"A9" name is already taken by ARM Cortex processor, which could also be found in mobile phones.
That's sad case of confusion, because it would take more time to google the A9-related things
That was just as true for the previous chips.
There was a Cortex A5, a Cortex A7 (VERY COMMON), a Cortex A8. All colliding with Apple names.
It's not clear to me whether or not there will be an A9X.
Pro: Everything points to the introduction of an iPad Pro, and the range from the "low-end" iPhone 6S to the iPad Pro is fairly large, especially in pixels that need to be manipulated.
Con: The iPhone 6+ pretty clearly could have used more GPU than it has. The iPhone 6S could also be fitted with a 3x screen (like the 6+) meaning it also needs more GPU. At which point, does the segmentation make sense? Just give everyone 3 CPUs and a big GPU, and segment by binning. Maybe the iPad Pro (likely to sell in smaller numbers, with bigger battery) just gets 300MHz over base frequency, iPad gets 100MHz over base frequency, and logistics is easier with only one chip?
The good thing is that Apple doesn't (IMHO) have to worry about artificial segmentation --- they don't care if you choose an iPad rather than an iPhone or vice versa (and would prefer you buy both) which suggests their primary concern will be what can be manufactured, not how to cripple one chips vs the other. My guess is that the A8/A8X split (which, remember, they had to decide upon maybe eighteen months before the iPhone6 shipped) was driven by concerns about how many CPUs they could physically fabricate, given (estimated, not certain) chip yields and how rapidly the fab can run. If those same estimates (how many will we sell? what do we expect yields to be? how many wafers can Samsung+TSMC process per month?) are different this time round, it may be quite feasible to accept the larger area of what would be the "A9X" and abandon the A9 as an unnecessary backup plan.
It's even possible that, right now, Apple haven't even made a decision yet --- they're waiting to see the Samsung yields and, perhaps, what TSMC can add to the volumes.