Generative AI and the Superstar Firm Effect

We examine how the emergence of Generative AI (GenAI) impacts the superstar firm effect across firms whose occupational tasks are more or less amenable to large-language-model automation. Using the release of ChatGPT as a market-wide shock to the availability of GenAI technologies, we find that investors value GenAI exposure among superstar firms. In the twelve months following the ChatGPT launch, high-GenAI superstars yield a 2.5% monthly characteristic-adjusted portfolio return, while high-GenAI non-superstars earn 0.6% and low-GenAI firms of both types cluster near zero. This premium among high-GenAI superstars continues to increase at the two-year horizon and is absent in the pre-ChatGPT period. Cross-sectionally, we find that this premium is concentrated among firms with high intangible capital, data capital, thick managerial hierarchies, and geographically dispersed workforces, consistent with GenAI creating value through complementary assets and coordination cost reduction. Lastly, we show that high-GenAI superstars exhibit higher profitability, as measured by return-on-assets, in the second fiscal year following the ChatGPT shock. Overall, our analysis points to a potential widening gap between superstar firms well-positioned to deploy GenAI and other firms.