These days, you can sense it even before anyone speaks aloud when you walk into a San Francisco software company office. The hum of the standing desks persists. Healthy retention figures continue to shine on the dashboards. Sales teams continue to meet their targets. Even though they don’t want to identify it just yet, the people inside are aware that something has broken.
The majority of the story is revealed by the numbers. Enterprise software lost about $1 trillion in market value between late January and early February, but the underlying companies continued to make money. HubSpot’s market capitalization dropped from $42 billion to less than $10 billion, a loss of more than half. Monday.com lost roughly 44%. ServiceNow lost 36%. The iShares Expanded Tech-Software ETF fell 22% so far this year, and traders at Jefferies began referring to it as the “SaaSpocalypse”—a melancholy, slightly humorous term that only really comes up when no one knows what to do.
The peculiar thing about it is that the companies themselves are still in operation, which sets it apart from previous sell-offs. Contracts are being renewed. The gross retention rate is approximately 90%. Eight- and nine-figure batches of cash continue to arrive every quarter. It used to be sufficient. Because the revenue seemed unkillable—sticky, recurring, predictable, the kind of thing pension funds and private equity sponsors could underwrite with ease—software was Wall Street’s preferred asset class for many years. These same investors are now examining the same income statements and concluding that the future may differ from the past.
If you want to focus on just one instance, the January 12 launch of Claude Cowork by Anthropic was the catalyst. A CNBC anchor created a productive Monday two weeks later.Using the new tool, create a com-style kanban board on screen in less than ten minutes. The stock fell 10% the following day and 6% during the same session on Monday. Software analysts believe that something psychological broke at this point. Not because a billion-dollar SaaS platform will be instantly replaced by a single anchor’s demo, but rather because investors could finally see what they had been hearing about for months during earnings calls.
The per-seat model is the deeper issue. SaaS economics were founded on the unspoken premise that, indefinitely, more people equals more seats equals more revenue. That is subtly reversed by agentic AI. An organization no longer needs to pay for five seats if it can use a coworker-style agent to manage workflows that previously required five analysts. It might cover three. Instead, it might cover the cost of an agent license. The discounted cash flow model can appear significantly different without the math collapsing all at once. Analysts at HarbourVest did the math and found that a 30% drop in seats combined with a slight price increase still results in a 23% decrease in revenue. Investors are currently imposing this uncertainty tax.
It’s difficult to ignore the fact that the industry’s most astute participants appear to be hedging. Agentforce was created by Salesforce. Before anyone else could, Microsoft forced Copilot into Excel and Outlook. The AI Control Tower is part of ServiceNow. The argument is that established vendors can capitalize on the trend because, under duress, consumers would prefer to purchase AI from the business that already has their data rather than take a chance on an unknown. That is true. Clients do express a desire for it. The question of whether the incumbents can truly deliver and at what cost remains unanswered.
The speed of the repricing is what seems novel. The dot-com bust and the 2022 rate shock are just two examples of the many corrections that the tech industry has experienced, but they were all related to changes in interest rates or clear declines in revenue. This one is both philosophical and structural. The money is still there. The clients are still present. The presumption that they will all still exist in 2031 has vanished.
A few of these businesses will change course. Some might even turn out bigger than they were. Others will learn the hard way that being indispensable does not equate to permanent indispensability. The software meltdown isn’t exactly a tale of failure. It tells the tale of how a market subtly stopped accepting the outdated narrative.
