*3 assuming Q3/4 2028 and more like 2 and a half assuming Q1/2 2028. It's Aug of 2025 now.
Chinese semiconductor thread II
- Thread starter vincent
- Start date
The technological gap in EUV spans from 2015 to 2028. We must close this gap within just a few years.
We're currently in a phase roughly equivalent to TSMC's EUV lithography research in 2015—a precursor to TSMC's formal introduction of EUV lithography machines for 7nm chip production in 2017.
Last edited:
To develop from the ground up a complete indigenous supply chain components and material for a EUVL is veryy veryyyy impressive, even the US and ASML can't able to do it themselves. NO other nation can unless your India with its Vedic knowledge and power.
Last edited:
Indeed, the U.S. embargo on China in EUV technology is far more comprehensive than externally perceived—confirmed by sources.To develop from the ground up a complete indigenous supply chain components and material for a EUVL is veryy veryyyy impressive, even the US and ASML can't able to do it themselves. NO other nation can unless your India with its Vedic knowledge and power.
Consequently, China must re-engineer every single link across the entire technology chain through intensive R&D. This represents a massive coordinated effort involving tens of thousands of researchers.
Currently, progress appears unfolding as planned—no major setbacks reported.
China is executing full-spectrum R&D at unprecedented scale, leveraging its vast engineering and scientific workforce with cost-no-object determination. At every critical technical node, multiple parallel tracks advance simultaneously across divergent theoretical and technological pathways. The sheer magnitude of this endeavor—when contemplated—reveals its mind-blowing ambition.
Last edited:
you are not stating anything that is not obvious so I would say that you should stop talking like that.
as for your stuff about tuning and process and such, that is kind of not relevant. As I said earlier, they absolutely do not need to wait until they are confirmed and yield is at a certain point before starting serial production using EUV. Since any EUV process for 5nm is pretty much better than a DUVi process. As such, saying EUV right now is where DUVi was at 2021 is entirely meaningless since their domestic machine doesn’t have to compete against ASML EUV.
I have in fact heard about anything from risk production later this year to serial production in 2028. Depending on what customer is willing to live with in terms of cost or yield/consistency, any number of years in between is reasonable.
This is false. EUV is not more preferable than DUVi. TSMC avoids EUV as much as it can.
It depends on the steps. If it didn’t TSMC wouldn’t use EUV at all. There are critical steps where DUVi has insufficient resolution even if there’s noise issues with EUV, and for those EUV is not only preferred but nonnegotiable.
Just seen this, which would appear to justify H20 ban in and of itself:
On the contrary, this is great news for domestic competitors. A major reason to buy domestic—rather than Nvidia—is because sales revenues are used to develop the same company's future offerings. You want that money going to your own companies instead of foreign ones, if at all possible. But if Nvidia is denied revenue because the US government is taking a cut, then it will obviously have less to reinvest in its own efforts.
And the current US government has already demonstrated its fiscal insanity, so there's not much to worry about giving it a little extra revenue. Is said revenue going to be invested in R&D? And even if it was, would the government do a better job of it than Nvidia? Not to mention that even a large slice of Nvidia revenue is only a drop in the bucket compared to government budgets. That is to say, assuming Nvidia doesn't raise prices (which would of course reduce its competitive edge), then this new policy means the US is getting less value out of the same money.
Last edited:
Seems AMD took the same "deal".