It is time to wrap this one up as well. Thanks for all the replies! As of time of writing, the poll shows 67% votes for "ride it out", with a combined 31% for Ryzen 2800X, and the remaining 2% for other hypotheses. Congrats to those who got it right!
Intel's 8-core i9-9900K has now been released, and indeed, there was no 2800X in response from AMD. It now looks like AMD will ride it out with their current line-up and lower pricing until Ryzen 3000 is ready.
With the latest EPYC 2 rumours pointing to an 8+1 chiplet design, the old speculations about a seemingly obvious progression to 12-core for Zen 2 and 16-core for Zen 3, by simply adding CCXs to the die, are all out the window. With all the focus on server, and little demand and need for more than 8 cores in the mainstream, the developments in the client space may look less exciting than before — although power, IPC and frequency improvements in 7nm Zen 2 may still get AMD in front. On the other hand, if Ryzen 3000 is based on reuse of the server chiplet with a separate IO chiplet in an MCM configuration, it may have compromises that keep it trailing Intel in areas where Skylake/CFL currently dominates, in particular games, Adobe software, CAD, database.
An interesting fact is that the prior SVP of AMD's client business, Jim Anderson, has left AMD since the launch of 2700X. This may be totally unrelated, and probably mostly is. But I have a feeling that he wasn't much inspired by the immediate future in the client space. The big splash for AMD was the return to high-end competition with the introduction of Ryzen. From here on forward, it will be less exciting, considering Intel and AMD more or less stay on par with 8 cores — especially if AMD continues to lose in gaming, which is the number one workload in the PC enthusiast space. Also, it must have been depressing to see the glacially slow adoption of AMD products with OEMs, despite very competitive offerings.