Intel to Outsource Chip Production to TSMC

AlphaAtlas

[H]ard|Gawd
Staff member
Joined
Mar 3, 2018
Messages
1,713
Amid reports of Intel having trouble supplying 14nm parts, Digitimes claims Intel is looking to outsource some production to Taiwan Semiconductor. The sources say that "entry-level H310 and several other 300 series desktop processors" will get outsourced to TSMC, and that higher margin products like Intel's 14nm Xeons will stay at Intel's fabs.

TSMC is already a contract manufacturer of Intel for SoFIA-series handset SoC chips and FPGA products, and makes Intel's baseband chips for use in the iPhone, the sources said. Motherboard makers expect the tight supply of Intel's 14nm chipsets to ease by the end of 2018.
 
Pretty sure the Desktop processor part is wayyyyy wrong, Intel doesn't design their CPUs to a standard foundry flow so it would take a very long time to relayout and resdesgin the processors for TSMC. Chipsets however is very plausible
 
Seems recently a lot of Intel's technologies just kind of flopped in the marketplace.

About 4 years ago the Intel Edison (SoC) was suppose to deliver real time speech recognition and no more having to be cloud connected.
I recall a whole lot of hub bub about it. Wonder what happened.
 
Pretty sure the Desktop processor part is wayyyyy wrong, Intel doesn't design their CPUs to a standard foundry flow so it would take a very long time to relayout and resdesgin the processors for TSMC. Chipsets however is very plausible

I agree. I don't know how TSMC could produce a chip designed on someone else's foundry lithography. I'm no EE but it doens't make sense.

Maybe the demand for super high quality binning is so low for the lower ranged parts can be "working" well enough that they're ok being fabbed on someone else's 14nm?

Any semiconductor pros on here that can weigh in?
 
Last edited:
Just how big is tsmc? There is only 24 hours in the day .. that i know.. how they going to keep up anyway?

they have at least 14 fabs. most of them are in taiwan but they also own wafertech which is in Washington state and a few others spread out over china.
 
TSMC spends close to $3B US a year on research into node shrinking processes. That is more than Intel and Samsung combined, and for more than a year TSMC has been known as a more advanced chip manufacturer.

TSMC recently completed a $20B plant and Intel played a roll in its construction the extent of which I do not know. But Intel has been working with TSMC in one form or another for 15 years or so.
 
I agree. I don't know how TSMC could produce a chip designed on someone else's foundry lithography. I'm no EE but it doens't make sense.

Maybe the demand for super high quality binning is so low for the lower ranged parts can be "working" well enough that they're ok being fabbed on someone else's 14nm?

Any semiconductor pros on here that can weigh in?

There's always a way, it's how much you're willing to pay to make it happen. It's a reengineer not sorcery. Some internal and cache latencies and clocks may change though.

If this is real they must be seeeeeriously boned for capacity. The heads that must have rolled here my god.
 
It's a reengineer not sorcery. Some internal and cache latencies and clocks may change though.

If this is real they must be seeeeeriously boned for capacity. The heads that must have rolled here my god.

a) "not sorcery" LOL that was great, I needed a humorous laugh today. As in actual humor, not sarcasm.
b) On capacity, either someone didn't plan well enough on and 14nm just really boned them in the long run, or their demand has really outstripped capacity, which is a double edged sword. You want to be in demand but you don't want to give customers time to think about it and turn to a competitor.
 
I agree. I don't know how TSMC could produce a chip designed on someone else's foundry lithography. I'm no EE but it doens't make sense.

Maybe the demand for super high quality binning is so low for the lower ranged parts can be "working" well enough that they're ok being fabbed on someone else's 14nm?

Any semiconductor pros on here that can weigh in?
Intel and TSMC have been working together for at least 15 years, and TSMC's new EUV Lithography processes have been shown to be extremely flexible I would not be surprised at this point if they could adapt or emulate Intel's process upon request.

Additional note they announced that they are on track for 5nm production for late 2019
 
Intel and TSMC have been working together for at least 15 years, and TSMC's new EUV Lithography processes have been shown to be extremely flexible I would not be surprised at this point if they could adapt or emulate Intel's process upon request.

