Speculation: Intel will become fabless

Page 9 - 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

TheGiant

Senior member
Jun 12, 2017
748
353
106
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 ...
well IMO Intel will become fabless by splitting to everything and manufacturing part
the pressures inside must be terrible for the CEO
they will soon have a big problem with talents..why wasting time as chip engineer when we can't make it and CEO is silent
if you look at Intel CEO talks then he for maybe a year making "softening" comments of the company status
 
  • Like
Reactions: Vattila

scannall

Golden Member
Jan 1, 2012
1,944
1,638
136
well IMO Intel will become fabless by splitting to everything and manufacturing part
the pressures inside must be terrible for the CEO
they will soon have a big problem with talents..why wasting time as chip engineer when we can't make it and CEO is silent
if you look at Intel CEO talks then he for maybe a year making "softening" comments of the company status
Between um 'early forced retirements' and heavy poaching Intel doesn't have the depth of talent they used to. Not saying they don't have some great engineers but the bench isn't near as deep as it used to be.
 
  • Like
Reactions: Vattila

Antey

Member
Jul 4, 2019
105
153
116
for that matter samsung is also in serious trouble. it's not like its 7nm node is widely employed. and its 5LP node is also facing difficulties. that only leave us with TSMC. remember 2011-2013? intel was unmatched, foundries like globalfoundries, samsung or tsmc were miles behind, nobody would have believed in that time that this could happen... but it happened. in 5 years intel could be ahead again like nothing happened... i don't think they will go fabless any time soon.
 
  • Like
Reactions: Vattila

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
@Vattila

@NTMBK

I . . . do not like Intel. But I like this even less. Intel either must form a strategic partnership with Samsung to save their foundry business, or they need to start licensing nodes from TSMC, even if that means yielding fab capacity to TSMC, or building fabs FOR TSMC state-side not only to address the security concerns of the US Gubmint (which they WILL have) but also to expand TSMC's wafer capacity. Intel is already capacity-constrained, so it will be difficult for their to yield fab capacity to TSMC as a part of a licensing agreement.

And I am changing my vote if possible. But I'm not sure Intel is going to go fabless so much as they are going to yield their (admittedly rather massive) fab capacity to someone else's nodes in one fashion or another.
 

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
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.

It's not just the market share they will lose, it's also that they will be forced to cut prices pretty significantly on the share they do keep. This rough stretch was going to be unavoidable, but with 7 nm now not an option it's going to be tough.
 
  • Like
Reactions: Vattila

PingSpike

Lifer
Feb 25, 2004
21,729
559
126
How well do you guys think porting Intel designs to another fab will actually work? My impression has been that if Intel has a secret sauce, its a tight coupling of design and process. When their process floundered the wheels fell off the whole thing. I know they've been trying to do more backporting (something competent leadership would have invested in back when 14nm had trouble getting off the ground) but that is still eating their own dogfood in the end, having a third party do the work seems like a different relationship.
 
  • Like
Reactions: teejee and Vattila

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,619
136
This is what I wrote in early May in another thread and which I feel still applies to this latest non-development:
Actually Intel's mass of die output will ensure its continued relevancy even having lost its process leadership. AMD can't and won't replace Intel's quantity. But Intel will have to get used to essentially becoming a commodity x86 chip manufacturer in the times they can't offer competitive process nodes.
x86 and foundry business is not purely about bleeding edge. It just happens to be against Intel's own image of itself to be a commodity manufacturer, so this will be a difficult transformation. Admitting their problems (and hopefully finally stopping with inane PR announcements like claiming industry leadership and reintroducing the new node every two years cadence) is a first step. This won't lead to Intel becoming fabless though.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
It worked for AMD.

I would ask @dmens to comment here (if possible), but Intel has a VERY different design culture than . . . anyone else. Not as many synthesized blocks, and other things. AMD had experience designing according to an outside fab's rules for years before Matisse.

@moinmoin

We're still talking about the semicon industry though. The world will move on from 14nm sooner or later. When it does, it'll be a mess. I really don't know how this will play out. How much more Cascade Lake can anyone really buy?
 
  • Like
Reactions: Tlh97 and Vattila

Vattila

Senior member
Oct 22, 2004
799
1,351
136
One analyst's view:

"On Thursday, Intel said it will outsource production to rivals like TSMC if it fails to meet its own targets – a humiliating prospect for a company that once led the industry in production technology. Susquehanna International Group analyst Christopher Rolland, a normally staid observer of the semiconductor market, said Intel needs even more drastic action. “We believe Intel has zero-to-no chance of catching/ surpassing TSMC (AMD partner) ... at least for the next half decade, if not ... ever,” Rolland wrote in a note to clients. “Intel should negotiate a sale of their fabs and equipment.”"


