That's matrix-matrix multiplication. Nobody disputes that Strassen etc. have sub-cubic complexity. What about one of the six decompositions mentioned, as GP claimed?
I responded to a post about the practicality of MM multiplication methods, though GEMM is quite fundamental.