Losing out on a big percentage of advanced chip making capacity could be catastrophic for the US economy and technological advancement. And military.
This is not an "investment" in future tax revenue via jobs, although jobs are nice.
Because I personally believe that this is a decent deal; I hate the idea and it still reeks of protectionism to me, but I think this serves US (taxpayers!) interests decently well.
I especially like the investment matching aspect, where the necessary commitment from the company is much larger than the grants, so in the end, they are dropping money into that fab (fudging numbers only gets you so far), and its in their best interest to make it work out.
I'd be very careful with characterising this as "wealth transfer", because it seems significantly less questionable than most defense spending to me and even that I wouldn't label as "wealth transfer".
Out of personal curiosity: Are you similarly opposed to agriculture subsidies, or do you consider those less problematic (magnitude is somewhat comparable at ~20billion$/year)?
European here, we could and should close our borders for agricultural goods that we can grow ourselves (and you Americans should ban alfalfa exports from California, while you're at it). Ag subsidies are only required because there's no way an US or European farmer can compete with Russian, Ukrainian, South American or African labor and land costs, but we still need farmers to ensure we're self-sufficient and don't end up like many developing nations did once Russia blockaded Ukrainian grain exports.
Just consider the Polish protests against Ukrainian grain, or the recent German protests against cutting fuel subsidies for agriculture. It just costs too much political capital to do anything about this...
The same applies to so many other types of agriculture. Nobody wants to let farmers go out of business for decades now in the Western world. And the rate of "natural attrition" hasn't matched the rate of technological efficiency improvements.
There is of course agriculture that isn't as efficient because higher human labor required for things like picking and sorting. But the automation for that is slowly coming.
For certain agricultural goods perhaps, but the Midwest is the most productive region for agriculture on the planet. Russia and Ukraine have vast areas of land, they focus on wheat because they mainly have marginal land and a limited growing season and wheat is perfect for that. Most of Africa and much of South America (Argentina being a notable exception) have terrible soil and require significantly more fertilizer to grow crops at scale. Labor is cheaper in those regions but that’s only one factor and it’s why so much farm labor has been automated in the US.
but they are the single largest exporter of grain overall, mostly corn from the midwest and California produce. the Green Revolution hit hard, and most farms are now huge consolidated enterprises. by comparison, in a lot of the world, most farms are smaller and often single family, and in many cases still using old world techniques -- e.g. non-trivial numbers of people plowing their fields with oxen, etc.
I personally think that its quite difficult to make protectionism happen in a way where you don't prop up an inefficent industry with continuous infusions of tax dollars in the end...
https://www.datacenterdynamics.com/en/news/ibm-sues-globalfo...
Another solution: Move the manufacturing to a near-shore and friendly country (ie: Mexico or Puerto Rico) and keep the design part in the US. But of course, that wouldn’t be as politically popular and wouldn’t create the extraction opportunity (which lots of companies will want to benefit from).
You think people will stop buying NVidia GPUs, AMD GPUs, PS5, XBox, Nintendo Switch, iPhone, Pixels, Samsung Galaxies, and a litany of major automotive parts from Microchip/STMicro/etc. etc.?
Because none of these are made in the USA. The only major chips made here are Flash/RAM by Crucial, CPUs by Intel, and some lower-tech (but still important) chips by TI. The vast majority of chips, well over 60% of the worldwide supply, is made by TSMC IIRC, and another huge chunk are made by other companies located in Taiwan.
--------------
Furthermore, Taiwan is a close ally who has extremely friendly relations with USA. Yes, there's a war-risk with China here, but Taiwan fairly competes with us. I don't think anyone has a single piece of anti-Taiwan hate inside of them.
Its not a protectionist thing. Its purely the "what if China attacks Taiwan" thing. Economically speaking, trade with Taiwan and USA is a good thing. Geopolitically, trade with Democracies is a good thing, and trade with a century+ old friendly nation is a good thing.
-------------
Taiwan can create this tech because they are smarter than us and outcompeted us on this subject. Its not "low tech", its extremely difficult chemistry, high-end PH.d stuff here.
Taiwan is basically the Vulkans from Star Trek. They're smaller, more technologically advanced, but weaker militarily. They're so obviously good allies that its unthinkable to do anything against them. We need a "just in case" plan if trade between our nations gets interrupted because chips are a strategic good, but... cutting them off would mean that they ally with someone else and that's even worse.
TI makes plenty of pretty high-tech chips, just not necessarily bleeding edge digital logic processors. Their real high-tech stuff are analog devices, something far less sexy to people on HN but still incredibly crucial.
Yeah, important but not as much of a headline as 20% more GHz or FLOPs for some reason.
