The SaaSpocalypse Is a Credit Event, Not a Tech Story

AI is destroying many SaaS business models and the pain is acute in the credit markets

Feb 17, 2026

The enterprise software selloff everyone’s calling the “SaaSpocalypse” has destroyed somewhere between $800 billion and $2 trillion in market cap from the software sector since late January. The iShares Expanded Tech-Software ETF (IGV) is down over 23% year-to-date. Salesforce and Workday are each off more than 40% from their highs.

Most of the coverage frames this as a technology disruption story: AI agents are coming for the SaaS stack. And that’s true, as far as it goes. But the more interesting, and more dangerous, story is about financial plumbing. The SaaSpocalypse is stress-testing a $500 billion debt structure that private equity built on top of the SaaS business model during the most leveraged decade in financial history.

The Mechanism Everyone Gets Wrong

The prevailing narrative that AI will replace enterprise software misses the more immediate threat. AI doesn’t need to replace Salesforce to destroy its economics. AI just needs to compress seat counts.

SaaS revenue scales with headcount. A company grows, hires people, buys more seats. This is the recurring revenue that the PE industry has salivated over for the past decade. But AI agents break that correlation. As SaaStr’s Jason Lemkin has explained: if 20 AI agents do the work of 20 sales reps, you don’t need 20 CRM seats.

The proximate catalyst to the SaaSpocalypse was Anthropic’s Claude Cowork launch in January. On February 2, Anthropic released Cowork “skills” for law. This tool automated contract reviews, compliance workflows, and legal summaries. The next trading day, $285 billion evaporated from SaaS and IT services stocks. Gartner fell 21%. S&P Global dropped 11%. Intuit lost more than 10%. Even PE firms themselves were hit: Ares and KKR each fell 10%.

Meanwhile, a January CIO survey showed IT budget growth decelerating to 3.4%, with funds being diverted from application software toward AI infrastructure. The hyperscalers are spending $660–690 billion on infrastructure in 2026, nearly doubling 2025, with ~75% of that total targeting AI-related infrastructure. That money is coming out of enterprise software budgets. The “best-of-breed era” of specialized point solutions is ending.

Follow the Debt

Here’s where it gets structurally interesting. During the 2010s and early 2020s, PE firms bought software companies at 10–20x recurring revenue and financed those deals with significant leverage. The thesis was bulletproof: sticky customers, high switching costs, predictable subscription revenue to service the debt. Over $500 billion in borrowing tied to software firms is now embedded in U.S. credit markets.

Software represents roughly 13% of the $1.53 trillion leveraged loan market tracked by the Morningstar LSTA index. Much of that paper is housed in CLOs—the same instrument class that amplified the 2008 financial crisis—and in Business Development Companies, the private credit vehicles run by the biggest names in alternative assets.

The BDC exposure is the flashpoint. Blackstone’s BCRED has 26% of its loans in software. Blue Owl’s technology-focused BDC has at least 29% in software debt; investors pulled 15% of their money out in a single month. Ares’s listed BDC sits at 24% software exposure.

The vintage problem makes this worse. Many of these buyouts were done at peak valuations right after Covid, financed with covenant-lite structures that give lenders limited early warning. Applying current public market multiples (3–5x revenue) to portfolios acquired at 9x+ implies the equity cushion is gone. And when these firms sit in levered vehicles, even modest markdowns cascade. BlackRock TCP Capital’s 19% NAV write-down in late January demonstrated the math: at 230%+ leverage exposure, an 8% portfolio decline becomes a 19% equity wipeout.

The Reflexivity Trap

The dynamics are now self-reinforcing. Software loans from private credit deals that recently refinanced into the syndicated market are selling off in secondary trading. Kaseya’s first-lien term loan dropped ~3.5 points. Team.Blue, a European software firm, had to halt a loan refinancing entirely. Apollo, reading the room, has positioned short against software loans.

If BDC redemptions accelerate and funds start gating withdrawals, the downward spiral intensifies: forced selling → lower prices → worse marks → more redemptions. This is the mechanism that transforms a software stock selloff into a credit event.

The Bull Case Is Real But Irrelevant to the Leverage

There are legitimate counterarguments. Deeply embedded systems of record don’t get ripped out overnight. Klarna famously ditched Salesforce and Workday for AI-built replacements, then admitted it “went too far” and started rehiring. Experienced programmers warn that 2026’s vibe-coded software will require extensive cleanup in 2027. JPMorgan and Goldman both argue the selling is overdone.

Maybe so. But the bull case operates on a 3–5 year time horizon, and the debt doesn’t care about long-term fundamentals. Floating-rate loans need to be serviced now, against revenue assumptions made in a pre-AI world. PE funds holding these assets need exits, and the IPO window is frozen for software companies.

The Economist compared this to the shale energy bust of 2014 — leveraged bets on a business model disrupted by a supply shock. The parallel is instructive, but there’s a key difference: the oil price eventually recovered. The per-seat SaaS model may not. Gartner projects that by 2030, at least 40% of enterprise SaaS spend will shift toward usage, agent, or outcome-based pricing. That’s a permanent repricing of the revenue streams PE underwrote at peak multiples.

Watch the BDCs. Watch the CLO marks. Watch whether gates go up on redemptions. The technology story is interesting. The credit story is where the systemic risk lives.

If you enjoy this newsletter, consider sharing it with a colleague.

Share Buy the Rumor; Sell the News

I’m always happy to receive comments, questions, and pushback. If you want to connect with me directly, you can: