Intel drops $20B into creating a foundry business.

Krenum

Fully [H]
Joined
Apr 29, 2005
Messages
17,741
Smart move. If you're going to bleed money in lackluster processor sales, you might as well invest the money in your coffers into something worthwhile.
 

zehoo

Limp Gawd
Joined
Aug 22, 2004
Messages
406
If their can foundry can catch up to TSMC then they would love to get contracts from Apple, Qualcomm and Nvidia etc.

It's also a backup diversification plan in case x86 fades out to ARM.
 

Red Falcon

[H]F Junkie
Joined
May 7, 2007
Messages
11,081
That's quite the tax write-off, let's hope they do something good with it in the coming years.
 

bonehead123

[H]ard|Gawd
Joined
Nov 27, 2011
Messages
1,247
If their can foundry can catch up to TSMC then they would love to get contracts from Apple, Qualcomm and Nvidia etc.

It's also a backup diversification plan in case x86 fades out to ARM.
Well, maybe someone should tell them that it's gonna take a HELL of a lot MOAR money than that if they ever hope to catch up to, let alone compete with, TSMC.... but at present, $20B is a proverbial "drop in the bucket" for them though..

And yes, I would say they could very well be preparing for the eventual demise of X86, as they know that the fruity boys (and nGreediya) are on to something now with their new ARM plans, and they would absolutely hate to see that as a singular reason why they die and or fade into the annals of history, which WILL happen if they don't get their collective shiitzu together and finally move beyond their pathetic 14nm++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ clusterfudgecrapola :)
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
3,873
Well, maybe someone should tell them that it's gonna take a HELL of a lot MOAR money than that if they ever hope to catch up to, let alone compete with, TSMC.... but at present, $20B is a proverbial "drop in the bucket" for them though..

And yes, I would say they could very well be preparing for the eventual demise of X86, as they know that the fruity boys (and nGreediya) are on to something now with their new ARM plans, and they would absolutely hate to see that as a singular reason why they die and or fade into the annals of history, which WILL happen if they don't get their collective shiitzu together and finally move beyond their pathetic 14nm++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ clusterfudgecrapola :)
I'm going to reserve judgment until I see their 10nm Xeons which should be hitting in the next month or 2, I really do need that work station and I don't want to pull the trigger on a 3970x at this stage because Intel and AMD should have something ready to go in the ball park any day now on the new series.
My poor old Dual E5-2620v4 is showing its age, and just not getting things done like it used to.
 

lopoetve

Fully [H]
Joined
Oct 11, 2001
Messages
31,199
Intel has been in the foundry business for a long time - this is a response to Elliott Capital trying to force them to divest that business (Elliott is generally clueless IMHO in the Tech space) like AMD did in the early 2000s (mostly because that would generate shareholder returns in the short term for Elliott, while ... not in the long run). Hedge funds. Fuck em.

Instead, Gelsinger is doubling down - which is probably a good idea, given that foundry space is going to be more and more in demand over the next 20 years - and going to expand their presence there instead.

edit: Given how long it takes to BUILD a foundry, this isn't about 14nm - but 7/5/whatever comes in the future.
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
3,873
Intel has been in the foundry business for a long time - this is a response to Elliott Capital trying to force them to divest that business (Elliott is generally clueless IMHO in the Tech space) like AMD did in the early 2000s (mostly because that would generate shareholder returns in the short term for Elliott, while ... not in the long run). Hedge funds. Fuck em.

Instead, Gelsinger is doubling down - which is probably a good idea, given that foundry space is going to be more and more in demand over the next 20 years - and going to expand their presence there instead.

edit: Given how long it takes to BUILD a foundry, this isn't about 14nm - but 7/5/whatever comes in the future.
Yeah they are saying they have their 10 and 7nm issues worked out and that the new fabs will be their 7nm which based on the fact that TSMC was able to pick up Intel's slack on the DoE supercomputer means that the Intel 7nm is roughly equivalent to the TSMC 5nm. Intel's previous foundry attempts were doomed because Intel would not rent out time on their current processes only ones that were a gen or two behind and nobody really wanted that at any volume, not when there were viable alternatives elsewhere that were cheaper as they weren't carrying the big Intel brand name behind them.
 

lopoetve

Fully [H]
Joined
Oct 11, 2001
Messages
31,199
Yeah they are saying they have their 10 and 7nm issues worked out and that the new fabs will be their 7nm which based on the fact that TSMC was able to pick up Intel's slack on the DoE supercomputer means that the Intel 7nm is roughly equivalent to the TSMC 5nm. Intel's previous foundry attempts were doomed because Intel would not rent out time on their current processes only ones that were a gen or two behind and nobody really wanted that at any volume, not when there were viable alternatives elsewhere that were cheaper as they weren't carrying the big Intel brand name behind them.
Bingo. As that business grows, renting space/time is going to be potentially hugely profitable. It's a GOOD idea. We're really down to TSMC/Samsung in the public foundry space (GF/Micron/etc are all pretty specialized or way behind). Intel being a third with that much demand? They'll bloody print money, if they can build a truly modern fab and process.
 

