You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
Some interesting comments from Samsung in this article would definitely be of interest to this forum:
For foundry services, Samsung’s investments last year were focused on 5-nm process technology using EUV tools at a fab in Pyeongtaek, South Korea.
“Although we did not generate a surplus in 2021, we believe our efforts will lead to significant future growth,” Ben Suh, Samsung’s investor relations spokesman, said during the conference call. “Even though increased overall supply resulted in a new high for revenue, profitability decreased slightly quarter-on-quarter due to a rise in costs associated with advanced processes.”
Still, Samsung has experienced yield problems with its existing production technology at its current 5-nm node.
The challenge of maintaining initial stable yield has increased, Samsung told industry analysts. Despite a delayed ramp to advanced nodes that lagged company expectations, Samsung expects gradual stabilization.
Feb 9 we'll probably find out about 4nm Exynos with AMD graphics IP block. Samsung 4nm is definitely in production.
Qualcomm is doing the Apple A9 thing with their 8 Gen 1: Samsung will do some, TSMC will do 8 Gen 1 +. Capa is getting tight, tight, drum tight. You could bounce a quarter off the tight capa at TSMC and Samsung.
The problem with Samsung foundry is the PDKS. They are much less stable than TSMC's at any given time. The big fabless companies cannot deal with unpredictability at that level if they want to get competitive products out in a timely manner. TSMC will again own 3nm as they did 5nm. The FinFET era is over, TSMC wins. It will be interesting to see how the Intel foundry PDKs do. Exciting times in the foundry business, absolutely.
The problem with Samsung foundry is the PDKS. They are much less stable than TSMC's at any given time. The big fabless companies cannot deal with unpredictability at that level if they want to get competitive products out in a timely manner. TSMC will again own 3nm as they did 5nm. The FinFET era is over, TSMC wins. It will be interesting to see how the Intel foundry PDKs do. Exciting times in the foundry business, absolutely.
There is at least FF 3nm at TSMC coming, is it? Samsung's first get GAA is expected to be worse in all regards to last gen TSMC FF, and only getting even in 2nd, or 3rd gen.
They moved to GAA because they never aced finfets as good as TSMCs, and thought a device transition will be an opportunity to have a fresh start against competition.
