Intel splits its struggling foundry division into an independent subsidiary

[Cal Jeffrey]

In a nutshell: Rumors regarding Intel partially divesting its foundry division have circulated since an insider leak last month. The company addressed those rumors this week, confirming its fabrication operation would become an independently run subsidiary. Head honcho Pat Gelsinger calls it "the next phase" of his plan to "transform" Intel into the powerhouse it once was.

On Monday, Intel CEO Pat Gelsinger informed employees that the company is transforming its foundry arm into an independent subsidiary, confirming insider leaks from August. According to the press release, disguised as a message to employees, the decision emerged last week after what Gelsinger described as a "highly productive and supportive Board meeting."

As an independent subsidiary, the foundry can accept outside investments and allocate funds within a separate budget. This separation will allow the foundry to run more efficiently.

"Collectively, these changes are critical steps forward as we build a leaner, simpler and more efficient Intel," Gelsinger said. "And they build on the immediate progress we have made since announcing our plan on August 1 to create a more competitive cost structure."

The foundry will keep its current leadership and form an independent board of directors. A confidential source told CNBC that the company is considering turning it into a publicly traded spinoff. Gelsinger says the foundry will expand its relationship with Amazon Web Services (AWS) with co-investments in new chip designs, including an AI chip for AWS and a custom Xeon 6 chip based on Intel's 3nm process.

Intel's foundry has been particularly troubled recently. After spending two years and $50 billion expanding its fab operations, the company turned in a dreadful Q2 2024 earnings report. Gelsinger called it "disappointing," but investors promptly launched a class-action suit claiming that executives, including Gelsinger and CFO David Zinsner, issued "materially false and misleading" statements about its foundry business.

The news of the foundry split caused stock prices to rebound nearly 22 percent from a low of $18.60 per share just five days ago to $22.66 during trading on Monday. However, it's still a far cry from the $49.55 stock price investors enjoyed in January.

Gelsinger remains confident that he can turn things around. In addition to the continued broadening of its AWS relationship, the company received a $3 billion cash injection thanks to the CHIPS ad Science Act.

However, the company still plans to cut 15,000 jobs by year's end. Nobody knows if they are getting let go, but the boss said the company would notify "impacted employees" beginning in the middle of next month. It is also divesting two-thirds of its global real estate holdings by the end of the year.

Permalink to story:

Intel splits its struggling foundry division into an independent subsidiary

 

[J95qRP7]

"As an independent subsidiary, the foundry can accept outside investments and allocate funds within a separate budget. This separation will allow the foundry to run more efficiently."

Translation: we want others to chip in for foundry costs and this lets us use TSMC for more designs, without losing as much face, just so we can compete. As an independent foundry, we can spend less on it, so it can fall further behind, as we no longer have the expertise and it's just too hard and too expansive to try and bother getting it back. Thank the lord of semiconductors that TSMC isn't as inept as us.

 

[passwordistaco]

Intel is circling the drain :-(. They never pulled out of the 10nm tailspin.

 

[Jack77]

"As an independent subsidiary, the foundry can accept outside investments and allocate funds within a separate budget. This separation will allow the foundry to run more efficiently."

Translation: we want others to chip in for foundry costs and this lets us use TSMC for more designs, without losing as much face, just so we can compete. As an independent foundry, we can spend less on it, so it can fall further behind, as we no longer have the expertise and it's just too hard and too expansive to try and bother getting it back. Thank the lord of semiconductors that TSMC isn't as inept as us.

Isn't being a foundry also a requirement to receive funding from the US government when it comes to that Chips act?

 

[user556]

Isn't being a foundry also a requirement to receive funding from the US government when it comes to that Chips act?

Intel still is for the moment. It doesn't stop being a chip fabricator until the two divisions are split for real.

 

[GhostLegion]

Well…I consider this to be the beginning of the end of Intel. It was doing poorly before but this is it. I have zero faith that this isn’t a stepping stone to fully spinning off the foundry arm. Which means TSMC really will be the only company doing leading edge nodes. Like I’m sure Intel as a company will survive, just like AMD has done. I just mourn this. What a shame.

