Speculation: Intel will become fabless

[ArchAngel777]

I don't see Intel going the way of AMD unless they severly botch 7nm. None of us quite understand (only an insider would know) why they failed so horribly at 10nm (lots of theories, but I am not sure we know for sure aside from generic: they were too aggressive) and yet they still have a shot at 7nm competing with TSMC 5nm at around the same time... But time will tell. Intel is a remarkable company when it has proper leadership. Hoping they can get their manufacturing woes figured out and be competitive again in manufacturing.

 

[thesmokingman]

I don't see Intel going the way of AMD unless they severly botch 7nm. None of us quite understand (only an insider would know) why they failed so horribly at 10nm (lots of theories, but I am not sure we know for sure aside from generic: they were too aggressive) and yet they still have a shot at 7nm competing with TSMC 5nm at around the same time... But time will tell. Intel is a remarkable company when it has proper leadership. Hoping they can get their manufacturing woes figured out and be competitive again in manufacturing.

One worth, monolithic.

AMD knew they could not compete on fabrication and process. And they could not produce or design monolithic die's like Intel. And as it happens we are nearing the technical limits thus they went to their modular design. Imo Intel doesn't have any shot at anything if they continue with monolithic designs. When crap happens, you cannot course correct with a monolithic design. They cannot even fix their mitigations in hardware, heck not even the ones that are a decade old. That should make their problem painfully obvious imo. They're gonna have to join the club imo if they are going to overcome their current situation. And for the record I don't believe they have 7nm. That's just talk. Talk like 10nm delays after delays all the while they keep making press releases about how 10nm is improving only axe the lineup in reality.

 

[jpiniero]

I don't see Intel going the way of AMD unless they severly botch 7nm. None of us quite understand (only an insider would know) why they failed so horribly at 10nm (lots of theories, but I am not sure we know for sure aside from generic: they were too aggressive)

They went too dense without using EUV is pretty much what happened, Intel admitted as much.

 

[DrMrLordX]

Or eat crow and license it from someone else, as GF did from Samsung for 14nm.

If Intel attempts to license tech, the only way they can do it is to effectively cede control of their foundry business to TSMC and/or Samsung. Give us your process and we give you time at our fabs. Plain and simple. It would be a less-drastic form of what IBM did with GF.

The alternative is to keep Intel's fabs on 14nm in perpetuity. If you were Intel, which would you want?

 

[soresu]

The alternative is to keep Intel's fabs on 14nm in perpetuity. If you were Intel, which would you want?

That sounds like a question of pride over profit, something tells me that their shareholders would not hold to similar opinions.

Their fabs are only useful while they help them turn a greater profit - there may well be a time when Intel decide to either go the AMD/GF route, or to beg TSMC for terms of licensing simply in order to maintain relevance in the marketplace.

Such a licensing deal is not quite so unlikely as you may think.

TSMC make no products of their own design (not for the wider global market anyway), therefore as long as Intel would agree to only fab their own designs with TSMC's licensed tech, it would only add to TSMC's bottom line.

Of course, that doesn't account for a possible drop in orders due to competition against TSMC's customers like AMD, nVidia and ARM/Qualcomm on various fronts - I imagine such a deal would be a fairly complex beast if it even could be worked out.

 

[chrisjames61]

Could somebody with technical expertise explain in laymans terms how a company with such a dominant postition as Intel got themselves into such a predicament? TSMC must be awash in money fabbing for Apple. Is the real reason Intel is in such a bad position due to the fact mobile is overtaking desktop cpu's and they basically have been locked out of the mobile business?

 

[NostaSeronx]

At this point, Intel is more likely to keep the Fab. While starting to acquiring more fabs for a larger pool of Custom Foundries.

Buying these fabs:
-> GlobalFoundries
-> UMC
-> HSMC India contract(that recently was revoked from receiving government funding)

 

[happy medium]

Intel will come back

They have good products and they have the X86 with them

Come back from what? Profitable, profitable, profitable, quarter after quarter. They own most of the market and by next year this time, will start to steamroll AMD again for another 5 to 10 years.
AMD'S 15 MINUTES of fame has about 5 minutes left

 

[Bouowmx]

I don't see fabless Intel likely in the near future, which rides on 7 nm coming 2022. Intel 7 nm projection(s) (edit: 1 2) show transistor density comparable to foundry 3 nm (200-250 M/mm^2).

 

[chrisjames61]

Come back from what? Profitable, profitable, profitable, quarter after quarter. They own most of the market and by next year this time, will start to steamroll AMD again for another 5 to 10 years.
AMD'S 15 MINUTES of fame has about 5 minutes left

Don't you realize that Intel has way more than AMD to worry about? If you don't think Intel has severe problems just because they are profitable you must have blinders on.

 

[ThatBuzzkiller]

What, steal India?

You know very well what I meant ...

Why do you imply that only Intel (or AMD) will be able to capitalize on a growing market such as India in the future ? China too has potential to create competitive x86 cores to capitalize these markets as well ...

 

[insertcarehere]

Don't you realize that Intel has way more than AMD to worry about? If you don't think Intel has severe problems just because they are profitable you must have blinders on.

^^^
If AnandTech's projections are accurate, A77-equipped SoCs will put serious, serious pressure on a lot of Intel and AMD's laptop cpus from next year onwards.

 

[Markfw]

Come back from what? Profitable, profitable, profitable, quarter after quarter. They own most of the market and by next year this time, will start to steamroll AMD again for another 5 to 10 years.
AMD'S 15 MINUTES of fame has about 5 minutes left

Profitable ? Yes, loosing market share YES Loosing volume ? Yes.
Like AMD did 10 years ago, you can only hide so long. 5 minutes left ? At least 2 years, and the way things are going, a very good chance it will be much longer than that.

