One Fifth of BDC Portfolios Is Software. That Is the Number, Not the Markdown

The figure is an exposure share, not a writedown. J.P. Morgan Asset Management puts business development company portfolios at approximately 20% software, against 15% in broadly syndicated loans and 4% in high yield ^[1][2], which makes private credit roughly five times more concentrated in the sector than public high yield ^[15].

That concentration was not an allocation decision made this year. Software and SaaS were a favoured private equity target over the past decade, and the buyout debt was largely financed by private credit funds and their BDCs ^[3]. The 20% is the residue of that vintage, sitting in long-duration paper that was underwritten against growth assumptions written before AI coding tools existed.

What it turns on is whether disruption converts into credit events. J.P. Morgan says the market has not yet seen widespread credit deterioration, that investment grade software balance sheets are healthy, and that displacement will not materialise overnight ^[6]. Prices moved anyway: initial spread moves of +10 to +35bps in a single day and +20 to +50bps week-to-date, per the firm's trading desk color ^[4]. Publicly traded BDC equity has been pressured, and non-traded BDCs already seeing increased redemptions may face further outflows ^[5]. That is the mechanism worth modelling. Redemptions force sales against marks nobody has tested.

The repricing is selective. J.P. Morgan calls the selloff warranted for names whose products are easily replicated by AI or "vibe coded" clones, particularly where offerings are rudimentary and R&D spend has been thin, while electronic design automation and risk analytics look more insulated ^[8]. On the equity side, Hg partner Benedikt Joeris says buyers now scrutinise whether a business has a real AI position and reward those showing actual products and adoption, and that Hg continues to exit above holding values ^[9]. Hg sold GTreasury to Ripple for USD 1bn two years after acquiring it for a reported USD 400m, a result largely attributed to an agentic AI buildout ^[10], or about 2.5 times ^[11]. Its Visma listing remains in limbo ^[12]. The other end looks different: Ropes & Gray partner Elizabeth Todd says distressed SaaS exits are likely to be valued on client books and data rather than revenue, a number that may not work for current owners ^[13].