Edit: Here is Bernstein analyst Stacy Rasgon, pretty fired up, in an interview with CNBC, discussing the implications of the delay:

 
Last edited:

PingSpike

Lifer
Feb 25, 2004
21,729
559
126
Who would/could buy Intel's fabs? Will it be global foundries 2 where the fabs are spun off just to eventually fail? Could the fab be turned around in 5 years if some one bought it, purged the management and poached as much talent as possible?
 

maddie

Diamond Member
Jul 18, 2010
4,722
4,625
136
A question.

If Intel does decide to and succeed in using TSMC, will TSMC allow them access to the latest technologies if Intel still is in the fab business? I can't see how TSMC will allow this.
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
One analyst's view:

"On Thursday, Intel said it will outsource production to rivals like TSMC if it fails to meet its own targets – a humiliating prospect for a company that once led the industry in production technology. Susquehanna International Group analyst Christopher Rolland, a normally staid observer of the semiconductor market, said Intel needs even more drastic action. “We believe Intel has zero-to-no chance of catching/ surpassing TSMC (AMD partner) ... at least for the next half decade, if not ... ever,” Rolland wrote in a note to clients. “Intel should negotiate a sale of their fabs and equipment.”"

I really don't think the US government would allow this to happen, especially if a non American entity were to be interested. That is, if it comes to that. What I see happening is Uncle Sam stepping in and invoking some national strategic interest clause to pump some $$$$ into Intel's process business to help bring them to speed.

Edit: That is not to say money is the issue with Intel's failures up to this point, because one thing Intel has is a ton of money.
 
  • Like
Reactions: Vattila

dmens

Platinum Member
Mar 18, 2005
2,271
917
136
but Intel has a VERY different design culture than . . . anyone else. Not as many synthesized blocks, and other things. AMD had experience designing according to an outside fab's rules for years before Matisse.

The main barrier to Intel using external foundries is political, as with everything there. Intel has huge teams whose sole job is to exploit their supposed process advantage to tailor dedicated circuit and test solutions for their products. Going external would render those roles obsolete, so the VP’s in charge of those organizations would never let it happen.
 

DisEnchantment

Golden Member
Mar 3, 2017
1,590
5,722
136
A question.

If Intel does decide to and succeed in using TSMC, will TSMC allow them access to the latest technologies if Intel still is in the fab business? I can't see how TSMC will allow this.
To support Intel's wafer volumes, TSMC would have to expand significantly.
I am not sure how this will work properly. That is like Intel financing TSMC to expand capacity and research.

They could use AMD's approach. Intel will manufacture IOD and TSMC the chiplets :innocent:
 

maddie

Diamond Member
Jul 18, 2010
4,722
4,625
136
To support Intel's wafer volumes, TSMC would have to expand significantly.
I am not sure how this will work properly. That is like Intel financing TSMC to expand capacity and research.

They could use AMD's approach. Intel will manufacture IOD and TSMC the chiplets :innocent:
Won't Intel have access to TSMC latest secrets if they use the most advanced node? Why would TSMC do this when they can alternately bet on AMD displacing Intel at a pace they can realistically expand fab space as needed?

It would be like allowing AMD access to Intel fabs when they were dominant in process tech.

So many possible futures here.
 
  • Like
Reactions: Tlh97 and Vattila

JasonLD

Senior member
Aug 22, 2017
485
445
136
Won't Intel have access to TSMC latest secrets if they use the most advanced node? Why would TSMC do this when they can alternately bet on AMD displacing Intel at a pace they can realistically expand fab space as needed?

It would be like allowing AMD access to Intel fabs when they were dominant in process tech.

So many possible futures here.

TSMC customers doesn't get access to their manufacturing secrets.

While it is very unlikely, even the idea that TSMC influencing the competitive landscape bothers me.
 

JasonLD

Senior member
Aug 22, 2017
485
445
136
It's not just the market share they will lose, it's also that they will be forced to cut prices pretty significantly on the share they do keep. This rough stretch was going to be unavoidable, but with 7 nm now not an option it's going to be tough.

