TSMC Gets $11.6B in US Grants, Loans for Three Chip Fabs | Hacker News Reader

TSMC Gets $11.6B in US Grants, Loans for Three Chip Fabs

bloomberg.com

113 pointsimpish920893 comments

It's nice to build chip fabs in the USA, but is it _really_ necessary to "gift" some of the most capable and productive companies on the planet billions of dollars (not even counting the tax freebies)?

The whole CHIPS act thing seems backwards and short-sighted.

As the US education system is backsliding into literally worse-than-third-world outcomes, can we expect that companies like TSMC will even be able to find enough US workers to staff these facilities at a scale which will be large enough to move the economic needle?

The Biden administration sees it as 2 birds with 1 stone.

Intel and TSMC are investing in swing states. It matters to Biden when it comes to the election.

The second is that the US genuinely wants chip fabs in US in case war breaks out.

It's most definitely inefficient use of dollars for mega corporations.

Without a doubt the biggest reason here is to ensure the tech is capable on their homeland. having most of modern tech being build in China/Taiwan and Japan is a HUGE security risk for US (and other "wesetrn countries")

>can we expect that companies like TSMC will even be able to find enough US workers to staff these facilities at a scale which will be large enough to move the economic needle?

Immigration.

Is part of the statecraft at play here to brain drain all the valuable engineers from Taiwan and China?

How else do you propose to get a company to build a factory at your preferred location rather than theirs? Paying companies is the usual way to get them to do things.

The economic effects are important but secondary to the national security effects of having TSMC chip manufacturing on US soil.

I wish it could be an investment instead of just grants and loans. Could be a way to jumpstart a national sovereign wealth fund.

I disagree. You don't want the government to own things because they would be slow and ineffective at corporate governance. It's better to simply subsidize and let the business owners make the important decisions.

Feels like a self fulfilling prophecy. There’s no reason government can’t be effective at decision making if things are set up the right way (their involvement could be distant). There’s also no reason to assume a giant company (like, say, Intel) is going to be particularly fast at it.

The people in government aren't incentivized to be effective largely because they have no personal financial reason to be. There's no stock options or things like that, no good bonus programs, etc.

What about a competitive salary they lose (i.e. get fired) if they underperform? We’re used to underpaying and overprotecting government employees but if you were setting up a new publicly owned company there’s no reason you’d have to do it that way.

> there’s no reason

The reason as always is politics.

For example, the Federal civil service ("general service") pay scale tops out around $160k. This really doesn't make sense and costs the government multiples of market wages for hiring for roles where market pay is higher than this. One route is to hire that person via a contracting firm with a huge (double or triple) markup.

And of course the reason we don't just revamp the pay scales and pay market wages is because it's political suicide to pay people half a million or more at market rates.

The American people, on average, don't like the level of wage inequality in society. For any particular issue where you shove it in their faces (like proposing to pay the project lead "competitive salary"), you will get shouted down.

That's also why we get strange arrangements like Congress members making around $170k a year with legal insider trading. The insider trading money they make looks like "free" and isn't easily quantified.

Great connection you make on the insider trading! That never occurred to me despite recently being fed up with local mayoral candidates; looking at mayor comp and immediately noping out because a substantial paycut for the pleasure of dealing with the biggest BS imaginable seems rather unattractive.

It's already difficult to fire people in large businesses, but firing people from government is in a whole different level. Especially for performance unless we're talking about extreme underperformance or job abandonment.

What I’m saying is that a publicly owned company has a lot of flexibility in this regard. Employees are not government employees. It’s like a regular private company except the government is the sole shareholder.

My broader point here about self fulfilling prophecy is reflected in your comment: it has been this way, therefore must always be this way. That’s not true.

Turns out that what happens in practice is corporations get chopped up for parts, and government usually does better.

Medicare for example is much more efficient than private health insurance.

The motivation for government employees is "because they care about doing the right thing". You should try it some time.

From working in government, “no, they don’t care about doing it the right way.” The people who get hired are, largely, the ones who will work for the lower salary paid, and plenty lack enough experience and/or capability to even do what they need to. At least in IT.

