How Semiconductors Ruined East Germany
Why Europe Lost Semiconductors
Europe’s Lost Decades in Semiconductors
It's great to get chip manufacturing back to Germany after all the struggles, even if it's not cutting edge.
I'm def not a fan of the CCP, so I loved watching the videos, but I felt like I was getting brainwashed.
Does anyone know how valid/factual their videos are?
Here's a video about China's Gallium & Germanium Export Controls:
I didn't find it anti-China, neither are his other videos in my opinion.
According to the transcript below Jon Y. is half Taiwanese/Hong Kong, make of that what you will.
https://www.manifold1.com/episodes/jon-y-asianometry-on-semi...
I think a lot of people in the Asian geopolitical theater just really don't like China because they do a lot of crappy things.
But it's not like they're designing them exactly the same way as years ago - some advancements are not strictly related to feature size and IIRC these new-old chips are considerably more power-efficient.
https://pr.tsmc.com/english/news/3049
"with the final investment decision pending confirmation of the level of public funding for this project. The project is planned under the framework of the European Chips Act."
Realpolitk wise they absolutely cannot risk regulatory capture to a few select firms on the west coast of the USA.
Altman himself threatened (for a brief period) to cut off GPT to the EU after Italy rightfully brought up issues with data privacy. Microsoft made him walk that one back real quick - he probably wasn't really aware that he is now one of their employees - but it hopefully got through to Brussels about how these guys want to operate.
We'll also probably get a lot better support for languges which aren't english (especially French, I really love their stubborn nature to stick with their langauge over the anglo one), an entire other scary effect of LLMs is the potential to accelerate language death
Very nice to see.
This has absolutely nothing to do with AI or LLMs. Generally these are automation chips for machinery or vehicles.
"Microsoft made him walk that one back real quick"
This sounds like a weird fan fiction. OpenAI openly negotiates with EU governments. No one walked back anything.
AMD is already there and Intel is building in Lower Saxony. Must be something about the region, perhaps the chemistry since the Erzgebirge has a history in mining. Meisner Porcelain for example beat Venice and replaced good old China (pun intended).
> No one walked back anything.
I'm not sure about that (haven't heard this story).
My impression is that EU sticks to discretion very much, and they aren't free from corruption. The US on the other hand is not known to play nice with matters of law in international affairs (Edit: shouldn't have said "international law").
Taiwan is a partner was the point, I believe. Of course they aren't going to give away the crown jewels either.
They will still have the same problem, but they don't think that far ahead.
With chip makers, it is different, because chips can be used in lots of different products and there is lots of demand from all corners. So car makers tried their usual strong-arm tactics and failed miserably. Chips being a sellers market, after all. After which car makers went whining to their politician puppets, getting them to OK billions in subsidies for "supply chain security".
This wouldn't be a big deal. GPT is not some one-of-a-kind leaps-and-bounds ahead of everyone else ground-breaking technology.
I have a hard time understanding the people who visit this website and havent used the technology.
from the press release itself:
> ESMC marks a significant step towards construction of a 300mm fab to support the future capacity needs of the fast-growing automotive and industrial sectors, with the final investment decision pending confirmation of the level of public funding for this project.
it's a 300mm fab.
> Realpolitk wise they absolutely cannot risk regulatory capture to a few select firms on the west coast of the USA.
not only is this 40 years too late, but the press release itself contradicts this comment.
Meanwhile new nuclear reactors cost 4 times as much as renewables and take decades to build.
So stop spreading FUD.
I think you're talking about the Québécois, because the french around here are swimming in a sea of _épouvantable_ mix of french and english words. But as afterthought you're still quite right, this mix usage is sliding downhill from ill inspired managers to general low class now.
These are not leading node semiconductors that will be used for training AI. These are commodity chips that will be used in the automotive industry.
Source?
These nodes are generations old and basically only in use in German cars with their abysmal 'software'. The only touching point these processors will ever have with AI is as an API endpoint to Google Voice or Siri.
