I think that another thing limiting it is the algorithm used to find the solution. Just from looking at the turns per second and the length of the machine's solve, it looks like it did about 75-100 turns to solve the cube. The God's number for 4x4 has a lower bound of 35 and upper bound of 57, which really shows that a better solving approach could definitely have been taken to reduce solve time significantly.
Is there currently an optimal algorithm for 4x4 like there is Korf's algorithm for 3x3? No. But an equivalent of an optimal solving algorithm for 4x4 created in the future would drastically improve this solver.