TSMC Announces Intention to Build and Operate Advanced Semiconductor Fab in U.S.
tsmc.com
tsmc.com
For the U.S., this is a potential disaster. If China marches into Taiwan, the U.S. will be left with no advanced-node manufacturing and it will have lost the technology race. It would take at least four to six years to recover, and by that time China would own the market. This is very scary for the U.S., which needs to develop AI chips at the most advanced nodes because density equals more accelerators and other types of processing elements. Even that isn't enough, which is why companies developing AI chips are stitching together chips that are larger than a single reticle. (https://semiengineering.com/the-risk-of-two-supply-chains/)
It goes further than that, too. The DoD needs AI chips, too. In the past, the DoD relied on older nodes. That's changed because it needs transistor density for AI. More MACs equals more performance. On top of all of this, TSMC owns the bulk of the fabless market. So this will happen. The big question is who's funding this move.
2050 Indo-pacific naval war is my prediction.
Real war starts when one party is unhappy about the outcome and unilaterally decides to no longer cooperate.
It's already inadequate against today's (~mach 5) supersonic missiles. And china has literally tens of thousands with thousands of kilometers range stationed along their coast. A navy that pokes its head out is not going to last long.
China's missile superiority in the (western pacific) region has long been a thorn in the US side, but there's little that can be done given that their territory is there, and the US can at best station some in allied territories nearby or put it on ships, which are easy targets.
So you pretty much have to keep your ships hidden at all times, but being hidden they are going to be pretty much useless except as missile platforms.
I don't think the PRC is in a position to finish off the ROC, and I don't think it would be in their interests, either.
But equally I don't think that the US want to intervene directly should the PRC take over Taiwan. This would be madness. The problem is thus to imagine how the US would manage to politically 'sell' their inaction. But at the same time, would the PRC dare to take the bet that the US wouldn't intervene?
Taiwan is not much different from Crimea on many aspects, but it is different in one key aspect: It's been given much more international prominence, mostly because it is a pawn to annoy the PRC. But the downside is that it makes non-intervention (which is the reasonable course of action, like in Crimea) more difficult to sell.
Then, should the PRC invade (which is unlikely) while I'm sure that the Taiwanese would put up a good fight they would lose anyway. There would be plenty of grand speeches and maybe sanctions, but the reality is that everyone would just be looking for the right time to recognise the situation and move on (that's what happened with the existence of the PRC itself, and I'm sure is what will happen with Crimea).
But even if you were contemplating a direct war the issue would be that direct wars against Russia or China are not realistic because they are nuclear powers and, in any case, too big to invade. Any direct war would certainly require a major aggression from these countries against the US themselves, which they obviously won't do in the same way as the US won't directly attack them.
It's all boils down to relative strengths and outcome of a war. There are wars that are never worth starting.
The question is, do we want to end this idea of "taking back" things that were "ours" in the past? What even is "ours" and when do you get to call the cutoff time? What is it, 1800 AD, 1200 AD, or 2000 BC?
China itself was made up of smaller civilizations many ages ago... so if I'm a present day decendant of one of them, do I get to claim that part of China as my own?
Also, when did the CCP ever control Taiwan? Never. So how can they take it back?
Again, this has been played to annoy the PRC, standard divide-and-conquer tactic, but the reality remains that no country is going to go to war against China over this, or in general.
It's ridiculous. The whole point is we don't want to have a strength-based order in the world. If so, it will become winner take all and lots of people will die.
If China wants to "finish" a civil war and start the attack then fine, they have the blood of innocents on their hands.
https://en.wikipedia.org/wiki/DF-26 - range of 2500 miles
The Exocet was made way back in the early 70's yet carriers and surface fleets in general still aren't obsolete.
https://globalnews.ca/news/4810853/china-railgun-warship-wea...
https://www.nationaldefensemagazine.org/articles/2019/3/22/i...
https://news.usni.org/2019/03/05/report-u-s-carriers-need-ne...
https://www.businessinsider.com/the-us-apparently-gets-its-a...
Of course a lot of this is self-serving stuff designed to drum up more money to spend, but defence against missiles is fundamentally asymmetric. The defender need to get lucky every time, while the attacker only needs to be lucky once or twice. And MIRV is a big multiplier in favour of the attacker.
Carriers are great against countries or factions without anti-ship missile capability.
If you use a nuclear warhead, you don't have to hit the ship directly.
(This is why I am a fan of nuclear weapons, but it still negates the idea of a naval battle.)
They're converted ICBMs, and require some kind of terminal guidance. Something traveling Mach 10+ can't use radar or infrared guidance. How does it hit a maneuvering ship?
Current thinking is drones or some other local surveillance would be used, radio linked to the DF-21. I wonder how well that will work alongside US electronic warfare capabilities?
Interesting stuff...
That is not clear.. at all. Europe, Canada, Australia, New Zealand -- the non-US western allies -- have all known exactly what China's end goal is and have down jack diddly squat to put pressure on them.
The only thing preventing China from taking out Taiwan is the United States. Europe and the rest of the world is too cowardly to even do anything, which is really unfortunate.
The problem is that Europe depends on (cheap) Russian gas for heating.
https://en.wikipedia.org/wiki/Budapest_Memorandum_on_Securit...
If we EU member states didn't get involved when we saw a fight right in front of us, the threat of us getting involved in a fight thousands of miles away isn't very credible as a deterrent.
More broadly, when I look at Taiwan I don't see the EU or US recognising Taiwan as a nation - let alone any tripwire forces or foreign airbases or nuclear sharing or NATO membership or joint training exercises which signal the fully fledged protection of a great power.
So it's no further from US bases than EU bases.
> Both Ukraine and Russia share borders with EU member states.
And Russia is a two mile walkable ice bridge from the US in winter.
> If we EU member states didn't get involved when we saw a fight right in front of us, the threat of us getting involved in a fight thousands of miles away isn't very credible as a deterrent.
Distance isn't the issue. EU countries/NATO will bomb any puny country armed with obsolete weapons, wherever it is.
What the US and EU will never do is get into a fight with a nuclear armed superpower.
The EU couldn't even put pressure on Erdogan let alone on Xi.
China has been buying influence here for years, especially through the poorer countries that were alienated by the strict austerity measures imposed by the bigger powers.
The nature of the EU politics where a single country has the power to veto the rest is a massive loophole that China is happily exploiting.
The fact that every EU country has different interests and agendas is a huge problem that results to the union never aligning on anything critical.
Just look how the refugee crisis was handled. Affected countries were crying for help and the others were like "sucks for you bro".
and German universities: https://www.timeshighereducation.com/news/china-influence-sc...
US has treaties with South Korea and Japan that make it it clear that the US will defend them if they are under attack. The Taiwan treaty is worded in a way that US may provide assistance or limited support in a war.
A quadrilateral nuclear disarmament treaty with Ukraine the US has signed has even more explicit wording, yet...
It is only since Xi became president, that China feels confident to have a more assertive foreign policy.
