New Mountain Capital's Steve Klinsky says the AI panic gutting software stocks has gotten ahead of itself, and he's not alone.
Roughly $300 billion evaporated from software company market caps in a single trading session in February, according to Bloomberg. Anthropic had just rolled out Claude Code and Claude Cowork. Investors decided AI agents could soon do the job SaaS platforms are built for. They sold first and asked questions later. That one session became shorthand for what the industry now calls the SaaSpocalypse. Five months on, it still hasn't stopped rattling nerves.
Steven Klinsky runs a firm built on exactly the kind of software companies this panic is supposed to gut. New Mountain Capital manages roughly $60 billion. Much of it sits in mid-sized enterprise software businesses, built up over years through add-on acquisitions. That's the portfolio at risk. So when Bloomberg Technology reported this week that Klinsky is pushing back on the SaaSpocalypse narrative, calling it overstated, it wasn't a neutral bystander weighing in. It was the guy holding the bag.
The case against panic
His argument, echoed elsewhere in the industry, is that AI's disruption of software will land slower and narrower than what public markets are currently pricing in. Vibe coding lets a founder or an internal team spin up a workflow tool in an afternoon. That's real. But most enterprise software isn't a workflow tool anyone wants to rebuild from scratch. It's a system of record. A decade of integrations sits on top of it, along with compliance sign offs and institutional muscle memory. Ripping that out and replacing it is a different project than prototyping a Slack bot over a weekend.
Marc Benioff makes a version of the same case. He runs a company that's already absorbed the worst of the AI fear. Salesforce shares fell more than 40% in the first half of the year, according to the Motley Fool, dragged down with the rest of the software sector on worries that agentic AI would eat its core business. Benioff's response, delivered back in February, was blunt. "People think we have our back against the wall when in fact the opportunity has never been greater," he told TechCrunch, framing AI agents as a new revenue line rather than an existential threat. Salesforce has been pointing to Pentagon Federal Credit Union as proof: the credit union now runs 76 Salesforce agents across its operations and says it's saving $1.6 million a year while handling more volume with the same headcount.
What the market is still pricing in
Whatever Klinsky and Benioff believe, public markets have kept pricing in the worst case. Software wasn't the only corner of tech getting hit this month. In early July, chip stocks lost more than a trillion dollars in combined value on fears that hyperscaler AI spending, which jumped 67% to roughly $650 billion, might be outrunning actual demand. Micron alone shed 13% and about $138 billion in a single session after Meta signaled it may be sitting on more AI computing capacity than it needs. That's a different fear than the SaaSpocalypse. Capacity glut, not software obsolescence. But it comes from the same place: a market that spent 2025 assuming AI spending would keep expanding forever and is now testing whether that assumption holds.
Not everyone in private equity shares Klinsky's calm. A Bloomberg Opinion piece published in April argued that private equity is actually worse positioned than private credit to absorb an AI shock, because so much buyout capital over the past decade went into exactly the recurring revenue software bets Klinsky's firm specializes in. If AI erodes seat based SaaS pricing even at the slower pace Klinsky expects, the debt loaded into those buyouts doesn't leave much room to absorb a valuation cut.
That's the real stakes if you're a founder raising money right now. Annual recurring revenue and net dollar retention used to be enough to get you a clean multiple. Not anymore. Buyers ask a harder question first: how much of your product could a customer replicate in house with an AI coding agent in a month? If the answer is not much, because your moat is proprietary data, deep integrations, or a workflow too regulated to reinvent casually, the SaaS metrics still work in your favor. If the answer is yes, your multiple compresses. It doesn't matter what Klinsky says about the industry overall.
Frankly, that's a more useful test than picking a side in the SaaSpocalypse debate. Klinsky is probably right that the timeline is slower than the February selloff implied - and he happens to be the person with $60 billion riding on that being true.
Also read: Satya Nadella warns you pay for AI intelligence twice, in cash and in secrets • Together AI raised $800 million on the bet that cheaper AI wins • Delaware Weighs a New Legal Entity Built for Companies Run by AI Agents
