D-instantons generate non-perturbative corrections to string theory amplitudes. These corrections are of particular importance because they encode non-perturbative effects in string theory and play a prominent role in moduli stabilization. In this talk, I will focus on D-instanton contributions, more precisely D(-1) and ED3 instanton contributions, to the superpotential of low energy effective theory of string compactification. Computing such contributions is challenging, because of the IR divergence due to zero modes of the D-instanton and the fact that some instanton contributions are expected to be generated in the presence of non-trivial RR and NSNS fluxes. To compute the D-instanton superpotential, one must compute the one-loop determinant around the saddle point. Due to the fact that these saddle point solutions have zero modes, the overall normalization of the superpotential contribution is ambiguous within the worldsheet formalism. I will discuss how one can use string field theory to regularize the amplitude to obtain an unambiguous answer. In part two of the talk, I will explain that the D(-1)-instanton contribution is expected to be generated in a non-trivial flux background, hence the direct computation of such effects is currently out of reach. I will show how one can use F-theory to compute the D(-1)-instanton superpotential and argue that in the global Sen-limit the D(-1)-instanton superpotential is absent.