How so? There hasn't been a sharp rise in military spending, they haven't invaded anywhere, and they've largely maintained their same course on most of the disputed territories.
edit may have misread your post, unsure if you're comparing or detailing two steps. China has invaded uninhabited islands and built buildings to stake claims for some time. Building the uninhabited island isn't some drastic change.
In comparison with China of the 90s and 00s, China of the 10s has definitely been more aggressive. In 1990s when the Philippines grounded an old navy ship on Scarborough Shoal in the south China sea to reinforce their claims, there was barely a peep from China. By the 2010s though, Chinese coast guard ships were harassing Filipino resupply ships to their outposts and China was making it's own island bases.
People think China's being more aggressive since their point of reference is the 90s and 00s, when China was frankly a pushover on the world stage. Chinese freighter gets accused of smuggling chemical weapons to Iran in 1993 and gets held hostage in the Indian ocean before nothing was found? Not even an apology from the US. Chinese embassy gets bombed by NATO forces in 1999? Nothing but words. A Chinese fighter pilot dies in a mid air collision with an American spy plane off of Hainan in 2001? No reparations for the loss of life and expensive military hardware. Anyone would think China was being more aggressive today if their baseline was that.
They engaged in combat with both Vietnam and the Philippines over islands in the 80-90's, and effectively took over Mischief Reef in 94 by building bases on an island. Not to mention what was a huge victory for them in the return of Hong Kong. That was their primary aim for much of those decades.
There probably is some growth in aggression, but nothing like "Xi's waiting for an election before invasion."
There are some that are pro-unification in Taiwan, but it's a minority. Most people are pro status quo, and the number that are for declaring independence is growing. The ruling party is very pro-independence (ideologically speaking, in practice they have to move slowly and maintain status quo to some degree), meaning releasing their claim over mainland China, changing name from "Republic of China" or "China" to Taiwan, seeking UN membership etc.
The other big difference is that Taiwan is quite capable of defending itself to some extent. China will win eventually, but not without devastating the island. This will affect everyone economically. China and the world. If TSMCs factories in Taiwan are destroyed, it'd have an enormous impact on world economy. The production of high-tech electronics would be stunted for years. Russia annexing crimea had absolutely no practical effect on the world.
The fact is that everyone, China included, is relatively happy with the status quo. The CCP has to make a scene now and then to appease the fragile ego of a large portion of the mainland chinese. But I really don't think they actually care that much about invading Taiwan.
Man, are saying this in your right mind? Where do you read this stuff?
People there abandoned their property, and ran! This is what they did.
Nobody openly welcomed it in their sane minds, besides few thousands elderly ex-party members, and alt-right weirdos.
The fact that I hear sentiments like this in the West is the direct proof that Russian psyops truly work.
Not sure how accurate it is though.
Ask any Russian abroad if he wants to come back to his "free" fatherland
It’s a bit more complicated than that. The historical and relatively recent mass starvation, persecution, mass murder and forcing into gulags is likely relevant. https://en.m.wikipedia.org/wiki/Crimean_Tatars
While I will not deny that a certain amount of bad blood still exists due to the deportation, I think most Tatars more concerned about their current livelihood (though I have talked only with 2 tatars, so my sample size is quite small).
You have 2 more personal sources than I do.
https://en.m.wikipedia.org/wiki/Deportation_of_the_Crimean_T...
Seemed like successful psyops by Putin to me. I guess if Ukraine had blocked VK sooner, not many people had been particularly much pro Putin.
Of course now Putin continues with psyops via Facebook but that's harder for him than via VK.
For example there were opinion polls done by western polling orgs in Crimea where the vast majority supported unification with Russia. It's not hard to understand. Salaries in Russia are ~10x what they are in Ukraine for the same job. Economically they're not comparable, and that's before you get to the whole conflict in Ukraine starting because the government was overthrown by rebels supporting a new government that amongst its first acts, banned the Russian language.
Remember that Ukraine's response to the peaceful annexation was to cut off Crimea's entire electricity supply. That's hardly the work of a friendly government trying to help its loved citizens. Russia had to send in generator ships, if I recall correctly.
Because those who have not welcomed Russians have disappeared.
Random link from google search: https://www.aljazeera.com/indepth/features/2014/12/disappear...
It's very easy to attribute such disappearances to the evil KGB to promote political agenda, while there are far more probable explanations. It brings to mind the "Heavenly Hundred", a significant number of those people haven't even died on Maidan, but to promote the political agenda they have been promoted to national heroes, even though their "heroism" is dubious at best.
I've personally met with people in Crimea who kept Ukranian citizenship and I haven't heard about any significant oppression from them. If you don't trust me, just read this article: https://tsn.ua/ukrayina/zhiteli-aneksovanogo-sevastopolya-pr...
Have the evil KGB killed and tortured those people? No, they have been simply ignored.
Now compare it with https://www.youtube.com/watch?v=atm0W5wA2y4
https://www.youtube.com/watch?v=WMKraUk0BSQ
And don't tell me that's organic that suddenly bunch of big guy "civilians" appear and start using force. We have heard about green little men. So there are gray little civilians also.
And Crimea does not need to loose 50% population that does not agree. It's enough to "disappear" few more prominent activists to scare the rest.
https://www.reuters.com/article/us-falklands-referendum/falk...
Not being tied to the economic basketcase of Argentina is a good thing, it seems.
It's kind of like how Spain will occasionally bluster about Gibraltar, despite the overwhelming majority of Gibraltar supporting British rule.
Pretty much the same situation, completely different reaction from the same people.
I believe both of these to be correct (Falklands pro-UK, Crimea pro-Russian), fwiw.
Well, I guess that’s what happens when you replace the population of Gibraltar by British settlers. They will probably be pretty happy to be under British rule. Are you going to apply this logic to the areas colonized by Israel too?
I give it between 1-5 years before they move in some way on Taiwan.
One thing I would not bet on is a direct military incursion - that would cause too many foreign headaches. Something messy that's hard to pull apart is better. The current events with social unrest in HK are a good example of creating ambiguity.
The one interesting thing to me is that China normally has a very long time horizon, but now they don't. Trump has effectively made sure that Cold War 2.0 has started, and China is the enemy. The problem for the Chinese in this situation is that a substantial amount of manufacturing will move out of China, weakening their economy. And a weak economy is usually not a good thing for leadership.
Right now, 1 country 2 systems means that PRC can't just dismantle HK's education system and institute their own until 2047. But the comments HK's education system indicate that the PRC thinks if they could just put HK's children through a state-approved education system, in 1-2 generations all the unrest in HK would be over. This is a worldview that people are ultimately malleable, or at least their children are, and that with the right childhood education you can get people to believe anything. Brainwashing, after all, is originally a word from Chinese that became an English loanword. In this worldview, everyone who's gone through schooling is brainwashed in some way to support the state that created the education system. And this worldview applies to any state-sponsored education system in the world. Take a Chinese immigrant family in America, put their 2nd gen kids through the American education system, by the 3rd gen you have Americans that can't imagine themselves as anything other than Americans.
