Microeconomics 2025, 1(1), 2005. https://doi.org/10.62617/me2005 1 Article Artificial intelligence agents and superstar effects Diego Lanzi Department of Economics, ALMA MATER STUDIORUM/University of Bologna, Bologna 40100, Italy; diego.lanzi@unibo.it Abstract: The paper models AI-services market competition in the presence of superstar effects. By modifying the original Rosen’s setup consistently with recent advancements of superstar theory, we discuss the role of scale-related technical change in the advent of superstar artificial intelligent agents and the main consequences of superstar effects on competition between AI-based systems. Furthermore, we outline how to extend the model for addressing three issues: (i) positive feedback loops from data accumulation; (ii) the co-evolution of artificial intelligence agents’ capabilities and market size; and, (iii) the “winner -takes-all” dynamics triggered by superstar effects. Keywords: artificial intelligence; digital innovation; superstar effects; technical change; “winner-takes-all” dynamics JEL: D31; L11; O33 1. Introduction Information technology allows a small number of talented individuals or firms to serve a large market and reap large rewards. In a series of papers, published in the second half of the 70s, Sattinger [1,2] and Rosen [3,4] called such a magnification “superstar effect”. Superstars are a small number of people (or sellers) earning enormous amount of money and dominating the activity they engage in. When the superstar effect operates there is a concentration of output and income among a few agents, and the distribution of rewards exhibits high skewness. Rosen’s first example [4] was that of comedians and television. More recently, Gabaix and Landier [5] and Terviö[6] have applied the superstar theory to labor markets. Workers with heterogeneous talent and employers of varying size are matched with some superstar workers earning enormous wages and big firms very high profits. From this perspective, superstar effects increase income inequality 1 . In a recent paper, Korinek and Ng [8] emphasize that advances in information technologies and digitalization have supercharged the superstar phenomenon. Superstar firms of the digital economy have had, between 1990 and 2015, rising market shares, rising mark-ups and an explosive growth of their intangible/physical capital ratio. Furthermore, the advent of artificial intelligence (AI) and artificial intelligence agents (AIAs) will increasingly generate superstar rents and absorb them in terms of investment, hence enhancing superstar effects. In the future economy, a few self-improving AIAs will likely dominate in terms of profits, notoriety and market size. A “winner-takes-all” dynamics, made easier by AI-based digital innovation, which allows us to replace tasks performed by traditional labor and capital; collect and process excludable information; and, replicate a technology at negligible cost [9]. In 1981, Rosen concludes that the main driving force of superstar effects is technological change that facilitates an increase in market scale. During periods of CITATION Lanzi D. Artificial intelligence agents and superstar effects. Microeconomics. 2025; 1(1): 2005. https://doi.org/10.62617/me2005 ARTICLE INFO Received: 9 April 2025 Accepted: 26 May 2025 Available online: 4 June 2025 COPYRIGHT Copyright © 2025 by author(s). Microeconomics is published by Sin- Chn Scientific Press Pte. Ltd. This work is licensed under the Creative Commons Attribution (CC BY) license. https://creativecommons.org/licenses/ by/4.0/