Semiconductors and Modern Industrial Policy
pubs.aeaweb.org
pubs.aeaweb.org
Who in the US makes surface mount resistors, capacitors, inductors, etc. in volume sufficient to meet all of our demand? (I'd love to be proved wrong here)
Adding to this point, don't underestimate passive components! For some applications, such as RF, you can't just change inductors with another inductor that have the same inductance value. Because they will have different parasitic capacitance and resistance, which will affect RF performance.
Passive components are surprisingly involved, they even have their own news site: https://passive-components.eu/
EDIT: Its not just RF, capacitor selection matters for power converters too. Do you know that multilayer ceramic capacitors (MLCC) can lose 70% of their rated capacitance at high DC bias voltage? This is what happened if someone doesn't realize that: https://www.reddit.com/r/embedded/comments/1cs2c5l/todays_re...
If I specced an inductor or MLCCs, I can't just change it. This means if any country wants to have electronics supply chain in house, they need to bring all of the passive components back, or they can't manufacture things.
Electronics is not just 5nm chips, those chips that always in the news need support components
Anyway, multiply this headache by nitric acid, hydrofluoric, bare wafers, several kinds of photoresist, arsine, phosphene, diborane, silane, cmp slurry, and dozens of other ingredients, plus oem spares for ten thousand different consumables for your equipment. Supply chain for semiconductors is no joke.
But cost is much higher than in China, and things like "bare wafers" are not included in that petrochemical output.
Are you familiar with the trace metals requirements of semiconductor processing? And the volume of an iso tank? And the volatility of the petrochemical refining industry? I wouldn’t write off the difficulty of maintaining qualified suppliers quite so cavalierly.
The legendary ATC (American Technical Ceramics), belongs to AVX which belongs to Kyocera. It also apparently manufactures all their high frequency capacitors in the USA.
So if we look at US manufacturing proper, it exists, though compared to the greater China, it is of course quite small, as one would expect, considering where most of the electronic assembly happens. However, if we look at the component manufacturing which is "not in China", then the situation is fine, with the dominant companies having hundreds of factories in Japan. Specifically, lots and lots of ceramic-based components (capacitors, resistors, filters, magnetics, etc) are manufactured in Japan (Murata, Kyocera, TDK) and to some extent by their subsidiaries in the USA and elsewhere. If we include Mexico, Eastern Europe, Israel, Philippines, a few factories in Western Europe, then the situation is quite balanced.
Here is an interesting article regarding capacitors: "Japanese companies hold a dominant market share of 56%, significantly ahead, while mainland Chinese MLCC manufacturers account for approximately 7% of the global total." https://www.cytechsystems.com/news/top-mlcc-manufacturers
Again, we should reserve that word for REAL war.
Not exactly.
China is federal, and funding comes from both local government and central government.
This can lead to local governments burning money on less successful ventures because of the close nexus between leadership and local politicians. China still has a severe problem with Access Corruption for this reason.
Also, at the Central Government level, ministries and SOEs will have their own funds and competing interests which leads to the occasional backstabbing
The collapse of Tsinghua Unigroup and much of the investments from Big Fund 1 are a good example. Big Fund 2 is only just closed recently, so it will take a couple years to judge the results of that batch.
After the 2015-16 market crash, China basically migrated towards a State Capitalism model with SOEs and Governments (local and central) acting as fund managers. This has pros in the sense that you can act quickly on political directives, but this has massive cons in that your incentives are aligned with keeping your direct managers who are political/party appointees happy.
Political Incentives (local or federal) might not always align with what's best for a product or company. EVs are a notable example of that, with private BYD and Tesla out-competing every other state funded EV and car manufacturer in China.
The problem you are describing is fundamentally different from that in the US.
The problem in the US is that in many cases the money is literally just pocketed and returned to shareholders. The problem with funding of groups who are likely to fail due to connections is significant of course, but at least even in that failure mode money is spent on R&D and infrastructure, which at least has a chance of some success and even in failure will develop talent.
Tsinghua Unigroup is a good example. Yes it failed and defaulted, but from it's failure there is YMTC which is a huge strategic success and UNISOC which is a moderate success.
