Speculation: Intel will become fabless

Page 8 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

With the loss of its manufacturing lead, will Intel become fabless?

  • Yes, Intel is a product designer at heart, and they will seek a more flexible fabless model.

    Votes: 24 13.4%
  • No, manufacturing is integral to Intel, and they will continue to invest to stay competitive.

    Votes: 155 86.6%

  • Total voters
    179

soresu

Platinum Member
Dec 19, 2014
2,612
1,811
136
Loongson/Godson has mostly flopped
Owing to the death of MIPS I would say.

ARM's ascension was catastrophic to the popularity of MIPS.
If you discount whole SoC development, though, then you get a narrower picture of what is going on in China
I'm not discounting it in general, just from the conversation I was having with BuzzKiller, which was more about uArch at the point you joined in.

As I said, I support HiSilicon/Huawei for keeping both ARM high end designs in use.

I have a Mate 10 myself and excepting the extremely annoying lack of headphone port and microSD slot (neither of which was made obvious at the point of purchase) it is a fairly decent phone.

I wish someone like Huawei would crank out a high end Kirin based Shield TV competitor, as the other options on the market are either painfully anaemic in performance or software constrained - even the new STV model seems to use an overclocked variant of the original A57 based TX1 chip (now 4.5 years old).
 

soresu

Platinum Member
Dec 19, 2014
2,612
1,811
136
China is going to go for what's ubiquitous first and effective second. ARM would more-or-less fit that bill.
More than certainly, as much as I love my x64 PC it's pretty obvious at this point that smartphones and tablets rule the world in sheer numbers (at least for consumers), and ARM is the name of the game in those markets.

What surprises me is the lack of Google ever having weighed in personally on ARM development when they seem to compete (or at least attempt to) with Apple.
I wouldn't expect them to try to replace Intel's x86 with a homegrown ISA, though.
Didn't they make a new ISA called Ucore or something similar sounding?

I remember it making news and then sinking without a trace.
 

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,618
136
I wouldn't mind seeing an Intel Core design on TSMC fabs, 7nm would be fine. Just as sort of a limited-run test chip, of sorts. Maybe they (Intel) could put them into a new lineup of NUCs.

Would they be cheaper than Intel's current 8C Skylake-derivative CPUs, or more expensive? (Tradeoffs, smaller die on 7nm, maybe they could enlarge the L3 cache slightly), more chips/wafer, but 14nm is largely depreciated, cost-wise, I expect.
Even when entertaining the possibility of Intel going to TSMC for its Core, a limited-run test chip would be the worst way to go about it. What makes new nodes increasingly costly are not the per wafer/die cost (those increase too, mind) but the huge upfront costs for design, mask, verification etc. The production run needs to amortize that, a limited run does the opposite of that. In comparison everything on Intel's 14nm is amortized since ages and much cheaper/more profitable as a result.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Didn't they make a new ISA called Ucore or something similar sounding?

Hmm, cursory searching reveals nothing. You might be right though. Regardless, that's the sort of things that happen to radically new ISAs, even those pushed by the Big Boys of Design (Itanium, heh).

@moinmoin

I think what you are saying is 99.99999% true. Intel would still need to do some work to see if they could port their allegedly-portable designs to other foundries should such a thing become necessary (such as Sunny Cove on Samsung 7nm or whatever). At this point I think Intel would rather just burn cash on 10nm and 7nm and see if that works out, though. They probably should have given themselves the flexibility to use external fabs just like they were supposedly doing with their now-defunct line of 5G modems. Alas, it was not meant to be.
 

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
So, great article on Anandtech regarding 10 & 7nm : https://www.anandtech.com/show/15032/intel-2019-fab-update-10nm-hvm-7nm-on-track

One salient point is this
Right now, the company produces all of its 10 nm CPUs in Hillsboro, Oregon and Kiryat Gat, Israel. Starting next quarter, the company expects 10 nm chips to also ship from its Chandler, Arizona, fab, which will increase supply of Ice Lake processors and will get Intel prepared for a broader use of the manufacturing process that had caused company a lot of troubles.

4 fabs will be running 10nm - Intel must have fixed most of their 10nm yield issues, otherwise there is no way they'd be dropping that kind of money on it.
 
  • Like
Reactions: Vattila

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
4 fabs will be running 10nm - Intel must have fixed most of their 10nm yield issues, otherwise there is no way they'd be dropping that kind of money on it.

Guess we'll find out soon enough wrt IceLake-SP and Sapphire Rapids. And the various Tremont variants.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
I will say one thing. I have NEVER seen so large a MFG facility as the Hillsboro one. Its is the size of Disneyland AND California adventure. An d some of the towers (not sure their use) at like 10 stories tall, and the building is so big I can't even estimate the size, but its like 6 stories tall. And the housing growth around here is insane. I feel like I am California sometimes, there are so many out of state plates (mostly California).

Now that I know its the 10nm fab, that explains a lot.

Here is a pic I googled, looks smaller from the air

And thats dated Sept 2016 !

I also found this:
"The company has hit a new high at more than 20,000 employees in the state, where Intel has long been one of Oregon's biggest private employers. "

And that 20,000 must virtually all be in this one fab !
 