It’s not that they are intrinsically superior Vulkans as you’ve described.
It’s a temporary distortion caused by Intel trying to rest on their laurels and drip feed advancements.
Even if we magically wave a wand and give Intel next-generation lithography today, they still don't have the capabilities to take orders from customers and translate them to designs.
-----------
Its not Intel who needs to catch up to TSMC. Its a 3rd party fab like GloFo who needs to catch up for USA to recapture that market share.
I'm happy that GloFo is getting some scraps and investment from this CHIPS act, its the right move. But I have my doubts that we'll reach the technological level of TSMC / Taiwan any time soon. But whatever we get its still better than nothing.
ASML began to ship its first High-NA lithography tool to Intel late last year ,and the machine will be fully assembled in Oregon in the coming months. [1]
[1] https://www.anandtech.com/show/21264/asml-to-ship-multiple-h...
And yet, it's Taiwan who won. We gotta give them props for that, and we can also thank the gods that it was a close ally of the USA who got so good at it.
In terms of nearshoring - that’s a good idea too but I think the workforce training would be even more difficult than it has been in the US and it would be less popular politically as you said. (Side note: PR is part of the US - maybe you meant Costa Rica)
Puerto Rico isn't a foreign country, it's a territory of the United States. Regardless of that, semiconductor fabrication is one of, if not THE most complex form of manufacturing.
It's not enough to simply build a fab in Mexico, the supply chain of materials and highly skilled labor that runs these fabs needs to exist there too. That's not to say that it can't exist somewhere like Mexico, but currently it doesn't and it would take longer to build up.
So while moving it to Mexico would remove the geopolitical risk problem, putting it there instead of the US would only give the advantage of cheaper labor but at the cost of taking much much longer to ramp up.
The big one everyone remembers is the set of grants giving to incumbent providers to expand access to broadband (25 Mbps down/3 Mbps minimum) to everyone in the country - a promise that, despite three rounds of grants in the billions, has not materialized.[1]
[1] https://www.jsonline.com/in-depth/news/2021/07/14/weve-spent...
The goal is not wealth transfer.
If you want that, move to Europe.
So what you would need to ask would be more like: Is the subsidiy higher than it needs to be to lure chipmakers into the US? And that is much harder to answer...
There are other economic benefits too.
But even if your simple calculation was correct and all there was to this story, what you’re missing in your post is that America as a nation has decided we need to produce semiconductors here for economic and national security purposes and so therefore subsidizing their manufacture is a normal thing that would occur.
> America as a nation has decided we need to produce semiconductors here for economic and national security purposes
a few decades ago we decided to move manufacturing jobs overseas because paying people who had families and bought "cars and lattes" here was just too expensive. Why reneg on that commitment?
Because it was dumb, changing away from a path when you see it is a dead end is a good thing to do.
Also I’m not sure “we” (to the extent that there even is an “we” here, there are always winners and losers with these sorts of policies so I bet the whole country didn’t agree) ever intended to move high-tech manufacturing away from the US. I think we want to outsource the bad jobs and keep the good ones. Of course it is a game everybody is playing, we’re shifting our strategy to hopefully more effectively keep the high-tech stuff.
If you want a more visible and recent datapoint, look to the perennial fights in the UN over the opposite, where a country that promised not to devalue their currency devalued their currency. Note that these discussions generally start from a point where everyone understands why it might be desirable to devalue one's currency, pumping exports & the real economy while dumping assets. This is macroeconomics 101 stuff.
The surprising thing is not that high tech manufacturing left, it's that for a while it was actually able to compete with a money printer for access to talent and resources.
Until the recent turnaround, it was understood the west was moving to a service industry. Someone Else(tm) (read: China) would manufacture everything (the "dirty", "cheap" work) and we would provide the service (the "clean", "expensive" work).
Of course, an economy can't be valuable if it can't actually make anything. As it stands, the US can't even make chopsticks by itself let alone microprocessors.
I actually have significant doubt CHIPS will be successful in bringing back microprocessor manufacturing, and more broadly bringing back manufacturing in general to the west at all, but we have to start somewhere if we don't want to be beholden.
Anyway, US manufacturing output still seems to be second in the world, around 1/2 that of China with 1/4 the population. Microprocessors are already made in the US, Intel (despite their struggles) is still making plenty of chips here.
I’m not really sure what the plan is with Global Foundries, in the sense that they presumably won’t keep up with TSMC or even Intel… but maybe 7nm will be the next or next-next generation automotive chip node or something.
This is what the west was trying to do before: Providing services like microprocessor design for the manufacturing industries overseas.
But of course, simply drawing stuff on paper for others to peruse doesn't actually produce value. It's the act of making something that produces value, and we finally realized that grave error.
