Extreme precipitation events have the potential to threaten physical infrastructure, property, and human lives, and are predicted to become heavier due to climate change. Understanding past, present, and future precipitation is important in analyzing how precipitation risks change spatially and temporally. The observational record, from which point precipitation frequency estimates such as NOAA Atlas-14 are derived, is limited by its lack of spatial coverage and it represents just one realization of past climate. When using a high-resolution large ensemble global climate model, we have multiple realizations of climate, consistent spatial coverage, and the added benefit of being able to incorporate climate change scenarios into precipitation risk analysis. Here, we use the GFDL (Geophysical Fluid Dynamics Laboratory) 50-km horizontal atmospheric resolution global SPEAR (Seamless System for Prediction and EArth System Research) 30-member ensemble to analyze how U.S. 24-hour precipitation extremes at various return periods change over the 1921-2100 time period. We quantify extreme precipitation risks across the U.S. and locally under different climate change scenarios (SSP2-4.5, SSP5-8.5, and natural forcings alone). With the large ensemble, we are also able to explore methodology and uncertainties in characterizing extreme precipitation risks.