By default, the power() function uses 64-bit integer dtype. It is too small for such a big number. You need to use float64 dtype that offers a larger but inexact, range of possible values.
Here is the fix:
>>> np.power(100,50, dtype=np.float64)
1e+100