Exclusive: U.S. moves to cut Huawei off from global chip suppliers

[PakFactor]

Wrong...You WILL lose far more than money. This is why it is so entertaining reading you fools.

There is nothing wrong with being technologically behind. The issue is whether you have CONTINUAL access to developing technologies to maintain your status quo. In isolation, you either have no or little access to those developing technologies. Indigenous development to catch up? Dream on. Commodity semicon products takes 2-3 months from wafer start to client.

https://www.zeiss.com/semiconductor...nership/zeiss-systems-in-a-wafer-stepper.html

You just developed a new stepper? Try at least 2 YEARS of engineering wafer runs before a single client is willing to accept engineering samples just to test wafers that came from your new stepper. That client will take one year of testing under his own criteria before publishing his findings. Now three years have passed and in that time, you cannot develop further that new stepper you just built. You have to lock in that technology to remove all the engineering variables that comes with continuous development. In that three yrs time, the client will compare wafers that came from your indigenously developed stepper against GLOBAL standards and if there are greater than %3 difference in all criteria, your stepper WILL be rejected. If one client reject, most likely others will without doing their own evals. In short, you have to wait 2-3 yrs to find out if your new hardware have any shortcomings under COMMODITY products, let alone using the new hardware on new 'cutting edge' products.

Right now, under autonomous autos requirements, 20 nm NAND commodity dies are being rejected left and right from all manufacturers and we are talking about major players like Samsung, Micron, Toshiba and so on. TWENTY NM...!!! How old is that NAND technology? Try at least 10 yrs old. In other words, autonomous driving requirements are rejecting %50 of wafers from established technologies as inadequate. That is NASA, ESA, and CNSA level.

Your China is slowly being severed from these global standards. The tool of that severance is not a scalpel but a dull spoon and that is why Huawei is terrified.

Question you work on logistics for a tech company?

 

[gambit]

Question you work on logistics for a tech company?

I work for a major semicon company. Probe section. Process Engineering.

Example...

https://www.mjc.co.jp/en/technology/column/wafer_prober.html

I interface between Production and Engineering. In other words, I develop processes that test wafers. Probe is the final fabrication section. Next are Extraction and Assembly and there are more details on those sections.

 

[PakFactor]

I work for a major semicon company. Probe section. Process Engineering.

Example...

https://www.mjc.co.jp/en/technology/column/wafer_prober.html

I interface between Production and Engineering. In other words, I develop processes that test wafers. Probe is the final fabrication section. Next are Extraction and Assembly and there are more details on those sections.

Nice, thanks for the info. Pretty solid career path considering semi's been on a bull run last 5+ years.

 

[gambit]

Nice, thanks for the info. Pretty solid career path considering semi's been on a bull run last 5+ years.

I have been in semicon industry for about 25 yrs after the USAF, F-111 Cold War then F-16 Desert Storm. I started as a Field Engineer for Santa Clara Plastics, a robotics wafer processing company.

 

[Nilgiri]

Wrong...You WILL lose far more than money. This is why it is so entertaining reading you fools.

There is nothing wrong with being technologically behind. The issue is whether you have CONTINUAL access to developing technologies to maintain your status quo. In isolation, you either have no or little access to those developing technologies. Indigenous development to catch up? Dream on. Commodity semicon products takes 2-3 months from wafer start to client.

https://www.zeiss.com/semiconductor...nership/zeiss-systems-in-a-wafer-stepper.html

You just developed a new stepper? Try at least 2 YEARS of engineering wafer runs before a single client is willing to accept engineering samples just to test wafers that came from your new stepper. That client will take one year of testing under his own criteria before publishing his findings. Now three years have passed and in that time, you cannot develop further that new stepper you just built. You have to lock in that technology to remove all the engineering variables that comes with continuous development. In that three yrs time, the client will compare wafers that came from your indigenously developed stepper against GLOBAL standards and if there are greater than %3 difference in all criteria, your stepper WILL be rejected. If one client reject, most likely others will without doing their own evals. In short, you have to wait 2-3 yrs to find out if your new hardware have any shortcomings under COMMODITY products, let alone using the new hardware on new 'cutting edge' products.

