Speculation: Intel will become fabless

[maddie]

I would be willing to bet its Apple. Apple has already hinted that they were designing their own chips for desktop use. Other than them, there aren't any companies that come to mind.

Nope. Apple is hundreds of billions market cap and according to Semiaccurate, the company is pretty much toast, having bet the farm on Intel.

 

[Gideon]

I would be willing to bet its Apple. Apple has already hinted that they were designing their own chips for desktop use. Other than them, there aren't any companies that come to mind.

No way, Apple is the no 1 company by market gap of $909.84B, they wouldn't be toast. This company is something in the line of Western Digital or HP (not that it would be either of them)

EDIT:
According to hackernews comments it might very well be Nokia.

 

[Thala]

No way, Apple is the no 1 company by market gap of $909.84B, they wouldn't be toast. This company is something in the line of Western Digital or HP (not that it would be either of them)

EDIT:
According to hackernews comments it might very well be Nokia.

Indeed Nokia would be toast if they placed a bet on Intel 10nm for their whole 5G line of base-station SoCs.

 

[LTC8K6]

Following sounds familiar? Maybe they think like you.

https://forums.anandtech.com/threads/speculation-intel-will-become-fabless.2549870/#post-39479276
"Intel will be just fine, as will their fabs.

10nm will turn out to be fine, even though the risk of trying to do it their way ended up as a train derailment.
Train wrecks get cleaned up, the train gets re-railed, and the train slowly gets moving again.

7nm on EUV will be very nice for Intel fans.

Soon this bad patch will be in Intel's rear view mirror, getting smaller."

And maybe it's all BS and the sky isn't falling every week...

 

[frozentundra123456]

Yea, looks like Charlie is off his meds again. As far as intel spinning off it fabs, the idea has been absurd from the beginning of this thread, as evidenced by the 9 to 1 vote that they will not. I dont have the slightest clue (well I do, but cant really say it) why a few posters keep trying to beat a dead horse.

 

[maddie]

Yea, looks like Charlie is off his meds again. As far as intel spinning off it fabs, the idea has been absurd from the beginning of this thread, as evidenced by the 9 to 1 vote that they will not. I dont have the slightest clue (well I do, but cant really say it) why a few posters keep trying to beat a dead horse.

I guess the we'll find out this year if Charlie is right.

 

[Stuka87]

Yea, looks like Charlie is off his meds again. As far as intel spinning off it fabs, the idea has been absurd from the beginning of this thread, as evidenced by the 9 to 1 vote that they will not. I dont have the slightest clue (well I do, but cant really say it) why a few posters keep trying to beat a dead horse.

You may or may not believe his story, but a lot of the evidence is there. 10nm has been delayed over and over. While competitors already have 7nm shipping out.

 

[Stuka87]

No way, Apple is the no 1 company by market gap of $909.84B, they wouldn't be toast. This company is something in the line of Western Digital or HP (not that it would be either of them)

EDIT:
According to hackernews comments it might very well be Nokia.

Yeah, I forgot just how much Apple's market cap was. Nokia would totally make sense.

 

[scannall]

Yea, looks like Charlie is off his meds again. As far as intel spinning off it fabs, the idea has been absurd from the beginning of this thread, as evidenced by the 9 to 1 vote that they will not. I dont have the slightest clue (well I do, but cant really say it) why a few posters keep trying to beat a dead horse.

I do have to wonder though, at what point to die shrinks keep making sense? The main purpose, from a business standpoint is lowering the cost per transistor. Since Intel bailed on mobile, then squeezing the most performance per watt isn't *as* important as it is for mobile customers.

 

[whm1974]

I do have to wonder though, at what point to die shrinks keep making sense? The main purpose, from a business standpoint is lowering the cost per transistor. Since Intel bailed on mobile, then squeezing the most performance per watt isn't *as* important as it is for mobile customers.

Well speaking for myself, but even for desktop CPUs I do not want a high TDP and would have to use a gigantic HSF to avoid cooking said CPU.

 

[scannall]

Well speaking for myself, but even for desktop CPUs I do not want a high TDP and would have to use a gigantic HSF to avoid cooking said CPU.

We're already well past the point of needing massive cooling though. Unless you overclock a lot. I value silence as well. My current computer has a winter overclock, and a summer one. For obvious heat reasons.

 

[alcoholbob]

I do have to wonder though, at what point to die shrinks keep making sense? The main purpose, from a business standpoint is lowering the cost per transistor. Since Intel bailed on mobile, then squeezing the most performance per watt isn't *as* important as it is for mobile customers.

I believe what's driving die shrinks at this point is AI research/machine learning. Cost per transistor isn't even close to being a concern in that market, perf per watt and raw number of transistors is. This is probably why Intel is trying to compete with nvidia with supercomputers, and trying to get into GPU market. Developing die shrinks for CPUs alone (spending billions every 2-3 years on a new node) at this point is making less than no sense. If they can't find enough of a market to fund development of their fabs they will eventually go fabless.

 

[LTC8K6]

You may or may not believe his story, but a lot of the evidence is there. 10nm has been delayed over and over. While competitors already have 7nm shipping out.

But Intel 10nm and everyone elses's 7nm are roughly the same. So saying "already have 7nm out" is misleading. It makes it seem like Intel is a gen behind, when they are really just late on the same gen.
These nanometer terms are largely marketing now.

 

[Thala]

I do have to wonder though, at what point to die shrinks keep making sense? The main purpose, from a business standpoint is lowering the cost per transistor. Since Intel bailed on mobile, then squeezing the most performance per watt isn't *as* important as it is for mobile customers.