Boy you like to troll AMD

 

[UsandThem]

https://www.tomshardware.com/news/intel-10nm-7nm-process-leadership-xe-graphics-gpu-dg1

It sure sounds like Intel has no intention of going fabless any time soon.

On the process side, Intel has three investments going on with 14nm, 10nm, and 7nm. Intel increased its capital expenditures forecast for 2019 by $500 million to $16.0 billion "as a result of increased 10-nanometer and 7-nanometer investments."

Intel has invested in record levels of capital over the last two years totaling over $30 billion, which consists of usual production capacity as well as additional capacity investments. Intel says it has increased its 14nm capacity by 25% this year while also ramping 10nm, and expects to increase capacity by another 25% in 2020.

 

[soresu]

China too has potential to create competitive x86 cores to capitalize these markets as well ...

Just as India has the potential to create RISC-V cores, and indeed they are already.

China only got this far so fast with x86 because of Hygon - obviously I'm not saying it can't be done independently, but given the minefield that is x86, it's alot easier when you have an AMD shaped space elevator to skip years of core R&D and ease ISA conformance.

The sheer amount of instructions, and instruction sets alone in x86 makes it a clusterfork of leviathan proportions to design for.

 

[NostaSeronx]

China only got this far so fast with x86 because of Hygon ...

Zhaoxin's 7nm x86 product is faster than Hygon though. It is also fabbed "natively" through TSMC.

 

[soresu]

Zhaoxin's 7nm x86 product is faster than Hygon though.

How does Dhyana compare to Zen1 though?

Not really fair to compare across a whole nodes worth of difference either.

It is also fabbed "natively" through TSMC.

?

Even if it's a TSMC fab in mainland China, it is not "native", it is owned by TSMC and not the Chinese government.

 

[chrisjames61]

How does Dhyana compare to Zen1 though?

Not really fair to compare across a whole nodes worth of difference either.

?

Even if it's a TSMC fab in mainland China, it is not "native", it is owned by TSMC and not the Chinese government.

Kind of like Huawei lol?

 

[soresu]

Kind of like Huawei lol?

Huawei is based in China, it IS a Chinese company - TSMC is a Taiwanese company - and yes, I'm aware of the relationship between the 2 countries, they are however separate national entities in the end, which is why China does not have a dozen native company fabs with state of the art process tech instead of being 2 nodes behind as they are.

So no, not kind of like at all.

You could have said "kind of like Tesla lol?", which would be more applicable to the situation.

 

[chrisjames61]

Huawei is based in China, it IS a Chinese company - TSMC is a Taiwanese company - and yes, I'm aware of the relationship between the 2 countries, they are however separate national entities in the end, which is why China does not have a dozen native company fabs with state of the art process tech instead of being 2 nodes behind as they are.

So no, not kind of like at all.

You could have said "kind of like Tesla lol?", which would be more applicable to the situation.

I am sure the tentacles of mainland China reach into TSMC.

 

[soresu]

I am sure the tentacles of mainland China reach into TSMC.

I'm sure they don't, but we are both blowing hot air with no purpose on that score.

The lack of IP security breach is self evident by China's continued native 2 node disadvantage vs TSMC and Samsung - if not, they would not need to fab Huawei at TSMC, or any other homegrown semicon designs (or semi homegrown as Kirin is with ARM supplied CPU and GPU).

 

[Ajay]

They went too dense without using EUV is pretty much what happened, Intel admitted as much.

After all the boasting about how awesome SAQP was going to be - sad. TSMC went with double patterning and got to 7nm with good yields and respectable electrostatics.

Anyway, I do hope Intel has success with 7nm EUV. It's going to be a tough ramp with all the demand pressure on ASML.

 

[ThatBuzzkiller]

Just as India has the potential to create RISC-V cores, and indeed they are already.

China only got this far so fast with x86 because of Hygon - obviously I'm not saying it can't be done independently, but given the minefield that is x86, it's alot easier when you have an AMD shaped space elevator to skip years of core R&D and ease ISA conformance.

The sheer amount of instructions, and instruction sets alone in x86 makes it a clusterfork of leviathan proportions to design for.

Don't even make a comparison between China and India since the latter is not even close to technological competence until another 2 decades while the former is already superior to America in some cases such as telecommunications equipment ...

China's independently developed x86 cores such as the Zhoaxin KX-7000 has IPC at least comparable to the Haswell architecture ...

 

[chrisjames61]

Don't even make a comparison between China and India since the latter is not even close to technological competence until another 2 decades while the former is already superior to America in some cases such as telecommunications equipment ...

China's independently developed x86 cores such as the Zhoaxin KX-7000 has IPC at least comparable to the Haswell architecture ...

The telecommunication equipment advantage is mainly due to intellectual property theft on a massive scale and Huawei being owned by The Peoples Republic of China with its colossal subsidies.

 

[soresu]

The telecommunication equipment advantage is mainly due to intellectual property theft on a massive scale and Huawei being owned by The Peoples Republic of China with its colossal subsidies.

Theft gives Huawei equivalence to competitors in tech - not the superiority which has certain countries spooked because it guarantees many countries telecom's companies will consider their equipment for national 5G telecom networks.

Huawei fiber cabinets are used in many places in the UK because they are both more reliable, with superior capacity to the competing Israeli (ECI) cabinets - this is straight from the horses mouth of British Telecom.