I guess it depends how ambitious you want to get. Maybe the later time to market implies a more ambitious design, rather than just integrating existing design on the same die. We will see...
"if pad-limitations allow for the latter to take ten dollars cost out of the platform."
Hard to be pad limited for flip chip. There isn't an I/O ring like for wire bonding. The "pads" are distributed across the face of the chip.
"Bottom line, this industry has ever been about integration, which will eventually end in a SoC"
Sure. But that will also mean that the race for performance and/or features has slowed. Now there are some synergies that can be exploited, but I suspect they aren't all that great.