Long term, is Intel planning to exit the consumer market?

NTMBK

Lifer
Nov 14, 2011
10,208
4,940
136
Is Intel's long game to exit the consumer market entirely, and move solely into high-margin servers and datacenter products? Intel has been consistently boosting investment in the datacenter and cutting back in the consumer market for years now. The consumer PC market is collapsing, and Intel has abandoned their attempt to break into its replacement (phones and tablets). Is the long term plan to abandon the low-margin consumer market to the ARM devices? Continue to slowly reduce investment in consumer parts, reap the (still very profitable!) long tail of sales, and use those profits to transition to a fully datacenter focused company.

If Intel gets their foundry business off the ground, I could see this being a viable strategy. Use ARM devices to fill their fabs and fund the R&D needed to remain a viable manufacturer, while keeping the high-margin datacenter stuff for themselves.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
No. HEDT is moving into the consumer as gaming keeps expanding. And then there is all the shared R&D.

People claimed Intel had given up on the desktop as well.
 

willfr

Member
Apr 27, 2016
33
5
41
Long term no, and I think so because they might have chips that are significantly better than the competition because of their process advantage. I don't know when anyone else really plans to 'ship' true 10nm, but intel really is doing that in the 2nd half of 2017. It could be a year or even longer before competitors ship true 10nm chips (and when they do they'll be marketing them as 3nm lol). It could be the same story with 7nm, I'm sure intel will get there but it's competitors will be at least a year behind, probably even more than a year behind. I think we'll see intel manufacturing more and more ARM chips but continuing their own too.
 

TheELF

Diamond Member
Dec 22, 2012
3,967
720
126
We can see ARM already going to octa and deca cores because it's impossible to get any more gains out of them.

We can also see intel investing a lot into iGPU and lower and lower power consumption.

Soon, as the consumer will ask for better performance and ARM won't be able to provide it intel will step in with desktop x86 cores,not i7 level ones at close to 4,5Ghz but with celeron/pentium level ones.
Al this running windows on arm stuff is the prelude to this,trying to get people get used to the idea.

The home PC market always was a niche market,it only got blown out of proportion because for a while it was the only (decent) means of accessing the internet and especially e-mail.
It makes sense that as the bubble deflates intel dials back on the home market.
 

Lodix

Senior member
Jun 24, 2016
340
116
116
We can see ARM already going to octa and deca cores because it's impossible to get any more gains out of them.

We can also see intel investing a lot into iGPU and lower and lower power consumption.

Soon, as the consumer will ask for better performance and ARM won't be able to provide it intel will step in with desktop x86 cores,not i7 level ones at close to 4,5Ghz but with celeron/pentium level ones.
Al this running windows on arm stuff is the prelude to this,trying to get people get used to the idea.

The home PC market always was a niche market,it only got blown out of proportion because for a while it was the only (decent) means of accessing the internet and especially e-mail.
It makes sense that as the bubble deflates intel dials back on the home market.
I think you are misunderstanding and underestimating the way ARM processors are designed. The number of cores is mostly for efficiency purpose. And it is not like they can't make use of all the cores as proved already.

ARM architecture is improving faster than Intel.
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
Intel even today makes more $ from the Client Computing Group than the Data Center Group. I dont see why they would like to exit the consumer market even if they will get half the profits they get today.

Q3 2016

CCG Revenue = 8892M
CCG Profit = 3327M

DCG Revenue = 4542M
DCG Profit = 2110M

- Client Computing Group. Includes platforms designed for notebooks (including Ultrabook™ devices), 2 in 1 systems, desktops (including all-in-ones and high-end enthusiast PCs), tablets, phones, wireless and wired connectivity products, and mobile communication components.
- Data Center Group. Includes platforms designed for the enterprise, cloud, communications infrastructure, and technical computing segments.
 

Atari2600

Golden Member
Nov 22, 2016
1,409
1,655
136
No. That's a mental idea.

At worst, they'll revert back to the old market. design a workstation/server CPU, then take a couple of cuts* off it for the consumer market. R&D can be much less than 5% or so of total and you can make a decent bit of money back off it.

*Obviously, you'll cut off the Xeon E3 line.
 

dark zero

Platinum Member
Jun 2, 2015
2,655
138
106
I can't see the Core Line die since it has unmatched performance compared to everyone. Only Apple A10 is near that performance.

However, I do see the Atom line dying due the sheer incompetence that even VIA can match them.
 

Qwertilot

Golden Member
Nov 28, 2013
1,604
257
126
Unless the consumer markets find a decent reason that they're prepared to pay for more performance than current Core chips, you'd have to think quite possibly - consumer grade CPUs are going to end up as pure commodities (they're quite a long way there right now and it is only going to get worse over time) - and Intel do rather like margins.

Especially if stuff like robotics/deep learning etc take off as a premium volume market on the same kinds of time scale(s).
 

NTMBK

Lifer
Nov 14, 2011
10,208
4,940
136
Long term no, and I think so because they might have chips that are significantly better than the competition because of their process advantage. I don't know when anyone else really plans to 'ship' true 10nm, but intel really is doing that in the 2nd half of 2017. It could be a year or even longer before competitors ship true 10nm chips (and when they do they'll be marketing them as 3nm lol). It could be the same story with 7nm, I'm sure intel will get there but it's competitors will be at least a year behind, probably even more than a year behind. I think we'll see intel manufacturing more and more ARM chips but continuing their own too.

