Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/intel-14nm-capacity-issue.10766/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Intel 14nm capacity issue

Fred Chen said:
There have been reports, such as DigiTimes: Intel To Outsource 14nm Chips To TSMC (Updated), of Intel 14nm capacity being insufficient, linked to 10nm delays, and even driving a second source in TSMC (using their 10nm?)

It seems to suggest some of Intel's original 14nm capacity got converted to 10nm and is now stuck there. Now with the rise of AMD at 7nm, could Intel even lose 14nm business there?
Click to expand...

Yeah I read this. Seems like AMD is getting what they need right now!! The first to 7nm for GPU and CPU will help them big time!
 
Fred Chen said:
There have been reports, such as DigiTimes: Intel To Outsource 14nm Chips To TSMC (Updated), of Intel 14nm capacity being insufficient, linked to 10nm delays, and even driving a second source in TSMC (using their 10nm?)

It seems to suggest some of Intel's original 14nm capacity got converted to 10nm and is now stuck there. Now with the rise of AMD at 7nm, could Intel even lose 14nm business there?
Click to expand...

Intel is already a big TSMC customer so this is not a big surprise. I doubt it will be TSMC 16nm or 10nm, 7nm would be my bet. Maybe GPUs or FPGAs, probably modems, not CPUs. If Intel is going to supply modems to Apple it would make sense to put them back on TSMC. Intel AI chip (Nirvana) is already TSMC.
 
I have 3 guesses.

Demand is so large that intel can't meet supply. There's an employee problem at the fabs. Too many chiefs not enough indians and intel can't find people to do the grunt work.
 
The Digitimes story is fishy, and implies something could change quickly. I think it would take years, 2-3 years minimum, for major manufacturing changes to occur at Intel. Their new modem was designed for Intel 14nm. Since prior modems were TSMC it could be they will stay with TSMC for the modems (maybe stick with 28nm for a while longer). Who knows. Intel has obviously been really conservative with launching products on new technology nodes lately. I can’t see them working on both “fab-lite” and “save 10nm” strategies at the same time, although you can see how they might be tempted. If they do both, it will raise costs exponentially.
 
It's hard to get reliable labor in manufacturing in the US. Americans in general don't want to do that type of work. No one should be surprised the work has been outsourced to asia, people use alibaba and that what happens.
 
So up until now, Intel was trying to get into the Foundry business because it needed to fill the excess capacity and now it cannot meet demand? Something is going on here. I would be more inclined to believe that Intel's plans had heavily relied on 10nm be up and ready a long time ago and this hasn't materialized so they are in trouble in more than just being able to produce 10nm chips to be competitive. With AMD getting more and more sales, is it realistic to believe that the overall market has grown that much that Intel cannot meet demand? This in my opinion can only happen if suddenly they have less capacity. By the way I think this is can mark the absolute end of Intel Foundry.
 
One can reasonably assume that the planned transition to 10 involved ramping down some 14, either by allowing the equipment to age out or by taking over the building for new 10nm equipment. So perhaps if they are then required to spin for a while longer at 14 the problem is they do not have the capacity they need to just do the same as before.
 
I dont know if we can call it "fabless" but there are several products (mostly through acquisitions) that are not manufactured at Intel fabs. These include several FPGA product lines, modems, AI chips etc. All these have not been integrated to use Intel fabs and by the capacity issues, they are probably not pushing them to use Intel manufacturing. I believe (correct me if I am wrong) that almost all of these externally fabbed chips are being made at TSMC.
 
I think intel may be going fabless in 14 nm in general. I don't think they have a dependable labor supply at their fabs and Oregonians in general don't want to do grunt work. It maybe easier just to outsource the work to TSMC and let them deal with it. AMD may have the capacity advantage because they're fabless.
 
It's still a victory for fabless. I would bet my lunch on TSMC doing desktop processors for intel, it's only a lunch.
 
And the testing facility is in vietnam. There's drama at intel.
 
I thought there was a vacant building in Arizona and I find it hard to believe that the facilities in Oregon aren't capable of meeting demand. There's something going on.
 
You must log in or register to reply here.
Back
Top