Well that is a fallacy. High efficiency is not just for mobile. It is important for servers as well and if you looking at HEDT with up-to 32 core CPUs like Threadripper you will acknowledge that high power efficiency is key here too. You do not have to look much farther than Intels showcasing of a 28core 5GHz CPU at Computex to understand how much todays computing is limited by power consumption even at the high end.

 

[Stuka87]

But Intel 10nm and everyone elses's 7nm are roughly the same. So saying "already have 7nm out" is misleading. It makes it seem like Intel is a gen behind, when they are really just late on the same gen.
These nanometer terms are largely marketing now.

If Intel is not shipping 10nm, but other fabs are starting to ship 7nm, then yes, Intel is a generation behind. As that means Intel is still on 14nm. TSMC meanwhile has been shipping 10nm parts for a year now.

 

[alcoholbob]

But Intel 10nm and everyone elses's 7nm are roughly the same. So saying "already have 7nm out" is misleading. It makes it seem like Intel is a gen behind, when they are really just late on the same gen.
These nanometer terms are largely marketing now.

Except we already have production dies of large monolithic high performance 7nm dies at TSMC (AMD Vega 20), and Intel is still struggling to eek out ultra low power dies at 10nm. They aren't a generation behind, but they are about half a generation behind, despite having a head start. And 7nm at TSMC has slightly better transistor density than 10nm Intel.

 

[Vattila]

Bits-and-Chips chimes in on the issue:

Intel may spin off its own FABs

"How? AMD and Intel will share almost the same nodes. Intel may sell to Global Foundries its FABs. Or Intel may create a Joint Venture with GloFo. We have to remember that Global Foundries has got the money of the Arabs and the Know-How of IBM, so it may be a good deal."

https://twitter.com/BitsAndChipsEng/status/1023305802197684225

 

[NTMBK]

Bits-and-Chips chimes in on the issue:

Intel may spin off its own FABs

"How? AMD and Intel will share almost the same nodes. Intel may sell to Global Foundries its FABs. Or Intel may create a Joint Venture with GloFo. We have to remember that Global Foundries has got the money of the Arabs and the Know-How of IBM, so it may be a good deal."

https://twitter.com/BitsAndChipsEng/status/1023305802197684225

No way in hell the US would allow Abu Dhabi to own a monopoly on PC CPUs.

 

[tamz_msc]

It's BitsAndChips, no need to take them seriously.

 

[NostaSeronx]

No way in hell the US would allow Abu Dhabi to own a monopoly on PC CPUs.

UAE and US are inseparable at this point.

So, a JV cross-funded by Mubadala Capital and Intel Capital. Would in turn fuse the US-owned triangle: Malta, Hillsboro, Chandler for leading-edge nodes. While also fusing Intel and GlobalFoundries fab IP and resources, etc. GlobalFoundries gets capacity utilizing Intel fabs and Intel gets working nodes utilizing GloFo cross-competitive R&D. Effectively, having a west coast versus east coast partner-competition for the best nodes.

Samsung and TSMC are shifting towards majority of Chinese-orientated customers. Intel and GlobalFoundries in most cases are stuck with majority of US/EU-orientated customers. Preventing another Japan-esque semiconductor catastrophe is key to a GlobalFoundries and Intel partnership.

Intel's Israel Fab and GlobalFoundries' Abu Dhabi unbuilt Fab might also be partnered up in the nodes. With maybe a bit of Dresden and Ireland involved as well. It would make sense on a global domination standpoint to fuse the fabs. GlobalFoundries is literally sitting on DRAM/NAND foundries somewhere in the Asia-Pacific. GlobalFoundries has partnered up with Micron before so... Intel/GlobalFoundries with Micron IP would be profitable for Micron via licensing. GlobalFoundries is full on laying on some NVM tech, but doesn't have the licenses to produce it. GlobalFoundries in turn won't allow that NVM to produce elsewhere. So, the joint aspect opens a lot of doors for Intel/GloFo/Micron.

 

[VirtualLarry]

Dude, you really need to get back on your meds.

 

[Kenmitch]

Not sure. Maybe I'll revisit the topic once they've finally managed to jackhammer thru the 10nm wall....That dang wall looks to be pretty dang tough. 2

 

[jpiniero]

They won't, and there is no need due to EMIB and similar technologies. The older nodes will have a lot of value for some time. But if 7 nm is a flop, moreso than 10, then yes that might be the end of the line for anything beyond that.

Don't be surprised if the Gen 12 (d)GPUs are fabbed at TSMC for instance, or at least parts of it.

 

[Vattila]

Intel to outsource 14nm chip production due to tight supply

"Intel is encountering tight 14nm process production capacity in-house, and is looking to outsource part of its 14nm chip production to Taiwan Semiconductor Manufacturing Company (TSMC), according to industry sources."

"Intel intends to give priority to its high-margin products mainly server-use processors and chipsets amid its tight 14nm process capacity, and therefore plans to outsource the production of its entry-level H310 and several other 300 series desktop processors to TSMC, the sources indicated."

https://www.digitimes.com/news/a20180910PD210.html

 

[LTC8K6]

Intel to outsource 14nm chip production due to tight supply

"Intel is encountering tight 14nm process production capacity in-house, and is looking to outsource part of its 14nm chip production to Taiwan Semiconductor Manufacturing Company (TSMC), according to industry sources."

"Intel intends to give priority to its high-margin products mainly server-use processors and chipsets amid its tight 14nm process capacity, and therefore plans to outsource the production of its entry-level H310 and several other 300 series desktop processors to TSMC, the sources indicated."

https://www.digitimes.com/news/a20180910PD210.html

"H310 and several other 300 series desktop processors"... I assume they mean chipsets?

So they are only outsourcing the chipsets then?

I didn't even think they made the chipsets.