This worldview that people can be molded is probably why China is in no rush to exercise full control over HK and Taiwan. After all, if people's national identities are so easily changed, any discontent in HK and Taiwan today can be brainwashed away in few generations. It's far more important to focus on the physical and economic levers that would allow China to exercise long-term control over those regions than social levers that determine whether people in those regions actually like the prospect. If they don't like it, their children will.
You mean other than the crippling sanctions that were actually working until Trump took office? I think you fail to realize how close Putin was to the precipice on that one - when the oligarchs lose faith, it's tough to keep your hold on power.
Moreover, the US is closer to Asia than Ukraine. I mean that honestly. The West Coast -- and the greater US in general -- has a lot of Taiwanese and Asians who would be upset if China took over Taiwan. Much more Asians than Russians or Ukrainians. It doesn't hit as close to home frankly.
And, as others have pointed out, they weren't trying to conquer all of Ukraine, just the parts with a "warm" water port, which also happened to have a lot of ethnic Russians. Limited scope, for a tactical gain (a seaport). The CCP, however, is trying to take all of Taiwan, and enforce their will on people who absolutely do not want it.
With a Nuclear India next door (and Nuclear Pakistan next to them), plus Japan, Vietnam, and a host of other countries that have a reason to be weary of China, it's a risky gambit. It is a good way to kickoff WWIII, for an island that is already, effectively, in the Chinese cultural and economic sphere.
The only thing preventing China from taking out Taiwan is their mutual economic annihilation in case hostilities intensify.
Taiwan's relationship with the US has been at best ambivalent. Do not forget US was close to waving a hand on the island just a decade ago.
https://www.armyrecognition.com/april_2020_news_defense_glob...
It seems like the US has been more willing to sell Taiwan better military equipment recently?
As a European, I have to agree. Here in Germany, I am sure the government would do nothing (other than symbolic protest), because it would be bad for German companies (short term).
The hostile rhetoric towards Taiwan is a CCP relic, it is Mao's legacy to close off civil war, not anything else.
It is like Abortion to GOP, which is all about identity, other than actual purpose of pursuing such ideal.
If occupying Taiwan is worthy of losing global trade and business and tens of millions of Chinese' livelihood, CCP would already act on it. The only reason it hasn't done anything is because it doesn't want/care to.
It is probably true that both China/Taiwan to stay in the status quo, with ever so often mainly symbolic bickering.
https://www.nytimes.com/2019/07/09/world/asia/taiwan-arms-sa...
https://en.wikipedia.org/wiki/List_of_US_arms_sales_to_Taiwa...
I don't know, I guess the entire field of diplomacy and international relations can be characterized as "symbolic bickering" - as opposed to "non-symbolic bickering" - which would involve dropping bombs over the other party's territory.
Under what timeline a regime won't need to arm itself? Even China and Russia are armed near the border, or China/North Korea.
Once China has "acquired" the technology, it doesn't need an export market anymore for economies of scale. China's internal market is more than large enough for that.
Just a detail to keep in mind.
However, China is not quite there yet when it comes to technology (hence the many Chinese students in Western universities), but they are well on the way.
1) Does the US have a base-of-operations off of China's coast? If it independent? Is it Chinese?
2) who gets Taiwan's manufacturing and technology? Does it supply the US military or the Chinese military? War is mostly about supply chains and manufacturing capacity. It's how the US won WWII.
These are big questions. They're not necessarily urgent, but they are big.
Can you explain this? If China takes over Taiwan, couldn't the US just take over whatever TSMC plants exist in America?
Wouldn't a TSMC plant in US be good? Since US would have the factory on its own soil? What am I missing?
If TSMC was concerned about the US taking over their factory, they could design lots of elements of the process to require a connection to HQ in Taiwan, and to stop working when that connection is broken.
When every piece of equipment has no working firmware, the factory is pretty much useless, and would probably take years to get running again.
Since 1949, you mean?
Indeed, hearing this was a shock to me.
$12B would be a double digit of their capex plan for the decade.
Them committing even to basic operation in the Arizona plant would mean to take double digit percentage off their existing 5nm, 3nm plant plans, or having this money come from the outside.
Also take this, their CapEx plan for the decade is entirety committed to the sprint towards 2nm. And even with all those money raised, and after going to bond market, they were already signalling that they have to raise even more money, or the 2nm may not happen.
Meanwhile, TSMC's 7nm process has been shipping since Q3 2018, and their EUV-based N7+ process is now in mass production. There's no question that TSMC is currently ahead of Intel both nominally and in all the important technical metrics. It's still entirely possible that Intel could re-take the lead if/when they get something post-10nm out the door, but that isn't going to happen this year. Intel's own CFO says it might take two more generations for them to regain the lead: https://www.tomshardware.com/news/intel-process-tech-lag-com...
Depends on which Intel 10nm you're talking about, and which TSMC 7nm. Comparing Intel's failed initial 10nm from Cannonlake against something that is in mass production on the TSMC side isn't exactly fair. With respect to the 10nm Intel product that you can actually buy, TechInsights has this to say: "Produced in Intel's new 10 nm 2nd Gen process that should offer comparable performance to TSMC’s 7 nm process"— and that's coming from the pros with electron microscopes who teardown these chips to reverse engineer process tech. Numbers reported elsewhere (eg. WikiChip) indicate that TSMC's 7nm allows for about a 15% smaller SRAM cell, and AMD's Renoir notebook processors seem to have about the same overall transistor density advantage over Intel's Ice Lake notebook processors.
And as far as I can tell, AMD's not even using the EUV 7nm process from TSMC yet.
Intel's current form of 10nm is roughly equivalent to TSMC's 7nm. What that basically means is TSMC is about a generation ahead.
> Meanwhile, TSMC's 7nm process has been shipping since Q3 2018, and their EUV-based N7+ process is now in mass production
So, to make this an apples-to-apples comparison: when did TSMC stop shipping new products on prior nodes? ;)
You can still go out and buy a new-production Ryzen 1600 AF chip right now, you know. AMD did not even finish releasing their 7nm lineup until basically a few weeks ago, Renoir products just started shipping last month.
Ice Lake actually beat the comparable AMD/TSMC chips to market by like, 8 months or something. They've been around in volume since like, november. I went to Costco before last thanksgiving and almost half the laptops had Ice Lake.
AMD can ship desktop processors sooner because they've architecturally managed to reduce the chiplet to smaller than the size of a phone SOC. Comparing apples to apples, or more specifically monolithic mobile processor vs monolithic mobile processor, TSMC isn't that far ahead.
Or for another data point, TSMC's big-chip 7nm yields are so bad that NVIDIA is disabling 15% of the processor on the Ampere GA100 even for the datacenter market that will be paying $25k a chip or something stupid.
TSMC is ahead, sure, but not as much as people want them to be. The ability to make phone SOCs and 70mm2 chiplets successfully doesn't mean they've got 7nm completely nailed.
The real disaster is the loss of untold lives defending an independent nation and full-fledged democracy.
Strong parallels to Europe in the 1930s. Looking back, no one cares much about the technology giants of that era being shut down, but rather the lives that were lost in the invasions, and the mass murder of civilians that followed.
