Private Equity's Software Fortress Is Being Repriced, One Loan at a Time
Thoma Bravo's $5 billion wipeout on Medallia is the emblem of a broader reset: AI anxiety has cut PE software dealmaking to a fraction of last year's record and turned recurring revenue from a bond into a question mark.
NEW YORK, August 24, 2026. For a decade, software was private equity's answer to every question. Sticky customers, contractual revenue, 90 percent gross margins: sponsors learned to treat a SaaS maintenance stream like a bond and lever it accordingly. That trade is now being repriced in public, and the clearest mark came in June, when Thoma Bravo handed Medallia to its lenders. The firm bought the customer-experience software maker for $6.4 billion in 2021 and lost its entire investment, roughly $5 billion, in one of the largest private equity restructurings on record. A group led by Blackstone, with Apollo and FS KKR, took the keys, injected $150 million of new capital and cut the debt load, with the deal expected to close by year end.
Medallia is a data point. The trend line around it is harder to dismiss. Software stocks shed more than 20 percent, over $1 trillion of value, at the worst of this year's decline, a rout traders took to calling the SaaSpocalypse. US private equity platform deal value in software ran at just $16.2 billion through the first five months of 2026, a pace that would finish the year at about a quarter of 2025's record $156 billion, according to PitchBook. S&P Global counted 3,665 sponsor-backed application software deals in 2025, down 21 percent from the year before and the third straight annual decline. The buyers who spent a decade paying up for recurring revenue are now the most selective people in the market.
The mechanism of the fear is specific. The vulnerable half of the software universe is the rule-based workflow: document review, tax compliance, expense processing, customer surveys. Those are precisely the tasks AI-native entrants can rebuild cheaply, and precisely the products whose renewal is a budget-line decision rather than an emotional one. Orlando Bravo told a Davos audience in January that AI will disrupt fewer than half of software companies but many of them, especially those whose core competency is technical. Coming from the industry's most prolific software buyer, fewer than half is not a reassuring number. It is an admission that nobody can yet say which half.
Lenders are doing the sorting in the meantime, and they are doing it with price. Refinancing software debt now costs more because credit committees are haircutting terminal value, and the anxiety has bled into the vehicles that hold these loans: concern over software exposure helped push public BDC valuations to their lowest levels since 2022 and fed the redemption queues now running through private credit's retail funds. The equity side and the credit side of the same trade are arguing about the same question, which is what a maintenance contract is worth when the customer can ask a model to do the job.
None of this means the software buyout is dead. It means the easy version is. Sponsors who can actually rebuild a product around AI, cut the cost to serve and defend the renewal have a genuine value-creation lever, and the capital now concentrating among fewer buyers will reward them for it. The ones who bought a workflow tool at 15 times revenue because the revenue was recurring are learning the clause that was always hidden in that word. Revenue recurs until the customer decides it doesn't. AI just gave every customer a reason to ask.