The issue you're describing is also present in the US, by the way, with local governments using tax rebates or direct incentives to lure corporations, and lobbying for subsidies to unsuccessful businesses due their locality/connections.
But local governments do not have a controlling stake and actual CapEx investments in those ventures in the US.
If Tsinghua Unigroup goes belly-up again or can't meet it's targets as part of the reorg, the Anhui Province is on the hook, as they have the controlling stake in it after the cleanup.
This is a risk that a lot of people on HN seem to ignore (I think because of the complexity of the Chinese federal system).
In most cases, it's provinces that are on the hook for these ventures, and if they fail, it's a significant chunk of cash that is lost. Unlike the Central Government, Local and Provincial Governments have weak financials because they traditionally couldn't raise bonds in the financial sector directly plus they have the added welfare and pension liabilities as part of the Deng-era financial reforms where welfare and pensions were devolved to Provincial and Local Governments.
> but from it's failure there is YMTC
Rewrite of history. YMTC was always a success as a BU, and that's why it was split off from Tsinghua Unigroup because the overall structure of Tsinghua Unigroup was inherently unstable and put Hubei's CapEx at risk as they gave the capital to Tsinghua Unigroup to create YMTC.
The impact is not very different as they grant billions in subsidies which they hope to recoup in tax revenues later. I don't see how this provides incentives that are significantly different except that there is no ownership, which is objectively a worse deal for the local government.
> If Tsinghua Unigroup goes belly-up again or can't meet it's targets as part of the reorg, the Anhui Province is on the hook, as they have the controlling stake in it after the cleanup.
Anhui Province will lose at most the ~5 billion it put into buying a stake of Unigroup. It will not be on hook for outstanding debts as far as I can tell, so I don't understand how this arrangement is any worse for Anhui than giving 5 billion in subsidies and tax credits as a US state would - for example Ohio provided an additional 2 billion in funding for the TSMC fab, NY around 10 billion, etc...
> Rewrite of history. YMTC was always a success as a BU, and that's why it was split off from Tsinghua Unigroup because the overall structure of Tsinghua Unigroup was inherently unstable and put Hubei's CapEx at risk as they gave the capital to Tsinghua Unigroup to create YMTC.
This doesn't change my point that the investment into Unigroup directly led to YMTC which is a massive success. I'm not saying that Unigroup is responsible for YMTC's success, I'm saying that investments into Unigroup - in this case including Hubei's investment - had good outcomes even if Unigroup as an entity failed.
Yes, but it's not a CapEx investment (as in upfront capital taken out of a treasury and invested), and is subject to public notices so there at least is an auditable trail. And unlike China, municipalities and States in the US can directly raise capital via bonds.
> so I don't understand how this arrangement is any worse for Anhui than giving 5 billion in subsidies and tax credits as a US state would
Because that stake in one venture is 10% of Anhui Province's entire Revenue in 2022 - Anhui's total revenue was only $49B in 2022 but it's expenditures were $115B in 2022 [0]. And that's just one venture.
A provincial government like Anhui has invested in dozens (if not hundreds) of large scale ventures such as the legacy automaker JAC Group and Volkswagen China. While these amounts might not necessarily at the same amount as SMIC (excluding JAC Group and Volkswagen China), they are still fairly significant. And Anhui is a middle of the pack government in China - all provinces (as well as the local governments within provinces) themselves have actual CapEx on the line in ventures.
This is a significant risk, as local and provincial governments also have a duty and requirement to provide public services, and ventures not doing well can have an impact on the financial health of provincial and local governments, yet they still have the developmental indicators of Ecuador or Cuba. Spend those billions actually alleviating the urban-rural gap instead of acting as a wealth creation mechanism for much richer Beijing, Tianjin, and Shanghai, where most of Tsinghua Unigroup's CapEx is spent.
> This doesn't change my point that the investment into Unigroup directly led to YMTC which is a massive success
It does though. Even though YMTC was under the Tsinghua Unigroup umbrella, it's primary capital came from a separate government and remained autonomous of Tsinghua Unigroup, and Unigroup's larger failures impacted actual deliveries and roadmap items for YMTC [1]
[0] - https://data.stats.gov.cn/english/easyquery.htm?cn=E0103
[1] - https://asia.nikkei.com/Business/China-tech/China-s-Tsinghua...
