In the last few weeks, a new term has shown up in financial headlines: “SaaSpocalypse.” It refers to the sharp drop in software company valuations that followed Anthropic’s release of plugins for its Cowork AI tool on January 30. Those plugins let businesses automate parts of legal administration, finance workflows, and marketing tasks. Within days, hundreds of billions in software stock value disappeared, and by mid‑February the selloff had spread to roughly a trillion dollars in lost market value.
The concern behind those numbers is straightforward: if AI can perform work that software coordinates today, why continue paying for that software? Why not build more of it yourself? That concern is not entirely misplaced. The CEO of Mistral AI recently told CNBC that more than half of enterprise SaaS could, in principle, be replaced by AI‑built alternatives. Andreessen Horowitz’s “Big Ideas 2026” report similarly suggests that traditional systems of record—the historical core of enterprise software—may start to lose their central role as AI shortens the distance between a user’s intent and the result. A report from AlixPartners put it bluntly in its title: “Farewell, SaaS.”
For anyone who attended the Hatch‑a‑thon, we saw this in person. We had around 50 builders creating applications that, in a different time, would have been packaged and sold as enterprise software. Given two weeks and the right tools, people were able to build what they needed themselves.
This raises a natural question: is enterprise software on the way out?
The widely discussed case of Klarna offers a more nuanced picture. In 2024, Klarna’s CEO announced that the company was ending its contracts with Salesforce and Workday and “replacing” them with AI. Headlines emphasized that their AI chatbot had already displaced hundreds of customer service roles and saved tens of millions of dollars. However, looking more closely, Klarna did not simply plug in a language model and discard its vendors. It consolidated data on an internal stack, used tools like Cursor AI to build new interfaces, and shifted to different SaaS providers. The company modernized and simplified its stack, but it did not eliminate software vendors altogether.
There are also structural limits on what can be replaced. A large share of enterprise software is off‑the‑shelf rather than custom‑built. Systems such as ERP platforms do more than move data between fields; they embed complex regulatory and financial logic. Those systems are not easily recreated by conversational prompts. The parts most exposed to AI‑driven change are the workflow and presentation layers that sit on top of established data infrastructure.
Taken together, this suggests a more balanced view. Simple workflow tools that mostly coordinate tasks, send notifications, or route approvals are under real pressure. AI agents can already handle many of those functions. At the same time, the systems that manage and govern organizational data become more important, because AI systems need clean, reliable data to be effective. Pricing models are also shifting, from per‑seat subscriptions toward usage‑based and outcome‑based arrangements. Instead of paying for a fixed number of licenses, organizations will increasingly pay for the results those systems help produce.
For people and organizations in Lower Alabama, the implications are practical. AI has not made enterprise software disappear. But it has reduced the need for a wide range of smaller, convenience‑oriented tools—especially for those willing to learn how to build or adapt software themselves. The distance between “I have a problem” and “I have a working tool” is shorter than it has ever been. A small business paying monthly fees for a scheduling app, a simple CRM, a project tracker, and a reporting dashboard may now be able to replace some of those with lightweight tools assembled in‑house. Nonprofits that could never afford custom development can realistically consider building tailored internal systems.
In that sense, the current moment is less a collapse than a reallocation. The “SaaSpocalypse” label captures investor anxiety, but on the ground what is happening looks more like a shift from buying generic tools to composing focused ones. For anyone willing to engage with these new AI capabilities, that shift represents an opportunity rather than a crisis.
More than a decade ago, Marc Andreessen wrote that “software is eating the world.” What we are seeing now is not the end of that trend, but a change in the menu. Software is still eating the world, but more of it will be software you assemble on demand, and less of it will be shrink‑wrapped SaaS someone else sells you.