This a proper way to disagree with or correct someone...
Chinese semiconductor industry
- Thread starter Hendrik_2000
- Start date
- Status
Okay guys we need to agree to disagree, Japan has the ambition to rekindle her past, good luck and wish them the best, BUT for the Chinese there is nothing to be embarrass about, nobody expect us to produce a 7nm chip even with the restriction. Now lets see If Japan can deliver cause there's a lot of problems to overcome, First IF the Americans allow the Japanese to use ASML EUVL 2) If Japan do produce a 7nm before Intel, will they allow them to sell it. 3) whose the customer? 4) shrinking market and intense competition, they will be competing against TSMC, Samsung and Maybe SMIC, cost wise can the Japanese be competitive?
Last edited:
I think overall saying that SMIC beat Intel for 7nm is a bit an overstatement. We know that Intel 10nm was similar to TSMC 7 nm. Intel will be producing also soon smaller nodes, I guess that if SMIC wasn't handicapped with the exclusion of EUV today, then it would come at similar level to Intel these days.
Why did you bring up Japan in this thread? Do you have difficulties parsing the title?
You are correct. I think some on this thread has been mis-informed about how bad Japan or US capability is while exaggerating SMIC's capability just a bit.
Using the exact same standard we use to declare SMIC has 7nm capability (technical capability, yield level, and output capacity) then Intel has 7nm capability for a few years already.
And as of this year, Intel has already shipped approximately 90K 7nm wafers. Below is a statement from Intel 2002 Q2 financial result:
"▪ Intel made significant progress during the quarter on the ramp of Intel 7, now shipping in aggregate over 35 million units. The company expects Intel 4 to be ready for volume production in the second half of this year and is at or ahead of schedule for Intel 3, 20A and 18A."
SMIC is in 7nm risk production but with very limited capacity. I'm hoping, in a few weeks, to hear good news from SMIC's Q2 financial result if it had improved to volume production level capability. But SMIC is not ahead of Intel like some on this thread tried to claim.
Bro, Japan has plan for a 2nm RD fab to be ready by 2025. This plan is already in progress.
And Japan's semiconductor capability is in fact better than SMIC's. Micron Japan (formerly Elpida) is making DRAM D1a/D1b node with overall capacity >100K wpm. Even though comparing logic and DRAM is like comparing apples & oranges, considering the minimum feature size and the patterning complexity, D1a/D1b are considered to be about equivalent to 5nm logic.
The capability to fab chips at this level of dimension, control the defect density to enable high volume production of 100K wpm would put them ahead of SMIC.
Whether Japan has a 7nm logic/foundry fab is a business decision that, like you said, involves consideration of market demand, competitiveness, etc. But don't let this confuse this to mean Japan is inferior in technical capability.
With the global geopolitical tension looming, if and when motivated, Japan will easily surpass China. This is especially true when we face restrictions while they don't. I had previously voiced my concern about the gap growing bigger vs. the world. Until domestic immersion and EUV scanner is available on a massive scale, I still believe my concerns are likely to happen.
Its just a plan by the US and Japan to hedge against TSMC. There is no guarantee that they will succeed. Probably the US is frustrated that TSMC is only willing to initially fab 5nm in its new US fab opening in 2025 while its Taiwanese fabs will already be ahead and fabbing 3nm or 2nm.
Japan is a country, SMIC is a company. The questions what company in Japan is going to make 7-2 nm chips? What companies are going to make the investment risk and who are going to be the clients of such edge nodes? I think even without EUV the Chinese can make a 5-2nm GAA research fab but a commercially viable enterprise is another matter, is not just technology but also the giant risk factor.
TSMC is hoping in its brand name to attract clients to their new fab in Arizona, so is Samsung and Intel. With designing chips costs going to roof exponentially for cutting edge nodes, I don't think many clients are going to jump to newcomer.
SMIC has the advantage of their brand name and years of experience with close to the edge nodes. With a market the size of China, the geopolitical situation and lithography problems solved, SMIC could leapfrog in the next coming future years.
@tokenanalyst bro If the Japanese gov't do achieve a breakthrough who will commercially produce them Intel? From what I read they had just cancel their orders of 3nm chips with TSMC and boy TSMC had been hoodwink, investing huge amount of money for an unreliable customer.
- Status