The Wrong Apocalypse [pdf] (ionanalytics.com)

A recent analysis has spotlighted a dramatic $2 trillion market decline in the enterprise software sector, primarily triggered by the introduction of potent AI tools from Anthropic, such as Claude Cowork and Claude Code. These AI systems demonstrate abilities such as drafting legal documents and automating accounting workflows, prompting fears among investors that human-driven software solutions may soon become obsolete. This skepticism intensified following Dario Amodei's essay, which warned of significant economic disruption and job losses due to these advanced AI capabilities. However, the analysis argues that the prevailing market narrative may be overly simplistic, conflating the potential for task automation with software replacement. The argument emphasizes that enterprise software's value is rooted not only in task execution but also in the coordination of complex organizational interactions. Unlike cognitive tools, which perform specific tasks, enterprise software serves to navigate and facilitate the “language games” of institutions—establishing norms, processes, and trust among varied agents. This perspective posits that while AI will likely disrupt the low-level cognitive functions of enterprise software, replacing the intricate institutional frameworks that underpin them will be a slower and more nuanced process. As organizations navigate these challenges, the transition to AI-native workflows may not happen as rapidly as some forecasts suggest, revealing a deeper complexity in understanding future disruptions within the AI/ML realm.