Interested in people's thoughts who have followed this field more than I have......
Basically, does Intel purposely limit its "advancements" in order to milk as much $$$, or are there truly R&D limits in terms of how quickly they can advance?
For example, if Intel wanted to, could it be producing 22nm or 16nm chips today, but it chooses to have a nice glidepath so it can get $$$ from a full generation of 45nm, then 32nm, etc? Or are they truly just not able to actually perfect a 22nm or 16nm processor today, but they just predict they will be able to do so in a few years?
I was just wondering as I read about their roadmaps out 3-5 years.......
I'm sure they do that to a point absolutely. Companies are already working on computers that use light or lasers to move information instead of electricity. I'm sure Intel has plans for it's future well in advance.
The answer to your basic question is YES (to both questions in limiting advances and having true R&D (as related to transitioning to production) limits).
Like any business, they spread out developments to optimize production runs. They also look at how long it takes to generate the revenue to cover expenses and how much is needed to produce the next good ideas (like transitioning to small production processes). They also tend to watch what the rest of the market is doing to take advantage of others figuring out how to do some things.
These practices are common and not specific to Intel.
I know - but the question is are the purposely taking every little STEP along the development curve even if they wouldn't need to. ie does one need to invest a gazillion dollars in a 45nm fab line in order to successfully develop 32nm technology a few years later? Or if they WANTED to, could the accelerate the pace of development by skipping a step....ie go straight from 45nm to 22nm and skip 32nm. So they could leap even further ahead, but then they have only collected 1x revenues whereas if they purposefully hit every step, they can sell everyone a 32nm chip, then 3 yrs later they can change the socket and sell everyone a 22nm chip.
If they rush ahead and skip a step, then they've furthered their lead against competitors, but they've also lost a revenue opportunity and gotten closer to the point of uncertainty (ie what happens if we can't figure out how to improve on our current offering).
Its probably a combination of both factors. I was just wondering whether, for example, any given step (ie 32nm) is actually unnecessary but they choose to hop on it for a couple years just to make sure they extract the dollars there before continuing down the curve.......
I'm sure Intel and many other tech companies could release newer, better, faster stuff more often, but that would be throwing money out the window. Where's the incentive to do that unless competition pushes it.
IMO, skipping a process generation (i.e., going from 45nm to 22nm) would be a pretty risky business proposition, as there are problems encountered in not only in the design rules but also ramping up production to the required volumes. If the bulk of your production is stuck on 45nm due to some problems with yields on 22nm, and in the meantime your competitor is cranking out 32nm cheaper and better performance than your 45nm lineup, you're in some plenty deep doo-doo, sorry for the technical jargon .
Also there's the matter of putting in years of R&D into design and fabbing - a company has to start getting some of that back as a return on investment, otherwise the stockholders would have the CEO's head, or other body parts..