Looking at FX versus core i7 benches makes it look that is impossible for AMD to achieve the IPC of any Intel recent core.
Zen's architecture is very different.
The biggest concern appears to be the strength of the 14nm process more than the design. If the chip is well-designed but can't get enough clockspeed at a reasonable power consumption level (while keeping variability between chips low enough) then AMD could be in for quite a bit of trouble, particularly in the server space where performance-per-watt is generally taken into greater account than on the desktop.
The mobile space also, of course, generally places a significant premium on that, although there is more wiggle room since budget models are a thing. In fact, when it comes to those, OEMs sometimes reduce performance-per-watt significantly by doing things like sapping performance with the choice of single-channel RAM.
I would watch for benchmark shenanigans aimed at the desktop space, too. If Zen doesn't do well in AVX 2 or whatever then expect the
de facto community benchmark to lean heavily on it. That could include separate code for older Intel chips to mask the AVX 2 focus, for the clever benchmark maker. Give those just enough of a deficit to make it look like the benchmark is completely natural while killing Zen performance. I have a feeling AMD went for a more Intel-like design for Zen to minimize this factor. Of course, it can also be attributed to Intel's design being better. It makes plenty of sense for a company with greater resources to produce a better design. After the Pentium IV fiasco execs are naturally going to be more alert to maintaining competitiveness so people shouldn't expect such a big opportunity again for AMD.