TSMC certainly "stole" the European chip industry. They had an IP transfer agreement with Philips.
Philips at that time also sold off its crown jewels like ASML and ASMI. This was the time when CEOs bragged about restructuring companies with inane statements like "the company should look like a board of asparagus, not like a board of spaghetti". Journalists thought it was brilliant at the time (enjoy your dinner!).
China and Taiwan benefited from this reckless outsourcing. I'm actually in favor of stopping any European or U.S. high tech exports and further outsourcing.
But it should be on an economic level, not on the silly warmongering level.
RCA also gave their crown jewels of semiconductor technology to what became MediaTek as RCA distracted itself with non-core acquisitions (transforming itself into "rugs, chickens, and automobiles," leading to their decline).
The scope of mergers & acquisitions in semiconductors is mind boggling & terrifying. It feels like there's so much less diversity, that there's a frailty of entirely Too Big To Fail players left.
Yah this is going to have the opposite effect, as it will only encourage China to develop their own EUV stepper and fab development. It might take 10 years to catch up, but 10 years isn't a long time in the grand scheme of things.
Actually it might take them less than 10 years to catch up. I was at ICCAD (https://iccad.com) last week and half the papers were from China. And some of the EDA CAD software is under export controls - specifically Gate-all-around technology used in 2nm and below. Right now the big EDA vendors have developers in China who are making the software they're not allowed to sell in THEIR OWN COUNTRY.
China is basically where Japan was in the 1970s and Korea in the 1980s. Only 10 years before they become dominant in the industry.
EUV (not High-NA) is much, much more complex than Immersion Lithography—and China doesn’t have have that either—the optics (mirrors), the light source, the masks, everything under vacuum… A blank check (which the government would be happy to sign) is not enough to catch up on everything in 10 years, integrate it and then make it work reliably at a decent throughput. Not impossible, of course, which is why they are going to try.
To top it all off, a lot of this stuff is patented. Not that China cares all that much about Western patents, but violations in this area would make the geopolitical situation even more tense.
hmm, china is perfectly happy to buy any semi fab tools from Western companies, 40% of ASML sales is to China. The fact that none semi tools company were any significance before the US sanctions is the proof of that. Not only are the patents, the market, talent, money isn't there to make it a sustainable business. But u block the sale, and go max pressure, u create a huge incentive. The engineers in these companies are super happy. Now, in 2024, Chinese semi fab tools companies are moving up to near top 10 global fab tool companies. I have no doubt we will see DUV and EUV fully running on commercial fab lines in the next few years.
"China doesn’t have have that either—the optics (mirrors), the light source, the masks, everything under vacuum" are you really sure about that?
Let me help you here, a lot of this stuff are in Chinese universities. But not commercialized, cuz there is no commercial fly wheel when there are perfectly fine products on the market. Now you just gave $1B dollars to these researchers to bring it to market. This will take time, but this is an engineering, iterate and trial problem, not inventing quantum mechanics. The demand is always there, people who get it get a life time of wealth, what do you think?
You underestimate the sheer complexity of it. You’re talking $1B, I am talking a blank check. That’s probably 100s of different research teams, each with a $?B, whose “research” would have to industrialized.
Then there’s integration and SW. Not to mention performance, metrology, diagnostics.
Now you have to actually integrate these new tools into a new leading node process. This by itself takes years, assuming you have the tools. China does have existing Immersion process nodes, but still.
Even with unlimited resources, I claim China can’t have EUV within 10 years. Maybe not even 20.
A ban and a trade war is more likely to make China just toss these patents on fire. Espionage is still a thing and for the right amount of $$, they might be able to find engineers who will switch sides.
Serious interest here. Can anyone else verify that the Chinese don't have the ability to develop and manufacture these chips?
That's stunning. I had no idea that was the case.
Or is this just a, "they don't manufacture computer chips" kind of thing. Which, strategically speaking, is critically different than, "they don't know how".
China can manufacture all but the best chips. They are continually improving, and I don't think anyone has offered a single reason why they couldn't do it.
Superiority complex. Which is interesting, given that tech companies have lots of Asians in them. Then the goal gets moved to cultural superiority. I don’t discount culture but money and power are strongest drivers in my opinion; and China is willing to offer you lot of that if you are willing to abide by the party line.
so in the event of effective sanctions on a country, where the tariff is effectively infinite making the amount supplied go to zero… this should be even better for the country affected?
Tariffs are always good for someone. Just not most people.
Is it good for anyone who wants to manufacture chips in China. Obviously, yes it is good for them. But it is not good, in the short term, for people who want to buy chips in China.
But it might be bad for everyone in China, because someone above implied that they actually don't have the knowledge to make these chips. Not sure how true that is? But assuming it is true, that would hurt everyone in China.
> so in the event of effective sanctions on a country
That's the point. You can't effectively sanction china.
> where the tariff is effectively infinite making the amount supplied go to zero
There is a difference between a complete blockade and targeted sanctions.
> this should be even better for the country affected?
One causes a complete collapse of an economy (north korea) while the latter helps a nation build up a sector of an industry (china).
