A technique for proving upper bounds on the matrix multiplication exponent by analyzing tensor operations.