Carlyle warns: Private credit is vying for trillion-dollar AI infrastructure financing, and concentration risk could repeat the software loan crisis.

Carlyle Group said private credit institutions are competing to finance artificial intelligence (AI) infrastructure buildout, and could repeat the credit concentration exposure seen in the software industry. A white paper published by Carlyle on Thursday noted that the industry may need to provide about $1 trillion in funding to finance AI computing infrastructure. That scale is equivalent to more than half of current total private credit assets under management. The white paper said that failing to set clear limits on concentration in the AI computing sector could become "the biggest mistake." Mark Jenkins, Carlyle's co-president and head of global credit and insurance, said in an interview: "We are in a period where the revenue model remains uncertain so far. In such an environment, as credit investors, it is hard for us to say 'okay, we're all in.'" Private credit managers are increasingly being asked to finance the massive expansion of AI infrastructure. It is estimated that related capital expenditures are expected to exceed $5 trillion by 2030. Financing forms vary widely, including data center construction and power financing, loans secured by chips that support the technology, and loans to special purpose vehicles. The white paper noted that, unlike software, the credit risk of data centers and other AI-related assets is more speculative and more likely to be correlated with overall economic trends, while many of the financing structures being used remain largely untested. For Carlyle, this does not mean avoiding AI investment. Jenkins said: "We want to take risk, but we want to take it in a balanced way." He said one of the biggest challenges facing lenders is that it remains unclear where AI's ultimate profits will accumulate whether with chipmakers, data centers, or application development companies. The white paper showed that the software industry experienced a similar boom between 2020 and 2022, accounting for about half of private equity deals during the same period. Lenders rushed into software companies, partly because their recurring subscription revenue was seen as stable and relatively less susceptible to recessions. But the rise of generative AI challenged that assumption, exposing software companies to a common threat of technological obsolescence. Since then, software loans have struggled in the syndicated loan market, borrowers have found refinancing difficult, and some private credit funds have encountered increased redemption requests. Jenkins believes AI infrastructure financing is repeating a similar lesson: what appears on the surface to be diversified financing projects may ultimately concentrate underlying funding in a handful of leading companies. He observed that the vast majority of underlying financing in the market is concentrated in seven or eight high-quality targets. He said it is particularly important to understand the ultimate counterparty, the contracts supporting the financing, and the value of the underlying assets. Jenkins said: "As investors, people need to think very, very carefully about what your counterparty risk exposure is, how the contract terms are written, and what the ultimate asset value is. In a crisis scenario, all of these will be crucial. And when everything is going well, they seem irrelevant."