InvestorsHub Logo
icon url

Unkwn

04/23/15 2:19 AM

#140479 RE: VeeCee #140478

I agree. The ones that have the most to lose are TSMC and Qualcomm. They had a great run and, as always, now have to fight to keep their big market share of a profitable business. Samsung is probably the most powerful in this game, though they have some issues with their mobile phones/end products, so as an investment, I don't know.

And then there is Intel, much to gain, but until now, they didn't prove they can do this. No success in the foundry business yet and losing money in mobile with no entry in the phone business so far. For me, this is a question of potential. Intel has the highest potential in this group, in my opinion, since they have the highest knowledge of the group in performance computing, where mobile is getting to, as well as the leading process technology. The potential is there, but execution has been lousy in recent years. We'll see if there's a changing attitude at Intel which can finally turn this company into a cash generating behemoth. Until then, the PC and server business will provide a bottom, so the risk isn't high in my opinion.
icon url

Dmcq

04/23/15 5:19 AM

#140480 RE: VeeCee #140478

TSMC is skipping 20nm


TSMC used 20nm for the Apple orders. They didn't skip it.

I think it was however always seen as a transitional step rather than something for the long term. Their 16nm processes will make them lots of money though not as much as if they had been already in production. Their schedule now is their original one, I think they tried to bring the schedule forward for Apple but obviously that didn't work out.

Samsung overtaking TSMC is impressive. The move of Liang Mong-song to Samsung can't have helped TSMC but it takes a lot more than something like that to achieve such an advance. Their fight over 10nm should be interesting, I think TSMC will take the lead again there but really it is all up for grabs.
icon url

mas

04/23/15 6:59 AM

#140482 RE: VeeCee #140478

Samsung 14nm is not equivalent to Intel 14nm.

This can be quantified.

TSMC 16FF has 90nm gate pitch, 64nm minimum metal pitch.
Samsung 14FF has 78nm gate pitch, 64nm minimum metal pitch.
Intel 14nm has 70nm gate pitch, 52nm minimum metal pitch.

If transistor area can be compared by the ratio of (gate pitch * minimum metal pitch) you get the following relative transistor area sizes

TSMC 16FF - 1.58
Samsung 14FF - 1.37
Intel 14nm - 1.00

It's obvious that the foundries initial Finfet processes are roughly a half-node larger than Intel's 14nm in density.