It may be true that TSMC and Samsung can meet demand during normal times, but it completely misses the point that NA and EU want significant production capacity locally not to reduce costs, but to secure a critical supply chain.
Is this a situation where he may be concerned that significant expansion of production capability away from Taiwan and SK will reduce the West’s reliance on TSMC, which may be a security risk for TSMC or Taiwan? I’m nowhere near an expert on this sort of thing, I’m honestly curious.
And frankly, this is everyone's motivation. EU wants to not depend on TSMC. US has gone all-in to not depend on TSMC. China has gone all-in to not depend on TSMC. And there are even rumors that Russia is trying to build their capabilities in this regard.
I think the leadership at TSMC is being a little naive. Alternatively, they may know very well what's going on here and they are being willfully obtuse to paint as bright a picture as possible of their long-term prospects.
You don’t become a leader in this industry by being naive. HN readers are being naive believing that TSMC really meant what they said. Anyone with a brain knows that the US wants to reduce reliance on TSMC
Yeah. It's not just supply chain, it's also national security.
> I think the leadership at TSMC is being a little naive. Alternatively, they may know very well what's going on here and they are being willfully obtuse to paint as bright a picture as possible of their long-term prospects.
It could be two things:
1. Some arrogance, "hah, this is harder than you think"
2. They're trying to discourage the competition because they know it will hurt their bottom line.
Even if they only lost 10% of their revenue, that can still hurt a lot. They could go from being a growing business to stagnating or shrinking.
I was thinking about that recently. China can't realistically invade (or intimidate/sanction) Taiwan as long as they are dependent on TSMC, because of those semiconductor manufacturing plants went down, it could be a global economic disaster, and China would be one of the most heavily impacted, given the proportion of everyone's electronics they manufacture.
China can't realistically invade TW until it can take on USN 10-15 years from now after more build up / military modernization anyway. Happenstance of TSMC dominance is IMO blessing in this regard, buys PRC 10+ years of military and semi catchup time. There's going to be theatrics and political posturing in the interim by US/CN/TW, but ultimately TSMC buying TW time is buying CN time to close gap with US.
The West should deal diplomatically openly with Taiwan: some limited production creation in EU/US in exchange for international recognition and protection of Taiwan.
That would be incredibly dangerous for Taiwan because China would react if it felt the military balance would actually change.
More importantly, the EU/US are not really in a position to make a credible threat to China. Let’s say they China declares its intention to forcibly ‘reunify’ Taiwan. EU sanctions on China would be incredibly painful... for the EU. The EU just signed a new trade treaty with China which probably tells you all you need to know about that option.
So down to the US. If the US didn’t deter China by trying to change the facts around Taiwan when it was considerably weaker it’s hard to understand why the US would take that step now.
That’s kind of the problem here. There isn’t anything that would credibly deter China at all so taking moves to antagonize China don’t accomplish anything. Even if other countries felt threatened by this they would much prefer to engage in ‘buck passing’ rather than taking actions that would directly hurt them.
From the Taiwan point of view, allowing any competition is opening the door to trouble in the long run. Political winds come and go, but building semiconductor capacity can't be done overnight, so the longer they can keep their edge, the longer they'll be relatively safe from their neighbor to the north.
It doesn’t make economic sense to build basically an entire new supply chain though.
You can’t just build a modern foundry with any amount of money. TSMC is part of a massive supply chain and engineering cluster. If you wanted to build something that looked like TSMC and in an emergency could produce something in six months you are probably talking about tens of billions per year at a minimum. Most of your costs are things TSMC gets paid for.
Since that doesnt really make sense you probably want a homegrown competitor. You just don’t have the domestic experience and expertise to replicate what TSMC has; It will never be cost competitive. TSMC can send billions on fabs and recover their costs and a domestic alternative will never be able to. Modern chips have to be adapted to each foundry; a homegrown alternative will have to eat this cost at first just to get orders.
If you want to be competitive you need to go big and stay there. Something like $100 billion a year for the foreseeable future would be the minimum. You could probably limp along with massive discounts.
Ah yes interesting point. Obviously TSMC and the Taiwanese government want to maintain their quasi monopoly for economic reasons. But there is also a military component: is Taiwan still worth fighting over?
TSMC is probably completely accurate. Their contention is that moving the supply chain would create "non-profitable capacity". However, national supplies do not need to be profitable if they are subsidized or fully funded by a state.
The state is free to, and often does, engage in non-profitable production to ensure essential domestic needs can be satisfied.
This doesn't make sense from a market perspective, but it can often make sense from a needs based perspective.
Won't this lead to innovation grinding to a halt? Right now TSMC has a decent profit margin to re-invest into newer technologies. If the margins become thinner, the industry will find it difficult to set aside R&D dollars for the newest technologies.
Of course, you can always have a newcomer who can disrupt the market but most Moore's law advancements appear to happen in government funded research labs and large semiconductor companies.
Not necessarily. Most innovation is state funded in general. Real innovation requires either monopoly profits or state funding because otherwise research must necessarily be directed towards short term payoffs.
