investment officer at Grey Value Management.

Those risks became more real last week when a major outage halted trading at CME Group Inc. and reminded investors that data-center customers can leave if there are repeated breakdowns. In the aftermath, Goldman Sachs Group Inc. paused a planned $1.3 billion mortgage bond sale for CyrusOne, the data-center operator.