AI as a Signal: Funding Effects of Key Generative AI Milestones on Startups

Abstract

This thesis explores how breakthroughs in generative artificial intelligence have changed the venture capital funding landscape for AI startups, particularly from a signaling theory perspective. It explores how two highly pivotal events – in 2017, the emergence of the Transformer architecture, and in 2022, the release of ChatGPT – have galvanized investors across the various stages of startup funding. By applying a difference-in-differences empirical framework to a very large sample of startups, the paper examines whether such events promoted or diminished the effectiveness of "AI adoption" as a signal in conditions of information asymmetry. The results indicated a somewhat bifurcated development in signal interpretation: on one side, there was not a broad-based funding surge on account of the Transformer milestone, probably because of its technical opacity; on the other side, the very visible and commercially resonant release of ChatGPT acted as a powerful signal for infusions of capital into AI ventures, most markedly at later stages. Thus, the results indicate clearly that investors indeed respond not just to technological improvements but also to how legibly those improvements are translated into commercial narratives. The study contributes to entrepreneurial finance and innovation literature by demonstrating how attention shocks reshape the signaling environment and alter capital allocation across stages of venture development.