Investors will not tolerate time lines of 10+ years and hundreds of millions to billions of dollars sunk.
Do you mean healthy profit margins?...because I'd rather prefer a combination of big profits, state subsidies and healthy competition than a monopolised/oligopolised market
Unfortunately, I did mean monopoly profits. In a non-monopoly situation, competition exists which forces market participants to constantly reduce costs and compete on quality or price. Eventually, a single or few participants will be left standing as the others fail and the the remaining ones engage in a process of buying their supplies and workers at discount prices. This is not an environment that supports long term thinking.
There might be an exception here and there, but there will be some structural reason why long term planning is possible (such as competition not being able to step on each others toes). It's not true in the general case.
True, that's because the system is designed to subsidize expensive research and let industry exploit cheap research. It makes no sense to think that state funding, having already done the heavy lifting, can't do the last mile.
Industry mainly provides access to the scalable means of production for tinkering, which can be owned by anyone, not just industrialists.
> Won't this lead to innovation grinding to a halt?
No more than Taiwan getting blown up. It's going to be about balance. No good to have all the eggs in a single basket, whether that is TSMC or some subsidised national attempt.
This is correct, but imo it is far more likely for the US (the largest military power in the world) to start something. No need to be hyper nationalistic about this.
It’s not obvious that the US will be the largest military power for long.
And china has been flying incursions into Taiwanese airspace with increasing regularity.
Plus, China has broken its treaty obligations in Hong Kong, and explicitly expresses the view that Taiwan is part of China, and it has no strategic semiconductor fabrication capability of its own.
Given all this, I’m not sure why you’d imagine the US would ‘start something’ purely because of the dollar amount it has spent on the military.
It depends on how you regard Taiwan. If you think of it like China's Texas, it's a domestic issue. If you think of it as a nation under threat, it's an international issue.
"Even as more than 1,000 people die from COVID-19 every single day in the United States, and the disease surges around the world, the US is preparing for a conflict that risks incalculable human suffering. Joining this offensive is the United Kingdom, with the highest COVID-19 death rate of the major European countries, which announced Tuesday a massive expansion of its nuclear weapons program, calling China a “major threat.”
It is not COVID-19, but China that the US has planted firmly in its sights. As Blinken made clear, “Several countries present us with serious challenges, including Russia, Iran, North Korea… but the challenge posed by China is different. China is the only country with the economic, diplomatic, military and technological power” to “challenge” the United States.
On March 10, Adm. Philip Davidson, commander of the US Indo-Pacific Command, told a Senate Armed Services Committee hearing he believes that China is likely to invade Taiwan within the next six years. “I think the threat is manifest during this decade, in fact, in the next six years,” Davidson said.
Given that the United States has, in the words of Defense Secretary Austin, “commitments to support Taiwan’s ability to defend itself,” to predict that China will invade Taiwan within the next six years is to predict a major Sino-American war within that same time period."
If governments are looking at Chinese-Taiwan relations, and gaming the result of an occupied Taiwan, those supply chains become profitable pretty damn quickly.
If China took Taiwan, they would be selling to the rest of the world most likely. However, perhaps they are planning for a case where the US bombed the crown jewels to maintain supremacy. That would create profits for a domestic sector.
The USG is perfectly happy to spend taxpayer money and foreigners' money to make their corporate constituency happy even if the financial equation is overall negative.
I don't think that there is any plausible scenario where China takes Taiwan chip fabs intact - if they are not disrupted during the conflict, I strongly presume they would all be intentionally blown up as soon as invading forces get a beachhead.
I think TSMC over estimates how state sponsored fabs would look like. I can predict now that it will be a disaster and many billions will be funnelled to bureaucrat friends through questionable procurement. If there is going to be a miracle and they actually build a fab, it will be lacking as they will not find specialists willing to work for EU wages and level of taxation.
To some extent it's true, but pretty minimal as ASML sometimes delivers machinery to their East Asian customers by rail to China then does the last bit by sea.
Sounds like someone wants to hold onto unsustainable monopoly.
Edit: chip shortage has been a known variable for over a year now. I think TSM overplayed its hands a bit. They didn’t want to expand and get caught when the demand would subside, which is very understandable. But what they didn’t foresee is that the US govt would get involved in propping up its competition in direct response.
A year is not nearly enough time to spin up a new fab. It is impossible to say with any certainty that TSMC did not respond correctly to a variable that has been known for only a year.
"We built enough capacity for the world so you shouldn't build any local capacity to ensure resiliency of supply chains because this will harm the current efficiency of our monopoly position."
These people are awful.
That's like saying you should never use anything other than RAID0 because anything else costs you more for the same performance. Yeah, sometimes cost/performance aren't the only important metrics!
The pandemic has literally been a real-life example of the downsides of the approach he is advocating.
