The only way the logic makes sense is if people think there's some massive use case for sub 5nm chips that literally can't be substituted by older nodes and it's not just power efficiency concerns on mobile devices. Where every year they delay 1nm chip production in China means China falling behind 5+ years in AI development and rollout.hpvc's point as I understand it is that by the time Chinese EUV enters production in such a timeline TSMC will probably be introducing 1 nm while China is at best *presumably* only still on 5 nm. I'm a bit skeptical of that kind of prognosis though, because what we're seeing with 3nm is that the development time got effectively delayed by a year, and because Apple is basically purchasing all of TSMC's 3 nm capacity for the next year (which btw implies yield limitations), anyone not named Apple won't be able to start getting 3 nm chips until 2024-25. That itself is effectively a 4-5 year cycle time going from 5 nm to 3 nm. TSMC has already started work on their 2 nm process but given delays to their 3 nm process I'm pretty skeptical that they won't see similar if not worse delays for moving onto their 2 nm process given the smaller feature size, more EUV multi-patterning, and new more complex transistor architectures needed, and that's if it even makes economic sense to try to maintain the fastest iteration speed possible given how small the customer base for 2 nm is likely to be given the ever increasing cost per node shrink.
It's a tech singularity fantasy and people want to believe.