UnknownSouljer

Supreme [H]ardness
Joined
Sep 24, 2001
Messages
7,059
As the current silicon shortage has shown us: we are going to need more foundries period. Which is also why TSMC is also expanding.
There are few companies in the world that can compete as a top end foundry - as obviously the cost of entry is so incredibly high with massive R&D expenses and facilities as well as just having the employee talent. And although it's obvious that Intel hasn't been there as they've struggled with 14nm for so long (as has been noted ad nauseum in this thread) Intel has the shortest road to being a tier 1 supplier again versus any of the others. I am definitely not holding my breath for Global Foundries to get in the game. You think Texas Instruments is up for it? How about ON Semiconductor? No? Yeah, didn't think so. Samsung is the current only other major provider of higher end chips - and frankly they need to expand too if they're going to increase supplying other vendors.

If Intel can actually get its foundry junk together, even if late, it would be welcomed by basically all silicon designers in general - especially if they open that foundry space to others, which seems to be their intention.
 
Last edited:

Sycraft

Supreme [H]ardness
Joined
Nov 9, 2006
Messages
4,931
As the current silicon shortage has shown us: we are going to need more foundries period. Which is also why TSMC is also expanding.
There are few companies in the world that can compete as a top end foundry - as obviously the cost of entry is so incredibly high with massive R&D expenses and facilities as well as just having the employee talent. And although it's obvious that Intel hasn't been there as they've struggled with 14nm for so long (as has been noted ad nauseum in this thread) Intel has the shortest road to being a tier 1 supplier again versus any of the others. I am definitely not holding my breath for Global Foundries to get in the game. You think Texas Instruments is up for it? How about ON Semiconductor? No? Yeah, didn't think so. Samsung is the current only other major provider of higher end chips - and frankly they need to expand too if they're going to increase supplying other vendors.

If Intel can actually get its foundry junk together, even if late, it would be welcomed by basically all silicon designers in general - especially if they open that foundry space to others, which seems to be their intention.
Also all the people acting like Intel will never be able to compete are forgetting that not long ago Intel was two nodes ahead of everyone else. They were way out in front for quite some time. So it is reasonable to think that they could get things together and become competitive again. Likewise one of the other foundries could trip up as Intel did.

Remember when AMD just beat the pants off of nVidia in the DX11 days, when the 400 series were massively delayed and power hungry to boot? nVidia had problems, and AMD had good chips... but then it changed years later. The 1000 series was extremely efficient and fast, AMD's competition was a pretty poor showing.

Just because a company does well, or has problems, doesn't mean it will be so forever. I'm certainly not going to bet that Intel will rush to the foundry lead... but I certainly wouldn't bet against them either.
 

lopoetve

Fully [H]
Joined
Oct 11, 2001
Messages
31,199
And in this line of business, competitive is as good as being best when it comes to revenue. Someone will use the line.
 

relapse808

[H]ard|Gawd
Joined
Apr 11, 2005
Messages
1,260
I am pretty sure one of these facilities is going up pretty close to my house.
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
3,873
I am pretty sure one of these facilities is going up pretty close to my house.
Depending on the distance either condolences or congratulations depending on weather it’s too close and tanks the property value or just right and doubles it.
 

///AMG

2[H]4U
Joined
Sep 19, 2012
Messages
3,568
My question is how many companies want to pay intel for the foundry business? How many companies want to pay THE competitor money to make your chip? Its also going to take them at least 2 years to get this one running. TSMC will never compete with their customers but Intel sure does.
 

DukenukemX

Supreme [H]ardness
Joined
Jan 30, 2005
Messages
5,478
Well, maybe someone should tell them that it's gonna take a HELL of a lot MOAR money than that if they ever hope to catch up to, let alone compete with, TSMC.... but at present, $20B is a proverbial "drop in the bucket" for them though..
Don't worry, I'm sure they're expecting a big fat stinky check from Joe Biden. They'll just triple the taxation to the poor to make up for it.
And yes, I would say they could very well be preparing for the eventual demise of X86, as they know that the fruity boys (and nGreediya) are on to something now with their new ARM plans, and they would absolutely hate to see that as a singular reason why they die and or fade into the annals of history, which WILL happen if they don't get their collective shiitzu together and finally move beyond their pathetic 14nm++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ clusterfudgecrapola :)
x86 has nothing to do with it. AMD is proof that x86 still has lots of life when there's a profit motive. Intel was so dominant for such a long time that they became complacent and let others catch up and surpass them. Not having competition means that the most profitable thing to do is not spend R&D, which is what Intel did. Intel now has no choice but to spend money and yes, $20B is not enough. Joe Biden better not help pay for this sorta thing.
 

