Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> it's going to be slow going like PRC EUV

Not even close. EUV lithography is as close to magic as it gets. By any reasonable assessment it shouldn't work but a few wizards somehow manage to pull it off.





Not even close in sense it's likely going to take west longer to build M/HREE at scale than PRC figuring out EUV + entire indigenize semi supply chain at scale.

The execution difference is PRC is generating enough semi talent to replicate EUV and entire semi stack sooner than later. They already have the most complete localized semi supply chain in single nation, i.e. they're doing ASML+5000 niche suppliers at once. Hence consensus estimate is they'll get there somewhere 2030-2035. Reminder EUV is basically a "tiny" ass effort from a handful of countries, for reference airbus/boeing each has 150k employees for commercial aviation, EUV was developed by 3k from Zeiss, 1k from Cymer, 13k from ASML... over 20 years of casual development. It's ultimately a hard but narrow specialization problem, hence PRC EUV prototype beating estimates/expectations. It's not magic, it's just people + cash + industrial vertical integration that PRC is uniquely well equipped to deal with.

VS west has "easier" M/HREE tree to rebuild on paper but lack both talent #s, and state capacity to execute. M/HREE is ~20 minerals each has it's own midstream extraction process that require dozens of plants and 100s of stages for 5/6/7+ sigma high end strategic use. It's a different monumental/gargantuan task, on top of the sheer fucking scale of infra involved. I noted Batou has 3 million residents for a reason, that's the scale of M/HREE industry west has to replicate. It takes 8-10 years to get a refinery up in the west, the chance of west getting 100s of highly polluting industrial chains up for M/HREE before PRC sorts out semi is close to zero. It's a mass scale industrial mobilization problem that west is uniquely not well equipped to deal with. I'd wager M/HREE more bureaucratic magic than even EUV technical magic for west.

Meanwhile, there isn't a single M/HREE plant in western pipeline that will do anything at scale until maybe 2030, only thing in pipeline is validating unproven lab extraction/refining methods by ~2028, if it works, will take years to scale extraction, and even more years to scale refining.


> Reminder EUV is basically a "tiny" ass effort

You illustrate a fundamental lack of understanding. 9 women can't produce a single baby in one month. That's just not how it works.

I think you really don't appreciate how utterly ridiculous the implementation details of the smaller lithography processes are. It wasn't merely limited to the west, it was limited to a single company.

> VS west has "easier" M/HREE tree to rebuild on paper but lack both talent #s, and state capacity to execute.

Wrong. The west currently lacks investors willing to shift focus to that extent and the state lacks the willingness to divert resources and step in themselves.

> It's a mass scale industrial mobilization problem that west is uniquely not well equipped to deal with.

It's not that the west is unable. We don't currently have sufficient motivation to overcome the political barriers that prevent speed.

I agree that retooling for that would take many years due to the scale of the physical infrastructure involved, and in practice will likely take multiple decades due to lack of urgency. Where I disagree is the comparison with EUV.


EUV is not a biological process on an immutable. This bad analogy on par with EUV is magic. Second mover advantage = compressing 20 year commercial cycle into 10 year strategic one viable. As it's been consistently done. Litho complexity wank needs to stop. ASML integrator of western expertise, it's not one company. We ended up having 1 integrator due to $$$. Meanwhile PRC generating more expertise with blueprint and poached many of the ASML implementers in the first place, while pursuing any EUV efforts simultaneously, stuff ASML had to ditch due to limitations.

Lack of willingness/urgency is just loser talk for last of system capacity, i.e. overcome political barriers, especially when it's been highlighted how strategic important it is to hammer out separate REE chain. Important to distinguish between unwillingness and simple inability. Easy to strong arm TW to TSMC Arizona for leading edge goals, but can't strong arm PRC to transfer M/HREE tech.

Note I didn't say M/HREE was "easier" than EUV in technical sense. I said in terms of execution, i.e. overcoming barriers, PRC is simply going to have easier working with EUV engineering problem than west with M/HREE engineering, massive infra, domestic politics problem. So it's going to be slow going, in terms of execution time.


Instead of continuing to parade your ignorance go read a whitepaper detailing the EUV process before telling me that it isn't akin to magic. Any other critical industry would have multiple competing techniques and implementors. There's even still more than one company operating cutting edge fabs despite the number dwindling as the processes got smaller.

An economic superpower identified cutting edge lithography in general as a national priority, allocated the resources, and after something like two decades of intensive research is _still_ trailing by many years. I can't immediately think of any other commercialized technology with a similar difficulty level.

As for REE, political willingness is entirely orthogonal from physical capability. A bunch of hot air on the evening news is irrelevant. If the politicians don't allocate the funds then they clearly don't see it as a top priority. If there were a pressing need then it would get done.

Where we really see the political dysfunction is the lack of planning for the future. By the time it's an urgent need there won't be enough time left for the buildout. But that's unrelated to the topic at hand.


I've read the white papers, that's why I have figures of company headcounts during EUV development off top of head. No, it's not magic. Magic fun simile, but thinking it cannot be recreated on accelerated second mover timeline because EUV "magic" vs science is bluntly, popsci cringe. EUV / semi wasn't recognized as critical industry at the time / there wasn't current geostrategic consideration over leading edge chips / hyperscaling. Hence market settled on single vendor.

PRC barely focused on EUV until trade war. Entire PRC semi push was unserious until like 2018 when they elevated semi to first class discipline, and already there's got prototype out, again years head of estimates.

For difficulty - M/HREE. World also settled on PRC as functionally sole supplier for 5/6 signma purity minerals that PRC process has functionally 100% dominance in. Competitors at PRC EUV lab tech scale. That's just how market forces equalized sometimes before geopolitical disruption creates opening for new entrants.

Ultimately west see priority on REE, they're are allocating funds, they are also finding out one can't buy capability, and wanting something bad doesn't translate to getting it done. Political dysfunction is precisely relevant to the topic at hand, because political will determine what's possible at what speeds even when nation has the expertise and money.


I think western companies and governments have ingrained into their own thinking that the optimization strategy of of minimal investments in fundamental sciences and engineering as real constraint. (actually in more that just that, but that goes off topic..) It's a short term focused fictionalization / profit extraction constraint, but because that's so built into the experience and performance companies in the west, many predictions completely misunderstand what is possible with a different focus. We'll see how fast this can be re-calibrated.



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: