Re:Intel Announces Retirement of CEO Pat Gelsinger
by ksec on Tuesday December 03, 2024 @09:08AM
(#42306152)
>1. Intel 18a is being delayed further and/or there are significant issues that will hamstring performance.
My guess is not 18A being delayed. It is simply not getting any / enough customer on board. And you cant simply spend tens of billions innovate to 18A, only to realise you dont have a customer to even recoup half the cost. Especially when TSMC is playing their card right. Which I know they will.
Getting into foundry business will always be difficult. And the benefits of using 18A with Intel, even if it was on schedule and 6-12 months ahead of TSMC, does not out-weight that TSMC will always be cheaper due to tooling familiarity etc, and even at the same cost, TSMC will have better volume to get go simply due to its size, and even with the same volume, TSMC is pure play, while working with Intel would mean a potential of IP theft. And even after ALL that, on a hypothetical Intel Foundry only, a lot of packaging and testing for your chip are all located in East Asia.
What I wrote above is essentially the same thing when I wrote it on Anandtech or SemiWiki 10+ years ago. The time when I got death threat from Intel Fan boys about TSMC becoming number one foundry overtaking Intel. Nothing has changed. Had Intel started working on it in 2014 / 2015, when Intel still has the CPU monopoly where it could used those profits to fund and change direction, Intel would be in a much much better position.
Unfortunately all those board wanted was dividends and buy back.
My guess is not 18A being delayed. It is simply not getting any / enough customer on board. And you cant simply spend tens of billions innovate to 18A, only to realise you dont have a customer to even recoup half the cost. Especially when TSMC is playing their card right. Which I know they will.
Getting into foundry business will always be difficult. And the benefits of using 18A with Intel, even if it was on schedule and 6-12 months ahead of TSMC, does not out-weight that TSMC will always be cheaper due to tooling familiarity etc, and even at the same cost, TSMC will have better volume to get go simply due to its size, and even with the same volume, TSMC is pure play, while working with Intel would mean a potential of IP theft. And even after ALL that, on a hypothetical Intel Foundry only, a lot of packaging and testing for your chip are all located in East Asia.
What I wrote above is essentially the same thing when I wrote it on Anandtech or SemiWiki 10+ years ago. The time when I got death threat from Intel Fan boys about TSMC becoming number one foundry overtaking Intel. Nothing has changed. Had Intel started working on it in 2014 / 2015, when Intel still has the CPU monopoly where it could used those profits to fund and change direction, Intel would be in a much much better position.
Unfortunately all those board wanted was dividends and buy back.
Yes. Fuck All Intel's board members.