Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/samsung-targets-chip-fabs-without-human-staff-by-2030.19396/
        )

    [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
)

Samsung targets chip fabs without human staff by 2030

Agreed , there is limited return on jobs already for Govts , so for it be zero would I imagine mean almost zero Govt subsidy for anything.
 
This is highly dependent on whether they can do that. 300mm is already extremely automated, and an extreme expense, and that only scores you saving after many years of non-stop operations even in an expensive labour country.
 
I don't agree with the part about no government subsidies. Fabs have become a strategic requirement for many high cost countries, and labor isn't the only factor in those countries that is high cost. Permitting, zoning problems, environmental laws, requirements for union construction labor and local content, and required community infrastructure upgrades. Where would you rather build a fab, even if labor is not a factor, Taiwan or Arizona?
 
What makes you think that a foundry can be run as automated as a bottling company? What does the semiconductor industry know about automation? Mr Blue and Mr. Chen, be reasonable.

Edit: Note.... I am being sarcastic. Building fabs in the US is a no brainer, unless US robots are lazier than Taiwanese ones.
 
Last edited:
What makes you think that a foundry can be run as automated as a bottling company? What does the semiconductor industry know about automation? Mr Blue, be reasonable.
Cliff, I'm not following you. I have no idea if "dark" fabs are possible or not. I have no expertise in fab operations. My argument is simply disagreeing with the notion that automating fabs will reduce the need to subsidize them in high cost countries. How am I not being reasonable?
 
Sorry... I was kidding (sarcastic) I should have written Mr. Chen. I was multitasking. Of course I agree with you. 100%
 
Also, labor rates can be reduced simply by training the fab workers in Spanish. The fab is in Arizona, correct?
TSMC's new fab is in Arizona. Samsung's is in Texas. Both states have large Hispanic populations. It sounds like you are suggesting that if all training material was available in Spanish and the trainers were fluent in Spanish, then TSMC could stop by Home Depot's parking lot and staff up. Nothing could be farther from the truth.
 
Not Home Depot. Put in a Craigslist ad and get Spanish speaking trainers. Mexicans dominate construction in Arizona and don't complain about the heat. Most work hard and would jump at the chance at getting a job at the wages the foundries will pay. Please explain where I am misguided.

Pick an example building. Construction, blue collar inside, and process. As far as process, sure, that has to be handled by experience personnel (I suppose). How about the rest of it. Please provide numbers.

Edit: Also, I gotta believe the construction of the site is more on the civil and industrial engineering side. I have to believe Mexico has a lot of experience in that area. Please educate me on what I am missing. I am an EE. I am not well versed in the manufacturing process, but I will tell you that almost all of my EEs and CS people came from Craigslist ads, most of whom had low GPA scores because they were taking the tougher classes. That is what I cared about.
 
Last edited:
This might be inevitable but potentially fabs in the quite near future would no longer be hiring or keeping staff, if companies follow Samsung's drive toward implementing smart sensors in fabs to fully automate them by 2030. https://www.msn.com/en-us/news/tech...fully-automated-chip-fabs-by-2030/ar-AA1mv4ba This could make fabs, which will eventually be AI fabs, more expensive (no govt subsidies) to build, sooner than we may expect.

Hopefully yield will be improved.
 
What makes you think that a foundry can be run as automated as a bottling company? What does the semiconductor industry know about automation? Mr Blue and Mr. Chen, be reasonable.
This is supposedly Samsung's intention. It could merely be an unrealistic roadmap goal.
 
I don't agree with the part about no government subsidies. Fabs have become a strategic requirement for many high cost countries, and labor isn't the only factor in those countries that is high cost. Permitting, zoning problems, environmental laws, requirements for union construction labor and local content, and required community infrastructure upgrades. Where would you rather build a fab, even if labor is not a factor, Taiwan or Arizona?
Agreed, these justify subsidies, but the old promise of building fabs meaning lots of jobs is weakened by a full automation scenario.
 
Agreed, these justify subsidies, but the old promise of building fabs meaning lots of jobs is weakened by a full automation scenario.
Yeah, I know, but I've thought the real justification for the recent worldwide fab construction party is, paraphrasing former AMD CEO Jerry Sanders, "Real countries have fabs".
 
Last edited:
I am not following the meaning of this thread. Fabs are already fully automated. Surely, no one is talking about automating construction. So what is left: maintenance, process, equipment engineering and yield. Some AI is indeed already in use to supplement process engineering but would anyone let a computer turn all the knobs on its own? Guard rails you say. Well who designs the guardrails? Process engineers of course.

The much more likely outcome: Large fabs will share their experts over many sites. TSMC is already doing this. So in that sense, labor will be reduced. Other likely outcome: the disappearance of maintenance, to be replaced by OEM and 3rd party contracts (TSMC and Samsung well on the way there).

What has not been automated cannot be automated but it can and will be outsourced.
 
Back
Top