A novel matheuristic approach for optimizing mineral supply chains under geological uncertainty will be presented. The algorithm combines local search heuristics and mathematical programming techniques. While local search heuristics are used to select the blocks to be extracted at each period of the life-of-the-mine, an exact algorithm inspired by Bender’s decomposition is used to optimize the flow of the extracted material through the downstream processes. In addition, the algorithm incorporates machine learning components to speed up the computational times and an adaptive component to guide the search. This approach should be able to handle the complexity of the problem and overcome the disadvantages of the standard approach, which consists of optimizing the upstream and downstream problems separately.