Do you think Intel will spin off their fabs?

[Phynaz]

So let's pick a timeframe, let's say three or five years. What do you think the chances are that Intel would spin off it's manufacturing into a separate company within that time, and why?

 

[witeken]

Maybe on the first day of April.

 

[ShintaiDK]

Never.

 

[frozentundra123456]

So let's pick a timeframe, let's say three or five years. What do you think the chances are that Intel would spin off it's manufacturing into a separate company within that time, and why?

You started the thread, what is your take on it?

To me, it Seems unlikely, since they invest so much and are so dependent on being on bleeding edge nodes. Personally I think this is a big danger for intel. Each node shrink seems to bring fewer benefits along with new problems, so I am not sure how long they can depend on this.

Maybe if they can integrate next gen materials they will start getting a good boost again from new processes.

 

[gdansk]

If Intel spins off their fabs it would be remembered as the beginning of their slow demise.

Intel should only spin off her fabs if near bankruptcy.

 

[witeken]

Any reasonable investor would be against such a decision. Being an IDM has quite an advantage. People like TuxDave will probably agree. And secondly, Intel is designed as 1 company. Every part of the company adds value to it and takes a bit from the burden of going into new markets (mobile, IoT). Big companies are always better in the semiconductor industry.

 

[ThatBuzzkiller]

Never.

The only answer that does matter ...

 

[erunion]

What's the more valuable part of Intel, its CPU design house or its fabs?

 

[DooKey]

Wow, I can't believe someone started a thread like this.

Answer: Unlikely any time within the next 20 years.

 

[DooKey]

What's the more valuable part of Intel, its CPU design house or its fabs?

Yes

 

[Stuka87]

Intel's fabs are the number one reason they are so far ahead of AMD. If AMD suddenly got access to Intel's fabs, things would change in the CPU industry in a hurry.

 

[MiddleOfTheRoad]

Never.

Downright scary:

I just agreed with something you said.

 

[frozentundra123456]

Hell just froze over!!

 

[alcoholbob]

Intel's fabs are the number one reason they are so far ahead of AMD. If AMD suddenly got access to Intel's fabs, things would change in the CPU industry in a hurry.

Aren't AMDs 28nm CPUs still slower clock for clock compared to Intel's 32nm and 45nm i7s? Even if The situation was reversed and AMD was on 14nm and Intel was on 28nm, Intel would still have the faster CPUs.

 

[TechFan1]

You mean spin off the chip design? How much revenue a year do the other companies that design chips like AMD, Nvidia, and ARM make?

 

[jpiniero]

I could see Wall St demand it if they think Intel is not being serious about the foundry business, especially in the era of Good Enough computing.

That doesn't mean it will happen though.

 

[Phynaz]

You started the thread, what is your take on it?

In the short term I don't see it. I view Intel as a manufacturing company first, and a chip design company second. But as with AMD, how long can one company's products support the economics of scale required to keep those factories running? Even Samsung needs external customers.

 

[III-V]

In the short term I don't see it. I view Intel as a manufacturing company first, and a chip design company second. But as with AMD, how long can one company's products support the economics of scale required to keep those factories running? Even Samsung needs external customers.

Well, they're trying to get customers on board. The fabless companies are basically waiting to see how things turn out with Altera. I heard Nvidia wanted to fab with Intel some years ago, but Intel was scared to death of the SEC, so that fell through.

They'll definitely be getting quite a bit of volume on 14 nm after they move their modems over, as well as moving SoFIA over. Their modems are already very impressive on TSMC 28 nm, and supposedly FinFETs are very good for RF, so we could see some very enticing products there.

I think Intel will essentially never spin off their fabs. The only way they'd part from them is if they no longer provided a competitive advantage (e.g. foundry tech is all but perfected, and there's no room for improvement), or if they get broken up by the feds.

 

[SOFTengCOMPelec]

On a serious note. What they could seriously consider doing, is opening up a "new named company" (and/or division), who specialize in designing/building the latest chips (foundry), to many external customers.

A bit like Toyota, use to be one giant car manufacturer. They eventually created the "Lexus" company/brand, which is really a (sub) division of Toyota.

