I totally agree. That's what impressed me about the strategy - look on the customer boards and see what's next to the FPGA and either buy a company or strike up a close relationship. A while ago it was OTN. Altera bought TPACK and Xilinx bought Omiino for their IP. I'm sure that Altera has neither the desire or expertise to get into RF, so they made a strategic investment (not an acquisition in this case).
As you say, the possibilities of bring together different products from various companies fabricated on a range of processes opens up the promise of 2.5/3D.
The new trend here may be in snapping up technology that DOESN'T end up on the same chip as the FPGA but can help construct system modules (boards on a chip) with FPGAs as the 'glue' that connects the various pieces. Power, RF, SoC FPGA and memory all on a single module with an integrated design tool? What's not to like?