What does the acronym MAC stand for please?
Sometimes it's also called FMA, for Fused Multiply and Add.
[0] https://en.wikipedia.org/wiki/Multiply%E2%80%93accumulate_op...
China marching into Taiwan could trigger a global war. They'd have to get Russia on-board. Hong Kong is reasonable as a first target. This would also frighten its SEA neighboring countries. Don't forget that China has lots of enemies (ie: Japan/South Korea).
Only 20,000 wafers per month is between mini- and megafab for TSMC. I suspect the production goes to US defense applications. Pompeo tweeting about it seems to confirm it.
https://twitter.com/SecPompeo/status/1261143980634509318?
>Secretary Pompeo >@SecPompeo
>The U.S. welcomes TSMC’s intention to invest $12B in the most advanced 5-nanometer semiconductor fabrication foundry in the world. This deal bolsters U.S. national security at a time when China is trying to dominate cutting-edge tech and control critical industries.
20k wafers (perhaps 450mm ones?) with unknown yield (likely 80ish) is quite low. For reference AMD 5700 has die area of ~250mm.
If the fab is located inside the US, AMD and NVIDIA can bid for many US government contracts that would go for Intel. After GlobalFoundries quit the race, US defense contractors have only Intel as supplier for many components.
If the consumer product is a toy then sure. If it’s a medical device then it could make sense. It’s all about margin.
> After GlobalFoundries quit the race, US defense contractors have only Intel as supplier for many components.
Maybe. If you want millions of chips a quarter then sure. But if you want 1000s a quarter then there are many options, depending on the node I guess. I’m not going to Google a complete list but one obvious example is Texas Instruments.
Small run 22nm is certainly available. These are fabricated in the US, albeit at an Intel site. https://www.mosis.com/pages/intel_sponsored_fab
What process node Google uses for TPU, one possibility for an AI chip, is not in the public domain since v1 but it’s very unlikely this chip is on the bleeding edge. Cost/benefit is likely not there for this use case. But only Google knows for sure since they run the cloud and get to define “benefit”.
https://www.nextplatform.com/2018/05/10/tearing-apart-google...
Maybe before Taiwan was sufficiently richer per capita that taking it over had something to do with its wealth, but now it's a almost exclusively a matter of pride and nationalism. If they get control of a deindustrialized husk, they can say "Look at all the US boot-lickers we chased out".
It doesn't. Just because he referenced "national security" does not mean this fab will be dedicated to producing chips for DOD contractors.
The DOD has been making a shift to COTS processing hardware over the past couple decades. Part of the national security implications of this fab is preventing supply chain disruption in the event something happens with China. This serves to both guarantee the ability to supply COTS hardware to the DOD and to reduce the impact on the overall economy, making it more difficult for China to hold the US over a barrel by threatening its technology industry.
Most of its chips are produced in Taiwan and over 75% are sold to customers in either the USA or (mainland) China, so TSMC understandably tries to remain neutral in trade disputes between the two. Tim Culpan at Bloomberg recently predicted that this neutrality may not hold for long, and TSMC will be forced to move some production to the USA. [1]
Especially if Intel continues to decline, increased investment by TSMC could be good news for American chip fabs and the communities that depend on them. On the other hand, the factory isn't built yet.
[1] https://www.bloomberg.com/opinion/articles/2020-05-11/tsmc-m...
About Intel... I heard multiple times, from different people that Intel's CEO is frequenting Hsinchu for some, yet unknown, business.
It's easy to guess though, there can only be one company there of interest to him.
If we can't move the supply chain closer we should at least decouple software from particular hardware manufacturers.
How and why should software be decoupled from hardware when there are so few hardware suppliers?
Some are ARM, some MIPS, some SH, some even are PPC.
I guess the availability of Android is making a lot of "dumb appliances" run on ARM even if they don't need a screen.
But for basic dumb appliances, PIC, MSP, Atmel processors (either the old 8051 or newer archs)
MIPS and PPC are still being used only because of cost ( vendors selling other chips ). The replacement cycles is very very slow, but the trend is pretty clear.
I'm less familiar with other vendors but I briefly worked with a Qualcomm-based device a few years ago and that was also ARM.
There might be some dedicated cores in those SoCs or devices for e.g. accelerators or Wi-Fi handling. I've seen ARC there but I have the feeling these are also moving to ARM.
That is why a third semiconductor company (besides Intel and AMD) creating x86 chips has come into existence: Zhàoxīn (兆芯)
https://en.wikipedia.org/w/index.php?title=Zhaoxin&oldid=949...
Personally I have plenty of arm machines. Part of the reason I switched from fvwm to cwm was because I couldn’t build fvwm on one machine and move it to the other.
In college they asked for x86 windows binaries for lots of the assignments and I always cross compiled from an arm machine (because that’s where I had the toolchain set up and didn’t feel like duplicating it on my x86 laptop.) If you feel stuck on x86 use more FOSS.
>[Taiwan] considers "one China" to mean the Republic of China (ROC), founded in 1912 and with de jure sovereignty over all of China. The modern-day ROC, however, has jurisdiction only over Taiwan, Penghu, Kinmen and Matsu. Taiwan is part of China, and the Chinese mainland is part of China as well.
Taiwan is de facto independent, not formally independent. Formally there is still unsettled civil war going on where each side claims the whole China.
Taiwan satisfies three of the four requirements of statehood at international law as stated in the Montevideo Convention.
The current government is very pro (formal) independence. But it's a bit hard to declare China and Taiwan to be seperate when China is threatening with invasion.
The goal of ambiguity is prevent unilateral declaration of independence by Taiwan and an invasion of Taiwan at the same time.
[1] Taiwan’s brain drain: semiconductor engineers head to China, https://www.ft.com/content/6eab0c1c-167f-11ea-9ee4-11f260415...
Edit: Original article was Taiwan loses 3,000 chip engineers to 'Made in China 2025' (https://asia.nikkei.com/Business/China-tech/Taiwan-loses-3-0...)
The geopolitical aspects are, to me at least, a bit over blown. Yes there has been growing concern about the integrity of the supply chain for critical resources, but that is the whole globalization problem in a nutshell. I find the calls to re-vertically-integrate the US to not be very persuasive in terms of practicality. Maybe if Canada and Mexico join the US in an economic union similar to the EU? Chips are just one part of a much larger puzzle and fabs are just one part of a supply chain that includes stops in Malaysia, Indonesia, and Vietnam for packaging and testing.
I do think it has been made abundantly clear how interdependent the world is these days. It is possible that the new clarity has caused some powerful forces to reconsider their world view. But has it changed it enough to create real and meaningful policy change? (For context, I see globalization as a means of maximizing profits, not about politics)
To my way of thinking, real change would be the ability to economically stand up a 100mm 5nm fab line profitably. That would allow the production of "jelly bean" chips (general purpose < 100M gates). That would be a pretty seismic change in my opinion.
Globalization is just a euphemism for hegemony. First it was the British Empire, then the American Empire.
Then again, history doesn't repeat, it rhymes.