Right now, under autonomous autos requirements, 20 nm NAND commodity dies are being rejected left and right from all manufacturers and we are talking about major players like Samsung, Micron, Toshiba and so on. TWENTY NM...!!! How old is that NAND technology? Try at least 10 yrs old. In other words, autonomous driving requirements are rejecting %50 of wafers from established technologies as inadequate. That is NASA, ESA, and CNSA level.

Your China is slowly being severed from these global standards. The tool of that severance is not a scalpel but a dull spoon and that is why Huawei is terrified.

Honestly its kind of hillarious how some severely uneducated troll-brats here just copy and paste the same thing and feel if they just believe it strongly enough, it will be true. Then they have +2/-19 rating to show for it on top, in spite of being "iron bro" for this forum....goes to show the root quality of their interaction even with severely stacked playing field.

Keep your quality responses going (As much as you have time/interest for of course) , its a gold mine of content.

 

[FairAndUnbiased]

Nice, thanks for the info. Pretty solid career path considering semi's been on a bull run last 5+ years.

test is a pretty low value added part and quite far removed from front end processes.

 

[letsrock]

Wrong...You WILL lose far more than money. This is why it is so entertaining reading you fools.

There is nothing wrong with being technologically behind. The issue is whether you have CONTINUAL access to developing technologies to maintain your status quo. In isolation, you either have no or little access to those developing technologies. Indigenous development to catch up? Dream on. Commodity semicon products takes 2-3 months from wafer start to client.

https://www.zeiss.com/semiconductor...nership/zeiss-systems-in-a-wafer-stepper.html

You just developed a new stepper? Try at least 2 YEARS of engineering wafer runs before a single client is willing to accept engineering samples just to test wafers that came from your new stepper. That client will take one year of testing under his own criteria before publishing his findings. Now three years have passed and in that time, you cannot develop further that new stepper you just built. You have to lock in that technology to remove all the engineering variables that comes with continuous development. In that three yrs time, the client will compare wafers that came from your indigenously developed stepper against GLOBAL standards and if there are greater than %3 difference in all criteria, your stepper WILL be rejected. If one client reject, most likely others will without doing their own evals. In short, you have to wait 2-3 yrs to find out if your new hardware have any shortcomings under COMMODITY products, let alone using the new hardware on new 'cutting edge' products.

Right now, under autonomous autos requirements, 20 nm NAND commodity dies are being rejected left and right from all manufacturers and we are talking about major players like Samsung, Micron, Toshiba and so on. TWENTY NM...!!! How old is that NAND technology? Try at least 10 yrs old. In other words, autonomous driving requirements are rejecting %50 of wafers from established technologies as inadequate. That is NASA, ESA, and CNSA level.

Your China is slowly being severed from these global standards. The tool of that severance is not a scalpel but a dull spoon and that is why Huawei is terrified.

Well Huawei is not selling to those clients. So their power of rejection makes no impact to huawei. Huawei itself is the consumer or the "client" of its chips as it uses them to power its mobile and laptops. The only client here in this case is the end consumer. And he has no 2-3 year process.

 

[Han Patriot]

Wrong...You WILL lose far more than money. This is why it is so entertaining reading you fools. .

Well genius, you can't lose something which you don't HAVE. When I say you will lose money, I meant losing money from investing in catching up tech. It's all about burning money continuous researching until the other party is exhausted. This is how semiconductor works, Samsung basically invested into a loss making venture for 10 years to catch up to TSMC and Intel and the Japs pretty much gave up. Semiconductor is basically a duopoly now.

There is nothing wrong with being technologically behind. The issue is whether you have CONTINUAL access to developing technologies to maintain your status quo. In isolation, you either have no or little access to those developing technologies. Indigenous development to catch up? Dream on. Commodity semicon products takes 2-3 months from wafer start to client.

https://www.zeiss.com/semiconductor...nership/zeiss-systems-in-a-wafer-stepper.html

You just developed a new stepper? Try at least 2 YEARS of engineering wafer runs before a single client is willing to accept engineering samples just to test wafers that came from your new stepper. That client will take one year of testing under his own criteria before publishing his findings. Now three years have passed and in that time, you cannot develop further that new stepper you just built. You have to lock in that technology to remove all the engineering variables that comes with continuous development. In that three yrs time, the client will compare wafers that came from your indigenously developed stepper against GLOBAL standards and if there are greater than %3 difference in all criteria, your stepper WILL be rejected. If one client reject, most likely others will without doing their own evals. In short, you have to wait 2-3 yrs to find out if your new hardware have any shortcomings under COMMODITY products, let alone using the new hardware on new 'cutting edge' products.

