AI’s data-centre gold rush is drowning in debt

As AI fever has propelled global stocks to record highs, the data centres needed to power the technology are increasingly being financed with debt, adding to concerns about the risks.

A UBS report last month said AI data centre and project financing deals surged to $125 billion so far this year, from $15 billion in the same period in 2024, with more supply from the sector expected to be pivotal for credit markets in 2026.

"Public and private credit seems to have become a major source of funding for AI investments, and its rapid growth raised some concerns," said Anton Dombrovskiy, fixed income portfolio specialist at T. Rowe Price.

"Although up until now an increase in supply has been met with relatively healthy demand, this is the area to watch especially taking into account large financing needs estimates," Dombrovskiy added.

The Bank of England warned last week that the growing role of debt in the AI infrastructure boom could heighten potential financial stability risks if valuations correct.

Christopher Kramer, portfolio manager and senior trader on the investment grade credit team at Neuberger told Reuters that the market has seen a structural shift as the largest technology companies finance their AI infrastructure ambitions.

"They really haven't been focal points in our market from a debt issuance standpoint, and that's obviously shifting really dramatically ... anytime you have that, it creates a lot of opportunity," he said on November 28.

"We're excited just from the standpoint that the market's changing. You're going to have a different dynamic, it creates an opportunity to take risks and create value for our investors," Kramer added.

Here are five charts that show how debt is increasingly funding AI's race for space.

1) ORACLE: CDS SPIKE REFLECTS INVESTOR CONCERN

Oracle shares fell almost 11% on Thursday, their biggest one-day drop since January, sparking a broader tech selloff as its massive spending and weak forecasts fanned doubts over how quickly big bets on AI will pay off.

Tech executives, whose companies long depended on strong cash flows to fund spending on new initiatives, have said the outlays are necessary for a technology that will transform work and make businesses more efficient, arguing the bigger risk is underinvesting, not overspending.

At their peak in September, Oracle shares had almost doubled in value year-to-date on the back of a $300 billion deal with OpenAI. But they have since fallen 42%.

In September, US credit rating agency Moody's flagged several potential risks in Oracle's new contracts, but stopped short of taking any ratings action.

Oracle's debt levels have been a focal point for investors, against a broader backdrop of more AI debt issuance and its credit default swaps(CDS), a form of insurance against default, closed Thursday at their highest level since 2009, according to data from S&P Global.

Boaz Weinstein's Saba Capital Management sold credit derivatives in recent months to lenders seeking protection on the likes of Oracle and Microsoft, Reuters reported last month.

2) AI INVESTMENT GRADE BORROWING SURGES

The investment grade (IG) debt market has seen a huge influx of tech issuance in recent months. Mega deals in September and October included $18 billion from Oracle and $30 billion from Meta. Google owner Alphabet also announced new borrowing.

JP Morgan estimates AI-linked companies account for 14% of its investment grade index, surpassing US banks as the dominant sector.

But Big Tech deals still only account for a fraction of the nearly $1.6 trillion in US IG debt issues expected in 2025.

3) MORE AI-RELATED 'HIGH YIELD' BONDS

There has also been AI-linked issuance in the high-yield debt market, where issuers have lower credit ratings yet investors are offered higher returns.

Overall, junk tech bond issuance is at a record high, data from Dealogic shows.

Al Cattermole, fixed income portfolio manager and senior analyst at Mirabaud Asset Management, said that as of November 25, his team had not invested in any of the AI-linked IG or high-yield bonds that had recently hit the market.

"Until we see data centres being delivered on time and on budget and providing the computing power that they are intended to - and there still being the demand for it - it is untested," Cattermole told Reuters.

"And because it's untested, that's why I think you need to be compensated like an equity ... not debt," he added.

4) PRIVATE CREDIT'S INCREASING ROLE IN AI FUNDING

Private credit - extended by the likes of investment firms, rather than banks - is also funding AI data centres.

UBS estimates private credit AI loans may have nearly doubled in the 12 months through early 2025.

Morgan Stanley estimates private credit markets could supply over half the $1.5 trillion needed for the data centre build out until 2028.

5) ABS MAKEOVER?

Securitised products, such as asset-backed securities (ABS), will also help fund AI industry growth, says Morgan Stanley.

They bundle together illiquid assets such as loans, credit card debt, or - in AI context - rent payable to a data centre owner by a Big Tech tenant, into a tradable security.

While digital infrastructure accounts for just 5%, or $82 billion, of the roughly $1.6 trillion US ABS market, BofA notes it has expanded more than nine fold in less than five years. It estimates that data centres backed 63% of that market, which it expects to add $50-$60 billion in supply in 2026.

ABS are viewed with caution since the 2008 crisis when billions of dollars worth of products turned out to be backed by soured loans and highly illiquid and complex assets.

Unlock more exclusive Cybernews content on YouTube.