https://www.digitimes.com/news/a20230316PD218/china-ic-manuf...
Chinese chipmaking technology development may stall at 40nm scale
With the US looking to broaden its chip tool ban against China with new rules set to be implemented in April, SMIC and other China-based foundries will have to halt the development of their sub-40nm process technologies, according to sources at semiconductor equipment companies.
---
I think in a few years the tech war will be obviously lost and people will look back at it as a massive strategic failure and own goal. Yes. China was always going up the value chain but Western companies were well placed to be a part of that as key subsuppliers, now they won't be because of idiotic cold war warriors.
Furthermore there are a whole swath of types of chips for which smaller litography isn't even applicable. China is developing GaN chips quickly, and these are power chips that will waste less power and become less hot.
And apparently building a 5G modem and a sattelite receiver chip aren't trivial things either. Apple failed to build a viable 5G modem for the latest iPhone while Huawei did it, showing that analog chip design (way more complex than CPU design) should not be underestimated. Apparently BW filters (a component used in 5G modems) aren't trivial either, with very few suppliers in the world. Existing sattelite phones has clunky antennas but the one by Huawei can't even be seen. And it looks like Huawei innovated heavily in heat dissipation technology, allowing its latest smartphone to perform competitively with other flagship phones despite the chip generating more heat due to bigger process node.
The list goes on. Just focusing on nanometers and litography is really a shallow view of what's going on.
Meanwhile in reality they already have at least 28m DUV steppers indigenously built and are rumored to have a few 7nm capable DUV steppers doing shuttle runs. And are doing really neat things in the EUV space taking a different direction than ASML, essentially using a particle accelerator rather than a tin laser.
It's been clear now for over a decade that every time we cut them off from a technology they produce indigenous versions. We cut them off of chips to go in a supercomputer and within five years they had the world's fastest supercomputer on indigenous designed chips with Tianhe-2.
> I think in a few years the tech war will be obviously lost.
I think the idea is to extend the no. of years lead US+ has. China will have to replicate one of the most complex supply chain ever seen. They will use all measures possible so would be interesting to watch.
this is like saying Tonya Harding had a fight with Nancy Kerrigan
Yet politic wonks talk the tech sanctions like they are doing an embargo on Cuba.
SMEE jumped straight from 90nm to 28nm and I think it caught a lot of the talking points off guard.
You don't assume the "enemy" will just bow down. Your policies must include their reaction. Like playing chess, you aren't just checkmating the king.
The best scenario for the west would've been to integrate themselves into the chinese high tech supply chain, such that their indigenous companies cannot exist without massive state subsidies (because they aren't in a position to compete), which drains other aspects of the state's budget, and in turn, allow the western companies the profits required for further innovation and new tech.
The export bans and the rhetoric around that is so short sighted. The loss in profit will directly affect the research into the next generation. The lack of western tech means the chinese indigenous chip makers will, by necessity, invent the tech themselves.
This was exactly the situation before the sanctions. That's why the Chinese semiconductor industry back then developed at a snail's pace: no Chinese company wanted sucky Chinese suppliers because foreign ones are better. Chinese suppliers developed competence slowly because nobody wanted to buy from them.
You're thinking of Sunway TaihuLight. Tianhe-2 is Intel, TaihuLight is ICC, Chinese domestic Sunway processor