In order to be successful AMD needs to fine tune their release. Aim too high and it might be bad for profits and / or their image, aim too low and they don't make enough of an impact. That's the dilemma right now. How high can they clock Zen 2 on the current 7nm process? That's the big question, because that will to an extent dictate what they have to release right now to make an impact.
Intel doesn't have anything to counter a 16 core mainstream-ish part, only a 10 core at most on the ring bus, more than that and it will be akin to Skylake-X and not as good for gaming. Besides, does mainstream really need a 16 core CPU right now? In my opinion it's too soon and the performance and power efficiency for a mainstream part is probably not there yet.
A 12 core will thus do for now at the high-end if the performance is there, at say $499. That also keeps their profits reasonable.
The 9900K competitor can come in at $329 for the low power, non X version, and $399 for the X one.
If I were AMD I would wait for 7nm EUV to release the 16 core part, it will improve performance by 10% or bring a 15% efficiency improvement according to this article
, and they can thus wait for Intel's Ice Lake and respond at a more leisurely pace and fine tune things with time.