 

[opckieran]

This is unsurprising, 18A by 2025 was just way too insane of a goal. Intel knew this so they decided to save face by spinning off their foundries in a way that would both minimize the hit to their stock for missing that insane target by claiming it’s a separate entity while at the same time allowing them to basically retain 100% control.

 

[Burty117]

When they were stuck on 14nm++++++++++++

You knew something was very wrong then, it just took a little competition to show how bad things really were at Intel.

I don't want them to go, competition is good, but they do deserve everything they're getting after practically doing nothing in the 2010's until AMD came back fighting.

 

[gamerk2]

Intel is circling the drain :-(. They never pulled out of the 10nm tailspin.

They never really recovered from their decision to not use EUV at 10nm.

 

[Felusy]

I bought Intel stock on the dip, will sell in a few years for xxx% gain.

You have to be a fool if you think Intel will go out of business. The state won't allow it. Chip act, go read. US is not letting asia have all advanced chip production. China can enter Taiwan any day and then TSMC is out of business.

Intel 18A and beyond are soon ready for mass production (2025)

Intel got like 10 billion dollars recently, chip act, military deals etc.

Arrow Lake and Lunar Lake looks impressive and AMD lost node advantage and failed with Zen 5 (For now, hopefully 9000X3D can save 9000 series).

The comeback has already begun. Intel up 13% in just 5 days.

 

[VariableSpike]

Of course Pat would position this as some grabdiose "next phase" of his plans, and what it actually is, which is rapidly trying to save face and please the investors by making the big loss making line go away in the financial reports and letting it be neglected and scapegoated if Intel keep treating it with the same attitude while giving them more cover to save face for using TSMC rather than their own foundries

 

[Cracken]

I bought Intel stock on the dip, will sell in a few years for xxx% gain.

You have to be a fool if you think Intel will go out of business. The state won't allow it. Chip act, go read. US is not letting asia have all advanced chip production. China can enter Taiwan any day and then TSMC is out of business.

Intel 18A and beyond are soon ready for mass production (2025)

Intel got like 10 billion dollars recently, chip act, military deals etc.

Arrow Lake and Lunar Lake looks impressive and AMD lost node advantage and failed with Zen 5 (For now, hopefully 9000X3D can save 9000 series).

The comeback has already begun. Intel up 13% in just 5 days.

Tbh I dont see neither arrow lake or lunar lake as particular impressive, it's decent products, sure, but nothing that will really set them apart from competition.

"You have to be a fool if you think Intel will go out of business"
I also do not think Intel will go out of business, it would be foolish to let them do that. Having an intel that doing poorly is better than not having them at all.
And that's what I see intel being for the forseable future, pretty bad - but still in the game. That does not mean there stock will bounce back to $50 or $70 tho, they might as well slump along as they do now. Plenty of reasons they could drop another 20% from more bad news, in fact im pretty sure there will be more bad news.

 

[Felusy]

Tbh I dont see neither arrow lake or lunar lake as particular impressive, it's decent products, sure, but nothing that will really set them apart from competition.

"You have to be a fool if you think Intel will go out of business"
I also do not think Intel will go out of business, it would be foolish to let them do that. Having an intel that doing poorly is better than not having them at all.
And that's what I see intel being for the forseable future, pretty bad - but still in the game. That does not mean there stock will bounce back to $50 or $70 tho, they might as well slump along as they do now. Plenty of reasons they could drop another 20% from more bad news, in fact im pretty sure there will be more bad news.

Lunar Lake beats both AMD and ARM in performance per watt. Early reviews of Arrow Lake are good as well and watt usage went down massively. We know more very soon, meanwhile AMD failed with Zen 5 and cheaped out on process node, using TSMC 4N/5nm instead of 3N/3nm which was ready (since Intel is using it)

AMD is chasing AI hard, money sink that will go on for years, however Nvidia is years ahead and is loaded with money.

Intel will bounce back for sure. Just wait and see. Did these kind of predictions tons of times in the last 20 years. Part of why I don't work anymore. Crypto and stocks, is what I do now and I am set for life really.