Here in Norway the government fully owns several "private" companies[1], like Statkraft[2]. They can have bonus programs and similar. Some are too good[3][4], others[5] not quite as wild.

[1]: https://www.regjeringen.no/en/topics/business-and-industry/s...

[2]: https://en.wikipedia.org/wiki/Statkraft

[3]: https://www.nrk.no/norge/statkraft-oker-inntekter-og-bonuser...

[4]: https://www.dn.no/energi/narings-og-fiskeridepartementet/lo-...

[5]: https://www.mesta.no/wp-content/uploads/2023/03/Mesta_Vedleg... (page 2)

> The people in government aren't incentivized to be effective largely because they have no personal financial reason to be. There's no stock options or things like that, no good bonus programs, etc.

I would hazard to guess that most employed people do not have stock options or get bonuses (only salary), but are still "effective" because they consider that doing a job well is its own reward: i.e., they have intrinsic motivation (rather than extrinsic).

I have worked in private sector, and in government (including academia/research), and most folks want to do a job well because they like the satisfaction of being able to know they did the job well.

I'd say it's not that clear cut. Do we have any facts (research) here or are these just beliefs? I can point to several but companies doing very well owned by governments, as well as many doing very badly.

Could be slow and ineffective. It's as likely a company would run it into the ground because of short-sightedness. There are so many examples of government services becoming private company enterprises and in short order everything got worse. But I guess there are equally many examples of the other way around as well.

They would be no worse and probably better than private equity or public markets.

The biggest owner of TSMC is National Development Fund, which is a fund owned by Saudi Arabia. I'm not sure if a US owned fund would be slower and more ineffective in corporate governance than a Saudi owned one.

I think it's actually National Development Fund of the Executive Yuan, which is a Taiwanese fund.

> wish it could be an investment instead of just grants and loans

It’s much more cleanly done with loans.

it is an investment

Another approach would be to apply tariffs to chips coming from adversary countries, until it becomes more economical to produce them in North America.

That would increase costs for US consumers drastically in the interim 20 years it takes to get those fabs online

Taiwan isn’t an adversary.

That’s the problem we are trying to solve: Too many important chips are made there.

>Another approach would be to apply tariffs to chips coming from adversary countries,

But Taiwan isn't an adversary of the USA so why would tariffs be "another approach"?

The Information Technology Agreement of WTO[1] eliminates taxes and tariffs for IT products, I'm not sure if chips belongs to this agreements?

[1] https://www.wto.org/english/tratop_e/dtt_e/dtt-ita_e.htm

I agree with you, and am excited for TSMC fabs in the US. At the same time, I also wonder if this would lead to _less_ defense of Taiwan if we neuter its strategic importance by building those crucial fabs somewhere else. Mind you, this is not an argument against building them. It just makes me wonder what the other effects will be.

At the end of the day it’s a trade-off decision. It’s probably cheaper to build highly subsidized factories in the States than to go to war over Taiwan.

US will still have plenty of reason to defend Taiwan, if only because it is a crucial part of America's containment strategy - China would have a much easier time operating in the Pacific if it controls Taiwan, and the US won't allow that. Further, if Taiwan falls a lot of countries will lose faith in America's ability to protect them from China which would thrust the entirety of South-East Asia into China's hands, also something the US can't afford. Though, with recent claims that China can build 1000 cruise missiles a day[0] it might not matter whether US defends them or not.

[0] https://asiatimes.com/2024/03/america-has-no-ukraine-plan-b-...

They have said they will not build the best and latest here in the US. Many of the fabless companies here in the US want access to that latest stuff and the US is vested in keeping that channel open for market/defense/security reasons.

The Chip Wars book is an excellent read and tells you us the supply chain is globally integrated. If we were to draw a country graph of technology dependence to create semiconductor technology T (tools, chips, etc) - well, there is no DAG. The defense department is not so happy to see all this free market sharing of essential technology going on as it benefits non-friendly nation(s).

The book also illustrated how behind Russia was/is with weapons technology and semiconductors in general (compared to the US).

