In addition to geopolitics, wouldn't overproduction of chips be beneficial to consumers in the form of cheaper chips and greater variety?
It's a great time to be a semi manufacturer: TSMC has customers like Apple funding their expansion, Intel has the U.S. government funding theirs. Can you blame them? A decade of record profits isn't what it used to be...
It has pictures!
https://upload.wikimedia.org/wikipedia/commons/9/9e/F15_bulk...
https://about.usps.com/who-we-are/financials/annual-reports/...
With a vote Congress destroys budgets and the narrative is the PO sucks.
Government incompetence is private interest driven propaganda
As individuals, Congress is doing fine by its members
Thanks for letting me know I can ignore everything you wrote after that idiotic statement.
The government has multiple branches... They're the legislative branch.
If we're going by your logic, the executive branch is the one that is not part of the government because they don't get to decide how to govern.
https://www.whitehouse.gov/about-the-white-house/our-governm...
https://www.usa.gov/branches-of-government
https://people.howstuffworks.com/three-branches-government.h...
> All but a few are millionaires who make more income from non-government sources than that paycheck. By virtue of their office, they're also well connecte
All of this is true for the executive branch too.
You can argue that govt need not be incompetent, but it's foolish to act as if the US govt is competent until it actually is competent.
That's true regardless of how good things could be if the US govt was competent.
"Hmm, I wonder who that sign is for..."
https://archive.ph/kvlnp / https://www.bloomberg.com/news/articles/2022-03-28/tsmc-sams...
It started with getting TSMC to build a fab in Arizona, but to be fair it's a small fab (25k wafers/month) and 5nm will not be leading-edge by the time it's operational. Presumably the point of that fab is traceability/domestic sourcing for the defense/intelligence community - I'd expect that to mostly be used for products to be sold to the USG, and that's how it's sized, just what they'll need for defense. But this new initiative looks like a sea change to me - both in US chipmaking and foreign policy.
I think the situation in Ukraine with neon supply (used in older processes, like the automotive chips that are already in shortage...) has finally struck the fear of god into US policymakers. It's increasingly clear that the supply chain is dangerously single-source in many many areas (EUV and high-end DUV litho machines, fine chemicals and high-end passives, ABF film, etc), and any problem in any area of the world can have huge implications, because the "build one part in every country" strategy means conflicts are going to disrupt something. You reduce the risk of a black-swan event taking out the whole industry at once, but you also increase the risk of at least some disruption happening, and at this point any disruption is catastrophic in itself.
But beyond that, China is taking advance of the world being distracted by russia to rattle the saber with TSMC again and that's ultimately just not a tenable situation. Not just on China's part either, Taiwan is working the situation too and that's a problem too.
Not only could china blow up the world's chip supply by starting a war with Taiwan, but TSMC has that leverage at any time too. A fab is a tuned swiss watch, even simply wiping all the computers used to run it (or getting cryptolockered like happened a year or two ago) or opening the fab up and allowing it to be contaminated would basically halt production for a period of at least months, and that's practically trivial to do. Actively destroying certain machines could put the fab out of action for years, and blowing up a couple EUV machines is easily within reach of nationalists/partisans at TSMC if it came to it. Even beyond physically having the fabs on US soil, that means that TSMC and Samsung (and thus their respective governments) have leverage over US policymakers in terms of US national foreign policy - and letting foreign corporations have leverage over US foreign policy simply is not an acceptable situation. It's been a boiling-the-frog situation getting here but China rattling the saber over Taiwan while Russia is invading Ukraine is beyond the redline and the situation can't be allowed to persist.
Even if the US gets enough TSMC-owned capacity to not be completely dependent on Taiwan itself - what if China invades Taiwan and the US decides to allow Taiwan to fall? TSMC could destroy (or seriously damage) their US fabs in retaliation and still blow up the US (and world) economy. And the only other cutting edge fabs in the world are Samsung and Intel... and Samsung is sitting under the barrels of tens of thousands of North Korean artillery pieces, and now they've got nukes and ICBMs too. They have older fabs in Texas but all their cutting-edge stuff is in South Korea. It's not a problem today, but these are strategic projects with a 20 year spin-up time (given all the associated industries that need to be replicated) and foreign policy issues can easily crop up within a timespan of a year or less. What happens if North Korea gets antsy and Samsung doesn't think the US is committed enough to the defense? Samsung could basically do the same thing as TSMC if they really wanted too, and now you've got two quasi-state enterprises with leverage over policymakers.
Global Foundries, too, is owned by an Abu Dhabi sovereign investment fund, they're not cutting edge, but they're yet another potential state-level player.
Fabs are strategic infrastructure and letting foreign nationals control delicate, easily-destroyed strategic infrastructure simply isn't acceptable anymore, period. And the US is explicitly and specifically choosing to disarm the Sword of Damocles that Taiwan has used to exert control over US foreign policy, going forward TSMC won't be able to use the fabs as leverage - which obviously has huge implications for Taiwan's foreign policy, this really calls into question the long-term commitment of the US to the defense of Taiwan. I don't think we're gonna abandon them tomorrow by any means, but in 10 years there is going to be a different balance of power. Both of those are huge foreign-policy sea changes.
But yeah, the idea of "the end of history" and that globalization means that nobody is ever going to get into any conflicts ever again because "gosh, we're all just so economically intertwined" is dead - it's obvious that conflicts will happen and the intertwining just means that we have tons of economic problems when it does. That's another sea change. It's not going to stop neoliberals from doing it anyway, corporations will still lobby for deregulation and offshoring and tax cuts because it makes them money and neoliberals will generally oblige, but this is a signal that policymakers are starting to be aware that they can't just handwave and "the free market will take care" of strategic industries with massive foundational implications to downstream products and the industry as a whole. Globalization doesn't mean strategic industries aren't strategic.
