All this tells me is that they are still having troubles with 10nm as they can only go up to 28w TDP max and 4.1 Ghz boost clock on 4C/8T parts. In other words, they still weren't able to get out of mobile territory, just like with the Cannon Lake affair. Now let's see if they can make these in volume anytime soon or if it's also smoke and mirrors like last time. If they can make them in volume, mobile is very important, so not all is lost.
The existence of the 9900KS tells me that higher core count with higher clockspeeds on the desktop is still many months away. 18% IPC uplift since 2015's Skylake arch (Coffee Lake refresh is still Skylake CPU cores under the hood) is decent, but they need the high clocks on the desktop too to remain competitive. Apparently only the 14nm++ process can do it as of now.
The existence of the 9900KS tells me that it's a higher binned chip than the already high binned 9900K processor. And that Intel made a mistake by jumping from stock speeds of the 9900K to all core boosts of 5 Ghz on the KS variant. The high factory clocks and the higher binning may yield a "efficient" chip in terms of heat output versus a 9900K clocked to that speed or more, but I theorize Intel made a mistake and hit a wall, whereas they could have offered 4.5 Ghz all core and then a 5-5.1 Ghz all core later to make even more money.
Calling it now, the 9900KS will be MSRP'd between $600 and 750 USD.
Whiskey lake with 4.8Ghz turbo already looks better then the next gen ice-lake (when u consider it's 14nm++ skylake) and soon comet lake is coming, so how good this 10nm CPU's gonna be against comet lake?