Moreover, Mistral was widely seen as a joke and testimony to the catastrophic state European tech is in.
Europe is done. I have no idea why they rushed to regulate anything. In the end the U.S. startups will tell us what to do. Europe is a joke.
Not that I see what interest I'd have in any kind of AI in my car.
Yes, these chips are for automotive and industrial use, what's the problem? Those are both important use cases.
You may think Europe is done, but who makes the EUV lithography machines used for TSMCs' cutting edge nodes? Who makes the leading ERP system used by a lot of companies around the world? Where was the Chrome V8 JavaScript engine developed?
The point is, Europe has plenty of innovation. But unlike the US, people don't go around boasting every chance they get. People just do stuff and are happy with that.
I do enjoy translating things into Sindarian or other conglangs from literature though.
I had a theory that giving it the custom instructions in latin might give better results, but i haven't had time to figure out how to benchmark something that's so non-deterministic and black-boxy
What do you mean?
there are also other potential benefits, e.g. "the" in english gives little information about what noun may follow. "die/das/der" in general cut the probability space to a third.
1: http://en.wikipedia.org/wiki/Simplified_Technical_English
2: Contrast particularly major rivals French and Mandarin
There's hardly a topic like this where almost everybody finds something to complain about ;)
A registered association (e.V. in German) bluntly named "Silicon Saxony e.V." (est. in 2000) with over 450 members headquartered in Dresden pretty much sums up the infrastructure and ambition.[0][1]
As pointed out the process TSMC’s 28/22 nanometer planar CMOS and 16/12 nanometer FinFET process technology (fabrication will begin in 2027) is not "cutting edge" but set to satisfy the (local) demand [2] mostly for the automotive industry. I guess a good conservative foothold given the current economic/geopolitical environment, but it's not like an innovation/investing boom or something.
However from a perspective of a young student, Saxony/Dresden - as for now - is pretty inexpensive as compared to Berlin, Hamburg or Munich and has ample opportunities (TU Dresden, Fraunhofer, MPI, Helmholtz Zentrum ...).
[From personal experience I find the people in Saxony (Leipzig, Dresden) have a more deep-rooted skepticism towards the federal government (or any centralized entity ruling over them) which can lead to a good thing - interesting perspectives/lively discussions or depending on your position less good - leaning more to the political (far) right.]
[0]https://en.m.wikipedia.org/wiki/Silicon_Saxony
[2]https://en.m.wikipedia.org/wiki/2020%E2%80%932023_global_chi...
I've witnessed visiting Indian colleagues of mine being threatened on the street for having dark skin and racist demonstrations are a weekly occurrence. I eventually moved away (among other reasons) because I didn't want my kids to grow up in this environment.
It was quite difficult to hire international talent there for the same reasons.
The city itself and university are super nice though.
While all of what you say is true and problematic, there are people from all over the world living in Dresden and some parts (especially Neustadt) are politically rather left leaning.
>The city itself and university are super nice though.
Then, I hope this will get more widespread :)
Is this an actual thing or just something pushed up because victimhood is idolized too much? The blue collar workers here are diverse af.
I went on a beach vacation with some French friends and was sort of stunned/apalled at the sort of non-stop "it is all the muslim/immigrants" fault talk.
So, like the US?
My take, if you want to have a bleeding edge semiconductor industry in your country, first start by trying to build a boring, vanilla, outdated semiconductor industry. If you can do that, then try incrementally making it competitive with global markets. (Yes, this requires tariffs initially to bootstrap the process, otherwise nobody will by your shitty overpriced semiconductors.)
[0] https://astralcodexten.substack.com/p/book-review-how-asia-w...
Like what if SK & Malaysia took the opposite approaches? How do we know that SK wouldn't still have been the successful one, and we'd be saying everyone should take the opposite approach because that's what SK did and it worked?
Source: Invisible China by Scott Rozelle which does a great job explaining economic development and comparing different countries.
