Gravitational lensing is a powerful tool to provide insights into dark matter, the expansion of the Universe, and supermassive black holes, that are inaccessible in other ways. Analyzing the features of extended lensed galaxies is sensitive to the overall matter distribution and its sub-structure, dark or luminous, and can constrain the population of dark sub-haloes. This is directly related to the properties of the dark matter particle (i.e. cold, warm, self-interacting, etc), and/or different galaxy evolution scenarios. Measuring the time delay in multiply imaged lensed quasars is an independent technique to determine H0, comparable in accuracy and complementary to other techniques, which can possibly lift the current disagreement between the latest Planck and type Ia Supernovae results. The onset of microlensing can provide a tomography of the quasar accretion disc and the direct black hole environment down to the event horizon, although it can contaminate the time delay signal if not taken into account properly. With the advent of all-sky surveys like Euclid and LSST, the number of known lensed systems is set to increase from a few hundreds to several thousands, promising to provide robust and exciting new results. In this talk, I discuss the new possibilities of gravitational lensing science in the large survey era and the associated challenges.