Some Context to this, its not entirely fanciful, in fact its very real.
https://www.globalfoundries.com/technology-solutions/cmos/performance/7nm-finfet
On the left their current 14nm LPP FinFet, its designed to run at 3Ghz operation, this does not mean 3Ghz is its maximum operating frequency, its simply that this is its optimal power to performance ratio, AMD have one CPU which operates at the 3Ghz frequency, the
R7 1700 This is a 65 Watt CPU, which for an 8 core 16 thread CPU with an IPC between Haswell and SkyLake/KabyLake/CoffeeLake is impressive, at least in my opinion.
Other CPU's in the lineup are effectively overclocked, to do that is actually very common these days for all 3 semiconductor giants, AMD/Intel/nVidia
AMD current problem is two fold, they do still have an IPC Deficit to Intel, all be it a small one to CoffeeLake and it seems only in low threaded situations, AMD actually has better Simultaneous multi-threading than Intel so where that is the deciding factor AMD's performance is right up at least on par with Intel's CoffeeLake.
The IPC difference between Ryzen and CoffeeLake really is give and take. personally i don't think that's actually a problem, i think people just like to point at performance differences and call it IPC when in fact that difference is nothing like a subtle as that, its blatantly a clock speed difference, even reviewers who i respect and trust do this when they really should know better but actually don't.
Which brings me to AMD's real problem Ryzen vs CoffeeLake, Intel's 16nm is far more mature than Globalfoundries, their operating frequency design is probably 4Ghz or more, so their CPU's are able to run at much higher frequency, as high as 4.7Ghz provided you have adequate cooling, that's a massive 30% higher clock speed out of the box than AMD's 3.6Ghz Ryzen 1800X.
AMD have said Ryzen is worst case scenario, I agree, its a brand new immature architecture built on a brand new immature processing node not designed for products like Ryzen, i think AMD did remarkably well to get it all to come together and work, a lot of experts said it never would, there was a lot of nay saying with Ryzen that it would never happen even close to AMD's claims, it exceeded those claims, a target of 40% IPC gain over Excavator turned into 52%, 70%+ over the FX series, 2.8Ghz on all cores turned into 3.6Ghz+, a maximum of 2.6Ghz Ram turned into 3.0Ghz to 3.2Ghz common and i have seen as high as 3.6Ghz, AMD's SMT wouldn't be as good as Intel's because Intel have 15 years experience with it, its actually better.
So what next? if you look on the right hand side you will see Globalfoundries new 7nm LP FinFet, this is what Zen 2 will be built on. its operating frequency is 5Ghz, thats way faster than CoffeeLake.
AMD have said there will also be a small IPC bump with Ryzen+
So the prospect is a CoffeeLake IPC+ 5Ghz out of the box Ryzen+ /2, the high end 'mainstream' may even be 16 Core 32 Thread as AMD seem to be moving to 16 Core 32 Thread CCX's.
https://forums.overclockers.co.uk/threads/canard-mentions-second-gen-epyc-if-true-nice.18798418/