Intel CEO: Intel open to fabbing for mobile competitors

[soccerballtux]

In the words of IDC, competition leads to inefficient allocation of resources. Multiple companies paying to research the same thing.

This would actually be a really interesting pro competitive move of they allowed AMD to use their fabs.

 

[Mondozei]

Considering that Intel's mobile node leadership is evoporating before their eyes it actually makes a lot of sense.

Also, Brian probably understands that Intel isn't going to win the SoC game. Mediatek's stuff are now better than their high-end stuff. Samsung's doing a lot better. It isn't just Qualcomm these days.

It's not that Intel is putting out bad stuff. It's just that the other guys move faster and iterate faster. And with Samsung's catch-up in mobile all but inevitable, Intel's last advantage are their node leadership for the moment being. Better exploit that while they can.

 

[AtenRa]

Why Intel's Mobile SoC competitors will fund Intels fabs when they can use pure play fabs without worrying of making Intel a stronger competitor ??

 

[Phynaz]

Why Intel's Mobile SoC competitors will fund Intels fabs when they can use pure play fabs without worrying of making Intel a stronger competitor ??

There could be scenarios where Intel's manufacturing upside outweighs the disadvantages of giving them funding.

 

[AtenRa]

There could be scenarios where Intel's manufacturing upside outweighs the disadvantages of giving them funding.

Since TSMC, Samsung/GloFo are spending lots and lots of money in Fab R&D to compete against Intel and against each other, i find it extremely difficult to find any scenarios that will benefit those big Mobile players to go and use Intel Fabs this days.

 

[III-V]

Since TSMC, Samsung/GloFo are spending lots and lots of money in Fab R&D to compete against Intel and against each other, i find it extremely difficult to find any scenarios that will benefit those big Mobile players to go and use Intel Fabs this days.

How do you find it difficult? They'd get a huge jump on their competition without paying any more for it (other than the overhead of switching and the learning curve involved).

 

[AtenRa]

How do you find it difficult? They'd get a huge jump on their competition without paying any more for it (other than the overhead of switching and the learning curve involved).

At what cost ??? Intels 14nm is extremely expensive and will remain more expensive against the competition for its entire life.

Unless Intel is willing to do another contra-revenue stand that will cost them more Billions than the first one.

 

[Khato]

At what cost ??? Intels 14nm is extremely expensive and will remain more expensive against the competition for its entire life.

Yup, design and mask costs for Intel's 14nm aren't really going to decrease with time. But then again, the same can almost certainly be said for any comparable process from Samsung or TSMC... at least not until there's a lithography advancement. Which is one of the many reasons why Samsung and TSMC are delaying the typical density increases past 20nm - if they can only get a few customers to use a node and it will be abandoned once it's not leading edge it's not as economically feasible. So instead slow down the density increase and add FinFET to provide incentive for high volume customers to switch from 28nm. Then just hope that EUV is viable for the critical layers of the next node... or as TSMC has indicated they might do, bring EUV in mid-node, which has the same effect of transforming a cost-prohibitive process to one that's more accessible to the customers who don't have a need to be leading edge.

 

[Phynaz]

At what cost ??? Intels 14nm is extremely expensive and will remain more expensive against the competition for its entire life.

Citation needed.

But you would have to ask the companies that are already using Intel's fabs.