Not to mention the people doing the manufacturing work will inevitably accrue the know-how to also just do the designing themselves too, and then we in the service industry really are worthless fools.
>I’m not sure what you mean by “can’t even make chopsticks by itself,”
I mean that quite literally. I know Georgia (the state, to be clear) has (or had) a robust disposable chopstick manufacturing industry, but nearly every disposable chopstick I see sold or given out are Made In China these days.
If we can't (or won't) even figure out the means to making some wooden sticks, we definitely aren't manufacturing microprocessors.
> But of course, simply drawing stuff on paper for others to peruse doesn't actually produce value.
Intel actually manufactures quite a bit in the US. The bulk of their fabs are here, and they ensure they can produce the wafers with good yields in the US. Once they've got the process down, the other fabs kick in. And even then, I believe a fair amount of the mass production is still in the US. It's the packaging that is mostly offshored.
So no, not just designing on paper.
I'm aware they have fabs (mostly R&D) in Oregon and have manufacturing fabs planned in Arizona, New Mexico, and I recall Ohio, but I've mostly heard praise of their Israel fabs and all the CPUs I've ever bought from them have said some variation of "Made in Costa Rica/Malaysia/Vietnam".
To be fair my latest purchase, a 14700K, says "Manufactured in US with global content." which is a slight improvement, but that could just as well mean "We shipped chips in from Israel and soldered a heat spreader on it." and it still ultimately says "Final assembly in VN." (Vietnam) right under it.
Made in USA this is not (yet?).
https://www.intel.com/content/www/us/en/support/articles/000...
>> I’m not sure what you mean by “can’t even make chopsticks by itself,”
> I mean that quite literally. I know Georgia (the state, to be clear) has (or had) a robust disposable chopstick manufacturing industry, but nearly every disposable chopstick I see sold or given out are Made In China these days.
But these are different things. “Can’t make chopsticks by itself” is clearly not true, I could just go make a pair of chopsticks if I wanted. You can find expensive hand-made chopsticks from the US. The US might not have a huge production of disposable chopsticks, but making really simple things at scale is a place where we don’t have a competitive advantage, and I wouldn’t be surprised if we have a relatively low demand for chopsticks, so this doesn’t seem surprising or indicative of anything in particular.
The point here was that your assumption of the economic effects of the subsidies didn’t include other meaningful factors.
This is about creating strategic independence.
We have national defense concerns and zero faith in intel.
EDIT: This makes it "defense spending" in a civilian line item.
[1]: https://www.intel.com/content/www/us/en/corporate/usa-chipma...
On the other hand they could just not complete what they promised and we would be in a bad position: https://www.theverge.com/2019/4/10/18296793/foxconn-wisconsi...
Others have said about trickle down economics (this can be debated).
Read Chip War by Chris Miller
https://en.m.wikipedia.org/wiki/Chip_War:_The_Fight_for_the_...
The other economic benefit is moving one part of the supply chain here may enable more of the supply chain to be completed here. For some of the hardware companies I work with, if you have to make a few parts in China, it's vastly cheaper and more flexible to do the whole assembly there.
We give giant science grants all the time, tons of money for infrastructure, etc, what makes this one immediately go into X dollars / jobs created * tax generated?
You'd need to split the pay between all 6.7 newly created roles, meaning they only need ~$60k take-home or so each - which is right around the US mean salary.
[1] https://www.oxfordeconomics.com/resource/chipping-in-the-us-...
Also, did you not account for capital costs? I know that at Intel fabs, labor costs are significantly lower than capital costs.
Consider, for example, that ASML charges $0.5B for a single unit.
Maintaining supply chain diversity is critical for national security. The US Military does not want to be dependent on foreign manufacturers for its equipment.
Furthermore the business and employees will be paying sales and property taxes, in addition to federal income tax, and will have 2nd & 3rd order effects from their spending.
Simply diving the subsidy by the 1,500 permanent employees doesn't capture the value that's created, along with the strategic value of bringing more chip manufacturing to the US.
Don't forget that TSMC was essentially a government enterprise when it started.
Be cynical if you want, but the one thing that has a 0% track record of effective results is calling every government function wasteful, corrupt, and untrustworthy and responding by dismantling government institutions and investments.
So assuming the construction jobs are temporary and the manufacturing jobs is what remains, this works out at around $2.066 million per employed personnel. Assuming a federal income tax rate of 20% and a 30 year tenure, this means the personnel should have a take home pay of $344k/year. And this is just for the scheme to break-even.
I think this IRA (the act) and CHIPS act are going to be just another wealth transfer where the "work" being done just to check boxes to be eligible for the money getting sloshed around.