Abstract: Systems and methods for autonomous device procurement are disclosed. A system can obtain point-of-sale (POS) terminal demand data and execute a machine-learning model to generate a forecasted inventory demand for POS terminal devices. The system can allocate the forecasted demand among devices based on ratios, simulate demand instances, calculate losses, and modify the allocated demand based on the losses. The system can generate a procurement plan based on the modified demand. The system may autonomously procure devices according to the plan. The forecasted demand may comprise a probabilistic distribution for multiple future time periods. The system may calculate a conditional value at risk (CVaR) based on the losses and generate the plan based on the CVaR.