Both the PS2 and PS3 launched onto fairly mature nodes. The Wii launched onto a mature node and so will the WiiU (at least for its CPU).
There's no clear pattern of needing to be in there right at the start of a new node.
As I said, the 28nm process isn't a roadblock.
If MS chooses to launch in 2012, they can.
If the timing for the new platform doesn't suit then the Xbox 720 will be different, I'm pretty sure of it. As I see things:
- You time the platform for the market.
- The market doesn't give a damn about the node.
- Shareholders do care about profits.
All well and good, but the planning process takes a while...
Need to get the design down (part of this is which node are we designing for?)
Need to get the Industrial design done
Need to get developers on board for the target design
Need to get suppliers in order
Need to get marketing team on it
It all starts up front with targeting a certain performance spec which is intimately tied to a node (or half node).
Granted, one can do all of the above, realize the timing isn't right (as the market or node process isn't ready for a next-gen console) and wait. Or they can wait even longer and design for a smaller node (pushing back the launch significantly so they'd better be damn sure not to open an opportunity for the competition).
But one can't say, "hey the market is looking for a new console now so let's get one together asap!".
As far as market receptivity to a high-end console at the moment, I'd say we are pretty much there.
(see anticipation for BF3 high-end pc demos)
I don't understand why your focus is on transistor density rather than chip size, frequency scaling, performance per watt and yield which all have either a very strong tendency or certainty of increasing significantly over time.
Density equals cost after yields issues are sorted out.
If the transistor density stays the same but the die grows by %30 and the frequency grows by 25% and perf/watt stays flat while yields increase by 90% then you're probably going to be rubbing your hands together, dancing round the campfire naked drinking moonshine.*
Better yields = cheaper ... yep.
But at some point, analysis needs to be done to determine yield costs saved by waiting vs profits lost by waiting and launching alongside or after your biggest competitor.
It's not just yields in a vacuum.
1 year of sales and mindshare = X $ lost from low yields
The yield factor is still a big unknown as TSMC hasn't shipped complex (gpu) 28nm product to analyze yet.
Take a look at AMDs fastest 40nm lineup and see how things went in the first 12 months. Now take a look at the 480 and then the 580.
ok ...
Now imagine what a 40nm to 28nm shrink plus 15 months of node maturity with bigger chips, faster clocks and better yields would bring. Even if you ended up in a PS2 vs Xbox situation shortly after, so long as you have the right platform you'll be laughing all the way to the bank.
Initial shrink showed near doubling density, unknown yields, and ~75% increase in transistor/watt
After a year on the market, 40nm showed a ~25% increase in transistor/watt
Waiting for that extra 25% seems a bit ... wasteful.
That's "a dollar waiting on a dime" as the saying goes.
Seems MS have had some pretty good luck recently.
(On a 90nm system!) 
That technology is already here.
It's neither here nor there in the discussion at hand of "when will they launch?" and "what will be in the box?".
My point was that to try and replicate the ability to launch a new console on the back of a new gimmick such as what Nintendo did with Wii without regard for cutting edge technology (or technology at all) leads to a risky proposition which paid of for Nintendo, but it was the first time in history and as you mentioned, MS has already traveled that road to the dead end and put up the detour sign for everyone else.
Now unless there is some other magical technology/gimmick that will get people to buy consoles in droves (an ipad knockoff won't do it, Nintendo!), MS and Sony will have to stick to the proven path to sustained sales and profits ...
Top-end Hardware.