Semi is oil now, there's more strategic/geopolitical considerations than supply/demand curve. US/EU/CN do not want Taiwan to have semi dominance, it's not in anyone's interest. Current TSMC position is happenstance due to poor industrial policies that countries are scrambling to address. IMO press releases and capex spending around Arizona announcement suggest TSMC wasn't prepared to built US fab, let alone 6. US pressured them, EU failed to. There's a good chance TSMC/Taiwan will try to delay their silicon shield evaporating for as long as they can. Probably not a coincidence big ticket US weapons sales are scheduled around when fabs would be up. Future will be priority sourcing from domestic fabs, let TSMC whither, by design.
Is Liu misunderstanding the Western plans or am I? I was under the impression that the US and Europe intended to get some onshore fab capacity as a second-source and backup in case Taiwan is blockaded or worse.
If the plan was to replace TSMC entirely, well, I'd think he was justifiably concerned as (a) a threat to his business and (b) evidence that the US was decoupling from Taiwan. But I don't think that's the case.
Arguably US and Europe absolutely intend to get some onshore fab capacity as a backup, and the choice for TSMC is to either do that (perhaps demanding some subsidies as enticement) or refuse with the expectation that this will inevitably cause attempts to decouple from TSMC even if it's not economically viable, purely for strategic reasons.
IMHO there are some similar trends with EU agricultural policy, where arguably one of the reasons (not much discussed to general public, but obvious if you look at what's being done) for sustaining farming subsidies in sectors where EU would be uncompetitive (due to climatic reasons and labor costs) is the strategic goal to ensure that EU does not become reliant on imported food even if it can be farmed cheaper elsewhere, so that in case of major geopolitical disturbances in coming decades (e.g. caused by climate change) there's local farmers, farmland, skills and infrastructure for that. The costs seem huge, but in proportion to the whole economy they're affordable - the same would apply for on-shore chip fab capacity that's needed for strategic self-sufficiency of key resources.
On the other hand, I fully agree with the TSMC statement that it would be economically unrealistic for EU to expand semiconductor fab capacity to fully satisfy their own needs - it's feasible to expand fab capacity so that you have some onshore supply of modern chips, but it's impractical (and IMHO not within plans of any politician) to ensure that there would be enough fab capacity to supply all the consumer consumption, the expectation is that most of the volume would still be supplied from Taiwan and elsewhere.
On the other hand, if I think more about it, it may be a key strategic interest of Taiwan to ensure that USA and EU don't get any reasonable onshore fab capacity in the near future - if Taiwan (and trade with it) becomes irreplaceable for USA and EU, possibly even more irreplaceable than the trade with mainland China, that would ensure that USA and EU would be motivated to defend Taiwan against any attempts of forceful reunification.
The goals aren't mutually exclusive, they just depend on time-frame and how well things go.
Regardless, even if it is the former goal, it is still a potential threat to TSMC profits. Personally, I would not look at it in terms of Liu (mis)understanding, but acting in his own self interest to maximize his profits.
Sounds like a tricky situation for everyone, don't want to alienate Taiwan or Apple or China. And TSMC is building capacity in the US. It isn't really a West vs. East game.
Doesn't this imply that it would be unrealistic to create large amounts of new supply to satisfy local strategic needs in a global economy?
for a large market like the US/Europe couldn't they choose to not participate in the global market for chips? or tip the scales such that local competitors are favored?
I don't see any reason why certain products couldn't be on a sliding tariff scale dependent on how much is imported. 20-40% imports = no tariffs, 80% triggers a sliding tariff scale. This would allow for strategic preservation and global competition. Once you fully outsource an industry it's hard for it to ever grow back - even if it makes sense for it too.
Tariffs like that can backfire and prop up non-innovative companies. The US steel industry is a good example. It’s still based on WWII technology, and there’s guaranteed income regardless of whether it’s competitive.
The question of unprofitability is interesting on large capital projects - it's often a question of what timescale we're talking about, and if nations become involved then the timescales can be both significantly larger as well as the dynamics of possibly loaned or direct grant money for the capital.
If it's a matter of subsidizing one industry to favor the overall economy (e.g. more chips so the automotive and other sectors don't slow down), then nations also get involved. China pretty much directly and openly, and the US/Europe more indirectly.
What's the US military budget? Let's say DOD decides it wants onshore chips. $10B is peanuts; $100B is easily doable if they say it's important enough- which it is.
the DoD budget was around 700 billion, no one is going to spend around 1/10th of their entire military budget on semi conductors at cost. They're going to assume some returns on these fab factories.
I couldn't find anything by site, but TSMC reported these stats in their 2019 Annual Report as a whole out of 51k employees: 5k managers, 24k professionals, 4k assistants and 17k technicians.
They reported highest level of education for their workforce of: 4.5% PHD, 44.9% Masters, 25.5% Bachelors.
I think Liu is trying to dismiss the west’s concerns on the only basis that he can: the short term economics. In reality there is a more important strategic priority at the core of these initiatives, but that’s a whole lot harder to play down.
The whole world economy relies on 3 countries in balistic rocket range of each other; not great. Having foundries distributed around the world sounds way more resilient to any conflicts or disasters that might arise in that particular region.