Commercializing Contrarian Ideas: Evidence from AI Contests

Most contrarian entrepreneurs – those who challenge conventional “best practices” – struggle to attract resources, yet a few achieve outsized successes once their ideas gain traction. One explanation is that contrarian ideas are simply riskier: they fail more often, but the few that overcome deeper technical hurdles generate exceptional breakthroughs. Another explanation emphasizes commercialization barriers: investors avoid unproven alternatives but rush in once it becomes clear that a contrarian approach outperforms the status quo. To provide evidence on these views, I examine hundreds of AI contests where either a contrarian or mainstream method is revealed to be best. Using a difference‐in‐differences design that compares the likelihood of founding a startup between winners and close runner‐ups, I find that contrarian victors benefit disproportionately from a public win – suggesting ex‐ante validation is crucial for commercialization. Importantly, even narrow victories yield large gains for contrarian contestants because contests do more than reveal intrinsic quality – they trigger herding among investors, who disproportionately bid for the newly validated approach. As such, skewed returns for contrarians, in part, reflect that investors support too few heterodox ideas and strongly reward successful contrarians. Finally, I show that mainstream researchers adopt these validated contrarian methods only after more decisive proof, allowing contrarian entrepreneurs to secure resources without facing immediate competition.