Maybe the long term US-Taiwan ties of factories in the US will make this worth it strategically.
I assume that all R&D will remain in Taiwan. I think this might still be a good move in the long run since it'll give more politicians stake in what happens to Taiwan (Arizona senators, etc).
https://www.globenewswire.com/news-release/2020/05/14/203353...
I find that a little confusing just reading the comments.
Nice that Intel will have more local competition.
Specifically TSMC has breakthrough CoWoS interposers, can’t meet demand, and the military is bottlenecked without a chiplet path made in the USA.
So it does appear like this is intended for a specific chip for a specific purpose.
How will it work out with their investors, only god knows.
And this has ultimately to pass by Taiwanese government, if it didn't already.
There is precedent for fabs though, so its not totally out there.
This seems like a move designed to mess with Intel's experienced employees and to try and combat potential tarrifs
1. By the time this fab is in major construction a lot of the lessons learned for 5nm will be known and can more readily be implemented in the USA without splitting precious engineering resources to other countries.
2. 3nm will be the most expensive parts and sell at a premium, which will limit demand.
3. This carries some risk, so giving up the bleeding edge tech might not be the best idea.
"tsmc could equip the Arizona foundry with American gear from its existing factories, freeing space in its Taiwanese operations for brand new non-American kit that can freely serve Chinese customers"
I don't care how it's accomplished, either carrots (tax incentives) or sticks (tariffs). Do what's necessary to bring the production of these vital components here.
Not responding is a competitive disadvantage, so it has to be either subsidies or tariffs and sanctions.
There's zero threat currently to that status and only China is capable of even plausibly posing a challenge in the next ~30 years. The Euro, Yen and Bitcoin pose no realistic challenge to USD dominance. It leaves only the Yuan, which still has no global position at all and isn't regarded as a currency you can trust. China will obviously do anything they can to alter that equation.
We're going into, at a minimum, a cold war with China that will last decades. There is no scenario where that is avoided. The US should become wildly aggressive about bringing manufacturing back, or otherwise pushing it out of China (including convincing allies to do the same), pulling every lever necessary.
Right now the world is peaceful and I stand in North America with an iPhone 11 Pro in my pocket and an advanced Nvidia GPU in my workstation. I’m not sure I see a problem with that state of affairs.
Reducing interdependence reduces the downside to stupid, selfish nationalistic moves. I don’t want that at all.
Also what you see as peaceful, others see as a stalemate unfortunately. Hopefully it lasts.
If a war breaks out now, they are in a better position regarding manufactured goods. Rare earth metals too.
https://www.google.ca/amp/s/www.nonpareilonline.com/news/spe...
edit: also, a minor quibble: I wish we'd get away from calling it "poaching", as if there is anything illegal or immoral about employers competing for employees.
Keep in mind, we're talking about PRODUCTION, not R&D. R&D happens elsewhere, this is just where the big plants go.
During the 80s & 90s, one of my besties worked on building fabs in and around PHX. Really exotic stuff that I didn't really understand. He'd joke that he had a PhD in welding.
He claimed at the time that fabs were migrating from California to PHX and ABX to avoid environmental regulation (enforcement) and cheaper water. While he was concerned about building future superfund sites, he also knew he'd never make that kind of money doing any other kind of work.
If you don't mind, I'd really appreciate your take. I still have family in Phoenix.
I've always been concerned about toxins (air and water). And I'm also grumpy about megacorps guzzling all the water at discount prices.
So while I'd love for TSMC to build their fabs in the USA, I'm worried about the downsides too.
Are my concerns justified?
Before water enters a fab it goes through a water purification plant to eliminate any impurities. As the water leaves the fab it goes through another process to decontaminate and recycle it. The process of purifying water is expensive and consumes energy so they try and recycle as much as possible. The number 96% sticks out in my mind but that is an old number.
Simply stated, I do not worry about local pollution from semiconductor manufacturing in the United States due to all the regulations. I would live next door to the Chandler plant happily. Pull open Google Maps Satellite View, type in "Intel Chandler Soccer Field" Look at all the green area around the plant, that's where the 4% that isn't recycled goes. You will also see a beautiful golf course and nice houses just east of the plant.
Side note: as I understand it, plants moved to AZ/NM because they wanted to eliminate process variance. Weather and water are a major source of variance. So they want places that reduce variance while simultaneously being favorable business environments.
Reference: https://ktar.com/story/76977/intel-plant-replenishes-water-s...
I could only assume they have much better monitoring of these situations nowadays.
System reliability is a big question.
> wasn't because they dumped toxic chemicals or water directly in the ground, but that [...] mechanism failed
It's not an isolated incident, leaks is actually the most common cause. Many fabs became superfunds not because the owners are evil, but because many safeguards, industrial or regulatory, were simply not in place at that time.
Let me repost an older comment.
I'd say early inexperience on semiconductor manufacturing was another major factor responsible for the pollution, many safeguards, industrial or regulatory, were simply not in place at that time. For example, many plants had storage tanks as parts of the production process, a common practice in the industry - most ended up having a massive leak, releasing various harmful chemical pollutants, including highly toxic TCE. And it seems the companies weren't doing it deliberately to save costs, many pollution events were "just" industrial accidents, and few foresaw the consequences of leaking tanks, not to say that some leaks were only discovered years later.
Now the companies are doing the cleanup under the supervision of the EPA, I think it's fortunate in a sense that these companies are the world's most powerful ones and they have the ability to pay, not some non-existent companies that already declared bankruptcy. And unfortunately, the price we are paying now for being ignorant in the early days is that the pollution still requires many years, or even many decades, of work to achieve even a basic, minimum level of cleanup. Some cleanup efforts have been active since the late 80s, and was still being continued working on in the 2010s.
Some examples:
* COMMODORE SEMICONDUCTOR GROUP
(Yes, the plant that produced MOS 6502 that we knew and love, and it was a pioneer for making the first widely-used microprocessor for personal computers, and also a pioneer for having a massive leak of TCE storage tank...)
> Background: Waste solvents were stored in an underground concrete storage tank on site until 1975, when it was taken out of service. An unlined steel tank was installed next to the concrete one in 1975. Inspections conducted by the Pennsylvania Department of Environmental Protection (PADEP) indicated that both tanks leaked.
> Cleanup: In 1981, Commodore excavated soils and pumped water from a contaminated well, then sprayed it onto surrounding fields. The volatile organic compounds (VOCs) dissipated into the air. Since 1984, air strippers have been in use to remove solvents from the groundwater.[...] Construction of the groundwater extraction and treatment system began in the Fall of 1999. In February 2000, pipelines and underground wiring were installed, pumps were installed at each of the extraction wells, and the groundwater treatment building was constructed. The treatment process equipment was installed in May 2000. Preliminary start-up and testing of the system began in August 2000.
* ADVANCED MICRO DEVICES, INC.
> Background: Two below-ground acid neutralization system (ANS) tank vaults were located at the northern and southern ends of the 901 and 902 Thompson Place buildings, respectively. Leaks from these ANS tanks appears to be the primary on-site source of VOCs to groundwater in this area.