This is also true for Anhui. Out of the ~5 billion about 2.6 billion was in direct funding, the rest is in debt/equity swaps which aren't going to be upfront costs.
Similarly in the US, for this kind of project you can expect about half of the funding to be a direct grant and the other half to be deferred.
> Because that stake in one venture is 10% of Anhui Province's entire Revenue in 2022 - Anhui's total revenue was only $49B in 2022 but it's expenditures were $115B in 2022 [0]. And that's just one venture.
Ohio's situation is similar, with ~8% of yearly revenue spent on that one single TSMC building. Ohio's direct revenue is only around 25B/yr, with the vast majority of the budget being funded by the federal government (mostly pass through, for example Medicare)
> It does though. Even though YMTC was under the Tsinghua Unigroup umbrella, it's primary capital came from a separate government and remained autonomous of Tsinghua Unigroup, and Unigroup's larger failures impacted actual deliveries and roadmap items for YMTC [1]
Again, the argument from the start was about outcomes in funding for the industry. I have not argued anywhere that Unigroup's leadership deserves any credit, just that some of the funding allocated to Unigroup - chiefly the one earmarked for YMTC - ended up with decent outcomes.
Those are still an upfront cost on Anhui's treasury as it is counted as a liability, and a liability with limited ability to service due to the relatively weak municipal and provincial bonds market due to the ongoing LGFV crisis.
> Ohio's situation is similar, with ~8% of yearly revenue spent on that one single TSMC [Intel, not TSMC - good catch selimthegrim] building. Ohio's direct revenue is only around 25B/yr, with the vast majority of the budget being funded by the federal government (mostly pass through, for example Medicare)
Ohio only gave $0.6B - ie. 2.4% [0] (and even that was controversial [1]). The rest of the $7.4B came from the CHIPS Act.
Furthermore Ohio has a AAA credit rating score [2] meaning it can borrow at low-to-no interest, which doesn't really exist as an option at scale in most Chinese provinces.
And finally, this investment by Ohio generated jobs within Ohio. The Tsinghua Unigroup bag-holding isn't allocating Capex for Anhui, as most of Tsinghua Unigroup's assets are not in Anhui.
That said, a capex-to-jobs case could be made for JAC Group and Volkswagen China which have Anhui government ownership stakes, but then again they themselves are losing marketshare to BYD - a privately funded company - like just about every other car manufacturer in China (the majority of whom are SOEs or have an ownership stake from Provinces or Central Ministries).
It's not so say American states haven't made similar mistakes before (eg. the NY Corruption indictments following the Buffalo Billions Scandal), but the fact that Buffalo Billions was news is itself a major deal - corruption and misallocation of capital within the BigFund and SOEs is sadly the norm in China, and everyday I can see CCDI arresting yet another person, while ignoring others until they piss off the wrong guy (and occasionally, even CCDI themselves are found to be corrupt).
> Again, the argument from the start was about outcomes in funding for the industry. I have not argued anywhere that Unigroup's leadership deserves any credit, just that some of the funding allocated to Unigroup - chiefly the one earmarked for YMTC - ended up with decent outcomes
And my argument is that this leads to a "to big to fail" situation which is extremely risky at the provincial and local level because of the lack of fiscal fallback options for local and provincial governments in China as well as the fact that all that capital could have been better spent by provinces to uplift their population's living standards instead of essentially acting as a wealth transfer to much richer coastal provinces or the 3 provincial level cities.
Tsinghua Unigroup is not a one-off example of this risk, plenty of similar crises and failures have happened in recently in China. And Anhui is not the only Chinese province faced with this situation - almost all are (except Guangdong, as usual).
Those tens of billions Anhui has spent not just on Tsinghua Unigroup in 2022, but the dozens of other similar ventures like JAC Group, Volkswagen China, etc could have been better spent on building it's human capital. For a government that should be the primary "decent outcome".