Right now, under autonomous autos requirements, 20 nm NAND commodity dies are being rejected left and right from all manufacturers and we are talking about major players like Samsung, Micron, Toshiba and so on. TWENTY NM...!!! How old is that NAND technology? Try at least 10 yrs old. In other words, autonomous driving requirements are rejecting %50 of wafers from established technologies as inadequate. That is NASA, ESA, and CNSA level.

Your China is slowly being severed from these global standards. The tool of that severance is not a scalpel but a dull spoon and that is why Huawei is terrified.

I think what you don't understand is China had always been under sanctions for semicon equipment, its always 2 to 3 gen behind due to supply restriction of semicon equipment, otherwise, we would have decimated the whole US Korean Jap and Taiwanese semiconductor industry. If US really wanted to cut us off just ban import and export from China outright, default on the debt, come on man we are accused of killing 90k Muricans and decimating US economy back to the 30s destroying wealth created since 2008. The best you can do is ban chip sales to ONLY Huawei? Lol, I would have nuked China.

Best process tech is 14nm, The best EUV equipment we have is now 22nm, DUV is around 10nm with double exposure, we need time to scale up the efficiency of these new tech, we pretty much produce the rest of the machines. Materials and gases are covered too. It's a matter of time my friend, the size will stop at around 1 to 2nm unless you go quantum. Koreans and Taiwanese can't produce their own materials gases and equipment, they are only good at pricess tech. Japs are good at equipment and materials but not in process, China is only one covering the whole supply chain after US. You are afraid...you can laugh threaten sanction us, but you know we won't give up.

 

[FairAndUnbiased]

Well Huawei is not selling to those clients. So their power of rejection makes no impact to huawei. Huawei itself is the consumer or the "client" of its chips as it uses them to power its mobile and laptops. The only client here in this case is the end consumer. And he has no 2-3 year process.

unfortunately he is mostly correct though. There may be an expedited process for Huawei but ultimately at the end of the day they need to make money. The customer is also not Huawei directly, it is the company that fabs for Huawei. SMIC for 14 nm or Samsung for 7 nm (one of the few choices if TSMC pulls out of their contract), basically. And SMIC or Samsung cannot take a risk on an unproven piece of equipment. There's a large gap between R&D equipment that can process 10 wafers reliably to something that can process 10k wafers reliably.

SMIC might accept the possibility of expediting evaluation of new equipment but it won't be quick. To get engineering data on a wafer you typically ship it to an ISO 17025 certified lab where each wafer will undergo ALOT of testing, to put it lightly. Each test costs ~1k and you need a statistically relevant number of samples. This is not automated probe testing, this is chemical testing to prove that your equipment actually works (mostly for etch and deposition chambers). You're looking at 500k-1M at the low end for just testing alone. it is not cheap or easy to prove that your equipment works and to put it in a commercial setting, it has to be proven to work reliably.

One alternative - again just spending money - is for Huawei to buy an existing fab.

 

[letsrock]

unfortunately he is mostly correct though. There may be an expedited process for Huawei but ultimately at the end of the day they need to make money. The customer is also not Huawei directly, it is the company that fabs for Huawei. SMIC for 14 nm or Samsung for 7 nm (one of the few choices if TSMC pulls out of their contract), basically. And SMIC or Samsung cannot take a risk on an unproven piece of equipment. There's a large gap between R&D equipment that can process 10 wafers reliably to something that can process 10k wafers reliably.