[0] (de) https://de.wikipedia.org/wiki/Max-Planck-Gesellschaft
"Silicon Saxony is a registered industry association of nearly 300 companies in the microelectronics and related sectors in Saxony, Germany, with around 40,000 employees. Many, but not all, of those firms are situated in the north of Dresden. [...]
[T]he area and the union — in many aspects — represent the only meaningful European center of microelectronics.", excerpt from https://en.m.wikipedia.org/wiki/Silicon_Saxony
Otherwise they have everything they would need for a great hub. Smart and well educated students. A great university and cultural scene.
As always its a little more then that..
If countries are less reliant on manufacturing within Taiwan, its one less reason to stand up to Beijing's one China principle.
Yes, they depend on ASML, but ASML itself was dependent on IP from TSMC engineers. TSMC already announced that they will stop applying for subsidies from the Chip Act[] in the US recently because the US keeps changing what confidential data needs to be shared with the US.
They seem to be at least slowly understanding that this is a dangerous game.
And for the people that keep talking about ASML you might well want to remember that not long ago the leadership was mocking China's capability to build their own domenstic supply chain only to flipflop shortly after saying that it would be foolish to abandon the Chinese market likely indicating that they might be concerned that their domestic supply chain might end show up faster than expected.
Taiwanese media last year was plastered with news about how the US hollowed out the Japanese semiconductor industry with its agreement in 1986 and how that will be potentially the fate of Taiwan.
But the sibling comments are correct about the potential blowback, let's not forget that the US has been talking about bombing TSMC themselves[]. I guess that probably explains why they just announced a delay to the construction of the Arizona fab.
[] https://www.koreatechtoday.com/south-koreas-semiconductor-ex...
[] https://www.aljazeera.com/economy/2023/7/27/samsungs-profit-...
Well, ASML is primarily and rightly very concerned that once "China" has their machines they'll attempt to reverse engineer them, ignore any western patents and IP and try to build 'Chinese chip machines'. It's happened to many industries, from Lego to fashion to hi tech. Hell, apparently there's even Chinese Knock-off Movies.
So, above everything else, there's a legitimate concern for ASML that once they move or deliver too much into China, they'll enable their own competition.
The U.S. said nothing of the sort. One Congressman, in the minority no less, did.
Is demand for electronics finally dropping? I want to be able to buy a 3 year old Nvidia low-range GPU at below MSRP...
> "“There is an insufficient amount of skilled workers” with the expertise to build a chip factory, TSMC chairman Mark Liu complained during a call with analysts. The executive warned the company might have to fly in “experienced technicians from Taiwan to train the local skilled workers for a short period of time.”"
https://fortune.com/2023/07/21/tsmc-complains-cant-find-enou...
The more appropriate term is probably foreign direct investment (FDI) in Germany from TSMC.
I find it funny that western people have issue with the idea that Asian companies can indeed also invest in western countries. It is akin to some discussion about Lega here on HN where some people talked about it as "onshoring" when they made a US factory – Lego is a Danish comapny.
TSMC alone has no control over if the Chinese government will commit such aggression, if they would do their whole company strategy based on fearing the CCP, there's no way they can be successful.
They are willing to burn a lot of money and time to reach their overall political goal of annexing Taiwan even if it takes a very very long time.
Ultimately it's unrealistic to assume that a country of 1.4 Billion people can not catch up eventually espescially if they would otherwise have to rely on a global competitor (US) or a military target which factories are rigged to blow (Taiwan).
The best thing Taiwan can do (once their semiconductor lead is no more) is to make themselves into a strategic ally located on Chinas doorstep.
https://en.m.wikipedia.org/wiki/List_of_countries_by_governm...
The Taiwanese government can attempt to prevent TSMC in participating, but what good will that do? Might as well ensure that TSMC at least have a hand in it and avoid customers leaving for another manufacturer. TSMC can then offer fabrication, maybe it's in the EU, maybe it's in Taiwan, the important part is keeping the customer and not allowing competitors to establish themselves.
