There is also one small exception to this. Yes, the AI 'cheats' as in it does not have to build ships. But the reinforcement points it gets are also some kind of resource. The reinforcement points are limited, and their effectiveness is determined by the number of guard posts and wormholes on a given planet.
What you can try on far away/uninteresting planets is to force the AI to spend reinforcement points there and not where you actually want to operate. The AI reinforces mainly planets which are on alert. So by putting planets far away on alert, you can 'suck up' a little reinforcement there.
Example (linear galaxy) A1--A2--A3--D--A4--A5--H--A6--A7--C
So you are on H, C is the AI core planet A are normal AI planets. Now assume the AI gets 4 AI reinforcement points per minute (or something to that effect). Then normally 2 would go to A5 and 2 to A6 which blocks your way to the Core planet. By destroying (or taking, the important thing is the destruction of the orbital command station) the planet D, the AI feals threatened on A3 and A4 too and will divert part of its reinforcements there. This means less reinforcements to the other planets (so in this example 1 point to each threatened planet).
With each reinforcement point, every reinforcable structure gets reinforcements (so guard posts, wormholes, command station, ...). By raiding the guard posts you will further diminish the amount of ships the AI gets. (remember: it does not get 50 ships, it gets 1 reinforcements). So attacking planets you are not interested in can be a valid strategy, but usually the effort is not worth the benefit if there is no further incentive to take that planet. (Well here you could take the planet and gather the knowledge around the planet and then abandon it)