I am happy that Intel has realized that the future is not just about the feature size of transistors but what the processor does. With its fixation on keeping the Moore's law alive, it got blinded and missed opportunities. I care not if the size of the transistor is 14nm if it can deliver excellently on mobile in power consumption.
This is really old news. Intel has been telling analysts and the press that it is cutting capital spending for a year now. If this "shocking story" is news to you, you haven't been paying attention. Go look up the stories that EETimes did last March. And again in June.
If this is supposed to be an indicator of more woes in the PC world, it's a year-old lagging indicator. Way to drive through the rear-view mirror, EETimes!
I understand that they plan now to invest for upgrades in the 3 existing fabs on site (Chandler). They had plans to upgrade the Ireland fabs which run at much older technologies. Upgrading these fabs to 14nm would be a major investment. It does make sense to us. It is cheaper to upgrade newer fabs which have newer technology nodes. The gap between the 22nm fabs and 14nm is much closer. Some tools of these fabs can be also used for 14nm. So, is Ireland still on the table and if so what is a good timing?
Intel does not need to build 14nm fab to compete. They simply need to make chips that combine processing speed with excellent power management. They can claim they are cutting capital investment, the fact is that Intel has a new leader who wants to pilot the firm to new business models
This is not a surprise. Intel has to balance their demand with their capacity. Building too much capacity causes excessive costs that show up on the bottom line. Upgrading an existing fab doesn't add as much capacity as bringing another fab on line. I'm sure they will bring it on line when the time is right. As for the older fabs, well in my experience, if they can't be upgraded, they will be shut down and sold off.
Is anyone else not particularly surprised by this move? It seems like Intel is doing what's encessary given the evolution of the market since they began their construction plans--but it's still a pretty signifcant indicator of the direction of that market nonetheless.
@Zewde: "Is anyone else not particularly surprised by this move?"
Well, consdering that Intel has been shifting to foundry business too --Altera, Achronix...--, it was clear that they were becoming aware of their productive capacity was already oversized for just building their own processors...
How exactly does this free up Intel to defeat ARM? It seems they are competing as much in their traditional PC space as they are in the lower-power and server segments where ARM is now moving with its 64-bit architecture.
>> How exactly does this free up Intel to defeat ARM?
It will be hard for Intel to defeat ARM. Amr is amorphous and you can easily beat it. It is not like AMD that you can attack its products directly. The business model of ARM makes it more challenging for attack just as Android in the mobile space will be hard for anyone to attack with its "open" source model.
Agree that ARM's business model provides a new challenge to Intel's. Though, for Intel to reclaim the market share (mobile space) from ARM is not impossible or far rearch. It just needs to play it right and leverage the advantages from their vertically integrated business.
And it's obvious, you don't attack just it's product (like Intel did to AMD). It's got to go after something more fundamental in this market. Collabration with other industry leader that is relevent in the entire system is the key.
Product means nothing without a strong and established eco-system.
I'll stop waisting time on this. It's not my bussiness to tell Intel what to do. Since, it's not like they don't know already. It's a matter of time for their plans to unfold.
>> I'll stop waisting time on this. It's not my bussiness to tell Intel what to do. Since, it's not like they don't know already
Absolutely. Yet, you are not wasting your time on this. Journalists, blogs, newspapers can build and destroy empires. Intel may know what to do but they continue to lose market share. Qualcomm commands more than 80% market share in the chipsets that power phones in U.S. I think honestly that Intel may not have a roadmap on this. If they have a plan, they would have shown us for the last 6 years. Recall that Qualcomm is in the 4th generation of their chipset and Intel is yet to launch one.
Finally, that is all we got. Just talk, write and comment. It does not change anything but that is life anyway.
Agreed. The problem is not one of comparison between products, it is a comparison of business models. That is a much bigger problem for Intel. If they cannot find a way to justify growth within their vertically-oriented, high-margin business model then they need to find a different one. The question is whether this is a 'shrink to grow' situation where they move to a new model or if it is just cutting costs to try to wring out a few more years on the old path.
"The question is whether this is a 'shrink to grow' situation where they move to a new model or if it is just cutting costs to try to wring out a few more years on the old path."
I think they should go the IBM way i.e. become a service company with high end fab capability (and spin out the bulk of their fabs as a foundry business). I think that's what's gonna happen in the end....
Wow Samsung seems to be everywhere in the race whether its smartphone or Fab. I guess their marketing strategy and the fact that they belong to other region makes lot of difference. But yes demand for PCs is decreasing and no one can deny that.
I am yet to see the real competitive advantage between 22nm and 16nm. All this quest to get down to the next lower feature size does not add much value. Transistor is free in the digital domain. The extra space saving is good but customers are not buying that. They are looking at great power performance which nanometer CMOS technology hampers by higher dyanamic and static power losses.
This makes much more sense. About 80% of 22nm equipment can be used for 14nm. To upgrade their 22nm facilities to 14nm appears to be a much better economical choice than upgrading their facilities which still run at 65nm or 45nm.