Hacker News Reader

Comment by 55873445216111 | Hacker News Reader

It takes >$10B and >2 years to build a fab which will be effectively obsolete ~2 years after it opens. If it doesn't open on schedule or is underloaded, the ROI goes negative almost immediately. (I somewhat oversimplified this of course but it's to make the point.)

View on HN

Obsolete? Automakers etc are huuuungry for old chips.

Yes but there is a huge technology gap. example: Automakers need a lot more small low-power MCUs built on ~130/90nm process. A 14nm fab that started 2 years ago is never going to be producing those 130/90nm MCUs. Furthermore, the semi toolmakers that produce the tools needed to build 130/90nm chips stopped making those tools years ago. This is a rabbit-hole that just keeps going. The problem isn't going away soon or easily.

Maybe three different 'bands' of production:

  * cutting / bleeding edge (as now)
  * 80/20 trailing edge (5 year LTS of cutting edge)
  * Cost focused 'lifetime' LTS

It seems like the cutting edge is built, and transitioned to the 80/20 already. It's that cost focused bulk components / entry level that lacks investment or even planning.

Maybe that's where some industry development and co-ordination from regulators could help establish a market that's healthy.

Not just automakers, and not just old fabs. Not everything needs the latest bleeding edge process. So, even then some of the old processes get updated to support larger wafers.

Ex: brand new 65nm fab:

https://www.electronicsweekly.com/news/business/bosch-opens-...

So, 10-15 year ROI easy, and I'm guessing that might stretch even further as the trend to ever smaller nodes slows. AKA I'm betting 2nm doesn't look nearly as aged in 2040 as 90nm looks today.

Reply on news.ycombinator.com