This would then benefit Intel, because they would be able to share the enormous, ever increasing chip fabrication plant costs, with other companies. Similarly, it would benefit other companies, as they would have access to some of the best chip production plants, in the world.

Intel would have to decide how to deal with competitor products, such as Arm and AMD stuff. As that is a political hot potato.

In theory, eventually chip plants may/will be so extremely expensive, that there can only really be one (set) of them, in the world. Gradually this already seems to be happening. I'm not sure of the exact number, but I think there use to be hundreds of chip manufacturing plants (probably more) round the world. Now there are only a handful (of big ones) and a few rather small (research) ones. I'm not sure of exactly what the situation is, partly because it keeps on changing.

 

[DrMrLordX]

In the short term I don't see it. I view Intel as a manufacturing company first, and a chip design company second. But as with AMD, how long can one company's products support the economics of scale required to keep those factories running? Even Samsung needs external customers.

Let's pretend Intel's revenues go completely south in 2015 which is not bloody likely. Somehow, they manage to pull in juuuuust enough money to pay for all activity NOT related to fab operation, process R&D (which is a sticky issue, since process R&D may be intertwined with other R&D costs in a company like Intel), and other fab-related expenses.

So basically, they start taking losses equal to all the combined costs of having a foundry operation and keeping said foundry operation up-to-date.

What is that loss, and how long would it take to eat through Intel's "dragon hoard" of cash they have bankrolled for . . . you know, rainy days?

Unrelated note: what are the odds of Intel letting Dunkelzahn sit on some or all of their money? Figuratively or literally.

 

[Arachnotronic]

In the short term I don't see it. I view Intel as a manufacturing company first, and a chip design company second. But as with AMD, how long can one company's products support the economics of scale required to keep those factories running? Even Samsung needs external customers.

Samsung doesn't have anywhere near the (logic) wafer volumes that Intel has.

Even a 2+2 Skylake will probably include a ~90mm^2 (+) SoC die + a 40-50mm^2 PCH die on 22nm.

Intel will sell these (+ larger variants) in 100M+ units per year. Intel also gets a lot of wafer volumes from server chips, and as they move things like ethernet controllers and the like to internal processes, that volume should only grow.

Just my 2¢...

 

[Phynaz]

On a serious note. What they could seriously consider doing, is opening up a "new named company" (and/or division), who specialize in designing/building the latest chips (foundry), to many external customers.

A bit like Toyota, use to be one giant car manufacturer. They eventually created the "Lexus" company/brand, which is really a (sub) division of Toyota.

This would then benefit Intel, because they would be able to share the enormous, ever increasing chip fabrication plant costs, with other companies. Similarly, it would benefit other companies, as they would have access to some of the best chip production plants, in the world.

Intel would have to decide how to deal with competitor products, such as Arm and AMD stuff. As that is a political hot potato.

In theory, eventually chip plants may/will be so extremely expensive, that there can only really be one (set) of them, in the world. Gradually this already seems to be happening. I'm not sure of the exact number, but I think there use to be hundreds of chip manufacturing plants (probably more) round the world. Now there are only a handful (of big ones) and a few rather small (research) ones. I'm not sure of exactly what the situation is, partly because it keeps on changing.

There's still lots of smaller fabs around, you just don't hear about them because they aren't huge companies. Even GM runs a fab.

 

[master_shake_]

don't they use old fabs to build flash and chipsets or something?

 

[SOFTengCOMPelec]

There's still lots of smaller fabs around, you just don't hear about them because they aren't huge companies. Even GM runs a fab.

Thanks for the correction. On reflection, there are so many chips that are available to buy, from single op-amps, NE-555 timers, Quad Nand gate logic (etc) up to ram/flash etc, that there must still be many fab plants, around.

Presumably China has a lot of stuff (fabs). I sometimes hear that they are going to bring out their own cpu, and then I never hear anything about it again.

 

[III-V]

don't they use old fabs to build flash and chipsets or something?

Chipsets, yeah. Intel doesn't make flash, though. They get their flash from their joint venture with Micron, IM Flash.