AI Debt Explosion Has Traders Searching for Cover: Credit Weekly

jmvntifid)1vp]lt1pqu}5w1_media_dl_1.png ICE Data Services

Article content

(Bloomberg) — As tech companies gear up to borrow hundreds of billions of dollars to fuel investments in artificial intelligence, lenders and investors are increasingly looking to protect themselves against it all going wrong. 

Article content

Banks and money managers are trading more derivatives that offer payouts if individual tech companies, known as hyperscalers, default on their debt. Demand for credit protection has more than doubled the cost of credit derivatives on Oracle Corp.’s bonds since September. Meanwhile, trading volume for credit default swaps tied to the company jumped to about $4.2 billion over the six weeks ended Nov. 7, according to Barclays Plc credit strategist Jigar Patel. That’s up from less than $200 million in the same period last year.   

Article content

Article content

Article content

“We’re seeing renewed interest from clients in single-name CDS discussions, which had waned in recent years,” said John Servidea, global co-head of investment-grade finance at JPMorgan Chase & Co. “Hyperscalers are highly rated, but they’ve really grown as borrowers and people have more exposure, so naturally there is more client dialogue on hedging.”

Article content

Article content

A representative for Oracle declined to comment. 

Article content

Trading activity is still small compared with the amount of debt that is expected to flood the market, traders said. But the growing demand for hedging is a sign of how tech companies are coming to dominate capital markets as they look to reshape the world economy with artificial intelligence. 

Article content

Investment-grade companies could sell around $1.5 trillion of bonds in the coming years, according to JPMorgan strategists. A series of big bond sales tied to AI have hit the market in recent weeks, including Meta Platforms Inc. selling $30 billion of notes in late October, the biggest corporate issue of the year in the US, and Oracle offering $18 billion in September.   

Article content

Tech companies, utilities, and other borrowers tied to AI are now the biggest part of the investment-grade market, a report last month from JPMorgan shows. They’ve displaced banks, which were long the biggest portion. Junk bonds and other major debt markets will see a wave of borrowing too, as firms build thousands of data centers globally. 

Article content

Article content

Some of the biggest buyers of single-name credit default swaps on tech companies now are banks, which have seen their exposure to tech companies surge in recent months, traders said. 

Article content

Another source of demand for the derivatives: equity investors looking for a relatively cheap hedge against the shares dropping. Buying protection on Friday against Oracle defaulting within the next five years cost about 1.03 percentage point, according to data provider ICE Data Services, or around $103,000 a year for every $10 million of bond principal protected. In contrast, buying a put on Oracle’s shares falling almost 20% by the end of next year might cost about $2,196 per 100 shares as of Friday, amounting to about 9.9% of the value of the shares protected. 

Article content

There is good reason for money managers and lenders to at least look at cutting exposure now: An MIT initiative this year released a report indicating that 95% of organizations are getting zero return from generative AI projects. While some of the biggest borrowers now are companies with high cash flow, the technology industry has long been fast changing. Firms that were once big players, such as Digital Equipment Corp., can fade into obsolescence. Bonds that seem safe now may prove to be considerably riskier over time or even default, if profits from data centers fall short of companies’ current expectations, for example.