Do I really need to quote you on what you've said? Okay then:
This sounds like you're saying that AMD will be "disrupting [Intel] repeatedly" this year while "rolling out their whole Zen 2 lineup on 7nm". Which just won't happen.
Your reading comprehension is really absolutely shitty.
Here, let me break it up that sentence for you:
"I'm looking forward to AMD disrupting them repeatedly..."
Self explaining. I expect AMD to repeatedly disrupt Intel. AMD already changed how Intel approaches the HEDT market, not only using their LCC die but for the first time ever also announcing the use of the HCC die for HEDT just to match/exceed what Threadripper bring to the table. Intel has not yet reacted to the treat that is Epyc.
"...until AMD is done rolling out their whole Zen 2 lineup on 7nm..."
Timescale until when I expect AMD to keep disrupting Intel. As part of its CPU roadmap AMD already announced Zen+ and Zen 2, the latter of which will be on 7nm according to the map. That came as a surprise as AMD is not done yet with the current roll out during which they are establishing new platforms, only desktop AM4 being done while Ryzen Mobile/mobile AM4 and APUs in general along it, Threadripper/TR4 and Epyc/SP3 platforms are coming in the coming months. The time needed is not so much for the chips themselves (which mostly rely on the same die) but building the ecosystem of the platforms that naturally are not there yet.
"...for which they apparently plan the tape out in this year already."
This was the second surprise when AMD CEO Su mentioned that they plan to tape out 7nm Zen 2 this year already, as the above mentioned CPU roadmap didn't come with a timescale and it seemed Zen 2 and 7nm would be farther off. The tape out of the current Zen core was in early 2016, so with that precedence a successful 7nm Zen 2 core later this year would have use see respective updates rolling out late next year. So their CPU roadmap covers quite a lot of action in rather little time.
You people seem to severely underestimate the actual complexity of the words you're throwing around. Look at how long it took AMD to build Zen1 and then apply the same time to what you're expecting from Zen2 because these "microarchitecture updates" aren't at all simpler than creating some CPU architecture from zero.
Of course from the ground up new designs (like Zen is) take much longer than incremental updates (like Zen + and Zen 2 are). AMD has been working on yearly refreshes for years now, AMD also has been working on modularizing all their IPs since the start of Bulldozer. The difference now is that Zen is actually competitive.
You also seem to assume that they'll continue to use the CCX strategy with Zen2 which actually makes no sense since the whole CCX strategy is a cheap way to make a whole lineup from essentially one CPU module - but the downside is the inefficiencies in both technical and economy sides of the equation. An expected way for AMD to evolve to Zen2 would be to produce a different module alongside optimizing the current one which would allow them to lessen the issues of cross CCX snooping in the most popular 8C configuration (for example) while retaining the older 4C CCX design for lower end of the market. So if AMD will use the same approach of 4C CCX for all CPUs of Zen2 lineup then they are automatically making Zen2 worse than it can be - and this would be a very strange choice as with Zen2 they will need all the additional IPC they can master as upgrading Zen1 in a meaningful way will be a whole different story compared to the insane boosts they've got with Zen compared to BD/PD/SR just out of thin air.
And here I think you are way off.
AMD's interest is not replicating Intel style micromanagment hell with over a dozen or so dies of different microarchitecture generations (they by far don't have the financial and man power for such redundant work), their interest is combining the high yield of small flexible but optimized modules with the capability to scale that across the full spectrum of their lineup. That's what they are currently building several platforms and their ecosystems for. And keeping all designs modularized allows them to refresh products across the lineup as soon as updates to the microarchitecture reach satisfying yields.
It's true that Zen is faced with design shortcomings regarding inter-CCX and likely also bigger NUMA (in Treadripper and Epyc) latency. But that's a tradeoff that both unavoidable (Intel's HCC and XCC dies with dual ring buses also fall prey to the same issue) and mostly negligible (datacenters worked with and optimized for NUMA for ages, and in the case of Ryzen DDR4 3200Mhz at low latency is able to mask most of the issue). That one tradeoff allows modularization that awards AMD the agility we see now despite it being way smaller than 1/10 the size and ever far lower the financial capability than Intel.