The damage happened fast. In the first week of February 2026, over $1 trillion in market capitalization was erased from software stocks in seven days. The IGV software ETF fell roughly 30% from its September 2025 peak. The S&P Software Index posted its worst monthly decline since the 2008 financial crisis. Salesforce dropped 26%. Multiples compressed from roughly 39x forward earnings to around 21x in a matter of months.
The catalyst was not a recession. It was a product launch. Anthropic’s Claude Cowork tool demonstrated that AI agents could handle complex, multi-step enterprise workflows without human intervention. The market’s conclusion was swift and brutal: if AI can replicate what enterprise software does, enterprise software is a melting ice cube.
That conclusion was mostly wrong. But it created one of the more interesting dislocations in tech in years.
What the Market Got Right
There is a real problem underneath the panic. The per-seat licensing model — the economic backbone of SaaS for two decades — is genuinely under pressure. The math is uncomfortable: if one person equipped with AI agents can do the work of five, a company only needs to buy one license instead of five. Enterprise customers noticed. The average number of SaaS applications per company declined as CFOs began cutting subscriptions that couldn’t demonstrate ROI. With 82% of companies reportedly reducing the number of software suppliers, rationalization was happening before the AI panic even started.
Multiple SaaS companies beat their quarterly revenue numbers and still watched their stocks fall. ServiceNow exceeded analyst expectations and dropped 11% intraday. The pattern repeated across the sector. Clearing the bar was no longer enough — investors were discounting the durability of future cash flows.
What the Market Got Wrong
Here’s the thing: global SaaS spending is not collapsing. Gartner projected worldwide software spending would grow 14.7% in 2026 to over $1.4 trillion — roughly $180 billion in net new spending in a single year. Enterprises are not abandoning software. They are spending more on it. They are just spending differently.
The companies with data moats, mission-critical workflows, and high switching costs are not going away. Enterprise software replacement requires staff retraining, data migration, permission re-architecture, and regulatory recertification. That friction is real and expensive. JP Morgan analysts called the software sell-off “overblown” and argued that fears were based on “broken logic.”
What actually happened — and this is the part the SaaSpocalypse framing misses — is a bifurcation. The selloff did not kill SaaS. It sorted it. The companies that could pivot to consumption-based billing, embed AI deeply enough to become essential to agentic workflows, and prove their data was irreplaceable… those companies found a floor. The ones that couldn’t are still in trouble.
The Survivors and the Inflection
ServiceNow repositioned itself as an AI orchestration layer — what some analysts called an “AI Control Tower” for enterprise workflow — and its Now Assist engine reportedly surpassed $600 million in Annual Contract Value early this year. Salesforce introduced “Agentic Work Units” as a billing metric, effectively shifting from per-seat to per-outcome pricing. Adobe’s Firefly and AI Acrobat Assistant reportedly tripled their ARR contribution year-over-year.
A February Goldman Sachs survey found that 49% of institutional allocators planned to increase exposure to software — the highest net figure since 2017. That rotation started quietly and is now accelerating into Q2 earnings season.
The part investors are still missing: vertical SaaS is a different story entirely. Companies serving healthcare, manufacturing, legal, and financial services — with proprietary, regulated data that AI cannot easily replicate — face structurally lower disruption risk. Forrester projects vertical software growing from roughly $133.5 billion in 2025 to $194 billion by 2029. That is not a market in decline.
The Setup Heading Into Earnings
Q2 earnings season starts mid-July. For software, this is the first real test of whether the AI monetization thesis is showing up in actual numbers — not just narrative. Watch for consumption-based revenue growth, AI tier expansion, and any evidence of seat count stabilization. The stocks that can show both adoption and monetization in the same quarter will likely re-rate sharply. The ones that still cannot answer the pricing question will face another round of multiple compression.
The SaaSpocalypse was a sentiment event dressed up as a fundamental one. The fundamental question — which software companies actually survive the AI transition — is still being answered in real time. And the answer is more nuanced than the February selloff suggested.
For informational purposes only.
