The SaaS-pocalypse: Why Investors Continue to Bet Against Software

For two decades, software-as-a-service (SaaS) was Wall Street’s favorite trade. Investors treated SaaS practically like a bond with a growth rate attached, prized for its sticky revenue, high margins, and predictable churn. But that story has now been aggressively, and possibly permanently, repriced.

The Seeds of a Shift

The warning signs of the current crisis were actually visible well before 2026. Last year, industry giants like Adobe and Salesforce lost 21% of their value, while HubSpot plummeted more than 40%. This divergence from the broader market suggested that something structural was shifting beneath the surface.

While analysts have spent the three years since ChatGPT’s debut warning that industries like legal services and programming were at risk, it took a specific wave of disappointing earnings and a new Cowork agent from Anthropic to spark widespread fear of AI displacement, resulting in a massive selloff.

Why the Per-Seat Model is Breaking

The damage to these "legacy" names has been swift and broad. Salesforce and Workday are both down more than 40% over the past 12 months. Intuit has fallen 34% this year, and even Shopify has dropped 29%. Perhaps the most striking moment came when Atlassian announced it was cutting 1,600 jobs (10% of its workforce) specifically to fund its own AI investments. The irony of a project management platform letting humans go to build AI was not lost on the market.

If a single AI agent can handle the complex professional workflows that software providers used to sell as core products, enterprises no longer need to buy 500 seats. They might only need 100 or 50. The immediate question for any buyer is: if an agent can do the workflow, why pay for the seat?

The Vibe-Coding Test

Not every software company is equally exposed. CNBC's "vibe-coding" test suggests that the most vulnerable names are those that "sit on top of the work": tools like Zendesk, HubSpot, and Smartsheet. Meanwhile, cybersecurity stocks like CrowdStrike and Palo Alto Networks are considered harder to disrupt because of their deep network effects.

Management teams are finding there is very little they can say to reverse this sentiment. A report from Bain & Company confirms that net revenue retention has stalled as buyer budgets shift from incremental software purchases to new AI tooling. While some bulls point to Gartner data projecting software spending will still grow 14.7% this year, the counterargument is that those dollars are now flowing to AI-native platforms rather than legacy, seat-based vendors.

What This Means for Investors

The market is currently split down the middle. Some, like Workday CEO Carl Eschenbach, argue the "AI is killing software" narrative is overblown, citing the deep data moats and high switching costs of enterprise tools. Longtime investors like Byron Deeter are even viewing this chaos as a "buy-the-dip" opportunity.

However, the bear case is harder to dismiss than it was a year ago. Even Orlando Bravo of Thoma Bravo, who has spent two decades building software businesses, publicly admitted that some of these valuation decreases are "very warranted".

For investors, the key theme is bifurcation. Infrastructure providers that store and process data are relatively insulated regardless of how many human seats are eliminated. But for those in the application layer, the clock is ticking. The companies that survive will be the ones that can pivot from charging per seat to charging per outcome before the market decides their fate for them.