If you take no risks, you don't ever win. Intel on a dip, is a sure-buy. Lets talk in a year

 

[Cracken]

Lunar Lake beats both AMD and ARM in performance per watt. Early reviews of Arrow Lake are good as well and watt usage went down massively. We know more very soon, meanwhile AMD failed with Zen 5 and cheaped out on process node, using TSMC 4N/5nm instead of 3N/3nm which was ready (since Intel is using it)

AMD is chasing AI hard, money sink that will go on for years, however Nvidia is years ahead and is loaded with money.

Intel will bounce back for sure. Just wait and see. Did these kind of predictions tons of times in the last 20 years. Part of why I don't work anymore. Crypto and stocks, is what I do now and I am set for life really.

If you take no risks, you don't ever win. Intel on a dip, is a sure-buy. Lets talk in a year

There are several things you got wrong in the picture you just painted. First of neither lunar lake or arrow lake wipes the floor, sure there are performance metrics where they do well, but the same can be said for the competition. Lunar/arrow are not groundbreaking or disruptive in a big way, it's good products for sure, but they are merely stepping stones for intel, they will need a succesion of those in order to firmly take the lead. Point is, this is not a Conroe moment for intel.
Also keep in mind that both lunar lake and arrow lake are much more expensive to make then any of their competing products, they simply cannot sell them at the same prices.
Zen 5 consumer products "Ryzen" has gotten a rocky start, that is correct, but Zen 5 isn't just Ryzen, in fact it's mainly EPYC, that goes for all Zen core products.
The most important markets for both Intc and AMD is the DC market, and there is nothing indicating that Intel is about to turn the tables and stop loosing market share to Zen 5. In fact it, judging by the roadmaps it seems Xeon's is in for another beating this year. Intel has lost around 30% market share to AMD/ARM over the last 7 years at an accelerating rate. It's this bloodbath intel has to stop in order to thrive again.

I don't believe we have to worry about INTC being knocked out, they will adapt even if they have to scale down and be less prominent. Regarding a real comeback for intel, it's surely possible, alot of things are possible, but the clear indicators of a comeback we have yet to see any of that.

 

[Burty117]

Intel 18A and beyond are soon ready for mass production (2025)

If that's true, why are Intel outsourcing more and more of its own products to it's competition (TSMC)?

 

[Durch]

I'm summary, let's take advantage of the taxpayers, split up divisions for more shareholders to invest, while we lay off staff while giving our executives a cushy bonus and deliver a subpar product.

Yep that sounds like Intel!

 

[Felusy]

If that's true, why are Intel outsourcing more and more of its own products to it's competition (TSMC)?

Because its 2024 and they need to launch a genereation now, not in 1 year.

Intel already have a bunch of customers for 18A lined up. Including the US military in a huge multi billion dollar deal.

 

[Burty117]

Because its 2024 and they need to launch a genereation now, not in 1 year.

Intel already have a bunch of customers for 18A lined up. Including the US military in a huge multi billion dollar deal.

So you're 100% certain 18A will be live and at full production in the next year?
Because the US Government doesn't appear to have set any timeline on using 18A, only date I could find is "prototype production in 2025".
On top of this, Broadcom was clearly looking at where they should be at this point in time if the node were as healthy as Intel claims, and their test chip results showed that not to be the case. Especially damning when Broadcom would only need mass production around 2026, if not later.

Intel have consistently failed over 15 years upgrading nodes, from 14nm++++++++ to 10nm was excruciating.

BUT, maybe you have some evidence to back up your claim that 18A will be in full production by the end of next year?

 

[Felusy]

So you're 100% certain 18A will be live and at full production in the next year?
Because the US Government doesn't appear to have set any timeline on using 18A, only date I could find is "prototype production in 2025".
On top of this, Broadcom was clearly looking at where they should be at this point in time if the node were as healthy as Intel claims, and their test chip results showed that not to be the case. Especially damning when Broadcom would only need mass production around 2026, if not later.

