Under-construction data centers tap ratings to fund AI buildup

The AI arms race is rapidly spilling into credit markets, as data center developers are now seeking credit ratings for projects still under construction — a sign of how urgently the tech sector needs capital to fund its multi-hundred-bn USD infrastructure expansion, the Financial Times reports. Back in 2021, tech giants self-funded roughly 80% of their data center development using their strong balance sheets. Now, however, 60% of developments involve third-party institutional capital.

To unlock these massive pools of capital, developers are rushing to secure investment-grade stamps of approval from agencies like S&P, Moody’s, Fitch, and KBRA. “It’s astronomical growth,” Fitch’s Roelof Steenekamp told the salmon-colored paper. Fitch alone evaluated over 35 private data center agreements in just nine months, averaging USD 3 bn per transaction, with the vast majority being newly constructed, hyperscaler-backed facilities. Moody’s and Fitch have also privately assigned investment-grade ratings to tens of bns of USD in loans tied to projects backed by Oracle Corporation. KBRA currently rates nearly USD 100 bn in data center debt and expects the figure to rise to up to USD 150 bn by mid-year.

Lenders are utilizing “credit tenant lease financing,” which essentially caps the project’s credit risk at the rating of its anchor tenant. For instance, S&P handed an A+ rating to USD 27 bn in debt for Meta’s Hyperion data center in Louisiana. To get the agreement done, Meta provided financial assurance, promising to begin paying rent even if construction is delayed, while also agreeing to cover any budget overruns. In many cases, lenders are underwriting the hyperscaler’s balance sheet more than the developer’s, according to S&P’s Dhaval Shah.

Still, the structure carries risks. Unlike traditional cloud data centers, AI-training facilities are often built in remote locations and can strand assets if a tenant leaves. Furthermore, the AI economy relies heavily on circular financing — hyperscalers fund startups that simply use the money to buy tech directly back from them. With rising costs and lagging end-user revenues, economists warn this fragile loop could trigger a massive industry-wide debt crisis, according to a report (pdf) from the Center for Public Enterprise.

The sheer size and opacity of these financing vehicles are beginning to attract Washington’s attention. Recently, lawmakers, including Senator Elizabeth Warren, have formally urged the Financial Stability Oversight Council to probe the “complex and opaque” structures funding data centers, Bloomberg reports. Warren warned that if AI companies fail to rapidly increase revenues to service their massive debt loads, it could trigger destabilizing losses across interconnected financial institutions.

MARKETS THIS MORNING-

It is shaping up to be another turbulent day for markets, with the tech selloff over on Wall Street and uncertainty surrounding US President Donald Trump’s tariff policy keeping investors on their toes. Things are looking better for Asia-Pacific markets, with most of them in the green in early trading this morning.

THE CLOSING BELL-

The EGX30 rose 2.6% at yesterday’s close on turnover of EGP 5.7 bn (8.6% below the 90-day average). Local investors were the sole net sellers. The index is up 21.6% YTD.

In the green: Rameda (+9.5%), Heliopolis Housing (+9.5%), and Raya Holding (+6.8%).