It's a middle-of-the-pack province in China with developmental indicators comparable to Ecuador, Cuba, and Peru and well behind Thailand's poorest region (Isaan). Think about how many more cars JAC could sell, how many more electric toilets RSD Group could sell, and how much more chicken 老乡鸡 could sell if Anhui's ytd median disposable income per capita was greater than $4,100 [3], and if Anhui's rural median disposable income per capita was greater than $2,500 [4].
You can argue purchasing power all you want but more of the high value goods that a company like Tsinghua Unigroup, JAC Group, etc are producing and intending to sell are at price points that are unaffordable at that level of disposable income.
By every standard it is a misallocation of capital, and sadly a very common one across China.
[0] - https://www.policymattersohio.org/files/assets/odod-intelons...
[1] - https://www.dispatch.com/story/news/2022/07/28/chips-act-fou...
[2] - https://en.wikipedia.org/wiki/List_of_U.S._states_by_credit_...
[3] - https://www.ceicdata.com/en/china/income-per-capita/disposab...
[4] - https://www.ceicdata.com/en/china/disposable-income-per-capi...
>Province in China with developmental indicators comparable to Ecuador, Cuba, and Peru
Disproportionate resources was concentrated to built Hefei, Anhui, new tier1 city with ~10 million people and per capita income of ~$20,000 build off 2600% growth in last 20 years because local gov focused on making it tech/innovation centre. The one elevated by Western Development Strategy, that built out industrial clusters and R&D hotspots like USTC to raise and retain human capital. The entire point is _UNEQUAL_ development so a few 100 millions of disproportionately affluent / educated urbanites with $20000+ income can buy high value goods that would have no market if everyone was elevated to 10000 per capita doing less advanced work, they already sell enough 老乡鸡. Still waiting on OLED panels from Ecuador, Cuba and Peru.
E:
And as I've responded in the past, you don't get to decide what I reply to. This is the internet, you are not the only audience in a conversation. I make effort to ignore you posts, and have, but when it needs to be called out on, I will. And again, my custom HN CSS doesn't even display screen names, but your identifiable writing pattern and questionable use of stats makes it obvious.
Questionable in this case, as in the past, you're using selective stats and omitting useful context to insinuate Anhui is somehow cuba tier shithole (no offense to fellow Cubans) who failed to develop / exploit human capita. An claim that can only be made by dancing around the development AND mere existence of Hefei, Anhui, one of PRC's innovation hubs that specifically and successfuly developed on the back of building human capital and high tech, which makes comparisons to Ecuador, Cuba and Peru even more perplexing. One simply doesn't drop stats about Anhui without knowing Hefei. You are not a stupid person. Your omissions in these arguments can only be interpreted as intentional by me, and while I try to avoid interacting with you (rather comments that reads like yours), I'm still going to call out bad analysis/arguments. I will edit original response for tone and stop relying here. But I encourage other readers to look up Hefei and how it grew in last 20 years, and understand that is the deliberate result of industrial policy (not just semi, but high tech).
I am having a discussion with GP, and while GP and I may disagree, we are still maintaining a civil tone with each other.
This has been a common behavior of yours. If there was block functionality or reporting functionality on HN, I would have done it in a heartbeat.
As I have told you before, stop replying to me. We do not agree and we have had heated conversations at multiple occasions.
----------
> This is a lot of words and numbers about Anhui to skirt around the existence of Hefei
A city with an urban population of 5 million people - so only 10% of Anhui's population.
What about the other 56 million who do not live in that urban core?
If I am to use your metric, then we should treat Turkiye, Brazil, Mexico, and Thailand as developed first world countries.
> The entire point is _UNEQUAL_ development so a few 100 millions of disproportionately affluent / educated urbanites with $20000+ income can buy high value goods that would have no market if everyone was elevated to 10000 per capita doing less advanced work
Yet the median disposable income per capita of urban households in all of Anhui is $6,500 a year [0].
> The entire point is _UNEQUAL_ development
This is completely contrary to Xi's policy of Common Prosperity (something which I agree with him about, but disagree about his implementation).
A first world or "developed" country is not only a country with skyscrapers or EVs. It is a country where the bottom 10% and the top 10% can drink tap water, have similar life expectancies, similar educational indicators, have private toilets, etc.