Intel have consistently failed over 15 years upgrading nodes, from 14nm++++++++ to 10nm was excruciating.

BUT, maybe you have some evidence to back up your claim that 18A will be in full production by the end of next year?

No but do you have proof that it won't?

AMD failed massively and abandoned GloFo as well. Without TSMC AMD would be nowhere either, TSMC is the sole reason Ryzen took off for real - Ryzen 1000 and 2000 were no threat for Intel.

Being fabless is easier yeah, but Intel is the only true competitor to TSMC. Sadly for AMD and consumers, TSMC is the reason price on Ryzen has increased big time since they left GloFo.

Samsung teases 2nm but they are lying about how advanced and dense their nodes are and done for years.

Samsung 8nm for RTX 3000 series were using their rebranded 10nm node, which in reality were closer to TSMC 12nm in terms of density.

Meaning Nvidia beat AMD with a process close to 12nm TSMC node, while AMD was using 7nm TSMC.

Intel has been delivering good performance for years, while using an inferior node. Now with Arrow Lake and Lunar Lake, Intel uses TSMC 3N aka 3nm, lets see what Intel can do with node advantage shall we? Intels only issue has been power usage and mostly on i9. Performance has been fine and they even beat AMD in tons of tasks, all while using a worse process.

AMD really dropped the ball with Zen 5. Cheaped out on process. Rebranded chipsets. Software bugs and reviews with no guidelines. Little improvement gen to gen, even tho Zen 5 is built ground up and a brand new arch. Lacking new motherboards on launch, meaning people have to flash firmware over USB to use a 9000 series on a older motherboard. No mention of 3D models even tho everyone knows they are coming. Just the worst CPU release I have seen in years. Techspot says this themselves. Go watch the video. Zen 5 sales are MISERABLE so far.

Intel pretty much has perfect timing here. Brand new arch on TSMC 3N. Node advantage for the first time in many years. Performance leaks look very good. Power usage is much lower. Performance goes up. New Z890 boards on day one, along with K models.

 

[Burty117]

No but do you have proof that it won't?

I'm not going to bother responding to the rest of your comment because you just go on a rant about AMD which is not what we're talking about at all.

If you need history on Intel failing to upgrade nodes, You can use Google yourself, but here's a list of their Nodes and what products are on them, as you'll see, they've been stuck on 14nm and 10nm for a very long time.

Evidence they won't be in full production by the end of 2025? As I mentioned but I guess I'll need to link an article, Broadcom is one of the first customers for Intel's foundry's and their testing went badly, their 18A node is simply nowhere near ready, 2025 is 2 months away.

Also, the Military contract you mentioned, did you only read headlines or did you read the full articles? Nowhere does it say they'll start full production next year, at most, prototype test runs next year.

 

[Cracken]

No but do you have proof that it won't?

AMD failed massively and abandoned GloFo as well. Without TSMC AMD would be nowhere either, TSMC is the sole reason Ryzen took off for real - Ryzen 1000 and 2000 were no threat for Intel.

Being fabless is easier yeah, but Intel is the only true competitor to TSMC. Sadly for AMD and consumers, TSMC is the reason price on Ryzen has increased big time since they left GloFo.

Samsung teases 2nm but they are lying about how advanced and dense their nodes are and done for years.

Samsung 8nm for RTX 3000 series were using their rebranded 10nm node, which in reality were closer to TSMC 12nm in terms of density.

Meaning Nvidia beat AMD with a process close to 12nm TSMC node, while AMD was using 7nm TSMC.

Intel has been delivering good performance for years, while using an inferior node. Now with Arrow Lake and Lunar Lake, Intel uses TSMC 3N aka 3nm, lets see what Intel can do with node advantage shall we? Intels only issue has been power usage and mostly on i9. Performance has been fine and they even beat AMD in tons of tasks, all while using a worse process.

