ASML is also part of this market, ASML current stock price is $789, where as TSMC is only at $114,Amazons stock price is currently $3,409, to put that in perspective.
That's how they do it over there. I remember watching a documentary about a factory complex in China with 15,000 workers that had restaurants, barbers, a hospital, places to get married, a daycare, apartments, etc.
@@glowingdawn9179 US businesses tended to be like this actually. Train companies and steel mills would build worker towns and much of the initial infrastructure and services. Disney might have still been one if it's founder lived forever. It's a paternalistic company mindset that's now lost in the US due to historical reasons like worker strikes in those company cities, decline of those industries that built the cities (trains and steel mills are a sunset industry in the US) and anti-trust legislation (to add that there is a general disdain for companies being too powerful, and a company city might feel dystopian for Americans), but still strong in Asia because of Confucianism.
Like in the US, Amazon probably has enough money to do a Samsung City if they wanted, but they don't. In Asia, I would reckon, that besides pure capitalism, being paternalistic like this, providing everything and appearing ubiquitous, it's a source of prestige.
I worked at a semiconductor equipment company and Samsung was a very good customer. They bought even more high end tools than TSMC. Gotta love having money to burn and throwing hardware at a problem. I also had the impression they were a big adopter of EUV, but apparently not based on the chart @ 3:24
I think TSMC is doing very well on node advancement and yields. However if Samsung does well with it's 3nm GAAFet they may pull ahead while TSMC stays with FinFet till 2nm. But TSMC has been investing in new material research, maybe they'll be first to a good silicon replacement.
@@zenova9926 All of that covers quality. GAAFET has better performance characteristics, so Samsung might have a better 3nm node, TSMC has had better node qualities than Samsung the last few years. New materials for either doping or replacement would absolutely be an improvement on factors other than density.
@@Jaker788 i ve read that samsung has yield problems on its 3nm GAAFET. Apparentely they cant go past 30% yield and this caused them to lose NVIDIA and even Qualcomm iirc There is a good article on this on a substack blog called "semi analysis"
Not really sure if sammy can catch up for while. To be honest, the yield rate on 3nm GAA is just too low to be commercially viable. And who is buying its 3nm at volume?
Another very well-researched and informative video. I think Taiwan and TSMC should be very happy that the United States has added China's SMIC to the Entity List (banning commercial transactions containing US technology unless an export license/exemption is first obtained). I think it's now very unlikely that China will be able to develop a world-class semiconductor fabrication facilities. Having Huawei on the Entity List directly hurts TSMC financially, but on balance I think it's well worth it. Side note: The strategy that economic engagement leads to democratization worked very well for South Korea and Taiwan (both were both brutal dictatorships until only 30 years ago). That strategy has of course failed spectacularly for China: as it got richer it's becoming more brutal. I really like the idea of democracies using their economic might to preventing brutal dictatorships from becoming proficient in high technology -- like the United States is doing with its Entity List.
I would not be so smug and play hardball with America against China. If and when China develops the most advanced chips, Taiwan may get burnt. Taiwan depends on China economically. This is a fact and any smart Taiwanese president needs to deal with China as an economic partner. South Korea will be happy to take over the Chinese market for chips if Taiwan makes an exit.
Eh so what the US did to Alstom and Toshiba is justified too? Narrow and prejudice view imo. But doesn't really matter as technology cycles happen throughout history and the future isnt written in stone. Its all about engineering talents and capability, and my bet is on China since they have a population of 1.4b people and a good analogy is comparing chip density to achieve better speed and capability u get the point, the more transistors u pack together the greater the chip's power....
"China: as it got richer it's becoming more brutal." I really have no idea what koolaid u have been taking? Have u been to China? Do yourself a favor go visit China, especially Shanghai go findout how the common chinese people life their lives, or go google and look at how westerners who have lived in China and findout what they have to say about living, working and studying in China! I myself have been to Shanghai, Shenzhen, Beijing, HK, Macau for holidays and the freedom and safety especially in Shanghai and Beijing I and my family was able to walkaround the city pass midnight in absolute safety in all parts of the city. If I were to do that in most of the US cities, I will probably have been robbed and mugged!
The advance node has less profit per wafer, actually, due to the investment in advance tools. Mature technologies have run orders of magnitudes higher WIP than advance nodes so the tools and facility cost were already paid back. That's why many old TSMC fabs even back to 8-inch production factories are still running. Those factory are still evolving and shifting their products. For example, old fab can maintain the same node but change from logic computing chip to pure analog or RF devices, or even micro-mechanic devices. That's a quite different business model and practice than intel and Samsung (they replace their HVM sites with latest mature node).