> Cleanup: After the two acid neutralization tanks were removed, AMD excavated and disposed of soil impacted with VOCs, and installed a groundwater extraction and treatment system. Extracted groundwater was treated with air stripping technology followed by carbon adsorption. [...] The groundwater extraction and treatment system began operating in 1983 and continued through 2002, when it was discontinued with State approval to allow for an in-situ bioremediation (ISB) pilot test. The full-scale ISB system, which was initially pilot-tested from 2002 through 2004, and then expanded in 2005, includes groundwater treatment by carbon filtration and injection of an organic carbon source to stimulate the growth of naturally-occurring microbes that break down target VOCs into environmentally-benign end products.
* INTEL CORP. (MOUNTAIN VIEW PLANT)
> Background: The site is one of three Superfund or National Priorities List (NPL) sites that are being cleaned up simultaneously. The other two Superfund sites are the Fairchild Semiconductor Corp. (Mountain View Plant) site and the Raytheon site. The three sites are located in the Middlefield-Ellis-Whisman (MEW) Study Area. Site investigations at several of these facilities during 1981 and 1982 revealed significant soil and groundwater contamination by toxic chemicals, primarily volatile organic compounds (VOCs).
> Cleanup: Under EPA’s direction and oversight, Intel has implemented the soil and groundwater cleanup program at the former Intel facility. The soil cleanup has been completed at the site and all the former MEW facilities.
* FAIRCHILD SEMICONDUCTOR CORP. (MOUNTAIN VIEW PLANT)
> Background: The 56-acre former Fairchild Semiconductor Corp. (Mountain View) site is located in Mountain View, California. A facility on site manufactured semiconductors. The site is one of three Superfund National Priorities List (NPL) sites that are being cleaned up simultaneously.
> Cleanup Activities: Under EPA’s direction and oversight, [...] Fairchild Semiconductor Corp. [...] implemented soil and groundwater cleanup programs that have included soil excavation and treatment, installation of four slurry walls, soil vapor extraction and treatment systems, and groundwater extraction and treatment systems. The soil cleanup by soil vapor extraction and excavation and aeration has been completed at all the former MEW facilities, including the former Fairchild facilities. Groundwater cleanup will continue to operate for many decades in order to meet the trichloroethene (TCE) groundwater cleanup standard of 5 parts per billion. The MEW site groundwater remedy has removed over 76,000 pounds of contaminants, and has reduced contaminant concentrations throughout the multiple aquifer zones.
> HEWLETT-PACKARD (620-640 PAGE MILL ROAD)
> Cleanup: Extensive investigation and cleanup at the HP Site has been conducted since discovery of a release from a 1,000-gallon underground waste solvent tank in 1981. Interim remedial measures included excavation and off-site disposal between 1987 and 1992 of approximately 10,700 cubic yards of soil, construction and operation (beginning in 1994) of a soil vapor extraction and treatment (SVET) system; and groundwater extraction and treatment beginning in 1987 and continuing to the present day.
> NATIONAL SEMICONDUCTOR CORP.
> Background: The National Semiconductor Corporation (National Semiconductor) previously manufactured electronic equipment at this 50-acre site. Underground storage tanks, sumps, and pipes are the suspected sources for contaminated groundwater and soil in Sunnyvale underneath the site.
> Cleanup: Beginning in 1982, contaminated soils, leaking tanks and equipment were removed. Subsequently, National Semiconductor initiated a program of pumping and treating the groundwater to contain the contaminated plume while further site studies were underway. In 1989, National Semiconductor began investigating the type and extent of contamination at the site. The investigation was completed in 1991, and the EPA selected the final remedy for the site. Construction of the remedy began in 1991. The final cleanup remedy, which includes soil vapor extraction and operation of the groundwater extraction and treatment system, is ongoing.
If you are asking about Intel ones, and ones owned by the defence hardware industry, it is the same reason why China is building their fabs as far inland as Xian, and other obscure and scattered across places: they are harder to bomb there.
Anecdote, since I've moved to AZ myself 5 years ago I've noticed massive growth in tech businesses in Phoenix and the surrounding areas. Cost of living for people here is also cheaper compared to to other cities I've lived in, and many people I've met who moved from California have told me how its cheaper here. Commuting is not too bad depending on where you live, and there are options which promote commuting or carpooling such as HOV lanes, Bike Lanes, and the light-rail system (the light rail is hit or miss depending on who you ask).
Good to see the American government still functions in one way or another...
... https://blogs.intel.com/csr/files/2019/11/Circularity-at-Int...
[Chlorine trifluoride] is apparently about the most vigorous fluorinating agent known, and is much more difficult to handle than fluorine gas. That’s one of those statements you don’t get to hear very often, and it should be enough to make any sensible chemist turn around smartly and head down the hall in the other direction.
The compound also a stronger oxidizing agent than oxygen itself, which also puts it into rare territory. That means that it can potentially go on to “burn” things that you would normally consider already burnt to hell and gone, and a practical consequence of that is that it’ll start roaring reactions with things like bricks and asbestos tile. It’s been used in the semiconductor industry to clean oxides off of surfaces, at which activity it no doubt excels.
https://blogs.sciencemag.org/pipeline/archives/2008/02/26/sa...
Read on to find out what happened when somebody accidentally spilled 1 ton of the stuff!
It’s not wildly surprising, given that all major USA chip players will now be using TSMC, including Intel.
Worst case, they will just fly in hundreds of existing, trained employees from Taiwan until the local workforce is stabilized. This is how Samsung bootstrapped their ATX factory.
For the doubters, this is just the beginning of Re-industrilization of America.
"Manufactured by TSMC-USA in Arizona". Super exciting to be quite honest.
Basically all of them, to some extent.
> But I can promise you if there is an option to buy an USA made AMD, I’d pay a premium for that.
AMD spun off their own foundry years ago (GlobalFoundaries) and still uses the for some things.
FWIW though TSMC already has a US fab, but all their high-end processes are going to remain in Taiwan.
AMD does make chips in the US via their spun off Global Foundries, and so does Intel. AMD switched some production to TSMC because they had much better defect rates at 7nm. You can buy US made AMD today if you want it just won't be the latest GPU's.
GloFo outright canceled their 7nm and smaller processes, back in 2018. All of AMD's leading-edge processors are now using TSMC 7nm, though the server and most desktop processors also incorporate a 12nm IO die. The days of AMD using GloFo are numbered.
Really, though, at this point it's easier to list what leading-edge chips aren't using TSMC: Intel's CPUs and FPGAs. All the other latest-and-greatest CPUs, GPUs, FPGAs and smartphone SoCs are made at TSMC. There's more diversity for memory manufacturing, and logic that doesn't need to be on a leading-edge process (eg. low-end smartphone SoCs, motherboard chipsets, SSD controllers).
They've operated in the US for a while now.
From the press release: "In the United States, TSMC currently operates a fab in Camas, Washington and design centers in both Austin, Texas and San Jose, California. The Arizona facility would be TSMC’s second manufacturing site in the United States."
