Indeed. If current frontier models are already so disruptive now, we may have vastly better models in a few years but also the current models will have become significantly cheaper. Exponential growth (or shrinkage) is definitely at play here, and should not be underestimated.