Intel Plunges as It Weighs Exit from Manufacturing Chips(bloomberg.com)
bloomberg.com
Intel Plunges as It Weighs Exit from Manufacturing Chips
https://www.bloomberg.com/news/articles/2020-07-24/intel-considers-what-was-once-heresy-not-manufacturing-chips
16 コメント
But there's already been a process slip.
> Intel’s current best technology, known as 10 nanometer in the industry, was scheduled to appear in 2017 and is only now making it into high-volume production. And when the company reported results on Thursday, it said the next iteration -- 7 nanometer -- would be delayed by a year.
So a three year delay for 10 nm and an indeterminate one for 7 nm all while TSMC already has 7 nm in production and is working on 5 nm.
If Intel does start outsourcing EUV, given the very large investments needed for that technology, I don't see how they would ever recover their lead.
Also this isn't just about Intel. The US invented the semiconductor industry and has been at it's forefront since the beginning. It looks like that is no longer the case and that the only company in the world with a cutting edge semiconductor fab capability is headquartered in a country China claims as its own.
> Intel’s current best technology, known as 10 nanometer in the industry, was scheduled to appear in 2017 and is only now making it into high-volume production. And when the company reported results on Thursday, it said the next iteration -- 7 nanometer -- would be delayed by a year.
So a three year delay for 10 nm and an indeterminate one for 7 nm all while TSMC already has 7 nm in production and is working on 5 nm.
If Intel does start outsourcing EUV, given the very large investments needed for that technology, I don't see how they would ever recover their lead.
Also this isn't just about Intel. The US invented the semiconductor industry and has been at it's forefront since the beginning. It looks like that is no longer the case and that the only company in the world with a cutting edge semiconductor fab capability is headquartered in a country China claims as its own.
> But there's already been a process slip.
They mean to say "another" process slip, but they've said that so many times it's starting to look bad. (And if their leadership was less clueless, they'd realize their mistakes there, but I've digressed....)
Another place the marketers have taken over is in the naming. It's widely accepted that TSMC 7FF "7nm" is the peer to Intel P1274 "10nm", and general consensus is that TSMC 5FF "5nm" is the peer to Intel P1276 "7nm".
They mean to say "another" process slip, but they've said that so many times it's starting to look bad. (And if their leadership was less clueless, they'd realize their mistakes there, but I've digressed....)
Another place the marketers have taken over is in the naming. It's widely accepted that TSMC 7FF "7nm" is the peer to Intel P1274 "10nm", and general consensus is that TSMC 5FF "5nm" is the peer to Intel P1276 "7nm".
> It's widely accepted that TSMC 7FF "7nm" is the peer to Intel P1274 "10nm"
I think there's one huge difference there though. As I understand it TSMC 7nm is an EUV process while Intel's 10nm is not.
So it looks like the bottom line is TSMC has mastered the next major fab technology and Intel has not.
I think there's one huge difference there though. As I understand it TSMC 7nm is an EUV process while Intel's 10nm is not.
So it looks like the bottom line is TSMC has mastered the next major fab technology and Intel has not.
TSMC's original 7nm process 7FF did not use EUV. They did introduce EUV for certain layers in N7+, their "second generation" 7nm process. Don't confuse N7+ with N7P, the performance-improved version of 7FF....
There's more information at https://fuse.wikichip.org/news/2567/tsmc-talks-7nm-5nm-yield... or https://community.cadence.com/cadence_blogs_8/b/breakfast-by... .
(This is why I always try to give the full name of the process, in as much specificity as I know it; it's very easy to talk at cross purposes about N7+ and N7P despite them being quite different.)
There's more information at https://fuse.wikichip.org/news/2567/tsmc-talks-7nm-5nm-yield... or https://community.cadence.com/cadence_blogs_8/b/breakfast-by... .
(This is why I always try to give the full name of the process, in as much specificity as I know it; it's very easy to talk at cross purposes about N7+ and N7P despite them being quite different.)
OK, but it still looks like N7+ has been shipping for almost a year so I think my main point still stands.
A friend of mine likes to quote the advice of an engineering professor: "The closer your product gets to the physics of a problem, the less money you're going to make."
