Absolutely fab: As TSMC invests $100bn to address chip shortage, where does that leave the rest of the industry? Taiwan Semiconductor Manufacturing Co., also known as TSMC, plans to spend $100bn over the next three years in response to chip demand and has advised its customers to expect to pay more. Word of the firm's investment plan comes from Nikkei Asia, which claims to have seen a letter from TSMC CEO C.C. Wei outlining the …
Given that China isn't doing so well on the home-front with chip fabrication, I don't know why they haven't started the Hong Kong playbook with Taiwan yet. Maybe they're waiting for civil unrest, so they can quash the uprising, swap out leaders, install a new National Security Law which supersedes their Basic Law where conflicts occur, require allegiance from Legislative Council members and smell like roses for the other half of the citizens who weren't protesting.
TSMC only has a single fab on US soil, everything else is in Taiwan except for two in China.
SMIC is doing OK, but they reported they're starting to struggle with 10nm designs due to US sanctions. And what was supposed to be China's next flagship foundry, Wuhan Hongxin Semiconductor, collapsed before they could even start.
There's been enough saber rattling between the US (and some other countries around the world) and China over Taiwan through the years, rising in a nice crescendo in recent times to wonder what China's next or final step will be to cement their hold on Taiwan. Xi does not like democracies.
But, I've taken this convo off course, so apologies... I'll grab my coat.
The US military has run scenarios over and over of the Chinese attacking Taiwan and the US loses every time. The US military thinks that they will invade in the "near" future.
I'm sure the Chinese read the same articles I do, and I imagine they see the new administration as a window of opportunity.
Do you think the US will commit their forces to what they see as a doomed endeavor? I don't.
But China doesnt win anything other than more mouths to feed. Any signs of a build up and those with the knowledge will be welcomed abroad and China has a nation twice the size of the UK with a lot of either destroyed or just unworkable tech. They cant invade Taiwan fast enough to have anything left to work with.
Ummm.....you think the US would defend East Germany against Russian annexation?
Why on Earth would they do that? It’s not 1992 any more you know.....
In the past decade Germany and France have been nothing but a strategic PITA for everybody.
There’s absolutely nothing in it for anybody to “defend” Germany against Russia, when most East Germans vote for old-style communism over Anglo neoliberal capitalism 10:1 any time they are given the choice. Which they rarely are.
Taiwan which is officially called the "Republic of China" (ROC) should never be confused with China which is officially called the "People's Republic of China" (PRC).
The names always remind me of Life of Brian:
"People's Front of Judea" (PFJ)
"Judean People's Front" (JPF)
"Judean Popular People Front" (JPPF)
"Judean Popular Front" (JPF)
They'd need to bring the full force of their military to do it. Taiwan buys weapons from the US, so they'd put up a good fight, but unless the US wanted to become directly involved in such a war China would have its way eventually as Taiwan would be outgunned and outmanned.
There is, however, a 0% chance that they'd get TSMC's fabs - and the equipment inside - intact, so this would do fuck all for getting them leading edge semiconductor manufacturing capability. Either the equipment would be packed up and airlifted out before the fighting reached that area, or Taiwan's military would make sure the fabs along with the valuable equipment inside were destroyed before surrendering. And China knows that.
If it happened it could be over in a very short time - one way or the other.
I truly doubt that major fabs have a plan in place for self destruction. It wouldn't even work unless they were already drilling (practicing) for it, and I've never heard of that.
Another thing - what if the invasion failed?
"I truly doubt that major fabs have a plan in place for self destruction. It wouldn't even work unless they were already drilling (practicing) for it, and I've never heard of that."
I doubt they don't. It's a serious strategic issue.
You don't have to level the building. Just toss one grenade at each important machine and be done with it.
One could call an airstrike, too. Take about one hour to drop bombs on the building, and not much can be done about it. Not on such short notice.
Just open all the dump valves between the backing pumps and the turbo/vapor pumps.
Near instant write-off for most kit when a turbo pump goes BANG from raw air ingestion, or a vapor catches fire from the same.