@@Asianometry ok so it is not overly dependant on apple. It wouldn't be great for the company but also for the business model. As a customer if I have to go to a company that might have to prioritize my main competitors need instead of me it would be a real concern.
@@delseras Bad yields are bad for them. tbfh, I'd rather be having bad yields than having Intels problems of CPUs that don't last six months.... :) One is bad, but not noteworthy, and one is absolutely destroying the brand.
@CelticKnight2004 just read stuff about intel and it's not looking good for them. SoCs manufactured by samsung usually burns battery fast and tend to overheat fast too but I guess that's better than what intel is having rn
@@delseras yeeah exactly. apparently Samsung & Apple are bidding to buy Intel. I suspect the US Govt would block the sale of intel (or takeover as the case likely will be) by any foreign company. Doubly so, if Trump is elected.
You missed the Japanese ban on sales of hydrofluoric acid to S Korea which will affect S Korean foundries' productions. TSMC is not affected by the Japanese acid ban,
Currently, TSMC is the leader in the foundry business but it will come a time when politics might bring down TSMC to second, behind Samsung, or less. In recent years, China has been very assertive in bringing Taiwan under its control and if this happens, TSMC will most likely become an entity with restricted access to US and other nation's chip technologies. If this scenario plays out, I would not be surprised if many of TSMC's top customers switch to Samsung for their chips need.
@Matthew Park It's ww3, Taiwanese and Taiwanese who works at and run tsmc are more than likely those who are anti CCP for sure, tsmc could stop operating overnight and the world economy instantly in ruin, too much economic stake at play for every countries with ties to tsmc to just stand idly.. With the world literally have only 2 major players who can manufacture these chips, there's no way out, Samsung can't even save the world even if they want to and would like it, it's physically impossible to dump it all to Samsung, global foundry is dead, umc whatever can no longer do it, Intel has never produced anything but its own. Even China know that assessment.
Taiwan won't fold like HK, Tibet, and Xinjiang as they were not armed and had no military. Taiwan is well-armed and the Chinese people will fight and die for their country as descendants of chiang kai-shek. The commies know this too. they will not goto war, but will use their economic and political power to suppress taiwan like they have been doing.
Actually i was hoping that this was also some sort of interpretation of future plans, not only a description of the status quo. You completely ignored the fact that Samsung is betting hard on an early GAA introduction to turn it all around, while at TSMC it is still unknown when they will draw level with that. This is especially important, because of all the Investments of Samsung you cite, most of it has gone into development of GAA rather than the short term nodes. Samsungs 8nm was a very cost effective evolutionary step from 10nm, same for 5LPE which is hardly an upgrade to their rather mediocre performing 7LPP. So all of that investment went not into competing with TSMC right now, but to leapfrogg them some time in 2022-23. No wonder the current processes are not 100% up to scratch when the investment has gone elsewhere. It is hard to say if this risky bet by Samsung is paying off and when, since GAA seems to be more complicated than anticipated. But still the electrical and scaling benefits of it are real and it is clear to everybody in the market that GAA will be an unavoidable thing for future node and performance scaling. @asianometry You completely missed these facts about the very different roadmaps of TSMC and Samsung.
@Sheng-fu Horng TSMC, Intel and Samsung have ALL said they're moving to GAA FET in trying to achieve a node which they all call a 2nm node. Their names are different but the transistor density is similar regardless of the name. In regards to getting to a 3nm node (these are advertising names not actual size of anything) Intel, TSMC and Samsung are doing different things. Actually I think Intel is skipping something they would call "Intel 3". Intel is going from "Intel 4" to "Intel 20A". On the other hand Samsung and TSMC are experimenting with different technologies getting to either N3 or 3N depending on which company you're talking about. So, until these companies say something different, they all plan to move to GAA FET to get to 2nn or 20A. What they're doing in between where they are now and getting to 2nm varies. Oh, and I think they are all still waiting for a new line of lithography equipment from ASML to move to these smaller nodes. I don't think any of these companies have this yet. My understanding though is Intel will be the first to get their hands on it. I think they own a larger share of ASML stock. But I could be wrong.
