Mining complexes are value chains where extracted material from different mines is transformed into sellable products through a set of processing streams. This value chain is governed by uncertainties in different aspects, from the pertinent geological attributes of the mineral deposits mined, to the different operational and processing components. Stochastic optimization formulations have been shown to maximize economic value and, at the same time, manage and reduce risk, thus providing reliable production plans and forecasts. However, related mine designs and production plans are static over the life of a mining complex and cannot include flexibility mechanisms to account for alternative, potential production and planning options. This paper presents a dynamic two-stage stochastic mixed integer non-linear programming formulation for modeling and optimizing a mining complex, including alternatives over capital expenditure investments and operational modes for different components of the value chain. More specifically, a dynamic decision-making mechanism is included, where mine production plans are allowed to branch, and parallel feasible plans are designed if a representative proportion of stochastically simulated scenarios of the mineral deposits mined conclude that it is profitable. This process generates new optimized plans that facilitate adaptation once more information is available. The practical implications of the proposed method are demonstrated through an application over a copper-gold mining complex comprised of one mine and six processing streams, where the dynamic model is compared to a traditional two-stage stochastic formulation, presenting a 10.5% increase in net present value.