Köhler theory for droplet growth is presented in a form suitable to describe stochastic activation/deactivation in turbulent environments. The supersaturation experienced by individual cloud particles is separated into mean (resolved) and fluctuating (unresolved) parts. The mean part dictates the shape of the so-called mean-field Köhler potential (that describes droplet curvature and dissolved substances effects). The fluctuating part of the supersaturation drives the “random walk” of droplets radii along the mean-field Köhler potential landscape. In parcels that are unsaturated on average, the Köhler potential exhibits a characteristic potential well. Under these conditions, supersaturation fluctuations may aid a haze droplet to climb the potential barrier and grow beyond its activation radius (i.e., to stochastically activate). In this framework, we explore the interplay between fluctuating and mean-field thermodynamics during the supersaturation build-up in a rising parcel. It is shown that the feedback on vapor of stochastically activated droplets (under unsaturated conditions) buffers the increase of the mean supersaturation driven by cloud updraft. This significantly extends the distance over which droplets are activated inside the rising parcel. Importantly, the scheme developed in this work is ready to be used as a stochastic subgrid-scale model of the microphysics in realistic LES of natural clouds.