Chinese semiconductor industry

Status
Not open for further replies.

antonius123

Junior Member
Registered Member
I believe this is a fundamental misunderstanding. first, the shortage is not in leading edge chips, it is in mixed signal and power chips for automotive applications.

second, the increased cost is happening at fab 21 (a leading edge fab) due to technical reasons. They are not going to eat the cost, they're going to pass it onto the customer. the customer is the fabless company. They're going to pass their costs onto the OEMs. The OEMS will then pass the cost onto the end customers. None of them will just eat the cost.

The final customer can then decide - I can simply not buy a new phone so quickly. I will wait. The OEMS then cut their orders. It's no big deal, most OEMs do multiple products at once. The fabless cut their orders. That's no big deal either, their fixed costs are relatively low and usually they have a large portfolio anyways since they only deal in IP. But the leading edge fab is in trouble. Every second it operates it burns money because it can't stop. It needs utilization to approach 100%.


This show that shortage is also happening with leading edge chips:

Please, Log in or Register to view URLs content!

That means all leading edge smartphone price will increase, and they will suffer some loss because some consumer will restraint buying high end smartphone. But TSMC wont suffer loss, because their fab is fully booked.
 

antonius123

Junior Member
Registered Member
7 nm 3D stacking chipsets from DUV multipatterning would absolutely destroy the markets for 5 nm from EUV. DUV is still far more cost effective due to the higher intensity light source and no need for vacuum. DUV is also transferrable to bigger process sizes like 14 nm, 28 nm while EUV is only cost effective for <14 nm.


Very interesting. But why all the foundry doesn't take this route? considering the cost is low and will be very competitive without hinderrance of the scarecity of the EUV lithograpy.
 

Vichysoy

New Member
Registered Member
Now the reality bites as the cost of producing 5nm chips at FAB 21 is Five times more expensive as they compare that of Taiwan. @antonius123 bro the mass production date is 2024, as reported in the video all FABS had raise their price due to the shortage, now picture this as China 28nm mass production ensued next year and that of 14nm in 2023 and 7nm at 2024, they will flood the market as @krautmeister had summarized, those FABS in Arizona will become a big white elephant project. If Huawei and SMIC can able to produced a 7nm 3D stacking chipset, then tech wise we are even with the US (5nm Arizona FAB).

Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!


Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!


Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!

6.9K views
Simply death for tsmc ! That all !
 

ansy1968

Brigadier
Registered Member
This show that shortage is also happening with leading edge chips:

Please, Log in or Register to view URLs content!

That means all leading edge smartphone price will increase, and they will suffer some loss because some consumer will restraint buying high end smartphone. But TSMC wont suffer loss, because their fab is fully booked.
@antonius123 bro that is a marketing ploy, how can there be a shortage in 3nm chips when it was schedule to be INTRODUCED and massed produce next year 2022.
 

ansy1968

Brigadier
Registered Member
Very interesting. But why all the foundry doesn't take this route? considering the cost is low and will be very competitive without hinderrance of the scarecity of the EUV lithograpy.
@antonius123 bro if I may answer your question even though I'm not an expert, simply they can purchase an EUVL machine, those that can not must find other means and 3D stacking is a way to circumvent it. Only China find it attractive due to its current circumstances while others didn't bother cause they are part of the US alliance system.
 
Last edited:

ansy1968

Brigadier
Registered Member
If the chip price increase due to shortage, then the winner is the supplier (foundry) at the expense on customer; client will always buy their product in spite of high price. This is what we call "supplier market". That means the chip business is very promissing, but the barrier to entry is very high.

The high production cost will be disaster only in the client's market, where no shortage but abundance instead and customer can easily choose the vendor that can offer the cheaper/cheapest price.
@antonius123 bro TSMC sounded an alarm last month of a possible glut in the market, cause they see all the expansion project in China may come to fruition in the next 2 years. They are relying on leading edge to see them thru and may cede the market of 28nm and 14nm to the Chinese. If Huawei and SMIC is able to mass produced its 7nm 3D stacking chipset in 2024 then its game over as the 5nm , 3nm and even 2nm all vying a small niche market.
 

xypher

Senior Member
Registered Member
Very interesting. But why all the foundry doesn't take this route? considering the cost is low and will be very competitive without hinderrance of the scarecity of the EUV lithograpy.
IMO:

Because all the leading companies like TSMC and Samsung already invested in <7 nm technologies & it would not make sense for them to pursue technologies that would collapse their profit margins. They need to recuperate the R&D money, money spent on equipment for fabs, etc. Therefore, they will keep going for thinner chips until they hit either technological or economic (when the whole process becomes so expensive that passing them onto customers will drastically reduce the demand) wall, while side gigs like packaging will be developed in parallel but won't be pushed over thinner processes as long as the latter can still bring money.

Most of the other fab companies are focused on the other sectors of the economy - automobile chips, sensors, industrial electronics, etc., where even 14 nm is rarely needed, hence there is no incentive to invest into development of packaging technologies. The individual packaging-focused companies are too small for now to start pushing their tech en masse.
 

FairAndUnbiased

Brigadier
Registered Member
Very interesting. But why all the foundry doesn't take this route? considering the cost is low and will be very competitive without hinderrance of the scarecity of the EUV lithograpy.
not all foundries have in house packaging, they outsource to a dedicated OSAT (outsourced assembly and test) company. In addition, the capability to develop advanced packaging techniques is very different than foundry. But yes, more and more companies are moving in this direction. The main thing with foundries is that they need to make their money back. They can't do that if they admit that you can get same results with different packaging. The other part is, you have to design from the ground up for this sort of chip.

Please, Log in or Register to view URLs content!
That is for outsourced processes ie not proprietary but even there there's plenty of advanced capabilities -
Please, Log in or Register to view URLs content!
The real heavy hitter is in proprietary packaging techniques such as
Please, Log in or Register to view URLs content!


YMTC is not the only company to do this btw,
Please, Log in or Register to view URLs content!


The important part for this sort advanced packaging design is that not all parts of the chip have to be manufactured on the same process. For example, you can manufacture ONLY the CPU on 7 nm, have memory on 20 nm, and all the I/O circuits on 180 nm, then stack them.

For example, YMTC manufactures
Please, Log in or Register to view URLs content!
, but they manufacture their
Please, Log in or Register to view URLs content!
. This means that they can use SMEE's low cost 90 nm DUV for the 180 nm and reserve the 20 nm for the (much higher cost) 28 nm DUV, or outsource their I/O wafers only to lower end fabs.
Please, Log in or Register to view URLs content!
, what this means is automatic 30% increase in memory density per die and cuts the number of steps in half (you just repeatedly pattern 1 wafer type).

The challenges, of course, are huge. It's an entirely different mentality than monolithic chip.
 

Tyler

Captain
Registered Member
If the chip price increase due to shortage, then the winner is the supplier (foundry) at the expense on customer; client will always buy their product in spite of high price. This is what we call "supplier market". That means the chip business is very promissing, but the barrier to entry is very high.

The high production cost will be disaster only in the client's market, where no shortage but abundance instead and customer can easily choose the vendor that can offer the cheaper/cheapest price.
Apple's, and AMD's chips will be 5 times more expensive if they are produced by TSMC in the US.
 
Status
Not open for further replies.
Top