[0] TSMC and Samsung are both quasi-state enterprises. TSMC is explicitly intertwined with Taiwan's foreign policy, they were the ones to negotiate for vaccines for taiwan for example, and Taiwan recently used the threat of cutting off TSMC's water (during shortages) as leverage in negotiations/etc. Samsung is a chaebol, one of the megaconglomerates that essentially run Korea's government shadowrun-style (this is how they industrialized from basically nothing to cutting-edge in 40 years, intense and tight corporate-state coordination). And this isn't just US-style "corporations donate to everyone", Korea's presidents have been repeatedly convicted of corruption for benefiting Samsung, and other actions are almost certainly happening behind the scenes that don't see the light of day on newspapers or can't be proven with hard data. Both Samsung and TSMC are either instruments (TSMC) or puppeteers (Samsung) of their states' foreign policies and cannot just be viewed as big corporations, they are players on the international stage in themselves.
* cutting / bleeding edge (as now)
* 80/20 trailing edge (5 year LTS of cutting edge)
* Cost focused 'lifetime' LTS
Maybe that's where some industry development and co-ordination from regulators could help establish a market that's healthy.
Ex: brand new 65nm fab:
https://www.electronicsweekly.com/news/business/bosch-opens-...
So, 10-15 year ROI easy, and I'm guessing that might stretch even further as the trend to ever smaller nodes slows. AKA I'm betting 2nm doesn't look nearly as aged in 2040 as 90nm looks today.
says who? the world's rivers and oceans are polluted with micro-plastics and heavy metals from E*Waste, while thirty years of "wild west" profits and dog-eat-dog markets, COMDEX and single-use devices will take 1000 years to degrade in Nature.
Calls for "more" to be viewed with earned-suspicion IMO
But, why not?
Imagine if video games' audio/video just stopped improving past the SNES. Imagine if we decided that 56k dialup was good enough. Imagine if we stopped improving cell phone cameras past the iPhone 3G. These things seem pretty low quality now because we've incrementally built so many better things in the interim. The unleaded gasoline of today is going to be as absurd as the leaded gasoline of yesterday within many our lifetimes. But we can't make big leaps without the interim nudges forward - the N64, the DOCSIS cable modem, modern phone cameras, hybrid vehicles. To do those, we need better microchips.
We also need more of them. Just because the richest areas of the world have theirs doesn't mean the less-well-off areas need to be shut out because they didn't get in on time.
No because that is a made up quote.
As someone whose daily drivers consist of decade-old flagship hardware, I don't need to. Outside software support, the main downside of my phone is battery. Maybe I should get a new one - it's from an era where these were still user-replaceable.
My similarly aged laptop? Same thing. I could upgrade the hardware; the small increase in compute power isn't worth the reduction in connectivity (ports) though.
Sure, some will have specific needs. But for the vast majority that doesn't need a 96 GB desktop or a 10K screen, flagship HW from 10 years ago is more than good enough. Just as for your examples where we need better chips: that was true 30 years ago. Nowadays, a smaller nanometer scale than the last 2 cycles probably matters far less than a better chip design for most use cases.
Such as the massive engineered screen addictions pandemic that afflicts most of the world today, especially the younger generation that never knew any other way.
Or the mountains, rivers and oceans of e-waste. Or decline of physical and mental health from lack of natural physical activity.
There are benefits of course, but it's wise to keep a balanced view on the pros and the cons as well.
A better world for people means education and healthcare, not a 1mm slimmer disposable phone.
More frameworks, more layers of abstraction and no confining force that forces anyone to write software that isn't a bloated mess.
DRAM was a boom and bust market and during bust/oversupply windows vendors that under invested in the next node technology got washed out when the next boom cycle happened. When the market is high volume, price per bit, and new nodes are every 2 or 3 years, you can't really compete it your are nearly 2x as expensive (30% linear improvement => 70% squared cost per bit).
Of course you need to survive the dry times, but you make maximum margin when you are on the next fab node and your competition isn't.
I think what you are saying makes more sense when applied to contract manufacturers than all semiconductor companies.
The vast majority of variety in product lines comes from who is ordering the chips, not the foundries where they are made. Adding more foundries will not introduce any more different chips, just chips that have subtly different performance or reliability, which is already a nightmare engineers deal with continuously.
China sells the cheapest chips, and companies only buy the cheapest chips. Making more foundries in the EU and Europe will not change that fact. China will have the huge majority of demand. If anything, a reduction in demand will give more power to counterfeiters and raise the floor price you have to pay to avoid them. It won't really impact prices at all (counterfeit chips are almost nonexistent for anything costing more than a couple cents) but it will make life significantly more irritating for people like me.
it depend on the craft smith's skill. everyone have access to Python but some is able to craft a robust app while others app is buggy and unusable.
TSMC consistently have the best yields and quality chips
Expertise and facility time are the limiting factors. Low volume purpose built SOCs will never be cheap (outside chiplet-based designs) because the ability to fill out a production line is critical to amortizing the immense startup cost of a design.
It is already possible to have completely custom chips built in lots of <100 on older manufacturing nodes. Runs cost low five figures to low millions of dollars. Some college courses involve having an entire chip made. Anyone can order one on 20-30 year old equipment. There is no demand and it will never get significantly cheaper aside from mass-produced chiplet ICs.
The West also wants reliable access to chips _despite_ war with China.
Another reason why TSMC would be against this