I don't think Agilent would be making an ASIC for the 2000x/3000x series if it wouldn't have a reasonable ROI within that series. In the last 5 years, the market hasn't been turned upside down that severely and it looks like they can reuse it for the lower end model now. That means they can probably make these ASICs for rather low cost in the 1000 series scopes, and probably push the higher series forward in a few years time.
FPGA's and in particular the Zynq are not cheap ICs. The Zynq is a 28Nm chip, the MegaZoom IV is a 90Nm chip. But I bet the ASIC will be probably a smaller die, less of a power hog yet faster than a Zynq. I'm not even sure if you could fit that much functionality at that performance in a Zynq; 1M wfms per second with hardware decoding, segmented memory and all the other bells and whistles? It was
published here that the chip contains 6M gates. Now some of that could be the embedded memory, but I wouldn't be surprised if the actual logic makes up a couple of Million gates as well.
The Zynq is a nice chip, but the lower end chipsets only compare to like a mid range Cyclone II in terms of memory and logic resources. Don't forget that because FPGA's are re-programmable, they got a bucket load of gates and transistors to make any kind of usable logic. This costs area, thus money, reduces speed, draws power and overall makes FPGAs unattractive in mid to high volume products.