I'm getting flashbacks to before the launch of Bay Trail. People argued that there was no way that the ARM vendors could possibly compete with a 22nm FinFET chip, when they were stuck on only 28nm planar (at best!). And yet, surprise surprise, Intel lost billions of dollars trying to break into the Android market, before eventually giving up.

Transistor technology is far from the only factor that matters.
 
  • Like
Reactions: Lodix
Mar 10, 2006
11,715
2,012
126
I can't see the Core Line die since it has unmatched performance compared to everyone. Only Apple A10 is near that performance.

However, I do see the Atom line dying due the sheer incompetence that even VIA can match them.

The Atom-based Pentium/Celeron chips pretty much wiped away more PC business from AMD than even Conroe was able to do, and did so very profitably. Not bad for a chip that everybody rips on.
 

Lodix

Senior member
Jun 24, 2016
340
116
116
Long term no, and I think so because they might have chips that are significantly better than the competition because of their process advantage. I don't know when anyone else really plans to 'ship' true 10nm, but intel really is doing that in the 2nd half of 2017. It could be a year or even longer before competitors ship true 10nm chips (and when they do they'll be marketing them as 3nm lol). It could be the same story with 7nm, I'm sure intel will get there but it's competitors will be at least a year behind, probably even more than a year behind. I think we'll see intel manufacturing more and more ARM chips but continuing their own too.
There is no more process advantage. TSMC is shipping 7nm in 1H of 2018 with better properties than Intel 10nm. And GF is scheduled to 2H of the same year.
 
Mar 10, 2006
11,715
2,012
126
You really joking right ??

No, I'm not. Intel said so in its investor presentations. They explicitly said that the margins on Bay Trail-based Pentium/Celeron chips for notebooks and desktops sold at solid margins and had a very good cost structure.
KyfXRPK.png

NZw5776.png


There is a difference between Bay Trail-T and the Bay Trail-M/D chips.
 
Last edited:

Rifter

Lifer
Oct 9, 1999
11,522
751
126
If AMD brings back some pressure with ZEN i dont think intel will back off. However if ZEN flops i can see intel pulling more R&D and resources from consumer products for sure. Why spend money when you have no one to compete with.
 
Mar 10, 2006
11,715
2,012
126
If AMD brings back some pressure with ZEN i dont think intel will back off. However if ZEN flops i can see intel pulling more R&D and resources from consumer products for sure. Why spend money when you have no one to compete with.

They aren't cutting the R&D in the core IP, all of that R&D is shared between the major business units.
 

NTMBK

Lifer
Nov 14, 2011
10,208
4,940
136
They aren't cutting the R&D in the core IP, all of that R&D is shared between the major business units.

Consumer and server core designs are diverging. They're using a different fabric. They need a competitive GPU architecture (and software stack to match). They need encode/decode hardware. There's a lot less overlap than back in the old days. The Atom has even less overlap, with KNL being very far removed from Silvermont.
 

Torn Mind

Lifer
Nov 25, 2012
11,567
2,626
136
Given how the mainstream i7 "K" chip is selling like hotcakes despite its price going up and people even buying that Broadwell 10 core, they'll keep on serving up chips to consumers. Consumers don't buy Xeons. Xeons give off the impression that it is for "servers" and that stuff won't work with desktops one way or the other, although Intel did make it that way again with the Skylake E3 line.

Fact is, people don't know crap about computers and buy based on guides like this. Yeah, that lawyer wasted lines of text but made no mention of SSDs or i5s having four single cores. That's because he doesn't know about these things. So the best his mind can say is "spend at least a minimum of $750". Pros who are clueless pay for the ignorance by lining the pockets of OEMs who build excellent workstations. By "pay", I don't mean it in a punishing sense. Just that since a person doesn't know what to look for, you're going have losses here and there.

Laptop parts are also client parts, and I doubt they just give up on the gamers and suckers who buy a quad core i7 just because they can.
 

mnewsham

Lifer
Oct 2, 2010
14,539
428
136
Consumer and server core designs are diverging. They're using a different fabric. They need a competitive GPU architecture (and software stack to match). They need encode/decode hardware. There's a lot less overlap than back in the old days. The Atom has even less overlap, with KNL being very far removed from Silvermont.
about the only difference in the architecture for consumer vs. server/workstation is the iGPU.

That and the super high end intel chips with FPGAs integrated.


For the general core architecture however they're the same and will likely continue to be the same for many years, there is simply no reason to change.
 

NTMBK

Lifer
Nov 14, 2011
10,208
4,940
136
about the only difference in the architecture for consumer vs. server/workstation is the iGPU.

That and the super high end intel chips with FPGAs integrated.


For the general core architecture however they're the same and will likely continue to be the same for many years, there is simply no reason to change.
Skylake X has four times as much L2 cache, AVX-512 SIMD units, and uses a mesh fabric instead of ringbus. It's very different.
 
Last edited:
  • Like
Reactions: Arachnotronic

willfr

Member
Apr 27, 2016
33
5
41
There is no more process advantage. TSMC is shipping 7nm in 1H of 2018 with better properties than Intel 10nm. And GF is scheduled to 2H of the same year.

Intel is the only company thats process is as advertised. Its 14nm process really is 14nm, and its 10nm process really is 10nm. The marketing departments of their competitors have made a mockery of this though. There are quite a few articles by experts that dive deeper into the weeds about this if you google it.

But overall it's not the most important thing, ARM chips will always be significantly cheaper so that will give them an advantage.