AMD really dropped the ball with Zen 5. Cheaped out on process. Rebranded chipsets. Software bugs and reviews with no guidelines. Little improvement gen to gen, even tho Zen 5 is built ground up and a brand new arch. Lacking new motherboards on launch, meaning people have to flash firmware over USB to use a 9000 series on a older motherboard. No mention of 3D models even tho everyone knows they are coming. Just the worst CPU release I have seen in years. Techspot says this themselves. Go watch the video. Zen 5 sales are MISERABLE so far.

Intel pretty much has perfect timing here. Brand new arch on TSMC 3N. Node advantage for the first time in many years. Performance leaks look very good. Power usage is much lower. Performance goes up. New Z890 boards on day one, along with K models.

Same as above, im not going to bother responding to all that, cause most of it is either false or misleading. It seems you didnt read earlier replies where you get corrected. Just stop man, you have no idea what you are talking about.

 

[Cracken]

No but do you have proof that it won't?

AMD failed massively and abandoned GloFo as well. Without TSMC AMD would be nowhere either, TSMC is the sole reason Ryzen took off for real - Ryzen 1000 and 2000 were no threat for Intel.

Being fabless is easier yeah, but Intel is the only true competitor to TSMC. Sadly for AMD and consumers, TSMC is the reason price on Ryzen has increased big time since they left GloFo.

Samsung teases 2nm but they are lying about how advanced and dense their nodes are and done for years.

Samsung 8nm for RTX 3000 series were using their rebranded 10nm node, which in reality were closer to TSMC 12nm in terms of density.

Meaning Nvidia beat AMD with a process close to 12nm TSMC node, while AMD was using 7nm TSMC.

Intel has been delivering good performance for years, while using an inferior node. Now with Arrow Lake and Lunar Lake, Intel uses TSMC 3N aka 3nm, lets see what Intel can do with node advantage shall we? Intels only issue has been power usage and mostly on i9. Performance has been fine and they even beat AMD in tons of tasks, all while using a worse process.

AMD really dropped the ball with Zen 5. Cheaped out on process. Rebranded chipsets. Software bugs and reviews with no guidelines. Little improvement gen to gen, even tho Zen 5 is built ground up and a brand new arch. Lacking new motherboards on launch, meaning people have to flash firmware over USB to use a 9000 series on a older motherboard. No mention of 3D models even tho everyone knows they are coming. Just the worst CPU release I have seen in years. Techspot says this themselves. Go watch the video. Zen 5 sales are MISERABLE so far.

Intel pretty much has perfect timing here. Brand new arch on TSMC 3N. Node advantage for the first time in many years. Performance leaks look very good. Power usage is much lower. Performance goes up. New Z890 boards on day one, along with K models.

Short time has passed, here we are.

How is that arrow lake doing ?
How is that 18A process node doing ?

It seems your hopes/dream/predictions is not going as hoped for.

 

[Felusy]

Short time has passed, here we are.

How is that arrow lake doing ?
How is that 18A process node doing ?

It seems your hopes/dream/predictions is not going as hoped for.

I am no fanboy. I look at reality and 9800X3D launch fixed the Ryzen 9000 issues. Best gaming CPU by far that still holds its own outside of gaming.

Arrow Lake was meh, yeah.
Lunar Lake was decent tho.

Intel will be back. I hope for competition as always. I have more AMD chips than Intel chips, so don't calle me Intel fanboy.

 

[Burty117]

don't calle me Intel fanboy.

It's in writing in this very comment section, you were telling us Intel's 18A process node would be online in the near future and that would fix everything instantly, you had nothing to backup your claims, it was just a feeling you had.

You only "feel" like a company will suddenly do things that haven't been confirmed or even rumoured, when you're a fanboy.

 

[Felusy]

It's in writing in this very comment section, you were telling us Intel's 18A process node would be online in the near future and that would fix everything instantly, you had nothing to backup your claims, it was just a feeling you had.

You only "feel" like a company will suddenly do things that haven't been confirmed or even rumoured, when you're a fanboy.

Sometime in 2025 I said. Same did Intel. We are in 2024.

Clearwater Forest is going to be 18A.

Yeah I am an Intel fanboy, with 6 AMD chips and 2 Intel chips in house. Haha.