News Intel mulls spinning off its manufacturing division

[Admin]

Intel is working with investment banks over strategic options regarding its manufacturing operations.

Intel mulls spinning off its manufacturing division : Read more

 

[Marlin1975]

More wow from intel. AMD did this as they were so far behind and needed cash bad, like near bankrupt bad.
How bad are things really at intel if this is on the board as an option?

 

[ezst036]

Intel is in deep trouble.

 

[Notton]

More wow from intel. AMD did this as they were so far behind and needed cash bad, like near bankrupt bad.
How bad are things really at intel if this is on the board as an option?

Yeah, AMD had a bad architecture in 2008, and their foundries were also garbage.
The spun-off side, Garbage Flounderies still remains, but it's pretty clear they are way behind and still produce hot running chips to this day.

 

[A Stoner]

I was an Intel fan for a long time. I think I owned Intel exclusively from around 1990 to last year. I have a AMD 7950X3D as my latest computer. I always wanted to switch to AMD, but they were always behind the bell curve by enough that I just was not willing to take the leap until the X3D processors came along.

While I am glad AMD and Intel are currently trading blows at the top end, it almost seems like Intel is on the way towards being second tier and we will eventually lose the competitive nature of the fight. I hope both companies can keep things humming along and keep the competition fierce so that growth is as fast as possible.

Maybe having Intel stick to just designing by getting rid of their fab will help. I guess time will tell. I think the thing that finally got me to step out of the Intel ecosystem was the thermals of their products. My computer room was just running way too hot.

The thing that attracts me most about AMD is the longevity of their platforms. My current motherboard and RAM should be valuable for several more years, whereas I would already be forced to upgrade at least the motherboard if I wanted to upgrade an intel processor.

gah lots of edits. 7950X3D not the unreleased 9950X3D LOL!

 

[MacZ24]

More wow from intel. AMD did this as they were so far behind and needed cash bad, like near bankrupt bad.
How bad are things really at intel if this is on the board as an option?

Maybe it's just that Intel foundries don't have enough clients because the prospective clients are weary of Intel having a look at their designs ? Going fabless would put a wall between design and production ? Just a theory.

 

[tracker1]

From my understanding of Intel's approach to CPU design, I'm not sure that this will help at all. They're to used to the mfg side hand holding them through the process. Will have to see how tsmc mfg chips works out for Intel.

Also if concern is the mfg itself. How well positioned are they to deliver say arm and other processors. They just literally offloaded their arm holdings which was a bad move in this light IMO.

 

[DS426]

If Intel completely fails, as in bankruptcy proceedings (and this would be because they don't let go of IFS), I'm going to assume that nVidia will acquire them to fill in the rest of the vacuum beyond the additional market share that AMD gains. It's no secret that nVidia wants more action in the CPU market, and although ARM ISAs or more broadly RISC appears to be their teenage crush, x86 isn't going to be dead anytime soon.

 

[thestryker]

Intel has gotten slammed by "investors" who don't have a clue for years now. The problem right now is that they've taken real financial hits as Gelsinger has tried to repair the damage done by having finance people in charge of a capital intensive tech company. I highly doubt spinning off the fabs would fix anything at Intel and would likely kill off the foundry as leading edge. From what I've been able to gather there's likely internal restructuring that needs to be done and all selling off the fabs does is potentially help stock price.

 

[Gururu]

Investors right now have merely slapped Intel's wrist. Wall street is very detached from the Vmin issue and any other enthusiast complaint. The US has too much intellectual property at stake with Intel and the market knows this. If Intel loses, China wins. With upside to the current restructuring with regard to profits and good rep with its new chip line up, they will be stronger than ever next year if only because the US bails them out.

 

[SonoraTechnical]

nVidia might want to buy Intel... but would the US justice dept allow it?

these guys are not doing well... rings of AMD selling fabs off to Global Foundaries. The relationship between AMD as the designer and GF as the manufacturer didn't endure.

GF got allot of money and built those fabs in Malta, NY when Sec H. Clinton was investing in NY's 'Tech Valley'. Now GF is stuck at 14nm processes and larger at that facility. No more super hi-tech stuff there. AMD has turned to others for years for production.