Sounds like simplistic regurgitation of data and statistics. Apple split its chip fabrication one year between Samsung an TSMC. The Samsung chips turned out inferior to those made by TSMC, they ran hotter and had higher battery consumption. Since then, Apple moved it's chip manufacturing to TSMC, never looked at Samsung again. Samsung is a competitor to most companies who need to have chips fabricated. Why would those companies give their business to a competitor?
foundry business is very expensive and less profitable than designing and selling actual semiconductors. thats why we have more chip designers than foundries. Samsung foundry invests a minimum in order to supply state of the art memory and APs for its own consumer products and businesses in the first line making Samsungs SC business a gold mine with a jaw dropping 40-50% operating margin outpacing TSMCs 30-40%. however the current SC crisis has turned the table in favor for TSMC boosting its op. margin to 50%. but this situation may not last as all foundries are going to expand their capacities and shrinking to smaller nodes is getting more expensive
Each time I watch your videos I ask myself - what does this guy do? He is technically too sound to be an investor. He puts out too many UA-cam video, an Engineer isn't suppose to have this much time. So, is John a CIA analyst, analyzing the Asian threat on US's dominance? Or is he just an engineer turned investor?
@@alexmartian3972 He's a financial analyst. That explains his deeper understanding of economics. His father was a microchip designer. That explains why he's technically so sound. I found both of this from his own comments in some past videos.
@@aniksamiurrahman6365 You at the end of comment asked if he turned investor. I assumed you wonder how he makes money and replied - he is a youtuber: patreon + ads provides at least some income.
This video doesn't even mention Intel. How Come? Isn't Intel as capable as Samsung and TSMC making that kind of advanced semiconductors? Anyone could answer me?
I think TSMC might fail not because of their technology or losing big customers. I think that the biggest problem is most people don't want to work at TSMC. Therefore, in the future, they might lack employees. TSMC's workers need to work 12h continuously four times in a week. They often need to start work from midnight. Even on the day-off or when they sleep, if a problem occurs, they need to go back to work. I'm now studying at a famous university in Taiwan, most of my friends in the university said they don't want to work at TSMC.
It will take years. If you follow SMIC's recent news, you will know the gap isn't small and TSMC continues to advance in the technology. Just recently, they have started planning for 1 nm and 2 nm factories when 3nm is not even in production yet.
@@carholic-sz3qv not everyone has the ability to go under 3nm. Currently only Samsung and tsmc as far as I know. Intel with that 20 billion fund might eventually get there in 3-5 years.
@@jethrolai its actually ASML and Carl Zeiss that makes the 7nm, 5nm, 3nm..... machines. tsmc and Samsung buys those machines and include them to their production line with other machines.
@@carholic-sz3qv if you mean EUV, DUV, yes, ASML is the only company that makes EUV, the entire supply chain including Zeiss, Cymer, Synopsys. Cadence Design System, and so on. The entire supply chain consists of more than 5000 companies. No one company can do this alone.
Although tsmc has a huge advantage in that it does not compete with customers and has more foundry experiences, it is a big deal if its technological progress is delayed. samsung will mass produce 3nm for the first time in the world, even in gaa. experience first means more skillful and reliable. market for less than 10nm is growing rapidly. since samsung has almost half market share of below 10nm, the overall market share gap will be reduced. if tsmc fails to mass produce 3nm in the second half of next year, it will lose apple to samsung, which will be a huge disaster. tsmc must mass produce 3nm until next year. people ignore Intel's foundry but actually only intel has possibility to compete with tsmc and samsung. intel, as a general semiconductor company and has the biggest sales among semiconductor companies, has very good semiconductor synthesis capabilities, so if it invests a lot and receives support from the us government, it can fast follow tsmc and samsung. foundry will be tsmc vs samsung vs intel. umc, global foundry and smic dont have ability and money.
I doubt it, Apple will never fully rely on Samsung for their high end chips, which means they'll never dominate. Although second place in a growing pie isn't all that bad. The question is who will be the 3rd largest contender that can ship similar products???
@@selohcin Long time? Few years I bet... especially since their government is supporting them as well. They're already talking about expanding to Texas with that $16 billion infras.
Should Apple build its own foundry(ies) on American soil? I personally think it is totally capable of doing so [in every way], but it seems like the CEO doesn't want to build one just yet.
I think you underestimate the necessary level of detailed process knowledge that is needed to set up a EUV chip fab. Look, even Intel is struggling and they have decades of experience and ample cash.
@@Martinit0 Probably hard at first but my point is that I don't want the U.S to rely too much on foreign supply chains. I'm leaning towards political issues that may occur. Who knows the future.
Others want them to build EV's too. If they could and make money, they would! not likely. Everyone is great at doing ONE thing. becoming a jack of all trades makes you mediocre at everything.
Apple M1 ARM chips are revolutionary in laptops-fast & 18-20 hours battery life. Their M1 based laptops are on back order & the growth will be amazing in next quarters earnings. The next low power/fast execution market is servers. Expect Apple & TSMC to smoke Intel & AMD in the future. I wouldn’t be surprised if Apple bought out Dell or HPE to incorporate their chips in servers. Also, GPU manufacturers like Nvidia are going to be hurting shortly since the GPU is in the M1 chip, thus making it super fast & energy efficient when compared to external GPUs.
