Zarathustra[H]
Extremely [H]
- Joined
- Oct 29, 2000
- Messages
- 38,961
Yeah, some of the issues may resolve by then, but Fab capacity I expect will be a problem for years to come. Having only two in the space is a problem since Intel started struggling, and GloFo just gave up, that's definitely problematic. Any expansion of capacity in terms of new fabs, or new players, or Intel fixing their stuff, etc. doesn't happen over night and will be a multi year affair.
Mostly agreed.
Developing new fabs and Intel fixing their process certainly takes a lot of time, but believe a lot of this work has already been ongoing for some time, so it's not like they are suddenly starting from scratch in 2021. If that were the case we'd be several years off.
TSMC has been looking to expand for some time, so some of the groundwork has already been done. They also already have the best working process in the industry up and running, so it's not like they are developing a new process from scratch. They already have a pretty good idea of what works and what doesn't.
Intel too has been working on backup plans ever since 10nm went south.
I have less enthusiasm that there will be any new entrants. In business we often talk about barriers to entry, things that make it difficult to enter a new industry, and it is tough to imagine any larger ones than in silicon fabs. The levels of expertise, and cost are immense, operating at the very edge of what technology makes possible. The costs are upfront, and you need to keep highly skilled engineers on staff for several years before you start seeing payoff.
I think this is why supply has trailed in recent years. If experienced fab owners like Global Foundries which are made up - among others - what used to be AMD's in house manufacturing are dropping out due to the extreme difficulty and cost, it seems highly unlikely we will see anyone new enter.
For latest gen process nodes I think we are stuck with TSMC Samsung and Intel for the foreseeable future, and I hope they expand capacity.