The Moat Problem: Why Most Pure-Software AI Startups Lack Durable Competitive Advantages

Short summary

Framework-driven analysis: most pure-software AI startups lack durable competitive moats. Only two survive—proprietary data anchored in physical processes (lab experiments, sensor logs) and longitudinal data requiring real wall-clock time. Everything else (switching costs, brand, data flywheels, network effects) erodes as foundation models advance.

• •Pure-software AI startups face dual squeeze: foundation models absorb value while replicating any pure-software innovation cheaply
• •Only two durable moats exist: proprietary data from physical processes and longitudinal data requiring real wall-clock time
• •Everything else (switching costs, brand, data flywheels, network effects) are temporary buffers that erode as models improve

Generated with AI, which can make mistakes.