The U.S. Animal Movement Model (USAMM) is a Bayesian model used to predict livestock shipments between agricultural premises in the USA. Shipment information is crucial in epidemiological modeling of infectious livestock diseases (e.g. foot and mouth disease, African swine fever, bovine tuberculosis, etc), and this is the main use of the predictions USAMM makes. The model predictions are crucial for this since no complete federal database of livestock shipments exist in the U.S. The current version of USAMM was developed specifically for the computational architecture available on the Tetralith cluster, and is highly optimized for this resource. Despite this, model runs routinely hit the maximum wall-time of 7 days, limiting how much complexity that can be added. With the new resources available through Arrhenius we see the potential to leverage GPUs to greatly improve the computational efficiency of the model. This project would develop a pilot GPU version of USAMM to evaluate the feasibility of porting the model over to the GPU partition of Arrhenius. If this turns out to be possible, it would allow further additions to the model that would increase its usefulness.