5 Debt Hotspots in the AI ​​Data Center Boom

AI For Business


  • AI data center bond issuance surges across multiple markets
  • Even more supply is expected from 2026 onwards.
  • Market participants focus on risks, concerns about AI bubble persist

LONDON (Reuters) – As AI fever drives global stock prices to record highs, companies are increasingly relying on debt to finance the data centers needed to strengthen the technology, raising concerns about the risks.

AI data center and project finance deals have soared to $125 billion so far this year, up from $15 billion in the same period in 2024, and increased supply from the sector is expected to be crucial for credit markets in 2026, according to a UBS report last month.

sign up here.

“Public and private credit appears to be the primary source of funding for AI investments, and their rapid growth is causing some concern,” said Anton Dombrovskis, fixed income portfolio specialist at T. Rowe Price.

“While supply increases have so far matched relatively healthy demand, this is an area to watch, especially given the large capital requirement estimates,” Dombrovskis added.

The Bank of England last week warned that the increasing role of debt in the AI ​​infrastructure boom could raise potential financial stability risks if valuations were revised.

Christopher Kramer, a portfolio manager and senior trader in Neuberger's investment-grade credit team, told Reuters he sees a tectonic shift in the market as the biggest technology companies fund their AI infrastructure ambitions.

“From a debt issuance perspective, these haven't been the focus of our market, but obviously that's changing dramatically… Whenever that happens, it creates a lot of opportunity,” he said on Nov. 28.

“We're excited just from the standpoint that the market is changing. There's going to be a different dynamic. It's going to create opportunities to take risks and create value for investors,” Kramer added.

Here are five charts showing how debt is funding the AI ​​space race.

1) Oracle: CDS surge reflects investor concerns

Oracle (ORCL.N)opens a new tab Stocks fell about 11% on Thursday, the biggest single-day decline since January. The company's big spending and weak outlook fueled doubts about how quickly big bets on AI would pay off, triggering a broad selloff in tech stocks.

Technology company executives, whose companies have long relied on strong cash flow to fund spending on new initiatives, say this spending is needed for technology that transforms work and makes business more efficient, and say the bigger risk is underinvestment rather than overspending.

On the back of a $300 billion deal with OpenAI, Oracle stock has nearly doubled in value since the beginning of the year to its peak in September. However, it has since fallen 42%.

In September, US credit rating agency Moody's warned of some potential risks to Oracle's new contracts, but took no rating action.
Against a broader backdrop of increased issuance of AI bonds, Oracle's debt levels have been a focus for investors, with credit default swaps (CDS), a type of insurance against default, closing at their highest level since 2009 on Thursday, according to S&P Global data.
Boaz Weinstein's Saba Capital Management has sold credit derivatives in recent months to lenders seeking protection such as Oracle and Microsoft (MSFT.O).opens a new tabReuters reported last month.
Oracle stock performance as a percentage against CDS for five years starting December 2024
Oracle stock performance as a percentage against CDS for five years starting December 2024

2) Rapid increase in AI investment-grade borrowing

The investment grade (IG) bond market has seen a large influx of high-tech debt issuance in recent months. Big deals in September and October included $18 billion from Oracle and $30 billion from Meta. Google's owner Alphabet also announced new debt.

JPMorgan estimates that AI companies make up 14% of its investment-grade index, outpacing U.S. banks as the leading sector.

But Big Tech deals represent only a fraction of the roughly $1.6 trillion in U.S. IG bond issuance expected in 2025.

Clustered bar chart showing year-to-date breakdown of tech bond issuance over the past three years in various U.S. bond markets
Clustered bar chart showing year-to-date breakdown of tech bond issuance over the past three years in various U.S. bond markets

3) Increase in AI-related “high-yield” bonds

The high-yield bond market is also seeing AI-linked issuance, offering higher returns to investors despite lower credit ratings for issuers.

Overall, junk tech bond issuance is at an all-time high, according to data from Dealogic.

Al Cattermole, fixed income portfolio manager and senior analyst at Mirabeau Asset Management, said as of November 25, his team had not invested in any of the AI-related IGs or high-yield bonds that have recently come to market.

“Until the data center is delivered on time, on budget, and provides the computing power for which it was intended, and there is still demand for it, it will not be tested,” Cattermole told Reuters.

“And since it's untested, I think it needs to be compensated like equity rather than debt,” he added.

Stepped bar chart showing annual issuance of high-yield tech junk bonds from 1995 to 2025. In the second half of 2025, there will be a sharp increase comparable to the peak in 2021.
Stepped bar chart showing annual issuance of high-yield tech junk bonds from 1995 to 2025. In the second half of 2025, there will be a sharp increase comparable to the peak in 2021.

4) Increasing role of private credit in AI financing

Private credit, provided by investment firms and others rather than banks, is also funding AI data centers.

UBS estimates that private credit AI loans could have nearly doubled in the 12 months to early 2025.

Morgan Stanley estimates that private credit markets could provide more than half of the $1.5 trillion needed to build data centers by 2028.

A pie chart that breaks down Morgan Stanley's estimate of how to finance data center expansion worldwide when it identified an $800 billion private credit opportunity.
A pie chart that breaks down Morgan Stanley's estimate of how to finance data center expansion worldwide when it identified an $800 billion private credit opportunity.

5) ABS transformation?

Securitized products such as asset-backed securities (ABS) can also help fund the growth of the AI ​​industry, according to Morgan Stanley.

These are illiquid assets such as loans, credit card debt, or, in the context of AI, rent paid by Big Tech tenants to data center owners, packaged into tradable securities.

Digital infrastructure accounts for only 5%, or $82 billion, of the roughly $1.6 trillion U.S. ABS market, but BofA notes that the market has expanded more than nine times in less than five years. The company estimates that data centers support 63% of that market, and expects to add $50 billion to $60 billion in supply by 2026.

ABS has been viewed with caution since the 2008 crisis, when billions of dollars worth of products were found to be backed by bad debts and highly illiquid and complex assets.

Stacked bar chart showing the breakdown of the US ABS digital infrastructure securitized products market.
Stacked bar chart showing the breakdown of the US ABS digital infrastructure securitized products market.

Report by Lucy Raitano. Editing: Amanda Cooper, Darla Ranasinghe, Alexander Smith

Our standards: Thomson Reuters Trust Principles.opens a new tab



Source link