After switching to ARM, expect Apple to buy TSMC, too
cringely.com
cringely.com
Furthermore, the United States government is pumping billions of dollars into domestic semiconductor manufacturing. If Apple wants a fab, they'll build one stateside, and they'll get Uncle Sam to pay for it.
But as has been seen by market realities of the last 15 years, affordable silicon manufacturing has been driven by massive scale across all market segments, not vertical integration. To the contrary, the most vertically integrated silicon manufacturers are the ones with the slower, hotter, and more expensive products. Acquiring a fab just for Apple products would probably not make better chips cheaper.
If anything, this business is too capital intensive for Apple to want to hold a foundry. Why lock yourself into one companies fabs like Intel did when you can pick between TSMC, Samsung, Intel and any other fabs that are competitive at the time?
AMD split off their foundry side. Intel is using Samsung's foundries more and more. Vertical integration does not make sense given the increasing cost of new fabrication nodes.
I am not sure what his motives are.
And worst of all, vast majority of the comments under the article agrees with him.
Every single day I am thinking, with or without Social Media, the Internet hasn't make more of the general public informed, which was one of its original intent and purposes, but the opposite. And in fact as I get older, it seems most of the world is very counterintuitive.
Well said, it was not better in the 'old days' but its also not better today. The Internet reminds me always at that scene in 7 (the movie) when he goes into the library and the watchman's are playing card.
GF has a decent 14/12nm node and they've basically said they have given up going smaller. That slashes their long term projections and valuation.
Apple comes in with a decent, but not great, offer while GF is still getting orders for their mature process. Then Apple does what only it can do, spend billions without blinking. Apple take 5 years to refine a 4nm node while using TSMC in the interim. R&D costs are offset with (1) whatever customers GF has, (2) building whatever it can on GF's 12nm node.