> This is a lot of words and numbers
You mean, I shouldn't make a thesis and argument based on quantitative as well as qualitative data?
I can make an argument based on vibes and still make it solid?
/s
[0] - https://www.ceicdata.com/en/china/disposable-income-per-capi...
Like rest of rural PRC, they can wait for the trickle down common prosperity. When it comes.
>disposable income per capita
That's dispoable per capita, double more for household. Move up to top 1-2 quantiles, and you have reasonble market for high end goods. And you know it's dispoable income.
>Common Prosperity
It is, but it's also reality, i.e. Xi not aiming a welfare state. But that's also secondary to national security issues like hammering semi conductor even if it means some provinces gets burnt. Hence also CCDI kick doors and take.
Xi's timeline for developed country is like 2049+, even then it's fully developed / modestly prosperous. In mean tie still has to balance common prosperity with strategic autonomy, and if that means farmers and informal workers have to take a back seat like they did under previous admins then so be it. Best he can do is throw them a few more bones. But not at the expense of growing "comprehensive power". A strong country can build high end tech to compete with other strong countries, even if parts of it developed to be strong.
>tap water
Yes, yes and developed countries definitely wouldn't neglect places like flint or first nations communities. I guess not so 1st world after all. Which is getting off topic, the point is planners need to plan for plan, somethings gets prioritized over others things, i.e. this is a thread about semi industrial policy, not poverty alleviation. Yes resource is limited, pouring money in chips can come at expense of common prospertiy. But there's a geopolitical competition happening, and if the competitor is not selling you chips, you need to prioritize making your own, even if at expense of building a lot of water pipes, for hollowing non productive villages.
> I am having a discussion with GP, and while GP and I may disagree, we are still maintaining a civil tone with each other.
> This has been a common behavior of yours. If there was block functionality or reporting functionality on HN, I would have done it in a heartbeat.
> As I have told you before, stop replying to me. We do not agree and we have had heated conversations at multiple occasions.
Your original post pre-edit had self-censored profanity. Your desire to not receive replies from a specific user seems entirely hostile to the HN guidelines and its assumption of continually exercising good faith, especially if you feel that replies to you lack it. If you have a problem with a user, notify HN mods directly. I don't see how you expect others to keep track of when you decree that they aren't worthy of your replies, or why you would subject the rest of HN to the degradation of public debate that such a block feature would result in on HN, but respectful replies should be the norm and the goal all the same in either case.
If you don't want to engage someone in conversation or debate, then don't reply to them. To seek to preemptively forestall debate before it even begins via blocking a two-way dialogue is contrary to entering into such debate by making your own first post to begin with. If the only winning move is not to play, why insist on having the first and/or last move/word?
The same is true for tax credits and indirect subsidies.
> Furthermore Ohio has a AAA credit rating score [2] meaning it can borrow at low-to-no interest, which doesn't really exist as an option at scale in most Chinese provinces.
So did Unigroup when most of it's debt was issued and now assigned to Anhui's investment vehicle, so in fact for this transaction they may as well have a AAA rating.
> Ohio only gave $0.6B - ie. 2.4% [0] (and even that was controversial [1]). The rest of the $7.4B came from the CHIPS Act.
Ohio really did earmark 2 billion dollars for the fab, not 600 million. The existence of one grant for 600 million is not evidence against the other 1.4.
https://www2.deloitte.com/us/en/pages/tax/articles/chips-act...
> For example, Ohio has publicly announced more than $2 billion of state grants, credits, and incentives to construct a new semiconductor manufacturing plant in Ohio.
>> And finally, this investment by Ohio generated jobs within Ohio. The Tsinghua Unigroup bag-holding isn't allocating Capex for Anhui, as most of Tsinghua Unigroup's assets are not in Anhui.
> And my argument is that this leads to a "to big to fail" situation which is extremely risky at the provincial and local level because of the lack of fiscal fallback options for local and provincial governments in China as well as the fact that all that capital could have been better spent by provinces to uplift their population's living standards instead of essentially acting as a wealth transfer to much richer coastal provinces or the 3 provincial level cities.
They are not too big to fail in the sense the word is commonly used. If they fail, the investment vehicle is bankrupt and Anhui doesn't incur any liabilities (remember that they can't issue bonds).
