As we continue to increase atmospheric greenhouse gas concentrations, surface temperatures continue to rise and hydroclimate patterns, including precipitation, are shifting. Making confident precipitation change projections is challenging due to large internal climate variability and the small spatial scales of precipitation processes. Here, we aim to understand the drivers of western North American mid-latitude precipitation change under large greenhouse gas forcing. To accomplish this goal, we analyze an initial-condition large ensemble from the Community Earth System Model version 2. We quantify the thermodynamic and dynamic contributions to precipitation change in each ensemble member. A machine learning method called self-organizing maps is used to identify common patterns of plausible precipitation change. While the thermodynamic contribution captures much of the forced precipitation response, the dynamic contribution arises from internal climate variability within the large ensemble. The differences in internal variability manifest as large-scale circulation patterns associated with variations of the North Pacific subtropical high. The largest difference across ensemble members is how far north the dry-to-wet subtropical to midlatitude transition occurs in the future. In sum, we show the spread of possible hydroclimate futures along the North American west coast that could occur within a single model initial-condition large ensemble. Overall, this work suggests that reliable projections of hydroclimate response to climate change will require constraints on the large-scale circulation response.