Yes, Vivek famously already said during his campaign that he'd tell the CCP that we need a few years to catch up on chips and then they can have Taiwan. I'm sure other share this (IMO inhuman, deprived of any morality and allegiance to our liberal western alliance and any foresight) sentiment

Being 2-3 years ahead in chip making(with other closing the gap over time) is not a defence against a potentially WW3 level event. Anyone who thinks millions of life is less important than few percentage increase in performance is delusional. I don't know how this discussion surfaces in HN so many times.

The USA - the most powerful and richest country in the world - could ... just ... build it.

The US really doesn't do things directly like that, at least not in the last 50 years. Maybe you could point to this and say its wrong, but going through a contractor or company is the way the government typically operates.

Isn’t that effectively what they’re doing? They’re using their riches to pay the people with the necessary institutional knowledge to build it in the USA.

I like this idea and think we should get back into federally building things too.

I think you’re completely missing the point. This has nothing to do with moving the economic needle.

Having 90% of the most important semiconductors in the world made on an island 90 miles from China, which China wants to take back, is a bit of a problem.

What happens in a protracted skirmish that lasts a year, for example, where we can’t get any of those semiconductors.

> which China wants to take back

That's Beijing rhetoric.

They can't take it back because it's never been theirs. It would be more accurate to say they want to conquer the last remaining territory from the legitimate government of China and wipe out the main alternative to their hegemony.

By the way, doesn't competition law say anything about these gifts?

The OECD has a good writeup about this very subject: https://one.oecd.org/document/DAF/COMP/GF(2022)6/en/pdf

> This is however not the case and subsidies are, with a few exceptions, mostly assessed or controlled using multilateral rules and free trade agreements.

I haven't read the full thing because I'm no economist, but it seems to be above board. For comparison, Intel is getting $8.5 billion from the CHIPS Act (https://www.intel.com/content/www/us/en/newsroom/news/us-chi... I can't find other US based chip companies like NVidia or AMD getting any from that pot of money though, which is weird. Although NVidia doesn't need it, given how their stock price and valuation rocketed up.

Have you ever played an economics game such as Victoria 3?

IMO the government should do a lot more subsidizing of important businesses. It would grow the economy faster in useful ways.

Yes, it is necessary if the world wants to have high-end chips manufactured more than a few miles from China. China's government has been spending billions and billions on an utterly massive military build-up. The US goal is to be able to continue to have chips, which are necessary to power almost everything else. I think what's going on is a geopolitical struggle. The US views this as a national security issue.

A good question is whether or not this will work. A big building is not the same as a working best-in-class fab.

It's about war.

The US wants on shore chips facilities with trained US staff so that if we end up in a shooting war with China we won't lose our supply of chips.

That's the whole thing.

>It's about war.

I mean yes, ultimately, but it's also about supply chain disruption that we just experienced with COVID. The US economy (and subsequently the Western economy) would probably get hit pretty hard with a chip cutoff and 5 years to onshore it. The economy is also about war; without a strong economy, we can't stockpile all our murder toys. It's also about leverage with China. With China on an aggressive footing and Taiwan so vulnerable, they have a leverage point we're now addressing.

We got this way because we thought if we brought China into the modern economy they would democratize. I think we came to the conclusion that it's not working out that way so we're adjusting and divesting from China. COVID was a wake up call in regard to having a single country be the world's production of goods. It ups the bet when that single country is becoming more of an adversary than a competitive partner.

Chip demand is cyclical, and idle capacity is extremely costly, so chip fabs are built around the bottom of the demand curve to always run at full capacity. If you want to avoid another chip shortage, especially for national security reasons, financial incentives are the only way to do it.

> ... but is it _really_ necessary to "gift" some of the most capable and productive companies ...

As I understand it, the gift is, for the most part, what we'd call 'funny money' - no one's really handing over the proverbial briefcase of USD$11B in return for them building out a fab plant on US soil.

While the figures are referred to as loans, I suspect in practice they manifest quite differently.

Well then maybe I can pay my taxes this year with the “funny money” you are talking about.

Forgone taxes are REAL money.

Or maybe you can pay 4-10x or even more the price for chips when something in the world destabilizes (remember that during covid?), was that not real money?

Libertarian types love to make models of the world that are so simplified they are totally incorrect. But hey, you'll be able to save on taxes in the good times, and starve in the streets during the bad.