Uhm nope apple M1 ARM chip isn't revolutionary. Its all about optimisation and apple only makes just one version of each procuts unlike Intel who makes chips for basically every companies and every applications. ARM chips by their nature are very powerful and efficient.
TSMC isn't going to smoke intel or AMD. TSMC is already making AMD chips and is going to make the next intel chips..... NVIDIA is also next tsmc customer with their tegra chips for example.
But who has shown better business practice in the past? Samsung or Apple? Answer: hands down Samsung 👏. Apple has cought my attention with its lawsuit against FB for selling out its users to the highest bidder. So I will be watching with interest as we move forward
No You obviously don’t even understand it was Facebook that filled the lawsuit not the other way around. Apple’s latest versions of IOS allow the hardware users to stop tracking on the device Facebook uses to target adds
@@johniii8147 If you are going to contest what I'm saying please use proper English and provide proof in the form of a link or I am just going to assume you are a troll. Thank you 🙂
You must have no idea how hard it is to build semiconductor chips... wow lol. China is over a decade behind.. even the US is falling behind. As long as technology improves the chips will have to improve.. even making graphics card takes decades of research because companies like NVIDIA has decades of experience.
If viewers liked this video I recommend this channel's "TSMC - Essays" playlist.
Here's a direct link to the playlist: ua-cam.com/play/PLKtxx9TnH76SRC7ZbOu2Nsg5mC72fy-GZ.html
@@shazmosushi thanks man, this channel needs more eyeballs.
Man, I work for ASML and love these videos that summon up the current industry!
Keep it up
ASML is also part of this market, ASML current stock price is $789, where as TSMC is only at $114,Amazons stock price is currently $3,409, to put that in perspective.
The amount of money spent on all of this is mindblowing. The facilities Samsung built are like a city within a city.
That's how they do it over there. I remember watching a documentary about a factory complex in China with 15,000 workers that had restaurants, barbers, a hospital, places to get married, a daycare, apartments, etc.
It’s funded directly or indirectly by the government Samsung is a source of pride for South Korea so they get plenty of power preference
@@glowingdawn9179 US businesses tended to be like this actually. Train companies and steel mills would build worker towns and much of the initial infrastructure and services. Disney might have still been one if it's founder lived forever. It's a paternalistic company mindset that's now lost in the US due to historical reasons like worker strikes in those company cities, decline of those industries that built the cities (trains and steel mills are a sunset industry in the US) and anti-trust legislation (to add that there is a general disdain for companies being too powerful, and a company city might feel dystopian for Americans), but still strong in Asia because of Confucianism.
Like in the US, Amazon probably has enough money to do a Samsung City if they wanted, but they don't. In Asia, I would reckon, that besides pure capitalism, being paternalistic like this, providing everything and appearing ubiquitous, it's a source of prestige.
@@johniii8147 That's what most of governments do. Nothing special.
I worked at a semiconductor equipment company and Samsung was a very good customer. They bought even more high end tools than TSMC. Gotta love having money to burn and throwing hardware at a problem. I also had the impression they were a big adopter of EUV, but apparently not based on the chart @ 3:24
I think TSMC is doing very well on node advancement and yields. However if Samsung does well with it's 3nm GAAFet they may pull ahead while TSMC stays with FinFet till 2nm.
But TSMC has been investing in new material research, maybe they'll be first to a good silicon replacement.
Is not the size but the quality of the node itself, guess you never do any research of these
@@zenova9926 All of that covers quality. GAAFET has better performance characteristics, so Samsung might have a better 3nm node, TSMC has had better node qualities than Samsung the last few years. New materials for either doping or replacement would absolutely be an improvement on factors other than density.
@@Jaker788 i ve read that samsung has yield problems on its 3nm GAAFET. Apparentely they cant go past 30% yield and this caused them to lose NVIDIA and even Qualcomm iirc
There is a good article on this on a substack blog called "semi analysis"
*semianalysis
Not really sure if sammy can catch up for while. To be honest, the yield rate on 3nm GAA is just too low to be commercially viable. And who is buying its 3nm at volume?
And it all started with some redstone and a button.
It’s not DRAM it’s DeeRAM, or S-RAM and not SHRAM.
I think he does it on purpose
Shut up
Great channel. Just subscribed today!
who would bet against samsung? just wait a decade who first then
2:36 where is this building
news.samsung.com/kr/2545
Yeosu, South Korea
Another very well-researched and informative video.