Last edited:

extide

Senior member
Nov 18, 2009
261
64
101
www.teraknor.net
That whole complex in Hillsboro, Oregon consists of several fabs -- D1B, D1C, D1D, D1X, RB1, and RP1 which produce stuff in 22nm, 14nm, 10nm, 7nm, and future research nodes. A massive site indeed.
 

Nothingness

Platinum Member
Jul 3, 2013
2,371
713
136
I will say one thing. I have NEVER seen so large a MFG facility as the Hillsboro one. Its is the size of Disneyland AND California adventure. An d some of the towers (not sure their use) at like 10 stories tall, and the building is so big I can't even estimate the size, but its like 6 stories tall. And the housing growth around here is insane. I feel like I am California sometimes, there are so many out of state plates (mostly California).

Now that I know its the 10nm fab, that explains a lot.

Here is a pic I googled, looks smaller from the air

And thats dated Sept 2016 !

I also found this:
"The company has hit a new high at more than 20,000 employees in the state, where Intel has long been one of Oregon's biggest private employers. "

And that 20,000 must virtually all be in this one fab !
There are half a dozen fabs on this site. Also the next large Intel fab will host less than 2000 employees.

And are design and software teams in Hillsboro. This amounts to several thousands of employees.

So no, there are not 20k employees in this fab alone :) But yeah the site is incredibly massive.
 
  • Like
Reactions: Markfw

Vattila

Senior member
Oct 22, 2004
799
1,351
136
Does this change the poll, maybe? (Currently at 7% for and 93% against the proposition.)

Intel 7NM Slip Causes Reassessment of Fab Model

"During the earnings call, management made it quite clear that they are looking at alternatives for manufacturing of future nodes. Whether or not to outsource and how much to outsource to TSMC. It seems from the tone of tonight's call coupled with the 7NM slip that Intel is on the slippery slope to give more of its manufacturing to TSMC and perhaps TSMC will get to do Intel’s most leading edge manufacturing as Intel falls further behind."


Also:


Just last quarter Intel presented a plan to get to parity at 7nm and regain process leadership at 5nm. With this delay, I presume that is now unrealistic. TSMC will be at N3 in the same time frame. Is Intel's historical manufacturing dominance lost forever? Will it lead to Intel transitioning to a fabless model? Will they be spinning off their fabs into a subsidiary joint venture with someone else, perhaps?
 
Last edited:
  • Like
Reactions: CHADBOGA

JasonLD

Senior member
Aug 22, 2017
485
445
136
Just last quarter Intel presented a plan to regain process leadership at 7nm. With this delay, I presume that is now unrealistic. TSMC will be at N3 in the same time frame. Is Intel's historical manufacturing dominance lost forever? Will it lead to Intel transitioning to a fabless model? Will they be spinning off their fabs into a subsidiary joint venture with someone else, perhaps?

I think Intel said parity at 7nm and leadership at 5nm. That doesn't sound realistic now.
Regardless of what Intel ends up doing, I hope every CPU vendors doesn't have to rely on TSMC exclusively on leading edge node and hope Samsung makes come back on 3nm.
 

Vattila

Senior member
Oct 22, 2004
799
1,351
136
I hope every CPU vendors doesn't have to rely on TSMC exclusively on leading edge node and hope Samsung makes come back on 3nm.

Perhaps, as I suggested in my OP, Intel and GlobalFoundries should get together and have a talk. GF did a lot of good research on 7nm that they have shelved.
 

awesomedeluxe

Member
Feb 12, 2020
69
23
41
I mean, how much confidence do we have in Samsung's 5nm node and Gate-All-Around 3nm node? What kind of volume can they realistically handle?

Because I am not sure TSMC would have capacity to supply Intel on N5P even after Apple has completely moved off of it. AMD already has their orders in. nVidia probably hasn't picked their 2021 darling yet but they have the financial capacity to compete with Intel if it turns out N5P is the only option. This is not a problem Intel wants to have.

It's also not a problem the US government wants Intel to have. The US government does not like the idea of there being exactly two competitive foundries with both in Southeast Asia. They are hedging their bets with the TSMC megafab but you'd better believe they are still rooting for Intel to get back in the game somehow. I think there will be opportunities for them to do so.
 
  • Like
Reactions: Tlh97 and Vattila

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Regardless of what Intel ends up doing, I hope every CPU vendors doesn't have to rely on TSMC exclusively on leading edge node and hope Samsung makes come back on 3nm.

That's not a bad thing for process. Volume is likely what determines how it performs.

Intel's process problems go back all the way to 22nm, when it came out 6+ months later than expected, and their bread-and-butter notebook/desktop chips didn't perform as well as expected.

I think the problems go even further back though, and that started with the iPhone kickstarting the Smartphone industry. They lost volume leadership against TSMC sometime in 2009-2011, and subsequently they started having issues with their process.

The more customers you have, the more flexible and diversified you have to be in your approach, which in turn enriches your process tech. Also if you are the leader, top tier engineers and graduates gravitate towards you, furthering the potential.

