Next generation air traffic management systems will be based to a greater degree on predicted trajectories of aircraft. Due to the iterative nature of future air traffic management algorithms, the success of these systems will depend strongly on the ability to rapidly generate trajectory predictions. By combining algorithmic improvements and high-performance computing hardware, Phase I research demonstrated significantly accelerated prediction of high-fidelity aircraft trajectories using the NASA-FACET software. Phase II research will build on the Phase I feasibility demonstration results to develop a full-scale computational appliance for rapid prediction of aircraft trajectories (CARPAT). The proposed architecture will combine the trajectory and airspace modeling capabilities of the FACET software with commercial, off-the-shelf high-performance computing technology. High-speed trajectory predictions and iterative computation of traffic flow management algorithms will be demonstrated under realistic traffic scenarios. The trajectory prediction appliance will commercialized during the Phase III work.