In this talk, we discuss the complexity of a class of minimax stochastic optimization problems. For random right hand side with partial distribution information on the first two moments, this problem is NP-hard. We review some techniques to tackle this problem.