The concordance model in cosmology, $Lambda$CDM, is able to fit the main cosmological observations with a high level of accuracy. However, around 95% of the energy content of the Universe within this framework remains still unknown. In this work we focus on the dark matter component and we investigate the generalized dark matter (GDM) model, which allows for non-pressure-less dark matter and a non-vanishing sound speed and viscosity. We first focus on current observations, showing that GDM could alleviate the tension between cosmic microwave background and weak lensing observations. We then investigate the ability of the photometric Euclid survey (photometric galaxy clustering, weak lensing, and their cross-correlations) to constrain the nature of dark matter. We conclude that Euclid will provide us with very good constraints on GDM, enabling us to better understand the nature of this fluid, but a non-linear recipe adapted to GDM is clearly needed in order to correct for non-linearities and get reliable results down to small scales.