One of the most striking things is that cartesian products of objects do not correspond to set-cartesian products. This to me was mind-blowing when studying schemes.