> You have earned a friend for life.
What a bizarre comment. If they leave, are they your enemy for life then?
So yeah, educate yourself and then behave reasonably.
I know. Where in my comment did I say otherwise?
> So yeah, educate yourself and then behave reasonably.
Okay, thanks?
Did you by any chance respond to the wrong comment? Such bizarre comments in this thread.
(New) Relationship with strangers can be in one of the 3 categories:
- Neutral - Friend - Enemy
OP is moving the company into Friend category (from Neutral). I don't know what he'd do if they decide to move back, but if they declare the company as an "enemy" then that creates a perverse incentive where people in Neutral category would refrain from becoming friend because that risks them becoming enemy.
So to be prudent OP must move the company into 'Neutral' category again otherwise they won't get future 'Friends'
I'm not sure that's fair. There's a lot of room between friend and enemy, here. When breaking up in a relationship under typical circumstances, the other person doesn't normally become an enemy, unless a person is not well adjusted mentally.
As far as bizarre goes -- bizarre is better than boring!
Edit: and this isn't actually a cutting edge fab. By the time it opens in 2024 the cutting edge process will be 3nm or beyond.
I am big fanboy of Robert Lighthizer, and with Phase 1 effectively ripped we are going to see a lot of moving things. May not happen immediately, first it will be slow then sudden.
Good call-out re: GF.
[1] https://wccftech.com/intel-abandoning-10nm-after-dg1-plannin...
Most of the deep tech in a fab is the equipment. The rest of the stuff is a giant box that's super clean and maintains that clealiness.
All of these Fabs[2] buy their equipment from the aforementioned companies.
Here is what an example supply chain looks like:
NVidia(US) has the IP and "soft" tech, they use software tooling(US) to make layouts, then contracts out the manufacturing to TSMC(Taiwan) who then buys their equipment from KLA/AMAT/LAM/TEL/ASML. TSMC(Taiwan) processes the wafers which are packaged at OSATS(OSATS are a collection of many companies[3]). The final product such as a graphics card is then further made accessible by writing insanely deep software driver stack (CUDA drivers for example) that are all developed by NVIDIA(US). For CPUs, there would be ISA and other "soft" tech involved which is Intel(US), AMD (US), ARM(UK, owned by Softbank(Japan)), Qualcomm, Broadcom, Microchip and Apple (all US, its a long list). I didn't even touch storage/memory. In addition to US, a lot of software and hardware talent and manufacturing comes from the EU, Germany in particular. I used to get on the call with a German company in the late morning and the Germans would be working their ass off at 9pm in Germany. We had an amazing team. It is not uncommon to pay $20k for a motor made in germany.
Somewhere in the middle to low end of the supply chain (motors, wires, connectors, may be even steel, copper, precious metals, chemicals) probably has Chinese involvement. You can imagine how a graphics card's supply chain gets very complex and deep.
Back to the central point - Semiconductor is a massive industry with complex supply chain. Depends on the level we are talking about. Also Global Foundries is not considered as among top dogs when it comes to semicon tech, although that's also debatable.
Edit: added tooling
[1] Source of the data (2011, old but still relevant): https://www.vlsiresearch.com/public/cms_pdf_upload/706001v1....
[2] https://en.wikipedia.org/wiki/List_of_semiconductor_fabricat...
[3] https://en.wikipedia.org/wiki/Integrated_device_manufacturer
The other side of the equation is EDA tooling, which is primarily US-based (Cadence, Synopsys, and Mentor Graphics).
Of course, the angstrom being the size that it is (.1nm), makes it unlikely there ever will be another scale. Whatever is next, it won't be silicon.
> You have earned a friend for life.
The Trump administration has shown that "friend for life" means "for as long as I'm getting exactly what I want, or think I am, and otherwise you can see yourself out."
> For the doubters, this is just the beginning of Re-industrilization of America.
What benefit is there really? This fab will have about 5 jobs, and a bunch of robots who could be anywhere on earth.
I'm not sure this provides any more national security than a stockpile would.
*With extreme automation and minimal employees. Modern Fabs have very few assembly line employee's. They use positive pressure rooms, overhead rail car system, robotic arms, and machines drop down from the fab floor to a maintenance floor when a human is actually needed.
Remember when Trump floated withdrawing from NATO?
Remember when Trump stopped funding the WHO in the middle of the worst pandemic of the past 100 years?
America First has no room for friends for life. Only friends for the moment.
I remember when Trump took no actions at all toward pulling the US out of NATO, and did nothing other than use it as a rather dumb prop to try to coax NATO allies into meeting their spending obligations.
> Remember when Trump stopped funding the WHO in the middle of the worst pandemic of the past 100 years?
I remember when the WHO cooperated with China to knowingly lie about the pandemic, including trying to forestall borders from being closed, trying to forestall the isolation of China, lying about the nature of the virus including the risk that it posed, and committed global genocide in the process.
And the previous president had already got the Europeans to commit to meeting their obligations over a period of a certain number of years which they were moving towards. Trumps nonsense achieved nothing other than making the Europeans and the rest of the world recognize that the US has become deeply unreliable and move further towards China.
A stopped clock is right twice a day, and sometimes an idiot gets an answer right. There is zero reason for any US troops to be in Europe or the US to remain in NATO. We’ve been burning our wealth for 70 years subsidizing European defense, and their economies.
> I remember when Trump took no actions at all toward pulling the US out of NATO
In times of peace, alliances exist only in rhetoric. NATO has strength only insofar as people believe that member states will come to each other's aid. Trump called his commitment to the treaty into question, and in so doing made the alliance less strong. Surely you see this.
>> Remember when Trump stopped funding the WHO in the middle of the worst pandemic of the past 100 years?
> I remember when the WHO cooperated with China to knowingly lie about the pandemic, including trying to forestall borders from being closed, trying to forestall the isolation of China, lying about the nature of the virus including the risk that it posed, and committed global genocide in the process.
I mean, if lying abut the nature of the virus or the risks it imposed was grounds for dissolving an organization, then surely the federal government and many state governments should be dissolved by now.
But for the sake of argument, let's stipulate that what you wrote was 100% correct.
The WHO does lots of things. Apparently one of them is facilitating "global genocide" (kind of harsh?). Another one is helping countries, rich and poor alike, survive this very disease we're stipulating it brought into being.
Is America or the world safer today because the WHO lost its biggest funder? What replaces it?
If we don't want the WHO being unduly influenced by China, how does defunding it advance that cause?
Stuff like this is just one step closer to that. I would happily pay a premium for it. Excellent news.
And to all of you currently looking at your rent in SF and wondering if you should stay there now that you can work remote: AZ is a pretty awesome place, and more and more it's looking like it actually could become a hardware hub in the US.
To be clear on my point about rent in Phoenix, btw: I am currently sitting about 400 yards from ASU (I live across the street from campus). I rented this house for about 10 years for <$1000/mo (2br, 1ba, now I own it), and there are still apartments/condos in this neighborhood for that price. We don't have an ocean, but we have some of the most beautiful hiking in the world, the city is extremely cycle friendly, and almost every biome you could want is a few hours drive max from here. It's really a great city.