Sure, and if Intel eventually leaves before Ohio makes money from the jobs, nothing is seen - this happens often. On the other hand if Unigroup is succesful, Anhui has equity.
> Those tens of billions Anhui has spent not just on Tsinghua Unigroup in 2022, but the dozens of other similar ventures like JAC Group, Volkswagen China, etc could have been better spent on building it's human capital. For a government that should be the primary "decent outcome".
The primary outcome of semiconductor investment is to prevent an external dependency that is already being exploited to cause a general slowdown on the economy which will affect everyone.
> It's a middle-of-the-pack province in China with developmental indicators comparable to Ecuador, Cuba, and Peru and well behind Thailand's poorest region (Isaan). Think about how many more cars JAC could sell, how many more electric toilets RSD Group could sell, and how much more chicken 老乡鸡 could sell if Anhui's ytd median disposable income per capita was greater than $4,100 [3], and if Anhui's rural median disposable income per capita was greater than $2,500 [4].
No matter what speeches Chinese officials make, the truth is that China is not and cannot at this stage execute equal development in economic terms. The absolute medium-term priority of the Chinese government is to create high quality development in key cities in order to move up the value chain before it is too late. In the meantime, redistribution is the most efficient way to deal with less productive regions - it is simply not viable to invest in rural communities at the current stage of Chinese development. China is not a unique countries, many other countries have tried either strategy and it turns out that investment in less productive localities in the name of equality simply does not work for a dozen different reasons.
If your understanding of either the CHIPS act or the Big Funds is that they are vehicles trying to uplift the median standard of life, there is no point having this discussion. It just isn't the purpose. The purpose is strategic, and cannot be understood using the same metrics as, say, countercyclical investment into medium businesses after a recession.
Besides their core business is profitable and they are publicly traded in Hong Kong.
The Big Fund is CHIPS Act scale - if not bigger, as it's 3 generations of funds and has reached almost $100B over a decade.
The $250B is the total capital from CHIPS and private investment. The other ~$200B is coming from private sector investments.
This is why the Big Fund is so big - it's been tough to raise private capital in China after the 2015-16 market crash so governments (central, provincial, and local) have had to step in to provide capital.
And if we're honest, the truly successful Chinese companies at the global scale (BYD, ByteDance, Tencent, PDD/Temu, Shein) always took private capital, and as such were able to outcompete state funded alternatives.
An anti-intellectual campaign in the late 1960s [1] that eliminated almost all university education, except for in engineering (as engineering served the Great Leap Forward's goal of increasing agricultural and industrial production) ? Combined with a deliberate purge of the previous ruling class?
Thus creating the conditions for the country to be led by Jiang Zemin (electrical engineering, National Chiao Tung University) then by Hu Jintao (hydropower engineering, Tsinghua University) then by Xi Jinping (chemical engineering, Tsinghua University) ?
It took VHSIC, Sematech, and a bunch of other initiatives to prevent the Japanese from collapsing the US semiconductor industry at the time. The subsidies from MITI almost crushed the US semiconductor companies.
Thsoe who fail to study history, etc. etc.
At the end of this, decades from now. Will it have been worth it? Will the US have created a bigger economic loss in China than the value of their own lost business (whether directly or indirectly by the local alternatives these policies foster)?
So far, the US has done massive damage, that is for sure, but ultimately, I am not sure how the tally will be.
Whether the policy is "There should be no barriers when trading with China" or it is "There should be many barriers when trading with China" - either way, decades from now. Will it have been worth it?
People are calling this war and it isn’t. It’s competition in markets and strategy to facilitate gaining marketshare in different markets.
And lo and behold, everyone wants to protect their own access to their own market.
Not everything is a war. I can tell you right now, with the new administration will come new policies that will have a segment of people in the EU and UK saying “economic war”. But it’s no more war than immigration quotas are “war”.
Let’s reserve that word for REAL war.
https://www.wired.com/story/jake-sullivan-china-tech-profile...
The American Who Waged a Tech War on China
China is racing to unseat the United States as the world’s technological superpower. Not if Jake Sullivan can help it.