If we're talking about algorithmic complexity: for some algorithms quantum computers won't be any better than classical ones; for others, the improvement will be exponential, not multiplicative, say from O(eⁿ) to O(n).
If we're talking about actual performance, we have no idea, because there is no general purpose quantum computer yet.
So the mention of "orders of magnitude" belies a complete ignorance of the topic.
> Fully useful quantum computers have the potential to reach computation speeds that are orders of magnitude faster than today’s supercomputers.
It's not talking about the present, where IIUC quantum computers are basically useless for anything but research, not performing as well as classical computers because of noise and shortage of qubits.