Saying comfortable to be 22-24C, 25 is just barely below zero. North Europe standard winter conditions are far lower...
Keep in mind Apple is sold around the world. Not everyone is American.
Because of all of my electronics being assembled in China, there have been millions of people that were pulled out of poverty.
In and of itself, what value does raising the living standard of millions of poor Chinese provide to Americans?
Rising living standard means more consumers that can buy American goods, and it also means that the likelihood of civil war/uprisings due to poverty goes down, which in turn reduces the potential for wars.
Taking into account how much the Afghanistan and Iraq conflicts have cost the US in money, how much lives and potential was lost, is actually easy.
Don't get me wrong, it was a good intention to get rid of the Taliban and Saddam, but the execution failed horribly. And that were backyard 3rd world countries. A conflict involving China would be an utter bloodshed, especially if it devolves into nuclear war, and it would totally destroy the US and probably EU economy.
We're seeing how much we depend on China now with the 'rona crisis, and that is a "China does their best effort to fulfill Western orders for anything from toys to medical PPE" scenario and not a "China is at war with the US and doesn't deliver anything to them and their allies" scenario. Politicians have also realized this but renationalizing even parts of what was outsourced to China will take decades. Until that happens China has the Western states by the balls, figuratively - and given how they acted with delivering PPE "donations" to Italy and other states that the EU left to fend alone they know this and are absolutely willing to use this as leverage to keep the US and EU from interfering in their eradications of Uighur Muslims and Tibetan people.
Globalization has put millions of people in the West into poverty. Globalization is a scheme by corporations to get cheaper labor and have the whole world compete against each other while destroying a healthy societal coherence.
https://news.ycombinator.com/item?id=23186174
I'd be taking bets on how quickly this is going to turn out poorly, but my last tiny ember of faith in any corporate actor being capable of rational thought just got extinguished, so I could be really wrong!
Long-term independence and stability of Taiwan is a political goal, not necessarily a corporate one. The political-corporate relationship in Taiwan is definitely different (or of a different flavor) than that in the USA right now, which makes this harder to parse for specific corporate objectives. In the aggregate, this could signal a willingness of the government in Taiwan to make bets (assuming they have to approve of things like this, I'm not 100% sure if that's true) that the long term goal (political independence) is best served by further entangling itself with the US. That's probably where the debate should exist.
Long-term independence and stability of Taiwan is necessary for TSMC to keep existing. It's in its best interest to keep existing.
I'm pretty sure the government doesn't need to approve of this sort of thing; VIA got away with it.
Further entangling itself with the US would be keeping itself for the most part in Taiwan. "We have drastically better tech than you, and your military, intelligence agencies & universities really desperately need these!" is a fantastic incentive. "Here's a factory that allows the US to be pretty much self-sufficient and no longer dependent on us for the next few years while you haphazardly try and get on top of the world again" is not nearly as good of an incentive.
Taiwan gets invaded yesterday, suddenly the US is in a sticky situation. Taiwan gets invaded when this factory is finished? 5nm vs 2-3nm isn't a huge difference.
I suppose that doesn't really address your broader point though, which is a good one, that Taiwan does/did have leverage in this relationship with their higher-tech fabs being located within Taiwan exclusively. R&D will still be in Taipei, I imagine. Very quickly this becomes, in my mind, a human capital question.
>If China can secure naval dominance within the second island chain
Skeptical that they'll be able to do this though, unless it represents a significant change in naval operations and they achieve this via drone or other submarine operations. PLA has had a difficult time getting the aircraft carrier concept to work, even after buying several Soviet era carrier vessels (if my memory serves). It might be a false signal, but that they haven't gotten the aircraft carrier concept to work (which would of course be required for an air bombardment of Taiwan) might mean they're working on other naval hardware that exploits US weaknesses, and thus crippling any US Navy activity via submarine or missile warfare in the Taiwan Straight is the ultimate objective for the above mentioned reasons.
Human capital creates a problem for Taiwan, because human capital is easy to obtain, and the US is really good at convincing the best and brightest of other countries to come to the land of the free, by coercion or just offers They Can't Refuse. This creates a kind of perverse incentive for certain pieces of the US government to hope the situation in Taiwan gets stickier, so they can offer a one-way ticket out of it for scientists they're interested in.
Hadn't thought of this. Still processing the thought. It reminds me of the US recruiting German scientists for NASA (absolutely not claiming that Taiwanese engineers are comparable to German engineers from Nazi Germany in any way shape or form). I'm expecting the TSMC fab to recruit local talent for operations, but I think you're right it'll definitely require some technical expertise that has to fly in from Taipei.
I think the bigger problem is the assumption that Americans can properly safeguard Taiwanese IP. Anything you send over here you might as well cc: Beijing on the email.
What I’m talking about is the capability of instantly kneecapping the US tech industry. A sudden differential advantage created by denying capabilities to your adversary. That’s what makes the current semiconductor situation a geopolitical tinder box.
This should terrify the US, which is the ideal. The US needs to go to bat for Taiwan, which won't happen when they have less incentive to.
China is willing to take independent countries for lesser and pettier reasons than semiconductors. Look what they've done to Hong Kong. Taiwan's been a thorn in its side for years. Distribute it around, China will still want to take Taiwan, the US has less of a reason to go to bat for it.
On balance you may be right, because China’s desire to control Taiwan is not exactly the result of a rational strategic calculation. Interesting times...
I have an actual reason, and it's in the thread I linked, and in my reply to another response to this comment.
The above comment was for two reasons: one, to publicly acknowledge that a guess I had made just a small time ago was wrong (that TSMC executives were rational actors, or if nothing else, not drastically short-sighted), and secondly to exclaim that I still believe that this is an incredibly short-sighted and quite frankly stupid decision.
EDIT: changed some words to be more generous to TSMC's executives.
Even if Trump is reelected, I could see them leasing an existing fab and have a ribbon cutting ceremony for Trump to get his photo-op and then just weigh their options with minimal additional expenditure (they've already commented on opening a fab in US won't be economical [1]) with the thinking that a few $million spent on a lease and avoiding antagonizing Trump might all be worthwhile in any kind of tariffs on their business.
[1] https://asia.nikkei.com/Business/Technology/TSMC-weighs-new-...
edited my post by adding these articles as well as editing the text to better reference the links
Wow Intel can't even get 10 nm working. Big win for Apple and AMD tho.
Its also difficult to say that "Intel 10nm competes with TSMC's 7nm" because in so many ways, it doesnt. Ice Lake server was supposed to be out this year with a new microarchitechture (the first since Skylake), but still is nowhere to be seen (though it certainly seems likely the big cloud players already have early silicon).
But is this not hurting Taiwan in a major strategic long term way? And risks turning the whole island towards mainland.
And for the ones celebrating "Re-industrilization of America". Is the America you envision really a America which does not have anything of importance produced outside America. Even if outside means traditional allies such as Germany, Japan, Taiwan ... ?
That only means the destruction of that island.