I see your point, but if the USA didn't encourage the company to set up fab shops domestically, then it wouldn't be receiving those taxes either - but in that scenario it also wouldn't be receiving the indirect benefits of having that company operating locally.

Your personal taxes are a different matter - the nation state is not trying to court you into operating within its borders with some tempting on-paper discounts.

In what world is the US education worse-than-third-world? The US has universal literacy, phenomenal higher education institutions, and primary and secondary schools that are comparable to peer countries on average.

We spend lots of money on schools in the US and generally get pretty good outcomes. Our math scores leave something to be desired, but average reading scores are better than peer countries. There are plenty of exceptions, especially in poor neighborhoods, and chronic absenteeism is a real problem, but it's worth keeping sight of the fact that the vast majority of American schools are quite good.

Some examples:

* PISA scores: https://www.oecd.org/pisa/OECD_2022_PISA_Results_Comparing%2...

* PIRLS/TIMMS scores: https://nces.ed.gov/fastfacts/display.asp?id=1

Per your first link, the US is below the OECD average with a -13 delta from 2018

That section is just for mathematics, which the commenter specifically called out. The reading and science are above average

Above average seems like a low bar if the US wants to stay the top economy.

Particularly so given how "wealthy" the US is.

I'm not trying to say that the US educational performance is great, just that it's not terrible. I think there's a lot of room for improvement, but there's a narrative that US schools are failing horribly and that just isn't true.

Wouldn't you agree that "failing" is relative? Given just how much money the US has, wouldn't you agree that we should expect better outcomes? Wouldn't it be a failure that despite our massive wealth, we are below OECD average in any subject?

That is a _huge_ shift of the goalpost from "US is worse than third world countries"

Wildly shifting goal posts from “worse than third world” to “above average isn’t good enough”

For math. Above average in reading and science.

There's also the matter of picking a definition for third world.

Does OECD include third world?

The PISA scores are below average in 2022 but not significantly so. Most countries have had a pretty bad decline in test scores due to covid, but that's (hopefully) a one-time problem and future cohorts will do better.

I'm not saying that's a good thing. We're a very rich country and we should do better.

But the US also has a lot of systematic problems that peer countries don't, like child poverty, that are likely a bigger cause of subpar test scores than bad schools.

There's a difference between "below average and rising" versus "below average and falling". The US, over this period, happens to be the latter: below average and falling.

Merely focusing on "below average" without acknowledging the trend -- declining -- seems to be disingenuous.

National averages are nothing like the areas of the US where a fab would go.

They need immigration to restore the values around education and work ethic. I hope we stay at it long enough to make these places great again.

Why are you pitting this against education funding?

The US spends more on defense than both put together. A lot more.

It is fun to imagine a future semiconductor supply glut in which Nvidia no longer has to charge an 85% markup on data centre GPUs to get demand down to the point they can (barely) meet it with supply.

More supply is going to create more demand of course, by unlocking more AI use cases.

https://en.wikipedia.org/wiki/Jevons_paradox

Industrial revolution is a fitting analogy too!

Would Intel survive if there is a glut? Their foundry unit reported a $7B loss this year.

Was it write-downs?

CHIPS act ($39 billion) pales in comparison to Chinese subsidies.

All Chinese semiconductor subsidies, both local and state subsidies, are estimated $60 billion __per__year__.

> Chinese semiconductor subsidies, both local and state subsidies, are estimated $60 billion __per__year__

Source? Note that several states, including Arizona, are topping up CHIPS.

Even if that's true, don't forget that the semi-conductor companies that are backed by the US are already state-of-the-art and profitable companies.

China has to establish a de novo industry, it's not surprising that to do that quickly and effectively it would require ten times as much funding.

True. China pays for technology. US pays for location.

Never doubt what China can achieve with 6.5 days of workweeks and pseudo-slavery work, as well as subsidies and currency manipulations.

Just look how they managed to scale up their car manufacturing business so quickly, now they can probably make enough cars to supply 100% of the world's demand, if they had appropriate demand.