Additional note they announced that they are on track for 5nm production for late 2019
5nm GTFO.. what is everyone else doing? Sticking their fingers up their butt?
Just what the hell is going on at Intel?... I still think its that forced ranking bullshit.. that basically broke GE.
 
It's getting to the point where specialization as an industrial concept has to be a reality of most companies soon. If it's not a core competency and outsource it. I would think if any company could, Intel could keep their internal foundries competitive. Samsung makes their own cell ARM processors AND Apple's, their primary competitor in the segment.

GM makes and designs many of their own engines and transmissions, but Chrysler buys ZF autos for their RWD platforms. I guess only the biggest of the boys can do their own shit in house.
 
5nm GTFO.. what is everyone else doing? Sticking their fingers up their butt?
Just what the hell is going on at Intel?... I still think its that forced ranking bullshit.. that basically broke GE.
TSMC spends more than 2x on research into node reductions than Intel, in fact they spend more than Intel and Samsung combined on R&D related to to their manufacturing processes. Their 5nm plant is expected to be just north of $25Bil and when complete it will be almost entirely dedicated to pumping out Apple products, they have also publicly stated that they are making good headway on their 2022 promises for 3nm.
 
Eventually only one company will be able to do this. Cause the die shrink will be so small.
 
Is this where AMD has their parts produced?
I was going to ask so is this how Intel screws AMD behind the scenes because Intel has the money to make a better deal and now AMD will have shortages or take longer to get to market? Hopefully, TSMC has the capacity for both, otherwise, at minimum I can see it affecting AMD's penetration into OEMs, the one reason being I've seen people in the past mention is because they could never supply the volume needed unlike Intel.
 
I would have hoped Intel would be willing and able to keep all their own chip production inside the company. This is super disturbing.

What do we need to do to make it competitive inside the US? Whatever it is, make it happen.
 
I was going to ask so is this how Intel screws AMD behind the scenes because Intel has the money to make a better deal and now AMD will have shortages or take longer to get to market? Hopefully, TSMC has the capacity for both, otherwise, at minimum I can see it affecting AMD's penetration into OEMs, the one reason being I've seen people in the past mention is because they could never supply the volume needed unlike Intel.
Different nano meter products :) AMD is going with 7 and Intel wants 14nm. Current AMD process used is 12nm.
 
wtf? super random HL video post?

Clearly you didn't bother playing that random clip. I put some effort finding a proper video and correct time stamp. Is everything in life handed to you? The confirmation might be in one of those clips. Valve left tons of clues.
 
Well, this beats their alternative plan which was to OEM Ryzen CPUs.
 
Pretty sure the Desktop processor part is wayyyyy wrong, Intel doesn't design their CPUs to a standard foundry flow so it would take a very long time to relayout and resdesgin the processors for TSMC. Chipsets however is very plausible

Yeah, I thought the same thing.

CPU designs have to be optimized with the process they are designed for in mind.

You can't just take a design made for one process and drop it into another one and expect it to work. Fab plants aren't 3D printers...
 
What on earth is a 300 series consumer CPU?
The H310 is an intel chipset, and I guess 300-series is referring to the other chipsets in that family. Technically they are processors (several different ones in a single chipset), although they aren't cpus.
 
What on earth is a 300 series consumer CPU?

The H310 is an intel chipset, and I guess 300-series is referring to the other chipsets in that family. Technically they are processors (several different ones in a single chipset), although they aren't cpus.

Other articles are saying desktop CPUs that go into the 300-series motherboards i.e. Coffee Lake. I wouldn't be surprised if the i7-9700k is a TSMC chip. People were saying that CPU designs can't just be ported over to a different foundry process easily. Would make sense for some of the 9th gen chips to be designed for the TSMC process especially since Intel knew that they had a shortage in capacity due to their 10 nm struggles.

According to our friends at "buy it now" Toms:

DigiTimes is reporting that Intel intends to outsource 14nm chipsets and "several other" 300-series processors, meaning the Coffee Lake chips, to TSMC to aid in production. The news comes as details of Intel's struggles with 14nm chip production surface from multiple sources.
 
