In mining complexes or mineral value chains, materials flow from extraction sites (mines) through crushers, stockpiles, waste dump and tailings, and processing plants to supply minerals to customers and market. New developments over the last decade have developed new advanced technologies for the simultaneous stochastic optimization of a mining complex, integrating upstream and downstream decisions into a single stochastic optimization model, that also manages the related supply (geological) and demand (market) uncertainties. This comprehensive approach addresses shared resources and operational interdependencies but results in large-scale, NP-hard problem that challenges existing solvers. To further tackle computational complexities, a novel framework leveraging contextual stochastic optimization is proposed to decompose the simultaneous stochastic model into interconnected upstream and downstream components. The proposed upstream model relies on pertinent block properties treated as random variables influenced by simultaneous model parameters, effectively capturing blending and processing activities. In particular, decision rules and end-to-end methods of contextual stochastic optimization are proposed. This framework enhances decision quality, manages risks, and offers a scalable solution for optimizing mining complexes under uncertainty.