Chinese semiconductor industry

Not just yield. Some machines have less output than others. Output is typically measured in wafers per month or some metric like that.
Imagine you have a machine with a weak light source. It will require more exposure time, which means you get less output.
Without knowing the specifics of the equipment it is hard to make a judgement on something like this.
But certainly having some output is better than having none at all so I think the investment on Chinese home grown machines and materials is more than warranted.

latenlazy said:

Domestic equipment might end up costing SMIC more money, not less. Cost isn’t a function of your equipment but your yield efficiencies. Worse yield means both lower quality and higher costs. If the quality of Chinese equipment isn’t up to snuff that makes that equation worse, not better.

Click to expand...

lets put side semiconductor, talk about things in general , equivalent stuffs in China versus US or EU is typically a factor 6 or 7 differences. Typical salaries also reflect that. Of course, you talk one is superior over the other.I am just saying when everything pretty much equal.

The bottom line is SMIC losing money at 28nm, so the prospect for more advanced node is not good. Dont even think about buying ASML EUV, prohibitedly expensive, they will not recover the cost.

only way out is make domestic equipment until its good enough.
If like you said its safer with foreign equipments for SMIC , if thats the case I bet there's good chance Huawei fabs using purely domestic equipments have a chance down the line to achieve greater scale with lower cost and expand into huawei foundry and put SMIC out of business.

Domestic equipment should be cheaper because the 1 time engineering recoup cost can be subsidied by the state. After that the equipment can be sold at low price to domestic players, mainly the material cost.

nastya1 said:

lets put side semiconductor, talk about things in general , equivalent stuffs in China versus US or EU is typically a factor 6 or 7 differences. Typical salaries also reflect that. Of course, you talk one is superior over the other.I am just saying when everything pretty much equal.

The bottom line is SMIC losing money at 28nm, so the prospect for more advanced node is not good. Dont even think about buying ASML EUV, prohibitedly expensive, they will not recover the cost.

only way out is make domestic equipment until its good enough.
If like you said its safer with foreign equipments for SMIC , if thats the case I bet there's good chance Huawei fabs using purely domestic equipments have a chance down the line to achieve greater scale with lower cost and expand into huawei foundry and put SMIC out of business.

Click to expand...

@WTAN thoughts?

nastya1 said:

lets put side semiconductor, talk about things in general , equivalent stuffs in China versus US or EU is typically a factor 6 or 7 differences. Typical salaries also reflect that. Of course, you talk one is superior over the other.I am just saying when everything pretty much equal.

The bottom line is SMIC losing money at 28nm, so the prospect for more advanced node is not good. Dont even think about buying ASML EUV, prohibitedly expensive, they will not recover the cost.

only way out is make domestic equipment until its good enough.
If like you said its safer with foreign equipments for SMIC , if thats the case I bet there's good chance Huawei fabs using purely domestic equipments have a chance down the line to achieve greater scale with lower cost and expand into huawei foundry and put SMIC out of business.

Click to expand...

Your equipment costs are one off capital expenditures. The cost drivers for your per unit production prices are typically operational expenditures. If your per unit costs are higher than your competitors that is usually a reflection of operational inefficiencies rather than expensive capital equipment. In this case it means the problem is not in the cost of SMIC’s equipment but the quality and efficiency of SMIC’s work process. Very likely the root issue here is that their yield still isn’t good enough to get good per unit pricing of product because the quality of their production process isn’t quite up to scratch. Scale efficiencies don’t drive down costs if your operational processes aren’t efficient or effective. Efficiency from scale is not just an automatic button that gets you better price performance by pushing out greater volume. If your production process isn’t good enough you tend to get inefficiencies rather than efficiencies from scale. The scale is the output, not the input, of those efficiency factors.

latenlazy said:

Your equipment costs are one off capital expenditures. The cost drivers for your per unit production prices are typically operational expenditures. If your per unit costs are higher than your competitors that is usually a reflection of operational inefficiencies rather than expensive capital equipment. In this case it means the problem is not in the cost of SMIC’s equipment but the quality and efficiency of SMIC’s work process. Very likely the root issue here is that their yield still isn’t good enough to get good per unit pricing of product because the quality of their production process isn’t quite up to scratch. Scale efficiencies don’t drive down costs if your operational processes aren’t efficient or effective. Efficiency from scale is not just an automatic button that gets you better price performance by pushing out greater volume.

Click to expand...

Yield and scale are more or less independent.
If your yield no so good, you can scale up and having more equipments with lower cost operating, that could compensate the lower yield.
SMIC can't do that.

nastya1 said:

Yield and scale are more or less independent.
If your yield no so good, you can scale up and having more equipments with lower cost, that could compensate the lower yield.

Click to expand...

If you add more equipment for the same yield percentage your production prices are still going to stay the same.

latenlazy said:

If you need more equipment for the same yield percentage your production prices are going to stay the same.

Click to expand...

not if your equipment is 7x times more expansive. The exact break down we don't know.
You can compensate for lower yield.
Currently, SMIC have higher cost . production cost does include capital expenditures.
We all know scale up will drive down the the cost per chip. SMIC doens't have enough customer, enough equipment , enough yield, everything combined.
increase number of equipment AND increase yield in tandem is way to increase the scale.

nastya1 said:

not if your equipment is 7x times more expansive.

Click to expand...

Chinese equipment may by cheaper, but it won’t be 7x cheaper. They may be 2-3x cheaper at best, but they might also only get you less than half the yield. Besides, as to my original point, the primary problem SMIC has is in the operational cost side of the equation. If SMIC is using the same equipment as other fabs and has lower yield/higher per unit pricing, using cheaper but inferior equipment doesn’t solve the root cause of their worse price performance. Using lower quality equipment with an inferior process may very well compound a low yield problem well beyond whatever price advantage they gain from using cheaper equipment. It’s much easier to ruin your efficiencies than improve them.

latenlazy said:

Chinese equipment may by cheaper, but it won’t be 7x cheaper. They may be 2-3x cheaper at best, but they might also only get you less than half the yield. Besides, as to my original point, the primary problem SMIC has is in the operational cost side of the equation. If SMIC is using the same equipment as other fabs and has lower yield/higher per unit pricing, using cheaper but inferior equipment doesn’t solve the root cause of their worse price performance. Using lower quality equipment with an inferior process may very well compound a low yield problem well below whatever price advantage they gain from using cheaper equipment. It’s much easier to ruin your efficiencies than improve them.

Click to expand...

I heard state of art for latest DUV equipment from ASML cost as much as $40+million each.
If Chinese subsidied the 1 time engineering cost, does the material cost $4million? I don't know .. It's tricky.
Half of the yield is terrible, that would be a total nonstory then.

what i saying is improving the yield is not enough, you got to scale up with more equipments to compete regardless.
lets say your yield is 100% and you have 1 main machine running in the whole factory, that machine would become the total bottleneck while all other equipments running in the factory
while your competitor have 30 main machine running at 100%.
who's going to win?