[FooBarWidget]

Definitely not limitless. Subsidies are only provided to get an industry to grow up. As they grow up and can increasingly fend for themselves, subsidies become progressively smaller.

There are a ton of other sectors where subsidies don't help nearly as much as the govt hopes. Until the US sanctions hit, subsidies for the semiconductor industry only had very limited effect.

[donny2018]

>There are a ton of other sectors where subsidies don't help nearly as much as the govt hopes. Until the US sanctions hit, subsidies for the semiconductor industry only had very limited effect.

Probably some sectors are easier to grow if it only involves copying something that already exists, and outscaling/outpricing everyone else. Then this is (in most cases) is a purely money+organization problem.

But if it isn’t just copying/outscaling/outpricing issue, like replicating ASML’s EUV lithography, then yes, we have a problem.

[FooBarWidget]

70% of the profits are generated by older processes, not sub-14nm, and so are perfectly well serviceable using mature DUV litography which is much easier to "copy". But even there, subsidies did not yield much progress.

On the other hand, new energy vechicles are... well... new. There is no mature ecosystem to copy, a lot of things have to be newly developed. In this sector, subsidies have been massively successful.

The problem is mainly one of market pressure, not one of technical ability. Chinese semiconductor companies wanted the best suppliers, so they chose international suppliers rather than domestic suppliers. Domestic customers didn't buy from domestic chip designers. Domestic chip designers didn't manufacture with domestic fabs. Domestic fabs didn't use domestic equipment. As a result, domestic suppliers never got enough customers to practice and improve their processes, which is why they remained low-quality. It was a vicious circle which the Chinese govt tried to solve for years without much success. Then US sanctions came and all of a sudden, Chinese semiconductor companies had no choice but to work together with domestic suppliers: it was either shitty domestic suppliers or die. Nowadays you see domestic semiconductor equipment companies have something like 150-200% growth YoY, something which they previously could only have dreamt of. Domestic DUV litography was at 65+nm for a long time but now 24nm DUV litography (still good for ~60+% of market demand) is around the corner because they finally get enough practice.

People ascribe too much to this simplistic view of China only being able to copy or that copying is easy, and totally underestimate economic pressures.

[tooltalk]

>> 70% of the profits are generated by older processes, not sub-14nm, ...<<

Not sure which chip segment you are talking about. The legacy chip manufacturing in general operate at much lower margin and much of revenue and profit comes from the cutting edge nodes: for instance, 2/3 of TSMC revenue and profit come from sub-10nm; likewise for SMIC in China, their biggest money maker is 14nm, their most cutting edge nodes.

>> There is no mature ecosystem to copy, ...<<

In the EV battery market, the established competitors/leaders in the market were LG Chem and Panasonic. But as explained earlier, they were excluded from participating/competing in China EV market which would have given them opportunity to further improve their process/yield and accelerated commodification of their tech.

>> Chinese semiconductor companies wanted the best suppliers, so they chose international suppliers rather than domestic suppliers. <<

There is little/no such "domestic" supplier in China's chip manufacturing -- over 90% of chip manufacturing equipment/suppliers are in the US, or Japan, or the EU. Even Taiwan and South Korea import 90+% of their equipments from those named countries and have very small domestic supply-chain of their own. And let's not forget that China has very little chip manufacturing talents of their own -- it's not surprising that former TSMC engineers were behind SMIC's 14nm/7nm. SMEE's first 28nm lithos, which is apparently still not ready for mass-production after having made release announcement two years ago -- likewise heavily depends on Japanese parts/engineering expertise. You can't just shortcut to 50-60 years of accumulative knowledge in making precision equipments by copying.

[FooBarWidget]

Okay. Sub-10nm don't make much money, and China can't produce sub-10nm (or any node invented in the past 20 years) anyway without western help, and they won't catch up because all they're good at is copying. You guys have won. So why is the US still throwing more semiconductor sanctions at China, including sanctions designed at hindering China to manufacture at older process nodes, going as far as coercing allies? Don't these guys have better things to worry about now that they've won and China have lost, like domestic issues such as violence and income inequality?

[tooltalk]

>> Sub-10nm don't make much money <<

sigh the cutting edge nodes get the vast majority of chip making revenue/profit.

>> China can't produce sub-10nm (or any node invented in the past 20 years) anyway without western help <<

China's SMIC already has 7nm, though efficiency, cost-effectiveness and yield are still unknown. As late as 2022, the US dept of commerce granted over $100B worth of licenses to China! so they got all tooling and lithos they need to make 7nm, or even 5nm, though they would be very expensive and poor yield.

>> and they won't catch up because all they're good at is copying <<

China could catch up organically, just probably not this decade, or next. The Japanese equipment makers such as TEL and the Dutch DUV/EUV maker, ASML, can still sell their equipments to China no problem; just not the most cutting edge stuff.

[tooltalk]

>> Subsidies are only provided to get an industry to grow up.<<

Those subsidies only applied to domestic companies -- remember that the South Korean battery makers such as LG had about 50% of the Chinese market share and 9 out of top 10 local EV makers as customers until in 2015, the CCP more or less forced them out to protect BYD/CATL.

[FooBarWidget]

Yes subsidies are meant to help domestic companies until they can compete with the incumbent. This strategy has been used by other countries as well.

Those who protest such subsidies are usually the incumbent, who would like to see no subsidies so that they can easily crush startup competitors.

[tooltalk]

>> This strategy has been used by other countries as well. <<

which "other countries" are you talking about? This practice is illegal under China's 2001 WTO access protocol, which prohibits any subsidy contingent upon "... or on the use of domestic over imported goods." I don't know any country that openly discriminated and excluded the use of foreign competitors' EV batteries -- especially in a market where as much as 40% of the EV inital purchase cost is subsidized.

It wasn't just the EV subsidies though. From 2015-2019, there was a slew of anticompetitive, discriminatory NEV policies that were blatantly illegal to eliminate competition and help domestic weakling, CATL/BYD. (see the EU's WTO complaint WT/DS549 filed in Jun 2018).

>> Those who protest such subsidies are usually the incumbent, who would like to see no subsidies so that they can easily crush startup competitors.<<

Biden/Mancine's IRA enacted last year in 2022 is partly modeled on this to counter China's unfairly gained market dominance. Quite interesting that China threatened action against ‘Discriminatory’ US EV Tax Break[1].

[1] https://www.bloomberg.com/news/articles/2022-09-22/china-thr...