Knowledge Spillovers Versus Competitive Dynamics: Evidence from Startups and Scaleups in Milan’s Tech Ecosystem

Abstract Moving beyond the traditional focus on geographical proximity, this study investigates the existence and nature of spillover effects within a non-geographical space. The core novelty of the paper is operational and methodological: we exploit text data scraped from corporate websites to construct a novel semantic proximity matrix within a spatial econometric framework, capturing firms’ industrial specialisations, technological adoptions, and cognitive competencies. Focusing on the tech ecosystem of Milan, the study examines how the balance between positive and negative externalities shifts across different phases of firm growth. From a conceptual perspective, the paper integrates growth with spillover dynamics, explaining the mechanisms through which firm maturity transforms knowledge-sharing opportunities into competitive frictions. We distinguish between two archetypes: younger, smaller startups operating within an exploratory innovation paradigm, and more mature scaleups driven by market exploitation and asset accumulation. The empirical results reconcile these perspectives by showing that the net effect of proximity is strictly contingent on the growth stage of neighbouring firms. Proximity to startups is associated with significant positive knowledge spillovers, consistent with their role as open, learning-oriented nodes. By contrast, proximity to scaleups triggers negative competitive pressures and “businessstealing” effects, as their aggressive market expansion imposes costs on technologically and industrially close neighbours. These findings offer fresh insights for researchers and policymakers, demonstrating that in modern tech ecosystems, the impact of proximity depends entirely on who is proximate.