I think Taiwan and TSMC should be very happy that the United States has added China's SMIC to the Entity List (banning commercial transactions containing US technology unless an export license/exemption is first obtained).
I think it's now very unlikely that China will be able to develop a world-class semiconductor fabrication facilities. Having Huawei on the Entity List directly hurts TSMC financially, but on balance I think it's well worth it.
Side note: The strategy that economic engagement leads to democratization worked very well for South Korea and Taiwan (both were both brutal dictatorships until only 30 years ago). That strategy has of course failed spectacularly for China: as it got richer it's becoming more brutal. I really like the idea of democracies using their economic might to preventing brutal dictatorships from becoming proficient in high technology -- like the United States is doing with its Entity List.
I would not be so smug and play hardball with America against China. If and when China develops the most advanced chips, Taiwan may get burnt. Taiwan depends on China economically. This is a fact and any smart Taiwanese president needs to deal with China as an economic partner. South Korea will be happy to take over the Chinese market for chips if Taiwan makes an exit.
Eh so what the US did to Alstom and Toshiba is justified too? Narrow and prejudice view imo. But doesn't really matter as technology cycles happen throughout history and the future isnt written in stone. Its all about engineering talents and capability, and my bet is on China since they have a population of 1.4b people and a good analogy is comparing chip density to achieve better speed and capability u get the point, the more transistors u pack together the greater the chip's power....
@linkzable - Unfortunately its not as easy to break thru
"China: as it got richer it's becoming more brutal." I really have no idea what koolaid u have been taking? Have u been to China? Do yourself a favor go visit China, especially Shanghai go findout how the common chinese people life their lives, or go google and look at how westerners who have lived in China and findout what they have to say about living, working and studying in China! I myself have been to Shanghai, Shenzhen, Beijing, HK, Macau for holidays and the freedom and safety especially in Shanghai and Beijing I and my family was able to walkaround the city pass midnight in absolute safety in all parts of the city. If I were to do that in most of the US cities, I will probably have been robbed and mugged!
You say that and the genocide continues, you can wipe as much as you can, china isn't clean and anyone who deep search can see.
The advance node has less profit per wafer, actually, due to the investment in advance tools. Mature technologies have run orders of magnitudes higher WIP than advance nodes so the tools and facility cost were already paid back.
That's why many old TSMC fabs even back to 8-inch production factories are still running. Those factory are still evolving and shifting their products. For example, old fab can maintain the same node but change from logic computing chip to pure analog or RF devices, or even micro-mechanic devices. That's a quite different business model and practice than intel and Samsung (they replace their HVM sites with latest mature node).
Depending too much on one single customer is often risky. How much percent of TSMC incomes are generated by apple ?
We don’t know for certain. The estimate is something like 15-20% of total revenue.
@@Asianometry ok so it is not overly dependant on apple. It wouldn't be great for the company but also for the business model. As a customer if I have to go to a company that might have to prioritize my main competitors need instead of me it would be a real concern.
25% in 2020
@@AxelCalvet that’s how the world works, money talks
Recent TSMC SEC filing says "Our largest customer in 2018, 2019 and 2020 accounted for 22%, 23% and 25% of our net revenue in the respective year. "
That's because primarily Apple. They reserved over 50% of TSMC's capacity. So Qualcomm went with Samsung.
Microphone stuff up around 1:41
Ugh you’re right. My error. I need to get back to my regular mike.
Healthy competition between TSMC and Samsung is great for all involved.
Competition shrinks prices and forces innovation.
Fastforward to 2024, samsung hasn't been producing good yields since then
@@delseras Bad yields are bad for them.
tbfh, I'd rather be having bad yields than having Intels problems of CPUs that don't last six months.... :) One is bad, but not noteworthy, and one is absolutely destroying the brand.
@CelticKnight2004 just read stuff about intel and it's not looking good for them. SoCs manufactured by samsung usually burns battery fast and tend to overheat fast too but I guess that's better than what intel is having rn
@@delseras yeeah exactly. apparently Samsung & Apple are bidding to buy Intel.
I suspect the US Govt would block the sale of intel (or takeover as the case likely will be) by any foreign company.
Doubly so, if Trump is elected.
Its not about market share now. There's way more than enough leading edge tech chip demand to keep every one growing.
What are your research methods to make this videos? I'm curious and amazed how youtubers put content together like this. Thanks.
Another great video
Hiya, just wanted to ask what source you used for TSMC alone makes 50% of Taiwans R&D spend. Thank you :)
Actually, 25%
178.7 billion
Very insightful
Will be interesting to see how DARPA' ERI tackles the issue of cost for these fabs.