All this bodes very badly for Intel's process team. Losing Apple, 7nm delays, their latest "big core" CPUs on par with Smartphone chips. Even their own Ponte Vecchio is defecting to the competitor which results in further volume loss.

Any more delays/problems with 7nm, and I can see sometime in the 5/3nm generation, them moving most of their product line to TSMC.
 
Last edited:

JasonLD

Senior member
Aug 22, 2017
485
445
136
I mean, how much confidence do we have in Samsung's 5nm node and Gate-All-Around 3nm node? What kind of volume can they realistically handle?

5nm could be alright if they fixed their issues from 7nm, but they are taking a major risk on GAA 3nm node. Samsung's 3nm is probably close to TSMC 5nm than 3nm in terms of density but theoretically, performance might be better with GAA, so I really hope they can get it right though I am not sure if they can deliver it on 2022.

That's not a bad thing for process. Volume is likely what determines how it performs.

Intel's process problems go back all the way to 22nm, when it came out 6+ months later than expected, and their bread-and-butter notebook/desktop chips didn't perform as well as expected.

I think the problems go even further back though, and that started with the iPhone kickstarting the Smartphone industry. They lost volume leadership against TSMC sometime in 2009-2011, and subsequently they started having issues with their process.

The more customers you have, the more flexible and diversified you have to be in your approach, which in turn enriches your process tech. Also if you are the leader, top tier engineers and graduates gravitate towards you, furthering the potential.

All this bodes very badly for Intel's process team. Losing Apple, 7nm delays, their latest "big core" CPUs on par with Smartphone chips. Even their own Ponte Vecchio is defecting to the competitor which results in further volume loss.

Any more delays/problems with 7nm, and I can see sometime in the 5/3nm generation, them moving most of their product line to TSMC.

Yeah, Intel not taking Apple's offer is probably the worst decision they ever made last decade.
Though TSMC's volume enables advantages on future node research, having a legitimate foundary competitor is good for entire industry. Here's hoping.

Another delay and you probably won't even see a 7 nm product make it to market.

Nah, TSMC alone can't handle entire Intel's volume. They will rely on their own fab for most of their volume product even if their leading edge chips are fabbed in TSMC.
 

JasonLD

Senior member
Aug 22, 2017
485
445
136
It's going to get to the point where Intel simply won't have the money to continue with the Fab R&D. That's really the problem.

Unless Intel's revenue hits heavy decline in next couple of years, I don't think that will be the case yet. Don't be so overdramatic.
 
  • Like
Reactions: Vattila

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
Unless Intel's revenue hits heavy decline in next couple of years, I don't think that will be the case yet. Don't be so overdramatic.

You should expect heavy revenue declines. And soon, I think we are at the point where the hyperscalers are starting to roll off of Cascade Lake and on to Rome. And it's only going to snowball from there. AMD's going to have to screw this up pretty badly to not gain serious market share.
 

JasonLD

Senior member
Aug 22, 2017
485
445
136
You should expect heavy revenue declines. And soon, I think we are at the point where the hyperscalers are starting to roll off of Cascade Lake and on to Rome. And it's only going to snowball from there.

Nah, that isn't the industry where dramatic shift happens. Rome will definitely get more marketshare, but not to the point of causing heavy decline on Intel side.
 
  • Like
Reactions: Vattila

ThatBuzzkiller

Golden Member
Nov 14, 2014
1,120
260
136
Intel has no choice as even Samsung is moving ahead of them in logic fabrication technology!

There will be plenty of capacity from TSMC and Samsung to make a transition from their dated 10nm process technology. Samsung could handle handle Intel's low-end products while TSMC focuses on Intel's high-end portfolio ...

The US government won't like leading edge process technology to be out of their control or the loss in American high-tech manufacturing industry one bit ...
 

Failnaught

Member
Aug 4, 2008
26
25
91
You should expect heavy revenue declines. And soon, I think we are at the point where the hyperscalers are starting to roll off of Cascade Lake and on to Rome. And it's only going to snowball from there. AMD's going to have to screw this up pretty badly to not gain serious market share.

Intel already lost Apple x86 business plus AMD is leading performance wise. Intel's profit margin dropped this quarter - this is telling. They are competing on price now. And then on the server side there are multiple ARM challengers. It's not looking good for Intel long term.

Nah, that isn't the industry where dramatic shift happens. Rome will definitely get more marketshare, but not to the point of causing heavy decline on Intel side.

I think it's somewhat like the decline of Blackberry. Once upon a time everyone had one. After iPhone came out, even more people had one because smartphones are cool. But then a few years later, Blackberry doesn't even sell phones anymore. These shifts don't happen over 1 quarter but over a decade it can completely shift.
 

JasonLD

Senior member
Aug 22, 2017
485
445
136
I think it's somewhat like the decline of Blackberry. Once upon a time everyone had one. After iPhone came out, even more people had one because smartphones are cool. But then a few years later, Blackberry doesn't even sell phones anymore. These shifts don't happen over 1 quarter but over a decade it can completely shift.

Well, While I don't like comparing phone industry to semiconductor industry, imagine if both Blackberry and iPhone has been running same OS, with same software ecosystem, outcome would have been little different. :)