Except you're better off designing for a time-frame than a process node and waiting until it's ready.
HW design is based on process node (power, heat, game software), not time.
You have to come up with your whole marketing strategy, games and services, manage your existing console, etc.
All of which can be planned with an intentional timeframe based on expected node availability.
So let's say MS plan for a Q4 2012 launch. Come Q3 it's apparent the yields wion't be good enough and they'll ahve to wait 9 months.
They should know by late q2. If by then things aren't ready, it won't be 9 months off.
28nm is set to have retail high-end gpus (high power, large, fast) available q1/2012.
28nm should be ready for console production 6 months after q1/2012 when these high-end gpus are launched. If not, a few months more should do the trick. If come q3/2012, 28nm isn't ready, and it will take 9 months more, then TSMC will have produced the worst node launch in history.
As is, TSMC customers are saying the node is coming along better than 40nm. The node should not be an issue for late 2012. If it is, worst case scenario is they have to delay a few months.
In comparison, 90nm gpus launched side by side with xb360.
150nm gpus launched in the same year as xb1.
What do they tell the software companies who were developing launch titles and will be sat around making no money for 9 months?
They won't be waiting 9 months (see above) and it isn't as though xb360 and ps3 consoles won't be able to play the launch lineup. I see no reason why the launch would be any different than the xb360 launch where the majority of the launch lineup was ports of xb1 games.
What about the marketing costs of making those Christmas ads which will be shelved?
The ads will still be usable ... they aren't like food where they have a limited shelf life.
What about the BRD printing runs they've booked?
I'm sure MS would take a hit for a reserved run if they have such a contract in place, but it wouldn't be a crippling blow to their budget to have these runs delayed a few months.
What about the empty hole in XB360's marketing campaign for Holiday 2012 because MS expected to be advertising XB3 but it's not ready?
I fully see xb360 and xb720 coexisting and actually xb360 being more targeted for Christmas sales (kids/family/casuals).
xb720 would be ideally launched around q4, but it wouldn't break the launch by missing this window as far as demand is concerned.
There aren't many products or industries I can think of that plan around the timeframe of 'when it's ready'. Things are planned many months if not years in advance, and although there are issues to work around, timeframes are known and peopel try to stick to them - hence crunch time in game development rather than taking another 6 months missing the deadline without earning any money and wasting all that marketing.
Agreed.
Plan for a specific time (q4/2012). If it is ready, continue as planned. If not, wait.
None of the measures necessary for a console launch will be spoiled by waiting 3 months.
In the case of PS360, had both planned for 65nm and then waited for when it actually became available (different to 28nm I know because that's started now), they'd have launched way late!
Indeed.
MS launched toe to toe with 90nm GPUs.
They couldn't have launched 90nm any sooner than they did.