Is that where Intel's fabs are now? perhaps too far behind others.

 

[vanadiel007]

I would have this message for Intel: never look for advice from banks, as they are going to provide advice on whatever makes them money.

Simply look at what went wrong, learn from and correct your mistakes. And make sure to keep your customers happy.

 

[vanadiel007]

They would be smart to contact Apple and offer them a deal on jointly using the Intel fabs for chip production. I am sure for the right price and conditions Apple would be interested.

 

[robbro9]

I hope no one is falling for the "If we sell the fabs the rest of the business is profitable" narrative. The fabs are losing money because the intercompany transfer pricing is apparently below cost (after all intel is its own largest customer for the fabs), which on the flip side means the other units are showing more profit than they realistically should. If the fabs are sold off and must stand on their own, guess whats going to happen to the pricing? It will go up, the fabs will become profitable and the remaining business units that are their customers (the remaining Intel) will be paying for that.

 

[passivecool]

Saying you are going to build it ain't the same as getting it built.

If intel does not stay the course on this it would be the end for pat as well and the start of an entirely new chapter, for better or worse.

 

[waltc3]

I think Intel is on the right track with a spin-off. If they spin off their FABs, they are also spinning off the rest of the company, as well, so to speak...😉 Intel's problem for many years has been that it's setup as a high-end x86 monopoly supplier, and Intel is no longer in that market position. And the company is simply too large to manage in one bundle. Intel FABs can still make and sell a lot of chips, just not cutting edge as we are approaching 2-4nm now for complex CPU designs, and TSMC seems to have the lead in that kind of fabbing. But, there are still lots and lots of people interested in buying larger chips, 90nm or 130nm, even, for various purposes. Spinning off would then allow both companies to hone their core competencies in a way that really isn't possible under the current Intel organization. I don't see Intel going broke because the company can spin off and downsize, and learn how to burn the midnight oil to compete--which is precisely what put AMD where it is today. Learning how to do more with less is critical--AMD has done it for years, so we know that it can be done. What Intel is still doing is doing less with more, and that will have to change. What is spent on R&D is only justified by the products it creates. The dollar amount expended is entirely secondary, imo.

 

[richardvday]

Real men have Fabs Jerry once said. Once meant as an insult but pretty much proven baseless over the years. It seems even Intel might have finally learned that lesson. Real men Don't have FABS? So much for vertical integration.

 

[thestryker]

The fabs are losing money because the intercompany transfer pricing is apparently below cost (after all intel is its own largest customer for the fabs), which on the flip side means the other units are showing more profit than they realistically should.

This isn't really accurate at all given that the first actual accounting for IFS has been done during huge expansion which is extremely capital intensive. As an example Intel just bought a $400m High-NA machine and spent the money to modify factory space to even install it and it's making them zero dollars. It's the nature of the industry as IFS needs more capacity, but Intel alone cannot saturate additional capacity which is why they didn't already have it. Effectively Intel has to expand capacity while developing new nodes and maintaining existing manufacturing contracts. The turnaround on revenue for IFS was never going to be fast and anyone who thought otherwise was living in a dream land not doing their due diligence.

IFS doesn't magically become profitable by spinning off it becomes profitable by more effectively using the capacity they have. Once they're able to largely retire Intel 7 things should get a lot better as all of their leading edge will be EUV which means node refinement and modification becomes a lot simpler when everything is using the same primary hardware.

 

[robbro9]

This isn't really accurate at all given that the first actual accounting for IFS has been done during huge expansion which is extremely capital intensive. As an example Intel just bought a $400m High-NA machine and spent the money to modify factory space to even install it and it's making them zero dollars. It's the nature of the industry as IFS needs more capacity, but Intel alone cannot saturate additional capacity which is why they didn't already have it. Effectively Intel has to expand capacity while developing new nodes and maintaining existing manufacturing contracts. The turnaround on revenue for IFS was never going to be fast and anyone who thought otherwise was living in a dream land not doing their due diligence.

