7nm is also a key node. It's where transition from DUV to EUV is needed; it's also a sweep spot node in terms of cost/performance ratio, after which the cost increases much faster than performance.
Chinese semiconductor industry
- Thread starter Hendrik_2000
- Start date
- Status
7nm is also a key node. It's where transition from DUV to EUV is needed; it's also a sweep spot node in terms of cost/performance ratio, after which the cost increases much faster than performance.
The next stage is 2nm GAAFET?
Good thing actually that you have something that can churn out FinFET down to 7nm albeit with imported tools and materials. That mean tool and material makers can learns instead of blindly try and errors.
Having multiple accounts for any reason is a bannable offense.
Quickie
Colonel
This comes from the U.S.'s delusion with the bull that China can't innovate and make things for itself without reverting to copying and reverse engineering. The U.S. historically has been trying to deny a lot of technological stuff from China for example trying to isolate China from international space exploration, specifically, the I.S.S.
What happened next is this makes China more determined to do things by itself and that so far in many cases has proved to lead to a much more successful outcome than it would otherwise be.
US decision making on sanctions relies heavily on a few career officials in their Commerce Department.
I remember a reading an article on how a select group of 'smart' academics there determined what sanctions to put on Russia with their own models and economic projections. I'll see if I can dig it up again.
As to their accuracy, the fact they have to keep doubling down on new chip sanctions as previous ones become ineffective can be interpreted several ways.
Echo chamber-like behavior makes mistakes more likely than not. And the stakes are pretty high here for the USA.
Sounds like a joke, I don't think they thought too deeply about this.
Wait until they realize it's a mistake and start threatening Samsung or Nikon for smelling a market opportunity.
Last edited:
The issue with the IP system is using a VPN service can result in different people having the same IP address. There have been past examples where this has been the case.
The IP system far from full proof, so the moderators do act with some caution and look at other indicators before banning members if there is suspected duplicate accounts.
Once you get into EUV things start getting weird. My opinion is that EUV is not as difficult as the move from planar to FinFET in terms of process alone. The equipment is hard but for the fabs, integrating EUV is not as hard as integrating stuff like high-k dielectric, tungsten gate, etc.
as you can see, initial costs after 28 nm start doubling at 16 nm, then triples at 7 nm. So that's 6x higher cost for 7 nm just to get started. , but , that is, kept off to keep within power and thermal constraints. So 50 million. You get ~3x higher transistor density for 6x higher startup cost. Per wafer processing costs are even higher since you need to introduce EUV, special EUV resists (which are all hard mask resists), more metal layers, etc.
This can only be justified through extremely high volume and price simultaneously. That's something like a mobile SoC or a gaming GPU. For everything else compute related such as business workhorse laptops, automotive/industrial CPU, TC, etc. 28 nm is absolutely fine (and indeed may be too much). Not going to get into opto/power/RF.
Thanks for posting this. In the article there is the link to the official draft, I tried to find it yesterday but without success.
For interested people, this is the official text.
Regarding the content you posted, unfortunately I expect all foreign toolmakers will be affected, not only US firms. It is impossible ASML will not be banned too. US will do anything to make ASML succumb: if they have to blow up a pipeline under ASML building or send marines there (considering ASML is in Holland, it would be quite convenient, they are already in the nearby) they will do it. If they fail to coerce foreign toolmakers, their actions will backfire hugely and they are well aware of this.
OTH the fact that they proceeded with announcing and publishing the new rules, before, by their own admission, an agreement with foreign toolmakers was found, it makes me think that they tried hard (we know they did), but as of today still did not succeed....or maybe they are just so arrogant and full of themselves that simply did not care.
If ASML, by nothing short of a miracle, eventually succeeds in resisting this wave of rage, I promise they will be my hero forever! I will hang an ASML factory's picture in my room!
- Status