China is the top trading partner of most countries around the world. Even with these sanctions. Go figure. It would be like if china banned exports of 'rare earth minerals' to the US. It would only make the US stronger because we'd just invest in mining rare earth minerals.
>Why does China’s near abroad matter to them? Why do they care about national unity? Why would they want unobstructed access to the deep-water Pacific?
Are questions that answer themselves. But they are the wrong questions.
It’s not China that’s the hostile actor here. It’s not China that’s meddling in distant affairs. This very thread pertains to an unilateral, unprompted hostile action by the US against China. And note that it’s not Trump’s doing. He’ll just do more of it, more openly.
And all of this is very much in line with America’s (at least) seven decades old strategic posture:
Taking out 95%+ of ledging edge semi that adds trillions to western hi tech and supports strategic industries seems worthwhile. Doesn't have to be invasion, but I'd expect grayscale shenanigans on island power grids etc now that there's less reason to hold back.
The US now has an isolationist regime. I think that's fairly likely to happen in the next few years if it is going to happen. Ukraine is also done unless Europe (probably Poland) steps up in a big way.
It doesn’t. But as the Empire’s grip on the world is slipping it will become more brazen, aggressive, openly selfish, erratic. Quod licet Iovi etc. A role Trump was born to play. But, as Europe’s economic woes after four years of D rule should tell you, it’s not just him who’s happy to cannibalize the Empire’s subjects.
I don't feel like China operates under the same geopolitical philosophy as Russia, they have other ambitions that I think are better served by avoiding wars. But who knows given the way things are going.
It's clear if you pay attention, that China's putting inordinate efforts into other places. Africa is one example. (Probably the principal example.) So they clearly have ambitions that lay outside of Asia.
No one fakes moves like that at that scale. They're serious about what they're doing.
Because war is economically expensive, they subscribe to certain philosophy and would prefer continue trading with Taiwan rather than turning it to rubble.
That being said, even if they would. Under that assumption, I find the idea to further provoke them to doing that morally repulsive. It's an egomaniacal move that disregards Taiwanese people.
The calculation is a bit different for a dictator though.
They won't be affected personally (unless the country rebels against them) so they can be fine to tank the economy if it gets them closer to some other goal (such as megalomaniac world domination).
A military fight wouldn't serve China's interests. Pouring money into their domestic fabs and using the same market-flooding and subsidies that they're known for in other fields to take customers from TSMC would be a smarter move. They probably wouldn't be taking over the high-end CPU/GPU marketshare anytime soon, but could put significant pressure in other areas while developing capability.
Two reasons why TSMC operations in Taiwan would cease: 1. Skilled workers to develop and operate the fabs. I doubt it that the Taiwanese would happily continue to innovate under a (possibly violent) Chinese regime. 2. Supply of fab equipment from ASML and other Western companies.
Hope what doesn't happen? The invasion? Or the embargo?
The Beijing regime is already in intense sub-threshold warfare on Taiwan and reiterate their false claim to the country's territory at every opportunity.
Denying them access to the finest silicon processes will no doubt trigger more narcissist punishment tantrums delivered through PLA military provocations, or sabotage, or trade attacks, etcetera, but it's hardly going to matter for whether they launch the invasion. They will do that once they've built so many new nukes and cruisers that they like their chances against the total of Taiwan's defense forces plus any other military that could be expected to intervene.
No, unless turning consumer products into data center use becomes relevant at large scale.
There was a time when companies and research labs bought Sony PS2 (and the PS3) to build supercomputer clusters. Japan placed export controls on the PS2 because they had militarily-useful computing power. According to DIA report 4,000 PS2 units had been purchased in the United States and shipped to Iraq in just 2-3 months during 2000.
Were the PS2s shipped to Iraq for computing or just for soldiers to play during their down time? I feel like the latter is far more likely, especially since we know that soldiers had access to game consoles for recreation.
I can't imagine what general purpose software they'd have wanted to run on a PlayStation 2 from 2000 that they couldn't run on a general purpose laptop in 2004.
You're being down voted, but it's the truth. PS2 were sold so much below cost it wasn't uncommon to see hobbiests and institutions building beowulf clusters with them.
But these days, consoles aren't sold below cost, and gigabit network isn't fast enough to make cluster computing make sense
The PS2 was a state of the art machine with similar hardware of a SGI workstation. Those computers started at 20k. While the PS5 has similar hardware as a low end PC.
If the PS5 had a Nvidia A100 the comparasion would make sense and people would be buying them to create clusters, like they did with the PS2 and 3.
The article says: "no longer manufacture AI chips at advanced process nodes of 7 nanometers or smaller." This was triggered when Huawei produced some kind of device with TSMC wafers of this class.
Maybe standard ARM cores are still approved for export to the mainland.
Philips at that time also sold off its crown jewels like ASML and ASMI. This was the time when CEOs bragged about restructuring companies with inane statements like "the company should look like a board of asparagus, not like a board of spaghetti". Journalists thought it was brilliant at the time (enjoy your dinner!).
China and Taiwan benefited from this reckless outsourcing. I'm actually in favor of stopping any European or U.S. high tech exports and further outsourcing.
But it should be on an economic level, not on the silly warmongering level.