What I wonder is why the world didn't cut their ties with China faster, as they are clearly killing the competition by playing a very unfair playbook. Only when China finally respect other countries by letting its population be real consumers, and also treat their population with dignity the west should be willing to talk with them.

The Big Fund (Chinese subsidy) has ongoing graft and corruption issues [0] all the way to the very top [1][2], so most of that money was never actually deployed competently. This is what lead to Tsinghua Unigroup's collapse, and a number of other corruption probes.

Furthermore, the Big Fund subsidy is an attempt to build a greenfield ecosystem, while CHIPS is basically reinvesting in a brownfield ecosystem.

[0] - https://www.ft.com/content/8358e81b-f4e7-4bad-bc08-19a77035e...

[1] - http://m.ccdi.gov.cn/content/8f/24/93463.html

[2] - https://www.reuters.com/world/china/china-watchdog-says-is-p...

US is only paying companies to locate. If you have to pay TSMC billions so that they build their factory in Arizona, US is not a competitive place.

At least Chinese are paying for technology development.

> paying for technology development

The US is not paying for that because that is already done in the US ecosystem. For example, DUV and EUV lithography is US DoE IP that was licensed to ASML to prevent anti-trust abuse by Canon in the 2000s.

There's a reason why I said "an attempt to build a greenfield ecosystem"

Basically,

TSMC:SMIC

Micron:YTMC

Hong Ha:Tower

There isn't an Nvidia, AMD, etc equivalent yet in the Chinese ecosystem, and that's why the Big Fund is so big (well, also because of graft as I mentioned)

Nvidia are AMD are fabless design companies. US has very limited edge there.

The Big Fund also funds Fabless design vendors like Biren [0], Ingenic [1], and Zhaoxin [2] in order to build a domestic Nvidia or AMD.

The Big Fund funds HUNDREDS of companies, not just SMIC. In fact, SMIC is basically the last man standing, as almost everyone else collapsed (YMTC and Hong Ha were restructured last minute to save Tsinghua Unigroup and other players), which is what triggered Xi's graft probe and hundreds of executives and government officials getting the ax.

[0] - https://www.birentech.com/

[1] - http://www.ingenic.com.cn/

[2] - http://zhaoxin.com/

> so most of that money was never actually deployed competently

IMO that' doesn't follow. When 100s of billions are involved, graft limited to a few percentage points while rest is deployed correctly. Access money hardly stopped other PRC industrial policy from succeeding. Regardless, small % is still 100s of millions / billions meaning it will get CCDI attention. The issue was coordinating greenfield ecosystem and domestic demand to play nice together, which they couldn't when western options were available. Now PRC domestic semi looks to be integrating at rapid rate, partly because most of the money did their job and build up enough domestic players across the semi supply chain so they're not starting from zero.

>All Chinese semiconductor subsidies, both local and state subsidies, are estimated $60 billion __per__year__.

The closest I can find to this figure is a Dept of Commerce report that estimates China gave $150bn in subsidies in total to its semiconductor industry in the period 2013-2023.

Do you mean that they give 60bn rmb per year?

No. They have increased subsidies dramatically after US sanctions and technology restrictions

Just a single package was over $143 billion. https://www.reuters.com/technology/china-plans-over-143-bln-...

That was because of debt restructuring due to Tsinghua Unigroup's collapse [0][1]

[0] - https://asia.nikkei.com/Business/China-tech/China-s-Tsinghua...

[1] - https://www.bloomberg.com/news/articles/2023-03-20/china-inv...

Did the proposed package actually happen? There seem to be no follow-up articles about it. If the CCP wanted to hide it, $140bn would still represent ~1% of their GDP, so there should be some pretty obvious markers indicating that it went through. In Sept 2023, the FT reported that the Big Fund (the main pot for semiconductor subsidies) was struggling to hit its modest targets of $40bn in total across a 5-year investment.

I don't see it as credible to say that China are currently investing $60bn/yr, or anything approaching that figure. They seem to currently be struggling to hit $10bn/yr.

https://archive.is/quo5I

Regardless of the reasons for the CHIPS act, any thoughts on how the new fab capacity will change the market?

Reduced costs for the consumer? All eaten up by AI chips?

Reply on news.ycombinator.com