Sorry, but this just seems like Intel doing what they do best: Screwing over AMD instead of trying to compete with them. TSMC only has a finite manufacturing capability, so If Intel were to out-bid AMD on EVERY slot, like book ALL the manufacturing for themselves, they would essentially cripple AMD's manufacturing capability on the 7nm node. The biggest product affected by this? The ROME Epyc chips, yeah, the ones that EVERYONE is predicting will slaughter Intel's lineup. So instead of producing a chip to compete with AMD's offering (they cant) and instead of lower their prices (HAHAHAHA) they instead will book every single manufacturing slot at TSMC and out-bid AMD on every future slot.

People will say I'm wearing a tinfoil hat, but seriously, back in the OG Opteron days, Intel payed OEMs literally billions of dollars per year to ensure AMD's opteron chips weren't used, and it ended up making Intel buckets of cash in the long run. How much money would it REALLY cost them to fully book-out ALL of TSMC's 7nm production? would it be billions of dollars? is that a price Intel has not ALREADY paid to ensure that AMD can't compete with them?
 
Other articles are saying desktop CPUs that go into the 300-series motherboards i.e. Coffee Lake. I wouldn't be surprised if the i7-9700k is a TSMC chip. People were saying that CPU designs can't just be ported over to a different foundry process easily. Would make sense for some of the 9th gen chips to be designed for the TSMC process especially since Intel knew that they had a shortage in capacity due to their 10 nm struggles.

According to our friends at "buy it now" Toms:
They're all sourcing digitimes, would be nice if there were some clarification from intel or digitimes on the matter.

In fact, digitimes have since edited their article like so: "and therefore plans to outsource the production of its entry-level H310 and several other 300 series chipsets to TSMC, the sources indicated."
So, pretty certain it's just their chipsets (at least for now).
 
Last edited:
They're all sourcing digitimes, would be nice if there were some clarification from intel or digitimes on the matter.

In fact, digitimes have since edited their article like so: "and therefore plans to outsource the production of its entry-level H310 and several other 300 series chipsets to TSMC, the sources indicated."
So, pretty certain it's just their chipsets (at least for now).

That makes more sense, but 14nm is still an unusually small node for chipsets.

Some sites are speculating that Intel is doing this to comply with new California power regulations. From TechReport:
Why is Intel manufacturing low-end chipsets on a leading-edge node to begin with? In conversations with other members of the media, I've come to understand that move is the result of the California Energy Commission's 2019 regulations. Indeed, Intel itself was involved in the state's rule-making effort and the timeline for the imposition of those regulations, and the company trumpeted its support for the rules upon their ratification. CEC 2019 imposes strict limitations on the power usage of computers and monitors in idle, sleep, and off modes.
 
Wow, just wow. I never thought I would see this day.

The management at Intel need firing.
 
well


we need to push California off into the ocean...
Yeah it's totally insane that one state is trying so hard to get ahead of the curve on saving our planet. We're all better off as complacent a-holes that care only about the convenience technology provides, and none of the downsides it brings with it.
 
It's getting to the point where specialization as an industrial concept has to be a reality of most companies soon. If it's not a core competency and outsource it. I would think if any company could, Intel could keep their internal foundries competitive. Samsung makes their own cell ARM processors AND Apple's, their primary competitor in the segment.

GM makes and designs many of their own engines and transmissions, but Chrysler buys ZF autos for their RWD platforms. I guess only the biggest of the boys can do their own shit in house.

Yes this is the model I use in my own business, it works great. Focus on what you do best and outsource as required to save time or money.
GM makes most of their own transmissions. Their manuals for V8s and V6s are built by Tremec however, which is most commonly a T56/TR6060/6070 platform which is actually derived from a Borg Warner design.
 
I would have hoped Intel would be willing and able to keep all their own chip production inside the company. This is super disturbing.

What do we need to do to make it competitive inside the US? Whatever it is, make it happen.

I get what you are saying but it's a double edged sword. If consolidation in the industry allows for advancements then that is a good thing. Like everyone keeps preaching in the industry making these chips is extremely hard and really, really expensive.

Your concerns come on the back end of how much will this effect the consumer in prices...which is the million dollar question!
 
Back
Top