The large D limit of vector theories and the large N limit of matrix theories have been invaluable tools to study non-perturbative aspects of both types of models. The large N limit and its associated sum over planar diagrams is unfortunately very difficult to study in the most interesting models. I will explain in this talk the recently discovered non-trivial relationship between large N and large D. It turns out that the maximum power of D of a given Feynman diagram has an upper bound that depends on the genus of the diagram. This remarkable property allows to define a new large D scaling, in which some couplings are enhanced compared with the standard large D scaling of vector models, which provides a non-trivial expansion of the sum over Feynman diagrams of a given genus in powers of 1/sqrt{D}. The leading large D planar diagrams are so-called “melons” which can be summed exactly. These results make the link between tensor model technology, SYK-like models and more standard matrix quantum mechanical models for black holes. The new large D limit is also strongly reminiscent of the large dimension limit of general relativity recently studied by Roberto Emparan and collaborators and may yield to a tractable holographic theory in an infinite number of dimensions.