That will depends on how competitive their 10nm products is. If their Golden Cove lives up to the hype, they can weather through times of node disadvantages, if not, they should already plan for TSMC 3nm right now. We will find out next year.
 
  • Like
Reactions: Tlh97 and Vattila

gorobei

Diamond Member
Jan 7, 2007
3,654
980
136
A question.

If Intel does decide to and succeed in using TSMC, will TSMC allow them access to the latest technologies if Intel still is in the fab business? I can't see how TSMC will allow this.
foundry customers dont get a full breakdown of every process involved.

they get a spec sheet telling them the best practices: min size features need to be, voltages, resistance, inductance of traces, etc that they should design towards. they have the masks made and tsmc makes some test wafers. the customer adjust the mask design from there. the customer doesnt learn the nitty gritty secrets tsmc uses to form the layers during lithography.

if intel does get some new information they didnt know before from that guidance, then somebody at intel needs to be sacked.
 

maddie

Diamond Member
Jul 18, 2010
4,722
4,625
136
That will depends on how competitive their 10nm products is. If their Golden Cove lives up to the hype, they can weather through times of node disadvantages, if not, they should already plan for TSMC 3nm right now. We will find out next year.
Almost double the transistor density with better efficiency by your competitor and you still see them staying relevant as compared to TSMC clients? The absolute best case for Intel now is hoping to keep the single thread crown. Surely you can't see this as being enough.
 

blckgrffn

Diamond Member
May 1, 2003
9,110
3,028
136
www.teamjuchems.com
I just voted No.

The world needs two players, if not Intel then who?

I mean, it's not out of the question that TSMC will stumble/hit a wall. There was an analyst quote that they didn't think Intel could catch TSMC in five years. Fine.

TSMC and everyone should have given up around 2010/2011 then because catching Intel was hopeless, right?

Can't wait to revisit this thread in 2030 :D
 

JasonLD

Senior member
Aug 22, 2017
485
445
136
Almost double the transistor density with better efficiency by your competitor and you still see them staying relevant as compared to TSMC clients? The absolute best case for Intel now is hoping to keep the single thread crown. Surely you can't see this as being enough.

Well, AMD's 5nm isn't until late 2021~early 2022 for their server and client parts respectively. If Intel actually manages to get 7nm parts out by late 2022~early 2023, they can weather through it.
 

RasCas99

Member
May 18, 2020
34
85
51
I would ask @dmens to comment here (if possible), but Intel has a VERY different design culture than . . . anyone else. Not as many synthesized blocks, and other things. AMD had experience designing according to an outside fab's rules for years before Matisse.

@moinmoin

We're still talking about the semicon industry though. The world will move on from 14nm sooner or later. When it does, it'll be a mess. I really don't know how this will play out. How much more Cascade Lake can anyone really buy?
Not dmens :) , but ill give my 2 cents , regarding the first part of your quote "Not as many synthesized blocks" , almost the entire modern CPU is synth , humans cannot compete with automation due to the scale and complexity of modern CPU design , there is of course CMOS custom design in Intel , as there is CMOS custom design in any respectable design house (standard lib for anything from arbiters to queues , fifo`s , memories) , and of course Analog is never a synth i,e always custom (again , as with everyone else) , maybe you can do some custom design to a specific wide adder or something like that if you have a strict timing budget you cannot hold with synth , but its usually not how we do things unless its really a must , the big change happened few years back where they stopped using their own tools for backend (tools that you use to verify your static/dynamic timing of the design) and are now using Synopsis as are most of the industry , so moving to a different Fab is not as big of a deal as it used to be , you use the Process kit you get from the Fab that support the industry tools and run your design flows with that , RTL (the way you write the design from a logic POV) is the same with a need to move from Intel std lib to TSMC std lib , but Analog might be more work due to bigger reliance on process behavior , but again not a showstopper.

Man , I cannot believe we are even discussing this .... my first job out of uni was Intel so many years ago , so many fond memories and good times , they get a lot of flack around the internet and some of it justified but a lot of it is just ppl hating on a corporate as an entity and gloat when those entity struggle , but I had a blast working there with some great engineers that taught me a lot and just good people all around , so many startups and companies spawned from Intel engineers , seems like a eulogy or something , but it might be that Intel as we know is no longer a viable business , time for management to earn their salary which they failed to do for a while now , they need to make some good decisions and they need to make them now , there is no magic solution here , so anything they decide to do will have some drawbacks , but something needs to be done , hope they make them as some good ppl and families rely on those decisions.