TSMC alone made 50% of R&D expenditure of the country.@ 5:25
Lol
No wonder. They are involved in anything with chip. They deserve it.
You missed the Japanese ban on sales of hydrofluoric acid to S Korea which will affect S Korean foundries' productions.
TSMC is not affected by the Japanese acid ban,
Korea already made it last year june.
Korea didn't make because because it is cheap. It's easy to make and now already make.
Currently, TSMC is the leader in the foundry business but it will come a time when politics might bring down TSMC to second, behind Samsung, or less. In recent years, China has been very assertive in bringing Taiwan under its control and if this happens, TSMC will most likely become an entity with restricted access to US and other nation's chip technologies. If this scenario plays out, I would not be surprised if many of TSMC's top customers switch to Samsung for their chips need.
You realize if China takes Taiwan by force it'll likely be WWIII.
@Matthew Park
It's ww3, Taiwanese and Taiwanese who works at and run tsmc are more than likely those who are anti CCP for sure, tsmc could stop operating overnight and the world economy instantly in ruin, too much economic stake at play for every countries with ties to tsmc to just stand idly..
With the world literally have only 2 major players who can manufacture these chips, there's no way out, Samsung can't even save the world even if they want to and would like it, it's physically impossible to dump it all to Samsung, global foundry is dead, umc whatever can no longer do it, Intel has never produced anything but its own.
Even China know that assessment.
Taiwan won't fold like HK, Tibet, and Xinjiang as they were not armed and had no military. Taiwan is well-armed and the Chinese people will fight and die for their country as descendants of chiang kai-shek. The commies know this too. they will not goto war, but will use their economic and political power to suppress taiwan like they have been doing.
Already building FABS in the USA to avoid the potential problem.
@Matthew Park Why should the US not care?
Why isn't tsmc private
Actually i was hoping that this was also some sort of interpretation of future plans, not only a description of the status quo. You completely ignored the fact that Samsung is betting hard on an early GAA introduction to turn it all around, while at TSMC it is still unknown when they will draw level with that.
This is especially important, because of all the Investments of Samsung you cite, most of it has gone into development of GAA rather than the short term nodes. Samsungs 8nm was a very cost effective evolutionary step from 10nm, same for 5LPE which is hardly an upgrade to their rather mediocre performing 7LPP.
So all of that investment went not into competing with TSMC right now, but to leapfrogg them some time in 2022-23. No wonder the current processes are not 100% up to scratch when the investment has gone elsewhere. It is hard to say if this risky bet by Samsung is paying off and when, since GAA seems to be more complicated than anticipated. But still the electrical and scaling benefits of it are real and it is clear to everybody in the market that GAA will be an unavoidable thing for future node and performance scaling.
@asianometry You completely missed these facts about the very different roadmaps of TSMC and Samsung.
@Sheng-fu Horng TSMC, Intel and Samsung have ALL said they're moving to GAA FET in trying to achieve a node which they all call a 2nm node. Their names are different but the transistor density is similar regardless of the name.
In regards to getting to a 3nm node (these are advertising names not actual size of anything) Intel, TSMC and Samsung are doing different things. Actually I think Intel is skipping something they would call "Intel 3". Intel is going from "Intel 4" to "Intel 20A". On the other hand Samsung and TSMC are experimenting with different technologies getting to either N3 or 3N depending on which company you're talking about.
So, until these companies say something different, they all plan to move to GAA FET to get to 2nn or 20A. What they're doing in between where they are now and getting to 2nm varies.
Oh, and I think they are all still waiting for a new line of lithography equipment from ASML to move to these smaller nodes. I don't think any of these companies have this yet. My understanding though is Intel will be the first to get their hands on it. I think they own a larger share of ASML stock. But I could be wrong.
When push comes to shove, Samsung will take back its capacity for Samsung products.
Sounds like simplistic regurgitation of data and statistics.
Apple split its chip fabrication one year between Samsung an TSMC. The Samsung chips turned out inferior to those made by TSMC, they ran hotter and had higher battery consumption.
Since then, Apple moved it's chip manufacturing to TSMC, never looked at Samsung again.
Samsung is a competitor to most companies who need to have chips fabricated. Why would those companies give their business to a competitor?
Bc they can't do it by themselves
The technology and factories that Samsung owns, Apple even cant dream of
foundry business is very expensive and less profitable than designing and selling actual semiconductors. thats why we have more chip designers than foundries.