SMIC might accept the possibility of expediting evaluation of new equipment but it won't be quick. To get engineering data on a wafer you typically ship it to an ISO 17025 certified lab where each wafer will undergo ALOT of testing, to put it lightly. Each test costs ~1k and you need a statistically relevant number of samples. You're looking at 500k-1M at the low end for just testing alone. it is not cheap or easy to prove that your equipment works and to put it in a commercial setting, it has to be proven to work reliably.

One alternative - again just spending money - is for Huawei to buy an existing fab.

My understanding is that US is banning chips made with its equipment. Not the lithography equipment itself. There are also other countries which produce lithography equipment (netherlands i think). i understood the "client" as client of chips themselves not chip making machines. Huawei designs some of its chips, fabs them with TSMC and uses them in its mobiles, laptops. It also purchases chips outright to be used in its mobiles and laptops.

it is those second chips which will be effected with the new ban reducing the competitiveness of huawei end devices in international markets.

 

[FairAndUnbiased]

My understanding is that US is banning chips made with its equipment. Not the lithography equipment itself. There are also other countries which produce lithography equipment (netherlands i think). i understood the "client" as client of chips themselves not chip making machines. Huawei designs some of its chips, fabs them with TSMC and uses them in its mobiles, laptops. It also purchases chips outright to be used in its mobiles and laptops.

They will prevent any chips made on US equipment from being shipped to Huawei. That essentially means that Huawei has to source from a plant that doesn't use much US equipment. the only choice would be SMIC or a Japanese foundry and Japan does not have a 7 nm foundry. I believe both Samsung and TSMC extensively use US equipment.

Lithography isn't the only equipment. There's various ion implantation, etch and deposition steps, all made with US equipment. There's some domestic substitutes but they're just ramping up. I have confidence that they will indeed ramp up, but it will still be a delay if they want to replace parts. Lithography is the big problem, but luckily there's Netherlands and Japan.

Like it or not Lam Research, Applied Materials and KLA Tencor are the kings.

 

[letsrock]

They will prevent any chips made on US equipment from being shipped to Huawei. That essentially means that Huawei has to source from a plant that doesn't use much US equipment. the only choice would be SMIC or a Japanese foundry and Japan does not have a 7 nm foundry. I believe both Samsung and TSMC extensively use US equipment.

Lithography isn't the only equipment. There's various ion implantation, etch and deposition steps, all made with US equipment. There's some domestic substitutes but they're just ramping up. I have confidence that they will indeed ramp up, but it will still be a delay if they want to replace parts. Lithography is the big problem, but luckily there's Netherlands and Japan.

Like it or not Lam Research, Applied Materials and KLA Tencor are the kings.

Ofcourse huawei's products will not be competitive in international market if access to the chips they rely on is taken away from them. But other chinese mobile companies can still buy them. And if china retailiates against boeing and apple then the effects can be disastrous. trade wars are lose lose wars atleast in short term.

And i dont know why trump is going all the way down on huawei, american farmers got a sweet deal in trade deal one.

 

[gambit]

test is a pretty low value added part and quite far removed from front end processes.

Are you serious in saying that? Test -- my part -- is low valued? Do I need to post a Captain Picard face palm image?

 

[tower9]

Ofcourse huawei's products will not be competitive in international market if access to the chips they rely on is taken away from them. But other chinese mobile companies can still buy them. And if china retailiates against boeing and apple then the effects can be disastrous. trade wars are lose lose wars atleast in short term.

And i dont know why trump is going all the way down on huawei, american farmers got a sweet deal in trade deal one.

Huawei is a target because they currently have the most advanced 5G network and are poised to dominate the 5G rollout globally. 5G is the backbone of the fourth industrial revolution and whoever masters it will control that platform and the future.

 

[Han Patriot]

Huawei is a target because they currently have the most advanced 5G network and are poised to dominate the 5G rollout globally. 5G is the backbone of the fourth industrial revolution and whoever masters it will control that platform and the future.

Well they can buy chips through a third party or create a new company call MIMO by firing the 5g research group. No more link to Huawei but curiously with all the whole 5G tech. Lol, there are many ways to an end. Ultimately I want China to be completely self reliant, we did that in satellites and military tech. Their last stand is advanced aircraft engines and semiconductors.