That is what makes that kind of AI dangerous. It just does not care and / or need a reason we could understand. It may decide to visit us in 250 years and reach us ~250 years later.
Other possibilities:
- The sentient / strong AI may have (developed) FTL travel / signalling capabilities which from our current understanding does not seem to be possible.
- The AI may have sent out scouts millions of years ago into multiple directions. The closest scout only needs to decide to conquer / harvest our planet.
- A combination of the above, for example: FTL signalling and activating scouts where needed.