Winds from massive stars have velocities of ~1000 km/s or more and produce hot, high-pressure gas when they shock. In the traditional, spherical model of these stellar winds, this high-pressure gas can act to quickly disperse the dense gas characteristic of regions where massive stars are born, acting to halt star formation. However, this classical theory is inconsistent with observations of wind-driven bubbles in the nearby universe and the observed high star formation efficiencies of super star clusters. I develop a new theoretical model for the expansion of stellar wind-driven bubbles that accounts for the turbulent structure of the surrounding gas. A key feature is the fractal nature of the hot bubble’s surface. The large area of this interface with surrounding denser gas strongly enhances energy losses from the hot interior, enabled by turbulent mixing and subsequent cooling at temperatures T ∼ 104–105 K, where radiation is maximally efficient. Due to this cooling, the solution is momentum-driven rather than energy driven, with resulting pressures in the shocked wind that are lower by up to a factor of 100. I explore the implications of such a theory and present a large suite of three-dimensional, hydrodynamical simulations that have been run to evaluate and test this theory. I also present simulations of self-consistently star-forming clouds where star formation is regulated solely by stellar wind feedback. These simulations allow us to test our theory in a more realistic context as well as track how wind material cools and collapses into subsequently formed stars.