With relevance to this situation, I guess the interpretation is that you can rarely make a lot of money when the competitive advantage is just that you have 18 months on a certain chip technology before others learn how to do it. Because the pure knowledge of how to do something can't keep you safe forever. It seems to be a constantly shrinking margin against others who are getting more and more sophisticated.
Maybe they saw this and recognized a losing game in part of their territory. Maybe more attention needs to be paid to what differentiating factor has become, even if it changes what Intel is about. I have no idea, but I hope they do!
With relevance to this situation, I guess the interpretation is that you can rarely make a lot of money when the competitive advantage is just that you have 18 months on a certain chip technology before others learn how to do it. Because the pure knowledge of how to do something can't keep you safe forever. It seems to be a constantly shrinking margin against others who are getting more and more sophisticated.
Maybe they saw this and recognized a losing game in part of their territory. Maybe more attention needs to be paid to what differentiating factor has become, even if it changes what Intel is about. I have no idea, but I hope they do!
The title is incorrect to say the least, Intel isn't weighing on Exit from Fab manufacturing at all. They are simply contemplating the idea of outsourcing.
I think it needs a little more context regarding the reaction from analyst.
>“You didn’t need to read any more,” Sanford C. Bernstein analyst Stacy Rasgon said. “Whatever little credibility they had is out of the window.”
Few months ago Intel CEO in an Investor meeting promised 7nm by late 2021, and to regain its lead in manufacturing by 2023 with 5nm GAA ( "Gate All Around" also known as Nanowire ). That is not entirely impossible considering TSMC is scheduling their take on GAA with 2nm in 2024. So Intel would be one year ahead again. Volume will be an order of magnitude smaller but at least Intel could get that headline win.
So in less than 4 months, Intel is now saying the 7nm is delayed by 6 months, and dont expect 7nm consumer product until late 2022 and 2023.
That is why they have loss all credibility. They burned most of it with 10nm, and whatever is left is now gone with this report as well.
Also note something that is not discussed anywhere, ( At least as far as I can tell ). Why a 6 months delay of 7nm, initially scheduled late 2021, then suddenly becomes "dont expect 7nm consumer product until late 2022 and 2023. " Which is 12-14 months delay.
I think it needs a little more context regarding the reaction from analyst.
>“You didn’t need to read any more,” Sanford C. Bernstein analyst Stacy Rasgon said. “Whatever little credibility they had is out of the window.”
Few months ago Intel CEO in an Investor meeting promised 7nm by late 2021, and to regain its lead in manufacturing by 2023 with 5nm GAA ( "Gate All Around" also known as Nanowire ). That is not entirely impossible considering TSMC is scheduling their take on GAA with 2nm in 2024. So Intel would be one year ahead again. Volume will be an order of magnitude smaller but at least Intel could get that headline win.
So in less than 4 months, Intel is now saying the 7nm is delayed by 6 months, and dont expect 7nm consumer product until late 2022 and 2023.
That is why they have loss all credibility. They burned most of it with 10nm, and whatever is left is now gone with this report as well.
Also note something that is not discussed anywhere, ( At least as far as I can tell ). Why a 6 months delay of 7nm, initially scheduled late 2021, then suddenly becomes "dont expect 7nm consumer product until late 2022 and 2023. " Which is 12-14 months delay.
This topic seems to be of prime importance to the US tech industry. I'm very surprised no post on this has yet made it to the front page.
It's been discussed pretty extensively: https://news.ycombinator.com/item?id=23932082
Well that's odd. I've been scanning the front page quite frequently since I heard about this on Friday and yet somehow missed this.
Will real estate next to their fabs be the next thing to plunge? There are going to be many unemployed engineers in a concentrated area with minimal transferable skills to local companies outside of the fab.
This is bad for our country.
In terms of security, Intel has been bad for a while. The last good chip was the Intel Core 2 Duo. Everything since then has had basically spyware on the processor via the Intel Management Engine. It doesn't matter if it was manufactured/designed in the US or not when you have exploitable proprietary software running on your processor. The only way to truly be free and have a decent insurance of security is with a fully open source, free system which includes your processor and bios.
> “To the extent that we need to use somebody else’s process technology and we call those contingency plans, we will be prepared to do that. That gives us much more optionality and flexibility. So in the event there is a process slip, we can try something rather than make it all ourselves.”