This in Germany and the one in the US are both Megafabs.They produce chips for special industries with strategic importance. In the US for military and defense, in the Europe for automotive industry and automation.
Having this beefy reliance on a small island that gives china the middle finger
-OR-
derisk the reliance on taiwan
Millions of lives would be lost if not more if it will be a full scale war. A chip fab is nowhere the same level of importance to be even discussed on the table.
"But they've spent X billions of dollars." Yes but this is irrelevant when facing off an existential threat to their business and the overall economy and security of Taiwan. Taiwan, like Israel, is run by cunning folks. They wouldn't have survived this long otherwise.
Mere reduction of the criticality of Taiwan's advanced manufacturing will not eliminate Taiwan's geostrategic importance.
Plus, the CCP's insistence on being an expansionist authoritarian state is reason enough to contain that expansion, to prevent further resource gains.
When you said China, which China are we talking about?. the fact is, Taiwan and China is still in a Civil War.
The United States did not explicitly state the sovereign status of Taiwan in the three US-PRC Joint Communiqués of 1972, 1979, and 1982.
The United States "acknowledged" the "One China" position of both sides of the Taiwan Strait.
U.S. policy has not recognized the PRC's sovereignty over Taiwan;
U.S. policy has not recognized Taiwan as a sovereign country; and
U.S. policy has considered Taiwan's status as unsettled.
The PRC has never, not even for a single day, controlled Taiwan, Kinmen, Matsu, etc.
The continued existence of the ROC (1912-present) truly causes tankies to lose their minds.
This feels more like TSMC grabbing from subsidies pots where it can hoping this will please the NATO nations and that TSMC hopefully survive long enough te reenter the mainland again on time before the mainland goes completely domestic with chips fabrication.
I am glad that Germany is coming to their senses and finally rejecting neoliberalism. Now if only Eurozone leadership (which was created during neoliberal heydays) would do the same across the EU.
Same with South Korea.
Second, why do you think a country like Germany (where the ratio of government expenditure to GDP is around 50%) can be classified as “neoliberal” in the sense you mean it?
The automotive industry made the mistake of cancelling their orders during Corona and were then surprised to find that the capacities were allocated elsewhere when they wanted to order again.
So they whined at the government to subsidise the fab.
1. Have many fabs and derisk by diversification?
2. Keep Taiwan safe and chest-bump china?
Germany needs CPUs for the industry. Cars, machines, planes, wind turbines and so on :)
But you cannot build Personal-Computers with them? That failure is caused by Siemens (Siemens-Nixdorf) and AEG. In GDR/DDR they were close to build a 386 compatible CPU. The remaining knowledge is still somewhat available, Global Foundries in Dresden.
But the CPUs for PC? Yes. Yes. Okay. Intel will build a Fab there, too ;)
There are plenty of application were structure size, performance and/or power draw doesn't really matter.
E.g. Automotive / Industrial applications
The Chips Act was marketed as an attempt to bring back "cutting-edge chips" to Europe, and to promote research and innovation. In the end it turns out to just be a huge government subsidy for a bog-standard fab which is at least a decade behind what TSMC is building in Taiwan.
Great for automotive companies who want to strengthen their supply lines, but pretty pointless when it comes to doing what it was actually supposed to do. Europe already has plenty of fabs like these which did not require billion-euro handouts.
And it can vary too, mass produced / generic chips can be more cost efficient if they can make more of them on a single wafer.
Not everything needs the 5nm cutting edge, and to be fair calling
> 300mm (12-inch) wafers on TSMC’s 28/22 nanometer planar CMOS and 16/12 nanometer FinFET process technology,
old is a bit unfair
22nm dates back to 2012, and TSMC has been doing 14nm FinFET mass production since 2014. So yeah, a decade behind cutting-edge can indeed be called "old" when the entire point of the Chips Act was to catch up with Asia and become a center for innovation.
No, TSMC 22 has no eFLASH, expensive SRAM, lacking of analog capability, lacks high temperature tolerance, and is tied to multiple patterning design flow.
