News Samsung 3nm Chip Mass Production On Track To Start in Q2

It would be interesting to check how TSMC positions its 5, 4 and 2nm nodes against 7nm to compare against Samsung's claims. I can't remember from the top of my head, but they were about 20% improvements per jump or so? So 7nm to 2nm (when it's ready) should be about ~50% better overall?

Regards.
 

InvalidError

Titan
Moderator
The question is how is the yield? Mass production with 20% yield is not quite “mass” production.
It can be if the things being manufactured have sufficient value to be economically viable at that level. Not an issue for things like medical, research, military, aerospace, telecom and industrial markets with nearly infinite budget for bleeding-edge stuff. Although it may not be billions of chips per year, the volumes can still be significant.
 
  • Like
Reactions: KyaraM

isofilm

Honorable
Oct 17, 2018
14
1
10,515
It can be if the things being manufactured have sufficient value to be economically viable at that level. Not an issue for things like medical, research, military, aerospace, telecom and industrial markets with nearly infinite budget for bleeding-edge stuff. Although it may not be billions of chips per year, the volumes can still be significant.

Samsung has admitted to lying about their process yields, which for the 4nm are 30% after more than a year of tweaking, even though they falsely claimed 70%.

Now you expect us to believe Samsung has miraculously resolved their 3nm yields with its never before fabricated GAA design problems, with a current yield of ~10%?

Wafer yields under 45% are a loss, given traditional foundry margins of 45%.