cybereality

Supreme [H]ardness
Joined
Mar 22, 2008
Messages
7,025
My question is how many companies want to pay intel for the foundry business? How many companies want to pay THE competitor money to make your chip? Its also going to take them at least 2 years to get this one running. TSMC will never compete with their customers but Intel sure does.
Competing companies can commonly work together if it's a win for both parties. For example, AMD owns the x86-64 spec and licenses it to Intel. But Intel owns x86 so they reached a cross-licensing agreement.
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
3,873
My question is how many companies want to pay intel for the foundry business? How many companies want to pay THE competitor money to make your chip? Its also going to take them at least 2 years to get this one running. TSMC will never compete with their customers but Intel sure does.
A fair number, there are a lot of markets that need chips produced not all of them are CPU's and GPU's the recent automotive stuff is a prime example of this need. The US is also very likely to start shoring up their own chip manufacturing tech, having almost all the worlds supply of high-end silicon coming out of Taiwan is proving to be problematic and in the next 5-10 years expected to become a lot more complicated for reasons that don't need to be mentioned in this particular forum. It is very much in the US's best interest to get some of that high end manufacturing localized, and they are certainly in the right to subsidize it going forward as a lack of access to those plants have shown it can cripple very large portions of the economy, and open up a pretty big security hole in the event of some global hostilities.
 

Thatguybil

Limp Gawd
Joined
Jan 21, 2017
Messages
134
Intel has tried to fab external customers IC but has never truly been successful.

Charlie D from SemiAccurate has cataloged Intels dabbles Fabinho other companies IC.
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
3,873
Intel has tried to fab external customers IC but has never truly been successful.

Charlie D from SemiAccurate has cataloged Intels dabbles Fabinho other companies IC.
But those past attempts were pretty half assed, keeping those fabs 2 generations behind while charging a premium for the Intel name was a bad call. They won’t have those options this time around.
 

///AMG

2[H]4U
Joined
Sep 19, 2012
Messages
3,568
But those past attempts were pretty half assed, keeping those fabs 2 generations behind while charging a premium for the Intel name was a bad call. They won’t have those options this time around.
But they will probably still be 2 generations behind TSMC. Also intel doesn’t only do CPUs and GPUs. But I bet you the big names won’t be going to intel anytime soon.
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
3,873
But they will probably still be 2 generations behind TSMC. Also intel doesn’t only do CPUs and GPUs. But I bet you the big names won’t be going to intel anytime soon.
Well behind TSMC perhaps but it will be Intel’s 7nm. Which seems to trade blows with TSMC 5nm. But in the old case when Intel was on 90nm they were going not fabbing 180 for the 3’rd parties, stuff like that.
 

OrangeKhrush

[H]ard|Gawd
Joined
Dec 15, 2016
Messages
1,673
great news it now officially means Intel competes against TSMC and the agreement is now avoidable as it is a conflict of interest. AND bumped to preferential
 

Red Falcon

[H]F Junkie
Joined
May 7, 2007
Messages
11,081
Don't worry, I'm sure they're expecting a big fat stinky check from Joe Biden. They'll just triple the taxation to the poor to make up for it.
Socialism hand-in-hand with corporatism... haha, I know, I know. :D
I mean, we wouldn't have a dark cyberpunk future (now-present) without megacorps greasing the wheels of a few contemporary politicians, now would we? :borg:
x86 has nothing to do with it. AMD is proof that x86 still has lots of life when there's a profit motive. Intel was so dominant for such a long time that they became complacent and let others catch up and surpass them. Not having competition means that the most profitable thing to do is not spend R&D, which is what Intel did. Intel now has no choice but to spend money and yes, $20B is not enough. Joe Biden better not help pay for this sorta thing.
You know, I thought x86/x86-64 would be dead by 2030 at the latest, but AMD has proven that ideology wrong, and have definitely shown that there is still a lot of life left in that ISA - agreed!
 

Mega6

2[H]4U
Joined
Aug 13, 2017
Messages
3,549
Currently valued at just shy of $95B
Market Cap of AMD is $92B, add in just acquired and yet to be finalized Xlinx of $29B and we have $121B. Typically, buying a company out offers a premium to shareholders in excess of market cap, so we are talking more than that.
 

rinaldo00

[H]ard|Gawd
Joined
Mar 9, 2005
Messages
1,682
AMD is worth a lot more than 20 billion, if nvidia wanted to buy arm for 40 billion then AMD is probably around 100 billion. didn't they buy xilinx for 20 billion?

https://www.macrotrends.net/stocks/charts/AMD/amd/net-worth
Interactive chart of historical net worth (market cap) for AMD (AMD) over the last 10 years. How much a company is worth is typically represented by its market capitalization, or the current stock price multiplied by the number of shares outstanding. AMD net worth as of March 25, 2021 is $92.68B.
 
Top