Even if not a full write-off, the cleanup from the oil/particulate contamination makes it super expensive.
I can't find the photo album now that I'm looking for it, but there was someone who posted the aftermath of a turbo pump that got some "unexpected rough air"... Lots of broken things.
This is some of the most sensitively calibrated equipment in the world. Dropping a single bomb on the roof would probably damage every machine inside from vibration alone.
Heck, they might not even need to destroy anything. I doubt it would be possible for keep ASML's EUV scanners, for example, in an operational state for very long without parts and service from ASML which China could not get.
> But other countries and foreign competitors have their own ideas about where chips should be made. ®
I would add that other countries and foreign competitors may have own ideas about supply chain security and the strongly politically influenced US production with its export bans might not be the safest choice. For the internal US and satellites market, makes perfect sense.
strongly politically influenced US production with its export bans might not be the safest choice
The US only imposes export bans on the highest-tech bits. You can still get lesser tech from the US even if you're on the enemies list.
It doesn't matter if fabs are in the US, or not. Countries like Taiwan and South Korea are close allies with the US, and go along with their export bans, too. And if they didn't, they'd risk getting cut off from US exports, themselves, such as the machinery those fabs need.
To really be secure from US sanctions, you'd need to put together a fab in one of the countries perpetually on US sanction lists, already. Iran, North Korea, Sudan, Cuba ans Syria come to mind. It would be a massive feat to get a world-class fab built in one of those countries, but if you could do it, they would sell to anyone and everyone, no questions asked.
>It doesn't matter if fabs are in the US, or not. Countries like Taiwan and South Korea are close allies with the US, and go along with their export bans, too.
That's exactly the point, today they are allies but tomorrow they may decide their own interests are more important. The US-based production does not have such privilege. E.g. at a certain moment in future Samsung may decide that the US market is not too important for them and they are willing to risk losing it e.g. in favour of the Chinese market. They will not have this choice if their production is in the US rather than in S. Korea or Vietnam.
>To really be secure from US sanctions, you'd need to put together a fab in one of the countries perpetually on US sanction lists, already.
These countries cannot even sell crude oil under the US sanctions let alone high-tech products. It is rather a rich, developed and mostly self-sustained country like China that can play against the US sanctions and hope to win in the long run (which I think they will).
Actually no.
It is simply good business to diversify your markets as much as possible in the event that one or more shuts you out overnight.
For this reason you should never rely on one single market.
The watchword is diversification. Any decent investor will be happy to tell you why you should spread your risk.
Samsung may decide that the US market is not too important for them and they are willing to risk losing it e.g. in favour of the Chinese market.
Zero chance of that. South Korea would come up with some reason to step-in and take over the company.
They're in the same boat as Taiwan. Without US support, their neighbors are likely to invade.
a rich, developed and mostly self-sustained country like China that can play against the US sanctions
China isn't self-sustained at all. They've done well making money on manufacturing low-cost western tech, but all their attempts to promote domestic alternatives have failed to overcome market forces. How have CBHD and Loongson been coming along? Why will SMIC do any better?
On the one hand, sure, and I certainly wouldn't want to rely on the US as the sole source of chips. But more competition would be nice. Given a choice between having all chips made in Taiwan, and having half there and half in the US, I'd prefer the latter. Obviously it would be even nicer to have them made in many different places all over the world, but given the costs involved that's unlikely, and at least having two or three choices would at least be better than only one.
"The only supplier of EUV equipment is ASML - if the Chinese want to force the US to stop semiconductor sanctions then buying ASML would give them effective control of the world's sub-10nm production."
That's the theory, but practically impossible. ASML is Dutch and publicly traded. EU would very unlikely allow it to fall into hands of China. Or any other nation.
ASML's market cap is about $250 Billion, and sudden stock hogging would likely push the prices very high. Instead of that Trillion dollar deal you could very well start your own ASML competitor.
Sure, it would take something like 10 years (at least) but China is playing the long game anyhow. As long as the Chinese can churn out decent quality EUV machines, they can start undercutting ASML and drive it out of business that way. That's how they've operated so far and seems to work well for them.
