Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/will-samsung-take-business-from-tsm-in-ai.20416/
        )

    [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
)

Will Samsung take business from TSM in AI

Isn’t it possible Samsung will take some business/market share (short-term) just by virtue of having foundry capacity?

AI is so hyped right now that I think any available fab space will find a product ..

(Long term is what matters to me, but I could see the article being correct temporarily)
 
‘I was at the Samsung foundry event yesterday and I would say no. In fact, Intel foundry will take away business from Samsung foundry so it will only get worse.
My read is Samsung is the anti TSMC. They’ve been caught stealing customer IP repeatedly
 
Isn’t it possible Samsung will take some business (short-term) just by virtue of having foundry capacity?
AI is so hyped right now that I think any available fab space will find a product ..

Samsung had a very good node at 14nm but they still sold on low price. Samsung 10nm had awful yield but at 7nm Samsung had another good node and did quite well. At 5/4nm yield was again a problem and Samsung 3nm was horrible. Who in their right mind would bet their product/company on that track record?

Traditionally Samsung sells on price. Paying 20% less for wafers is a great deal if you can get the wafers with reasonable yield. But the risk is very high which is why TSMC says Trusted Foundry.

At the leading edge there is only one choice now and that is TSMC. I am told that Samsung 5/4nm is yielding better so they are getting business which is good. Intel may also be getting foundry business at 4/3nm but it is not the big runners. Samsung 2nm (optimized version of Samsung 3nm) is not competitive with TSMC N2 or Intel 18A but they may get some business if yield is improving as Samsung said it is.

As I have said many times before, Intel and Samsung foundry today are battling for the NOT TSMC business, the customers who don't want to use TSMC for whatever reason. Hopefully Intel and Samsung can win business from TSMC based on competition (a better product) but, unfortunately, I do not see that happening today, but hopefully it will happen soon.

Samsung's new pitch of being the "AI Foundry" is more proof that Samsung is following the industry trends not leading, which I think is a mistake. As Wayne Gretzky said "Skate to where the puck is going, not to where it has been". Very applicable to the semiconductor industry, my opinion.
 
When major AI chip companies are printing money, foundry discount is practically nothing. For example, NVIDIA is making twice more profit than TSMC right now, so for NVIDIA, risk of not getting good dies outwieghs benefits of cheaper foundries. So, there's TSMC, as usual.

Samsung will get customers since not everyone can access TSMC for whatever reason, but they need to help their customers win their businesses first for some consecutive years.
 
Back
Top