Historic Data

I think that absolutely the PS4 'Neo' will be the default/only PS4 model within a year at most. It simply makes no sense and is actively counterproductive to continue to offer both variants unless they had a ludicrous amount of leftover parts. In a mark-to-market world of 'lean manufacturing', smart corporations avoid that kind of thing like the plague though, for obvious reasons.

It's not like a PC universe where say Acer can bulk buy $40 Pentium Dual-Cores for a low end box, $90 for i3s, $135 i5s, etc. Sony has to get their APUs from Global Foundries (and perhaps now Samsung as well if they have any new fab time to spare), and there simply isn't a lot of sense in the semiconductor industry in continuing to run old-gen fabs. So 28nm 1st-gen X1/PS4 APU production has been EOL'd, and probably for a while now. 

If there was a replacement for the original spec APUs that were simply die-shrunk as in previous gens AND the new higher-performance APU, then I could see there being two models run for a while just to see how that would play out, but that die-shrunk basic APU got cancelled a while back :

http://www.extremetech.com/gaming/219859-samsung-globalfoundries-to-fab-next-gen-amd-gpus-apus

The flip side of that is that the X1 will need a new APU revision as well, as depending on EOL'd 28nm is a no-go. If it weren't for the ESRAM situation, they could get away with nearly the same design, but they will have to be more creative in their solution.

Anyway, there are many other intricacies of the industry that cement my perspective on this, but I would give the following predictions :

(1)- PS4 'Neo' will be the definitive/only PS4 sold within a year of debut, and probably less than that. I'm talking base hardware, obviously different bundles and HDD sizes could/will be offered. (99.5% certain on this)

(2)- Sony will release the PS4 Neo in a very understated manner in terms of marketing/announcements. I expect not to hear a whole lot of specifics about gaming in particular, but more on VR/Media/4K Video. I believe that this serves fairly closely to the 'slim' revision of gens 6 and 7 for them more than it does a new way of doing things. Practicalities and realities of production and fabrication made this decision for them rather than a grand design. (XX% certain because this is somewhat intangible)

(3)- PS4 'Neo' will launch at no more than $399, and quite possibly at $349, maybe 50/50. Depending on their timing in getting PS4 OG's cleared from the pipeline and inventory system, it may even drop directly in as a replacement over a matter of weeks at the exact same price of the OG PS4 that it is replacing, with the OG's getting a temporary $279 bundle or something. (99% sure $399 or less, 60% sure $349 on debut)

(4)- People will complain about this, that, and the other thing as per usual, and 'Neo' editions of older PS4 titles may take a while to appear, if at all. The ones that do will be heavily scrutinized by the likes of DF, and a new wave of console tech warring/arguments will emerge after basically fading out from the craziness of 2013-2014 times. Double this madness if Xbox 1.5 also sports a beefier APU and has patches/modes to improve 1080p/framerate/etc on some titles. (10,000% certain of continued internet hijinks and trololo moments, lol)

Potatoes!

What's the historic data for the PS2? Not that I'm suggesting you're being selective.

Point was to show that it becomes harder to sustain growth rate as you go higher.. Sony managed to make a jump from 10.06 to 13 million (~30% jump) but from 13 they only managed to go upto 14.3 million (~10% jump) and never managed to even break 15 million.

So in %age term this 20 million target may be only 15% higher then last year but i think its a massive jump.. And i personally 18.5 million is the top PS4 can reach at $299..

Well, with Sony saying they're going to try and emulate the success of the PS2 and the PS4 is more or less selling as well as the PS2 so far, I thought it would be fitting to compare it to the PS2. It just seems to make the most sense to me.

While NEO decreases in price, PS4 will decrease in price as well. Even when NEO drops to $349 or $299, PS4 will still be $50-100 cheaper and thus NEO won't be the go-to model. Pricing it $100-$200 more was a mistake imo. And if they underprice NEO too much to make it the go-to system, then they will be losing money. 

Who's going to keep making them 28nm APUs for vanilla PS4s? You do realize that even with a more powerful APU, the Neo will be cheaper to produce than the 1st gen model? That's how die shrinks work, if the yields aren't horrendous, and industry reports on GF's 14nm process have been outstanding :

http://wccftech.com/globalfoundries-14nm-finfet-amd-zen-2016/

That's obstensibly about Zen, but it's the same fab generation that GF uses for all 14nm products from GPUs to whatever, they're an independent specialist who collaborate to make the processors for various major buyers from around the world. You and I could contract with them if we had a design and the right gigantic pile of cash. A contract would be drawn up for X amount of processors, including iniital prototype spins, respins, final samples, and finally a go to full production with a specific yield and volume target. We would have to agree to pay for the period of the contract, and they would fulfill the production for us through that period. As the contract period went through, we would be in contact with them about their next gen process (now 10nm on the table for ~2020ish), and have to make the decision to die-shrink our design, or to use the opportunity to fit more gusto onto the processor at the same or less power/heat profile on the new node. Sony has gone for a middle ground approach.

In previous gens, contracts were made to remake 360 and PS3 processors using new process tech, thus reducing heat, reducing price dramatically, and making it so you could fit smaller and smaller cases.

