The Bank for International Settlements published a bulletin on business development companies' exposure to software firms in US private credit, highlighting a large concentration alongside limited signs of current market stress. The bulletin finds that BDCs have lent about USD 115 billion to software companies, equal to roughly one fifth of all BDC lending and more than 80% of their technology portfolios, while generative AI has increased uncertainty around those borrowers' future revenues without yet changing loan pricing or investor valuations. By late 2025, credit spreads on software, other information technology and non-information technology loans had broadly converged, with the sharpest compression in newly issued loans and in loans originated by non-listed BDCs. Reported credit quality remains benign, with less than 1% of software loans behind on payments and no clear deterioration in other backward-looking measures, but the bulletin says these metrics may lag underlying risk and that narrower spreads leave less protection if losses rise. The exposure is also concentrated and interconnected: about 60% of software lending now goes to firms borrowing from seven or more BDCs, while the five largest BDCs hold around 37% of all software loans and the 10 largest hold more than half. The bulletin adds that some structural features could limit spillovers, including low statutory leverage, a high share of senior secured lending and the closed-end funding model of listed BDCs. At the same time, it points to vulnerabilities in non-traded BDCs with semi-liquid redemption structures and to the broader opacity of private credit beyond BDCs, which could make any AI-related repricing or credit shock spread more widely than current metrics suggest.
Bank for International Settlements2026-07-14
Bank for International Settlements examines USD 115 billion of BDC software lending, finds AI disruption risk is not yet reflected in pricing
The Bank for International Settlements said BDCs have about USD 115 billion of exposure to software firms, but generative AI-related revenue risk is not yet being differentiated in loan spreads or BDC equity valuations. Spreads have narrowed, reducing loss buffers, while exposures are concentrated across a small number of lenders and shared borrowers. Low leverage and secured lending may help contain spillovers, but non-traded BDC funding structures and opaque private credit exposures remain a concern.