Ryan@ohryansbelt
$46.9 billion in software debt trading at distressed levels.
$25 billion in software loans below 80 cents on the dollar.
Apollo cutting its software exposure in half.
Blue Owl freezing redemptions.
Morgan Stanley and BlackRock capping withdrawals. Goldman pitching hedge funds on ways to short software loans.
And a Goldman executive telling clients on a call today that private markets firms are "glad" the Iran war is providing a "distraction" from questions about their software loan exposure.
AI is systematically dismantling the thesis behind $600-750 billion in software loans, and the $3 trillion private credit market that financed a decade of software buyouts is starting to crack. Here's what's happening:
> Over 1,900 software companies were acquired by private equity between 2015 and 2025 in deals worth over $440 billion. The thesis was irresistible: sticky recurring revenue, high margins, high switching costs. AI is stress-testing every one of those assumptions simultaneously.
> Software EBITDA multiples have collapsed from 30x at end of 2022 to roughly 16x today. Revenue multiples went from 10-12x to about 4x. The collateral value behind these loans has been cut in half.
> JPMorgan is marking down software loans in private credit fund portfolios and limiting how much it will lend against them. Jamie Dimon drew parallels to the years before 2008. JPMorgan's lending to nonbank financial firms tripled from $50 billion in 2018 to $160 billion.
> Troy Rohrbaugh, co-CEO of JPMorgan's commercial and investment business, told analysts "I'm shocked that people are shocked"
> The real exposure is hidden. Bloomberg found at least 250 software loans worth $9 billion+ were categorized as other industries by BDCs. A pricing-software company labeled "business services." A restaurant software company classified as "food products."
> Deutsche Bank got stuck holding $1.2 billion in loans backing a software acquisition it couldn't sell to investors, a rare "hung deal" signaling how fast lender appetite has evaporated
> UBS estimates default rates could hit 13% for US private credit if AI disruption accelerates, more than 3x the projected high-yield default rate
> Casualties are piling up. Figma down 80% from its high despite 40% revenue growth. Navan IPO'd with $657 million in debt and is down 60% in four months. Pluralsight was a "keys handover" where lenders took control from sponsors. Finastra's loans fell from high 90s to ~93-94.5.
> 23 out of 32 rated BDCs have unsecured debt maturing in 2026, totaling $12.7 billion. Golub Capital, with 26% of its portfolio in software, already cut its dividend 15%.
> Orlando Bravo of Thoma Bravo, arguably the most important software investor alive, said at Davos that AI will disrupt "less than half" of software companies. That's his base case. He's the bull.
> The S&P North American software index fell 15% in January alone, its biggest monthly decline since October 2008
As Blackstone's Jon Gray put it, "Everyone's focused on these bubble risks. I think the biggest risk is actually the disruption risk. What happens when industries change overnight."
The history of financial crises is not a history of greed or stupidity. It is a history of collective belief in a model of the future that turned out to be wrong. Subprime assumed home prices would never fall nationally. Private credit assumed software margins would never compress.
The difference is that nobody built a technology that made houses worse. AI is actively dismantling the asset the loans were built on.
