CVX does not allow the formation of that product.
However, depending on what you want to do with the resulting covariance matrix, there may be a way to accomplish that without explicitly forming the covariance matrix. For instance,x'*w*w'*x = square_pos(norm(w'*x)), which will be accepted by CVX (presuming w is a CVX variable and x is input data). And if that term appears by itself in the objective function. there is no need to apply square_pos.