IFS doesn't magically become profitable by spinning off it becomes profitable by more effectively using the capacity they have. Once they're able to largely retire Intel 7 things should get a lot better as all of their leading edge will be EUV which means node refinement and modification becomes a lot simpler when everything is using the same primary hardware.

Just to be a bit nitpicky, expenses for capital expansion (depreciation) should not hit P&L until they are in service. The cash needs are there sure, but they are not what is driving the losses.

 

[TheSecondPower]

AMD spun off their manufacturing as Global Foundries, and Global Foundries played a role in Bulldozer's failings and later Global Foundries stepped out of bleeding-edge production. Boeing spun 737 fuselage manufacturing as Spirit AeroSystems, and afterward the 737 Max 8 had two crashes and a 737 Max 9 lost compression in flight.

Splitting doesn't fix anything, but it does open up a lot of options for accounting tricks. Outside of accounting deception, at best it makes a company smaller and more focused. I think splitting off Solidigm made sense because SSDs are a separate business. But Intel foundries and Intel design are integral to making processors and they have to work closely together. They could work separately but for years they clearly worked better together than AMD and Global Foundries did separately.

And Intel foundries losses can't be entirely from 3rd-party customers not signing on. Intel foundries still isn't meeting the supply needed by Intel design, so it's not like they have supply for 3rd-party customers that they could sell to be profitable. Maybe Intel design isn't paying enough, as I think robbro9 said.

 

[robbro9]

And Intel foundries losses can't be entirely from 3rd-party customers not signing on. Intel foundries still isn't meeting the supply needed by Intel design, so it's not like they have supply for 3rd-party customers that they could sell to be profitable. Maybe Intel design isn't paying enough, as I think robbro9 said.

Add in if I'm not mistaken that in one of Intel's recent financial releases they expected to take a P&L hit from using TSMC to fab certain products because the 3rd party charges more for fab services than internal... sounds like internal is undervalued/priced again. Not exactly apples to apples as they are getting a more advanced node I believe. But in general, sounds like an competitive independent foundary would be able to charge more in arms length transactions roughly speaking...

 

[TheSecondPower]

I think the failings of Global Foundries + AMD actually played a role in Intel reaching this point. Moore's Law worked for years partly because the foundries believed that their competition would keep up with it. But after Intel released Ivy Bridge in 2012, Intel no longer needed to keep up. Many people said Moore's Law was dead and Intel was way ahead of TSMC, Global Foundries, and Samsung. Intel lost the need to push hard or even run efficiently. Now the need is back, and it'll take time to learn how to meet it.

 

[TheSecondPower]

But in general, sounds like an competitive independent foundary would be able to charge more in arms length transactions roughly speaking...

It's possible that Intel foundry is priced below its competition for reasons other than capability, such as yield, capacity, design tooling, familiarity with design tooling, or like of established rapport. Some of those reasons don't affect the value to Intel design, since they have rapport and have always used their own foundry tools.

 

[robbro9]

Its also possible its priced below to help write a narrative...

 

[thestryker]

Just to be a bit nitpicky, expenses for capital expansion (depreciation) should not hit P&L until they are in service. The cash needs are there sure, but they are not what is driving the losses.

The part you're not getting is that this is the equivalent of an idle fab. They didn't install this (or the future ones) in some empty building this isn't currently being used. These aren't being used to create new fabs (it would immensely help of they were) because it's not viable to roll out new nodes in a new fab. They also had to move equipment from Oregon to Ireland which certainly isn't the way they wanted to ramp EUV capacity. All of these costs used to be attached to the workgroup producing products on the respective nodes.

Add in if I'm not mistaken that in one of Intel's recent financial releases they expected to take a P&L hit from using TSMC to fab certain products because the 3rd party charges more for fab services than internal... sounds like internal is undervalued/priced again. Not exactly apples to apples as they are getting a more advanced node I believe. But in general, sounds like an competitive independent foundary would be able to charge more in arms length transactions roughly speaking...

There are no competitive fabs period which is a good chunk of why TSMC can charge whatever they want for leading edge. The other part being Apple consistently buying up huge portions of leading edge which drives up overall cost. These things wouldn't occur if Samsung and Intel had equivalent nodes/capacity on them.