TSMC 40 on the other hand is the last portable node, with MRAM, eFLASH, high temp, some mixed signal, RF, no MP, and everything else your soul desires.
Having the bleeding edge tech sounds great, but where are the customers that are going to pay for bleeding edge output and the countless billions of investment it takes to get set up to produce it?
Decade old technology sounds very sensible.
It's all good and nice if private companies want to do this, they should know best after all. Spending another 5 billion on taxes on this is flabbergasting, especially in an area of the country that is incredibly foreigner-unfriendly (to say it mildly).
Many chips don't need fancy new tech.
You don't re-fit a fab, you build it from scratch. There is very little technological overlap where it matters due to the shift to EUV. You'd essentially have to scrap >75% of the fab to fit in the new equipment.
I'm no expert though so might be off by a mile here.
It’s good to see new fabs in Europe, but I think it’s a bit late and with “old” technology. On the other hand there are many other areas in need of chips that will benefit from this in the coming years.
I know where I work we have customers that run our products at -70C and +50C, we can't run a lot of modern chips because they fail our tests at those operating temperatures.
Another reason was Russia cutting cheap oil during the 70s after the oil shock. The GDR had imported cheap oil from Russia, was refining it and selling it with high margins to the West for $/Deutsche Mark. That source of money dried up in the 80s.
With sparse resources, the GDR was between a rock and a hard place. For modern production it needed chips, reverse engineering western chips e.G. Z80s (reverse engineered chip called U880 in the GDR) or smuggling chips was very expensive (~5-10x the price). Building chips on their own was also expensive. But they decided to invest billions into home grown chip production (they smuggled a whole factory via Austria, see "Red Fini") leading to the famous 1mbit chip "U61000" - which was old at that time, also the Japanese produced the same amount of chips in a day that the GDR was able to produce in a year. So things went nowhere.
But those billions were not available for other things.
Without money ($, Deutsche Mark), in the 80s to pay back credits to western banks a lot of vegetables, fruits and consumer products like clothing and washing maschines were exported to the West (most of the fruit production of the GDR around West Berlin went to West Berlin, many products in the West German QUELLE catalogue - like Sears - were made in the GDR). These products were missing in local shops, shelves were often empty, especially outside the large cities like Berlin, Leipzig and Dresden.
Additionally factories in the 80s in the GDR were in a very bad state (my father in law was head of production in a large machine factory in East Berlin during that time), lots of stuff was being stolen on a constant basis, maschines were old and lacked computer support (see above). This again reduced available consumer goods.
Because consumer goods were scarce, when things were available people bought more than they needed, leading to more scarcity (as seen during the pandemic). It is said that every household in the GDR had a second car as spare parts (another example are agricultural factories hoarding harvester spare parts, so when broke down, spare parts where hard to get).
Missing consumer goods in the GDR led people to leave their jobs in factories whenever something was available, e.G. rain boots, which reduced productivity even more.
In the end so many things were missing that people mass fled the GDR when Hungary opened the border.
All this together made the Central Committee believe everything is lost (book to read in German "Das Ende der SED - Die letzten Tage des Zentralkomitees" with meeting notes from the Central Committee), and the Russians not willing to come to help with tanks this time [0] (like they did in 1953 to kill Germans to keep the communists in power in East Germany) power was transfered to the people leading to reunification.
[0] For comparison: When Poland went bancrupt in 1980/1981 Russia and the GDR were near an invasion - which Poland prevented with a military coup.
Unless it’s for cars?
Chip fab locations are about water and talent. Electricity sources are pretty fungible. The former not so much.
You need stability, skilled and productive workers and it wouldn't hurt to have a chemical industry, Germany has all that.
Let's say TSMC does manage to build a fab before 2027 (the year China is expected to complete the modernization of their military and launch their campaign against the island of Formosa's governing body, aka the Taiwanese government). How might such an extravagant outlay of investment funds (which could significantly improve the economic outlook in Germany and take a lot of pressure off the existing government) impact Germany's relationship with China in the crucial near term (the next few years)?