Samsung foundry invests a minimum in order to supply state of the art memory and APs for its own consumer products and businesses in the first line making Samsungs SC business a gold mine with a jaw dropping 40-50% operating margin outpacing TSMCs 30-40%. however the current SC crisis has turned the table in favor for TSMC boosting its op. margin to 50%. but this situation may not last as all foundries are going to expand their capacities and shrinking to smaller nodes is getting more expensive
Each time I watch your videos I ask myself - what does this guy do? He is technically too sound to be an investor. He puts out too many UA-cam video, an Engineer isn't suppose to have this much time. So, is John a CIA analyst, analyzing the Asian threat on US's dominance? Or is he just an engineer turned investor?
He is a youtuber that is we can know for sure
@@alexmartian3972 He's a financial analyst. That explains his deeper understanding of economics. His father was a microchip designer. That explains why he's technically so sound. I found both of this from his own comments in some past videos.
@@aniksamiurrahman6365 You at the end of comment asked if he turned investor. I assumed you wonder how he makes money and replied - he is a youtuber: patreon + ads provides at least some income.
@@alexmartian3972 No, I wasn't pondering about how he makes money. I was pondering about what he is.
This video doesn't even mention Intel. How Come? Isn't Intel as capable as Samsung and TSMC making that kind of advanced semiconductors? Anyone could answer me?
Ask intel. They lost out long time ago in manufacturing.
Apple and Qualcom switched to Saumsung before and really got burned.
Uhm nope thats totally wrong.
@@carholic-sz3qv well 5nm Samsung is equal to 7nm tmsc.
@@siesta9525 nope Samsung 5nm is similar to tsmc 5nm and they use the same machines from the same supplier ASML and other optimisations in house.
@@carholic-sz3qv so why apple don't use Samsung?they always do that in the first place, Because tmsc just better everyone know that.it
@@carholic-sz3qv even the news about sd 900/895, Qualcomm will use better technology to manufacture sd 900/895 lmoa
you can wow you clients by calling 10nm as 7nm+ for example
Tsmc have over 200 technologies and over 10000 products??!! is it real? what is mean of the technology? design rule?
I think TSMC might fail not because of their technology or losing big customers. I think that the biggest problem is most people don't want to work at TSMC. Therefore, in the future, they might lack employees. TSMC's workers need to work 12h continuously four times in a week. They often need to start work from midnight. Even on the day-off or when they sleep, if a problem occurs, they need to go back to work.
I'm now studying at a famous university in Taiwan, most of my friends in the university said they don't want to work at TSMC.
Lol i work as engineer in one of the semiconductor fab in taiwan.. If i have ever a chance to enter tsmc.. I would not hesitate to enter..
I'm surprised Japan not a fab place. Only a matter of time before China takes up share.
It will take years. If you follow SMIC's recent news, you will know the gap isn't small and TSMC continues to advance in the technology. Just recently, they have started planning for 1 nm and 2 nm factories when 3nm is not even in production yet.
@@jethrolai everyone is working on 3nm and smaller nodes, not just tsmc
@@carholic-sz3qv not everyone has the ability to go under 3nm. Currently only Samsung and tsmc as far as I know. Intel with that 20 billion fund might eventually get there in 3-5 years.
@@jethrolai its actually ASML and Carl Zeiss that makes the 7nm, 5nm, 3nm..... machines. tsmc and Samsung buys those machines and include them to their production line with other machines.
@@carholic-sz3qv if you mean EUV, DUV, yes, ASML is the only company that makes EUV, the entire supply chain including Zeiss, Cymer, Synopsys. Cadence Design System, and so on. The entire supply chain consists of more than 5000 companies. No one company can do this alone.
Intel : respect, i need respect!
Make videos on Micron Technology
I like my chips with a little salt, and vinegar?
Tasty
Although tsmc has a huge advantage in that it does not compete with customers and has more foundry experiences, it is a big deal if its technological progress is delayed. samsung will mass produce 3nm for the first time in the world, even in gaa. experience first means more skillful and reliable. market for less than 10nm is growing rapidly. since samsung has almost half market share of below 10nm, the overall market share gap will be reduced. if tsmc fails to mass produce 3nm in the second half of next year, it will lose apple to samsung, which will be a huge disaster. tsmc must mass produce 3nm until next year. people ignore Intel's foundry but actually only intel has possibility to compete with tsmc and samsung. intel, as a general semiconductor company and has the biggest sales among semiconductor companies, has very good semiconductor synthesis capabilities, so if it invests a lot and receives support from the us government, it can fast follow tsmc and samsung. foundry will be tsmc vs samsung vs intel. umc, global foundry and smic dont have ability and money.
Wow
One will now be able to also use intel maybe 😆
more like TSMC is the leader of Taiwan. That company calls the shots.
what the hell is a dram?!
