I will discuss the current understanding of the power of holographic algorithms with matchgates. The focus will be on the Universality of this class of algorithms for those #CSP type problems that are #P-hard in general, yet solvable over planar graphs.