My prediction: PS5 will release Holiday 2019. A slimmer (cheaper) PS4 will release either 6 months before or months after the PS5 if not in 2018. Seeing as how 4K adoption rate is higher (at least it was) than 1080p was when it first started I don't see 4K adoption rates being an issue really. But then again there is a pretty big shortage of movies in 4K...

October-November 2019:

PS4 Super Slim $200 (maybe $150 on big sales)
PS4 Pro Slim $300 (maybe $250 on big sales)
PS5 $400 (may go 'premium' at $500 but less likely)

Give or take based on manufacturer schedules. PS4 still fully supported for a couple years but may gradually get dropped as 9th gen goes on.

November 2019. I feel certain about it.

I think Holiday 2019 is very plausible, 7nm Zen+/Navi will have been optimizing yield for more than year by then.
They can still in parallel sell low price point Slimmed Pro with x-gen compatability ala Scorpio/Bone or PS4/3,
so a higher than optimal price needn't overly tank unit sales, hardcore early adopters will be ready for new gen.
I don't see later than Holiday 2020.

Likewise interesting to see what MS will do, will they match PS5 intro date or wait a year+ and target higher $/spec?

depends how good scorpio sells. 2018 or 2019

MS won't really have a choice. All this realease stuff really all boils down to availability of tech.

Once the 7nm chips come along, sony can do one of two things with the PS5. Make an APU or make a seperate CPU and GPU which will be significantly more powerful but also much more expensive. Sony will not do the latter.

Now say 7nm chips come to market around mid to late 2019. Then it means 4nm chips will be at least 4-5 years away. If sony releases a 7nm based onsole in 2020, MS will have to follow suit as it would make no sense not releasing at the same time. Cause even if they waited a year or two after the PS5 released, they would still end up with a console thats similar in power to the PS5.

And lastly, the focus on power we have today will become very irrelevant come the next gen. Both console will be able to handle 4k games and at least at 30fps.

Isn't Pro basically PS4 Jag with higher clocks and Pit X2 with a few Polaris and Vega features on 16nm?

I don't see why PS5 couldn't be Ryzen + Vega/Navi, all in a 14nm APU. The node is smaller and more efficient, and the tech in the CPU and GPU is much more advanced and considerably faster than Jag and Pit. They could shoot for 8 to 10 Tflops, 2-3 years after Pro, which would be just another step like PS4 to Pro. I don't think PS wants to have such big power jumps like previous gens, just constant iterations of consistent power steps. This makes FC and BC easier.

This would most likely mean that PS wouldn't be able to cram all that into a tiny box, since the heatsink, fan, and supply would all likely have to be larger than what Pro has now. They would probably have to build PS5 bigger than Pro, so similar in size to the XB1/PS3 (internal supply).

*Edit (forgot to make my Scorpio point)

Depending on whether or not Scorpio is the same or newer tech than this gen, and what node it uses (14/16nm?) XB will most likely want to wait for 7nm to become stable with high yields before launching their next console. If PS5 by chance ends up being on 7nm, than XB will want to launch at that same node as well. Not to mention XB bashing PS about not wanting to launch consoles every few years and how they don't believe in that.

Agreed, and I don't think Sony is no siting around waiting for MS. I think Sony have their plans for the next 10 years when it comes to PS regardless of what MS and Nin is doing.They most likely sat down and determined they want a console every 4 years no matter what. That have been Sony's position all gen "in our own world, doing our own thing". That have worked out perfect for them, they are not responding or reacting to anything.

Tech solely dictates release date, yet Scorpio will launch on same process as Pro a year later?  

None of this is based on achieving "totally different class of power", any more so than Scorpio,
MS used this exact strategy with Scorpio, and as I indicated they do that with higher $ as well.
Because the extra year doesn't help with relative cost (vs Sony), although ABSOLUTE cost drops.
It is not just about "beating Sony", but making the most out of their existing platform base.
Xbone was selling half as well as PS4, so a year delay made sense to make most of platform.
Given Scorpio launched a year later than Pro, keeping the year delay is just the same gen length,
and trying to launch at same time would mean a shorter gen length on weaker selling platform.
Each platform has it's fans, so exact power parity at all times is just not be-all-end-all for either one.
The multiplats which are biggest sellers will likely continue targetting Scorpio even after next MS release,
never mind in first year when PS5 base would hardly justify AAA development on it's own.

I'm not saying MS must necessarily launch a year later, it just looks like it could go either way...
And the deciding factors have as much to do with financials and install base as "just tech availability".

I do agree the next gen will demonstrate the "decreasing returns" on perceived benefits vs numeric performance increases.
(which further underlines why it isn't catastrophic for MS to have a year launch delay with Scorpio or it's successor)
I would guess VR would be area where those performance increases would have highest impact on perceived benefits.

No.

PS4pro is using an underclocked polaris (36 compute units) GPU. And those only come in 14nm. The CPU could be built on a 16/14nm process but its definately not still 28nm. And yes, the CPU is still jag but the smaller fab process allows them overclock it while still mainting whatever heat margins they have.

One of the downsides with the Jag CPU is that its never really been higher clocks friendly. Which is why they probably couldn't get it much higher than where its at now.

While Ryzen is supposed to be a 14nm chip, Vega 20/Navi is designed to be 7nm. 

What you see in the PS4pro box, is the best they could do using the currently available 14nm tech applied to both the jaguar CPU and the polaris GPU. Yes, they could definately clock all that higher, but that would require an even bigger case and bigger cooling solution. And of course cost more. 

And it makes no sense shooting for 10TF.... reason being that polaris and vega will peak at 5-8TF on a 14nm process. So unless they go the extreme route of using a dedicated GPU and CPu (which is really expensive) there is no way they can hit 10TF in an APU chip form factor. 

Lastly, that 2-3yrs after pro you are talking about? Thats right around 2020. Ryzen+/Vega20/Navi which are all 7nm chipsets should be on the market around 2019. Cerny himself said it, a new console isn't defined by just a jump in GPU power but by the addition of a new CPU and signiicntly more RAM. All of which will be available around 2020 and that can be made to fit into a $400 box. Sony ending up with a 14-16TF GPU console and a significantly beefier CPU with next gen memroy and storage solutions would end up having nothing to do with what they want or not want. Its just gonna be down to where the market is at.

Look at it this way, If sony and MS were to make a 10-12 TF console in 2020, it would be the equivalent of them going out of their way to make underpowered hardware. Or unless of course they wanted to come in at a $250 price point.