In the phase retrieval problem one seeks to recover an unknown n dimensional signal vector x from m measurements of the form yi = |(Ax)i| where A denotes the sensing matrix. A popular class of algorithms for this problem are based on approximate message passing. For these algorithms, it is known that if the sensing matrix A is generated by sub-sampling n columns of a uniformly random (i.e. Haar distributed) orthogonal matrix, in the high dimensional asymptotic regime (m, n → ∞, n/m → κ), the dynamics of the algorithm are given by a deterministic recursion known as the state evolution. For the special class of linearized message passing algorithms, we show that the state evolution is universal: it continues to hold even when A is generated by randomly sub-sampling columns of certain deterministic orthogonal matrices such as the Hadamard-Walsh matrix, provided the signal is drawn from a Gaussian prior.