Armour doesn't become invincible when A>P, it just takes half damage.
Dev dude, is this a binary test to apply the 'half damage', or a scaled linear test...? i.e.
IF: A<P
THEN: Damage
IF: A>P
THEN: Damage x 0.5
Or
func: Damage x (0.5+(A-P)) [or similar scaling function]
A = Armour value
P = Penetration value
Because, if it is a binary test, then that tipping point where A is approximately equal to P, means that for very little change in armour value from a tech upgrade your entire army goes from being able to attack 'once' to being able to attack 'twice' with a single tech upgrade either way. However as I'm sure many people will appreciate, the ability for anti-tank guns to piece various grades of armour was also as much to do with the engagement range of the gun with the tank, as the gun itself. It wasn't a simple case of 'it could penetrate the armour or not' (except with the most armoured variants).
Therefore in that context, the game mechanic if its binary could in principle be 'game breaking' since it forces you in effect to prioritise both AT and Armour values to the highest degree, because you are talking about a bonus of +100% effectiveness for your tank units, or not.
In that sense, it is totally gamey, and you are doing yourself a great disservice in not prioritising and not researching to full. Because this is the optimum strategy, all players will follow this research path, thus netting a 'zero' overall effect, thus making this mechanic worthless between majors, and only serve to nerf minors and the AI yet more.
-------------------------------------------------------------------------------------------------
On the other hand, I hope it is a linear scaling of effect, so that as Penetration passes Armour, you gain that 'double damage' mark, but say you were just below that mark, say 5P to 6A, then your 1.666 'double damage' for example.
This would be a better mechanic, because it would account for the fact that AT gunners could hold fire to closer ranges before firing, or very heavy armoured tanks could literally have shells bounce off them, much like the Japanese found to their horror against the T-34s and other tanks during the fall of Manchuria because they hadn't upgraded their AT guns since about 1934.
It would also mean that non prioritised AT guns can still have an effect, albeit at a lesser value, rather than AT guns being either "yes, can do!" or "No, your screwed!" if it is binary. Not to mention, CA bonuses as well as needing HA attack as well all adding to completely nerf minor armies...
I do strongly hope that it is a scaled test for the sake of the game.