Jorge Costa Oliveira
In the first six weeks of 2026, the S&P 500 Software & Services Index saw nearly $1 trillion in market value evaporate. On February 3 and 4 alone, the Nasdaq 100 shed almost $300 billion in software and data stocks in what analysts and investors dubbed the “SaaSpocalypse.”
The long-anticipated structural shakeout of the software industry began on January 12, 2026, with the official launch of Anthropic’s Claude Cowork, a desktop artificial intelligence agent capable of managing local files, navigating complex corporate interfaces, and deploying “parallel sub-agents” to handle massive data entry and analysis tasks.
On January 30, during what is now known as the “Plugin Event,” Anthropic released 11 specialized open-source plugins enabling Claude Cowork to perform functions such as legal document discovery, tax accounting, and automated sales prospecting. Claude Cowork not only assists human workers; it independently executes complex workflows. The resulting “licenses compression” – in which companies require far fewer software seats to accomplish the same volume of work – triggered a sweeping repricing across the technology sector.
Providers of specialized software saw their shares plunge. By early February, the contagion had spread to general-purpose SaaS companies, as corporate procurement departments announced plans to right-size their software stacks, citing efficiency gains from autonomous AI agents.
As of mid-February 2026, the panic has led to a fundamental re-rating of the sector. Forward earnings multiples for the software industry, which averaged 39x a year ago, have collapsed to 21x. Venture capital firms and hedge funds have begun forced liquidation of seats-dependent stocks, fearing that the growth engine of the last two decades – adding more users – is permanently broken.
Investors expect widespread adoption of AI agents such as Claude Cowork to accelerate the dismantling of the SaaS business model. This SaaSpocalypse marks the transition from the “cloud era” to the “agents era”. For two decades, the industry followed a “system-of-record” model – software was where humans entered and stored data.
The “system-of-action” model introduced by Claude Cowork flips that logic: the AI agent performs the work, while the human supervises. This shift alters the human-to-software ratio that has governed technology companies since the 1990s.
If an AI agent replaces five employees, the total addressable market for license-based software shrinks by 80 percent – unless vendors find new ways to capture value.
The primary losers are established cloud-era giants: (i) Salesforce, at the center of the storm; despite launching Agentforce, the stock has struggled as investors question whether $0.10-per-action AI pricing can offset lost $150-per-month user seats; (ii) Workday and Adobe, which have suffered sharp multiple compression; for Adobe, the threat is existential as autonomous creative agents reduce demand for Creative Cloud licenses; (iii) specialized players such as Thomson Reuters and LegalZoom, whose information moats are challenged by AI reasoning.
Conversely, winners are in infrastructure and physical-reality sectors: (i) NVIDIA remains central to the AI era; (ii) utilities and other so-called “boring stocks” have rallied as investors recognize that agentic AI’s power demands make electrical grids a safer bet than software licenses. Companies with a physical backbone – homebuilders and industrial manufacturers – are viewed as AI-resilient because physical assembly cannot be automated by a desktop agent.
linkedin.com/in/jorgecostaoliveira
No Comments