I'm not trying to downplay the technology advantage ASML has - Intel with all their resources failed miserably without ASML to get below 14nm - but since ASML cannot fulfill the global capacity needs there is a profitable market waiting - for those who can afford playing this game.
I understood that a contributing factor in the World shortage of ICs was the drought in Taiwan which is impacting on TSMC production.
I understand that Arizona, USA, is largely desert so where is the water necessary for two fabs to come from ? Or is it another plan to grab a few billion in subsidies from Uncle Sam without actually doing anything productive to justify them ?
Chris Cosgrove
Froim what I recall, Las Vegas doesn't get power from Hoover dam ... it (and the Shasta dam in N California) were built mainly to power the pumping of water along the irrigation canals in California's central valley and to power Los Angeles and Southern California ... when the power distribution agreements was drawn up Las Vegas didn't really exisit - it grew due to the need to house all the construction workers nearby.
Also, for water, the anmount of water that each state that the Colorado River passes through can abstract is tightly regualted and I doubt there's any spare capacity .... especially as the US have had to backtrack from the Mulholland credo that "any drop of water reaching the ocean is a wasted drop of water" and allow some portion of the river to flow through Mexico to the pacific.
Like that is ever going to happen.
The Hoover Dam is the end of the chain. There is so much water being siphoned off the Colorado River it's pure insanity to build a semiconductor plant in one of the driest places on Earth.
For Heaven's sake, can't you go and build in Montana ? Or Washington State ? They've got water up there.
State subsidies won't help you maintain production when the draught comes along.
Compared to the cost of a fab, a seawater desalination plant is peanuts. TSMC could easily afford to build a desalination plant (and if necessary a power station to power it) in Taiwan for a small fraction of the cost of a single fab. Unlike Arizona, Taiwan is an island with plenty of access to seawater.
Given Arizona's water problem - choosing somewhere by the great lakes instead would seem to be a better choice (unless tax breaks etc decide the positioning).
TSMC has stated they are spending $20 billion on their next generation fab. I would imagine that is a significant hurdle for U.S. companies. TSMC gets all sorts of subsidies, which the US doesn't do. I think this will require some law changes as well, to allow US companies to cooperate without fear of all being sued for anti-competitive practices. I know Intel has said they are looking to build two fabs in Arizona; I don't believe it, I think they are just trying to exert economic pressure on TSMC and others. Intel has already said their lower end processors are being farmed out to TSMC. These fabs are just too expensive for individual US companies to undertake without subsidies or other financial support.
So, living in the US, I don't care one bit if the chips are made in Taiwan or US. But, the simple fact is there are ongoing shortages, and having production come online in Taiwan, US, and Europe, will help with these shortages more than just having a single fab open up in Taiwan.
Side note, I don't worry about China taking over Taiwan either; physically, they could, but China does try to avoid taking actions drastic enough they would affect their world trade, and wholesale taking over Taiwan would be one of these. Second reason, I have heard there is a large amount of highly profitable arbitrage (essentially, making cash by carrying goods and services back and forth between China and Taiwan... not illegal goods, just goods where the market price in China and Taiwan are drastically different). The politicians wetting their beaks in this would not want it disrupted, and probably do have enough power and clout to lean heavily in that direction.
Concentrating so much of so many critical cutting-edge fabs is insurance against invasion since the price the rest of the world would have to pay if all of those fabs got disrupted/destryed is too high. Therefore, cheap whenever compared with the cost of ensuring Taiwan's defense against mainlaind China.
Another consequence is that even if TSMC builds something elsewhere in the world, it will still keep the vast majority of the advanced fabs in Taiwanese land and never tease the option of its allies pondering if it is worth or not letting China retake Taiwan.
TSMC's fabs in the US/Europe/China will allways be smaller ones or less advanced.
The SK Hynix with the recently announced $100B expansions will follow the same line, ensuring the large and continuous military presence of the US nearby against NK and China.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
