Fast Company's Karandeep Anand argues that the next era of entertainment is sub-fandom, the idea that the real value is no longer in capturing mass audiences but in serving the obsessive core. It reads as a business thesis. It is actually a cultural one, and it has been true since at least 1977. What's new is the infrastructure built around it.
Baby Keem, Casino, and the Economics of the Loyal Few
Consider Baby Keem's rollout for Ca$ino. It is not a crossover campaign. It is a depth campaign. The artist is betting that the people already inside the universe will spend more, share more, and sustain longer than a casual listener acquired through a playlist algorithm. This is the sub-fandom thesis in practice. Kyle Chayka's work on algorithmic homogenization maps the inverse: when platforms optimize for the broadest possible reach, they sand down the edges that make sub-fandoms form in the first place. The tension between those two forces is where contemporary music culture actually lives.
When Synthetic Audiences Mirror Real Ones
Here is where it gets stranger. A 2026 arXiv paper by Kinzinger and Hartmann tests how well LLMs can mimic individual respondents using socioeconomic microdata, essentially asking whether synthetic personalities can replace real audience research. The answer is: better than expected at the aggregate level, worse than expected at the individual level. Which is precisely the sub-fandom problem in reverse. Mass audience simulation works fine. Simulating the specific obsessive, the person who has every Baby Keem lyric memorized and would pay $400 for a box set, is where the model falls apart. The superfan is, almost by definition, an outlier that statistical personas cannot capture. TurboFund's piece on predictions rewarding accuracy over luck touches this exact dynamic: the investors and operators who win in niche markets are the ones who stop averaging their audience and start treating outliers as the signal. Sub-fandom is not a marketing strategy. It is an epistemological stance.