In this gen, plans were initially on the table for an interim direct die-shrink of the APUs, but that got scrapped when the entire node was dumped. So there IS NO die shrunk vanilla X1/PS4 APU that exists. To beat the dead horse : that means that only existing stock of old APUs is available for vanilla models, unless they decided to use the full 'Neo' and 'X1.5' model 14nm APUs to power them, and just feature disable or laser cut them to intentionally reduce the performance to vanilla spec. And THAT would still involve more effort and expense than just dropping vanilla models completely.

The 'Neo' APU on 14nm is a lateral move between a pure traditional die-shrink that sees the same number of transistors but a corresponding reduction in power/heat requirements and physical package size, and a pure performance-oriented generatioonal evolution where a maker decides to shoot for the maximum gains in performance by using a larger transistor count but still aim for the exact previous-gen power/heat properties. With the 'Neo' APU, they appear to have kept the CPU cores basically identical to the original spec (so they will be directly shrunk 50% in size, and correspondingly drop in power/heat requirements, but clocked higher, which uses some of that new headroom in power/heat), while increasing the GPU portion of the processor in both size AND clockspeed. The result is straddling the line somewhat, hypothetically this design should use a bit over half of the 1st-gen APU's power and heat profile, while being a fair bit smaller in square MM in size.

The games media thus far has looked at it as a separate new product, which makes sense to an extent. With the adoption of HDMI 2.0 and the imminent release of BD4K media, it's an ideal time to 'refresh' the PS4 into a 4K content portal. It's also a pre-emptive choice in the event that Xbox also had to make the same choice in APU replacement, and gives them a way to combat any potential threats to their perceived/real performance in games should it come to that. A pure 14nm die shrink with no increase in CPU/GPU performance might have left them open to Xbox marketing their 1.5 model as a superior performing model. That's a separate issue, but I very much doubt that either Sony or MS want to re-open the performance marketing battle. I expect a very understated marketing campaign for both that focuses only minimally on gaming improvements, and speaks more generally about the improved user experience and capabilities.

Suffice to say though, that 'Neo' should already be less expensive to make than PS4 vanillas, and that's about to go form 'less expensive' to 'the only thing they can make without getting someone to make ANOTHER new 14nm pure die-shrink of APU 1.0 for them'. And just the validation/testing/production of that along with running another PCB for 2nd gen vanilla PS4s would STILL be more expensive in total than just using their produciton capacity at GF for ONE APU.

I'm looking at it from a perspective of having followed semiconductor developments from the manufacturing and design side since the late 1980s. From this side of the industry it's a little easier to see how production realities affect what is possible, practical, and advisable with products through the years. And it's true from desktops, to laptops, to tablets, to phones, to set-top boxes and so forth. It's all stuff that now only a handful of fabs in the world can really make for you. Intel, Global Foundries, TSMC, and Samsung are the only premiere ones, and only GF and TSMC are the real 'guns for hire' in the fab side that can truly make volume for companies like Sony and AMD. Samsung keeps most of their production capacity for their own products, Intel is basically fully closed to everyone else, and then there are some bit players like UMC who are generationally behind the others, but who can make perfectly fine products for devices like vehicle ECUs, refridgerator/appliance processors, yadda/etc.

TL:DR Summary :

PS4 Vanilla 28nm APU is EOL with no die-shrunk replacement

PS4 Neo 14nm APU, even with more transistors/larger GPU area IS STILL SMALLER (thus more per 300mm wafer) than vanilla 28nm APU

Result : PS4 'Neo' will inevitably be the only model on offer in fairly short order AND it is a less expensive model to make than 28nm PS4 1.0

If that's the case, I think it should release the same price as the PS4, and be advertised as a PS4 slim with a slight performance boost. I don't think pricing it $100 - $200 more and then having to rely on excessive price cuts to make it the default model is the right way to go about it. It confuses customers and segregates the userbase whereas if it were advertised as a PS4 slim with a bit more power and was made the default model from day one that would make a very smooth transition.

Unless this is just a short-sighted move to get good profit margins off console sales at the risk of consumer trust and marketshare. 

I have a hard time believing NEO will be cheaper to manufacture than vanilla PS4 on day one in the first place, but I also don't follow hardware specs and all of these technical details at all like you clearly do, so just for sake of argument, I'm going to assume what you said is mostly correct. But then again, I also wouldn't have believed that a gfx card more powerful than a Titan X would retail for $379 lol

Are you also saying that you don't really believe that NEO is making any kind of statement about where Sony or the industry as a whole was going, but rather was just making the natural, cheapest choice in reaction to the shakeup at production within AMD? 

With Gran Turismo Sport, Horizon Zero Dawn and Final Fantasy XV all launching this year, combined with a decent pricecut and hype surrounding VR, it's certainly possible. We're already up YoY through May, so I can't see why the trend can't continue with a healthy string of releases.

This is without counting surprise release announcements... We might actually see Resident Evil 7, Dragon Quest XI and Mass Effect Andromeda launch this FY.

I don't believe that was ever confirmed, was it? I've heard various conflicting rumors, but no official confirmation.