Dee-ram or s-ram don't know what is this
@@jjy-g4n that sounds like DRAM, but what's a dram?
DRAM a type of memory Dynamic Ramdom Access Memory
$1.5 billion... 😱😱😱
The miracle of the Han River...will dominate the chip world...
Probably not...at least not for a very long time.
YEAH SAMSUNG ALWAYS SHOWED MIRACLE.
I doubt it, Apple will never fully rely on Samsung for their high end chips, which means they'll never dominate. Although second place in a growing pie isn't all that bad. The question is who will be the 3rd largest contender that can ship similar products???
@@selohcin Long time? Few years I bet... especially since their government is supporting them as well. They're already talking about expanding to Texas with that $16 billion infras.
@@aoh4905 TSMC investing 100B in next 3 years and building fabs right next to Intel in AZ.
Should Apple build its own foundry(ies) on American soil? I personally think it is totally capable of doing so [in every way], but it seems like the CEO doesn't want to build one just yet.
I think you underestimate the necessary level of detailed process knowledge that is needed to set up a EUV chip fab. Look, even Intel is struggling and they have decades of experience and ample cash.
@@Martinit0 Probably hard at first but my point is that I don't want the U.S to rely too much on foreign supply chains. I'm leaning towards political issues that may occur. Who knows the future.
Others want them to build EV's too. If they could and make money, they would! not likely. Everyone is great at doing ONE thing. becoming a jack of all trades makes you mediocre at everything.
@@jjy-g4n tsmc is building fabs right next to intel in arizona. samsung is building a fab in texas.
Samsung, TSMC, Intel going to Dominate.
The sound-quality, presumably of the microphone used for recording this video is bad. Telephone-quality really.
Anddddd they lost all those clients
Apple M1 ARM chips are revolutionary in laptops-fast & 18-20 hours battery life. Their M1 based laptops are on back order & the growth will be amazing in next quarters earnings. The next low power/fast execution market is servers. Expect Apple & TSMC to smoke Intel & AMD in the future. I wouldn’t be surprised if Apple bought out Dell or HPE to incorporate their chips in servers. Also, GPU manufacturers like Nvidia are going to be hurting shortly since the GPU is in the M1 chip, thus making it super fast & energy efficient when compared to external GPUs.
Uhm nope apple M1 ARM chip isn't revolutionary. Its all about optimisation and apple only makes just one version of each procuts unlike Intel who makes chips for basically every companies and every applications. ARM chips by their nature are very powerful and efficient.
TSMC isn't going to smoke intel or AMD. TSMC is already making AMD chips and is going to make the next intel chips..... NVIDIA is also next tsmc customer with their tegra chips for example.
@@carholic-sz3qv AAPL bought out the euv space from TSMC to make the
Please buy a good mic, what you making is fine but audio quality sucks.
Apple has a huge account with TSMC. TSMC dont need Apple
😯
one europe foundry and SMIC will soon join on this stage
But who has shown better business practice in the past? Samsung or Apple? Answer: hands down Samsung 👏. Apple has cought my attention with its lawsuit against FB for selling out its users to the highest bidder. So I will be watching with interest as we move forward
No You obviously don’t even understand it was Facebook that filled the lawsuit not the other way around. Apple’s latest versions of IOS allow the hardware users to stop tracking on the device Facebook uses to target adds
@@johniii8147 If you are going to contest what I'm saying please use proper English and provide proof in the form of a link or I am just going to assume you are a troll. Thank you 🙂
@@ICEMAN-fp9zn lol I am the troll. Very correct English. You will just have to Google the stuff I’m not gonna do your homework work for you.
@@johniii8147 If you say so troll..
fabless semi players with "their" x nm process. JUST LOL
I'm waiting their nm measure to go smaller than plank length.
$ASX to the Moon📈🚀 long term Investor 💰
Please buy a better microphone. :)
Samsung is not as reliable as TSMC, ask Apple and Qualcomm!
Samsung sd888 and s8gen1 were trash
I believe Samsung will defeat TSMC soon. They have done so far. If US weighs on TSMC, it will regret soon.
@vox Lennart Loser's dream
Don't waste our time with BS generic graphics picked out of some crappy picture library. So lazy.
It's a matter of time that China will lead that industry for 2 reasons: it's located in Asia and the government has the money to make it happen.
You must have no idea how hard it is to build semiconductor chips... wow lol. China is over a decade behind.. even the US is falling behind. As long as technology improves the chips will have to improve.. even making graphics card takes decades of research because companies like NVIDIA has decades of experience.
Samsung sucks