Let's not forget, for decades the CCP worked strategically on bringing about the reverse: isolating the ROC government by engaging in lending practices that sometimes resulted in situations where they could use this leverage to promote a gradual deprivation of the ROC's status as an officially recognized governing body (which has shrunk over the years to a mere 11 or even fewer nations).
Of course it isn't as if Germany's support would significantly affect the CCP's (meaning, Xi Jinping's) plans of Chinese Rejuvenation (wherein bringing Taiwan under heel is a non negotiable component), considering that even the US' increasingly overt support has had no effect on what the CCP has always viewed as an inevitability. Conversely, as the Ukraine debacle has indisputably demonstrated, even indirect support (especially from a Western ally) has a profoundly disruptive effect on the expected development and eventual outcome of any military engagement.
Also, as some may still remember, one of the principal factors that led to China's significant manufacturing modernization was a somewhat similar deal, wherein China "bailed out" a prominent US manufacturer (Boeing?) that was close to bankruptcy, by agreeing to purchase several billion dollars' worth of civilian airplanes, and in return being handed the high precision tooling infrastructure that's indispensable in the manufacture of high tech machinery (because of the military/national security implications, this required political support at the highest levels of government, meaning presidential and bipartisan support). I am suggesting this could be analogous, differing chiefly in the type of benefit that is gained by the other party, as a result. For China, it was high precision tools, while for the 23 million inhabitants of the island of Formosa (who as a group, if not a sovereign nation already dominate the semiconductor industry) it may well be a tactically swift reversal in international standing (something that barely registered at all until very recently, but then rapidly expanded into an all encompassing existential crisis rivaling their defeat at the hands of the Chinese communists, 70 years prior).
Everything except simple current business need.
For Germany, it's an insurance against TSMC possibly bailing out in the future.
The 22nm node is also a very strange choice. It's absolutely not a thing which most of German industry needs, which are microcontroller, high voltage, mixed signal, niche memory, and ASIC friendly nodes.
Easy: Infineon is German, NXP is Dutch, Bosch is German. Germany is not exactly going to give a massive subsidy to a foreign company, but giving one to local companies is "bolstering local economy". And TSMC gets a fab at a 50% discount out of it.
Curious why they've chosen Dresden though.
Anybody knows how much the government is willing to invest? Missing some tangible figures.
The region is known for its semiconductor industry, at least in Germany. Quote from Wikipedia about Dresden:
> Silicon Saxony Saxony's semiconductor industry was built up in 1969. Major enterprises today include AMD's semiconductor fabrication spin-off GlobalFoundries, Infineon Technologies, ZMDI and Toppan Photomasks. Their factories attract many suppliers of material and cleanroom technology enterprises to Dresden.
So it just makes sense to build fabs in Dresden because there is already a lot of know-how there.
I believe this started in GDR times and continued (at massively reduced capacity) after the reunification.
Intel, AMD, Infineon and others operate (or plan to build) factories in Dresend (or saxony in general) and its reasonable to assume that experienced workers and local suppliers make the area a preffered location.
Given Eastern Germanys weak economic situtation these are very welcome investments.
Edit: see bellow for corrections
Because many foundries are located there like GlobalFoundries and Bosch
5B, according to German news: https://www.tagesschau.de/wirtschaft/tsmc-dresden-100.html
Did Tesla ever finish that German factory? Last time i read about it no one wanted to work there...
OTOH Netherlands would make a much better option in almost every angle. And add having ASML engineers around the corner.
Germany has around Dresden one of two regions in Europe with a large cluster of chip industry.
https://en.wikipedia.org/wiki/Silicon_Saxony
ASML buys key equipment in Germany: optics from Zeiss and lasers from Trumpf for EUV lithography.
Intel builds up chip production in Magdeburg, Infineon in Dresden, Wolfspeed in Saarland, Bosch in Dresden and Reutlingen and now TSMC in Dresden.