Enterprise Software: AI Agents Shift Decision-Making from IT Buyers to Automated Systems

Executive Summary

A16Z's Martin Casado identifies a fundamental disruption brewing in enterprise software procurement: AI agents are beginning to make infrastructure decisions previously controlled by IT teams and central buyers. This shift threatens to reshape a multi-trillion dollar market where human decision-makers have historically controlled vendor selection and technical architecture choices. Casado observes that developers using AI coding tools like Cursor or Claude Code are no longer consulting IT-approved infrastructure lists—the AI is selecting databases, APIs, and services directly. This represents a structural inversion where consumption patterns bypass traditional enterprise sales channels. The implications extend beyond individual tool selection to fundamental questions about compliance, security, and vendor relationships when decision-making becomes algorithmic rather than human-controlled. Casado's contrarian view challenges the prevailing narrative that AI will simply enhance existing SaaS interfaces. Instead, he argues that while coding becomes democratized, engineering complexity actually increases, requiring more sophisticated operations and infrastructure management. The regulatory bottleneck thesis provides additional conviction: despite massive demand, supply constraints are purely bureaucratic rather than technical, creating artificial scarcity in data center capacity and power infrastructure. This analysis suggests traditional enterprise software vendors must rapidly evolve from seat-based to consumption-based pricing models while maintaining their core value proposition of encoding business processes and compliance frameworks.

Key Insights