Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/what-would-you-do-if-you-are-the-ceo-of-intel.5710/page-4
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

What would you do if you are the CEO of Intel?

Intel's problem is too many chiefs and not enough Indians and they are far from the only ones.
 
Agree with count, the Loeb Letter points to a dismantling of Intel, ending IDM status. Probably a spinoff of MFG and design units, my opinion. It’s not up to Loeb of course, it’s up to the new CEO.

I’m assuming a new CEO.

Agree with count on “carnage”. Intel cost structure was supported by leading technology. In the current state they are in, trailing by 3 nodes, no customers other than internal, I think the manufacturing spin co will need lowest cost operations; UMC or SMIC would be the points of comparison. Humbling.

GloFo is the more obvious point of comparison, I could see a wafer agreement being part of a spinoff as a sweetener. An actual combination with GloFo would be really interesting as an Intel Fab/GloFo combined business should in theory be viable... except for the fact that the thing would be run by private equity.
 
There must be a plan to downsize the site in oregon and let people go.
 
GloFo is the more obvious point of comparison, I could see a wafer agreement being part of a spinoff as a sweetener. An actual combination with GloFo would be really interesting as an Intel Fab/GloFo combined business should in theory be viable... except for the fact that the thing would be run by private equity.
I'm wondering if the multi years wafer agreement between AMD and Globalfoundries greatly restrained AMD's ability to compete after their split. After Globalfoundries announced stopped development on 7nm or smaller process technology, AMD was finally free to make decision based on what's best for AMD without worrying about their contract commitment to Globalfoundries.

If a similar wafer agreement is required on splitting Intel into a fabless Intel and a new foundry company with Intel's existing manufacturing division, I'm not sure it's good for the new fabless Intel. It can be even worst than not to spilt them into two independent entities.

While AMD had spent several years to sort out the relationship with their old brothers and sisters at Globalfoundries, Intel doesn't have such luxury to do so at all.
 
I'm wondering if the multi years wafer agreement between AMD and Globalfoundries greatly restrained AMD's ability to compete after their split. After Globalfoundries announced stopped development on 7nm or smaller process technology, AMD was finally free to make decision based on what's best for AMD without worrying about their contract commitment to Globalfoundries.

If a similar wafer agreement is required on splitting Intel into a fabless Intel and a new foundry company with Intel's existing manufacturing division, I'm not sure it's good for the new fabless Intel. It can be even worst than not to spilt them into two independent entities.

While AMD had spent several years to sort out the relationship with their old brothers and sisters at Globalfoundries, Intel doesn't have such luxury to do so at all.
Agree the chaotic way AMD spun off Global Foundries made their competitive strategy worse for a while. I think the Intel spinoff will learn from that mistake though, unless they are really that dumb (can't rule it out).

Changing topics slightly, unless AMD starts building more integrated-graphics chips, I think enterprises with remote desktop setups for work from home will continue to buy Intel because of integrated graphics, regardless of AMD advantages in node and performance. Enterprises don't pay for a graphics card as a rule, and graphics cards are hard to buy at any price currently. So Intel has that market sewn up for now. I wonder how much hay Intel can make off things like integrated graphics, Ethernet networking, and Wifi, all of which they own near-monopolies on.
 
According to Bloomberg, Intel is in talks with TSMC and Samsung to outsource some chip production. Article is found here.

Key points:
1. TSMC is offering Intel a 4-nanometer process with initial testing using the older 5-nanometer process. Test production of the 4-nm chips would be available in 4Q2021. Volume shipments would begin the following year. Products wouldn't come to market until 2023 at the earliest and would be based on established processes used by other TSMC customers.

2. Talks with Samsung Foundry is currently at a more preliminary stage.

3. Intel is still hoping for last-minute improvements in its own production capabilities & have yet to make a final decision less than 2 weeks ahead of a scheduled announcement of its plans
 
According to Bloomberg, Intel is in talks with TSMC and Samsung to outsource some chip production. Article is found here.

Key points:
1. TSMC is offering Intel a 4-nanometer process with initial testing using the older 5-nanometer process. Test production of the 4-nm chips would be available in 4Q2021. Volume shipments would begin the following year. Products wouldn't come to market until 2023 at the earliest and would be based on established processes used by other TSMC customers.

2. Talks with Samsung Foundry is currently at a more preliminary stage.

3. Intel is still hoping for last-minute improvements in its own production capabilities & have yet to make a final decision less than 2 weeks ahead of a scheduled announcement of its plans

Intel will start at 3nm otherwise they will be behind AMD. Intel has already started hiring foundry teams in Taiwan and Korea so outsourcing will happen, absolutely. I would bet it will be a fab-lite situation versus fabless for Intel. Especially with a corporate raider in the mix. TSMC is upping CAPEX to $20B and Intel cannot keep up with that pace. Samsung can but they can't yield so it's TSMC's business to lose.
 
I don't think ishrak would submit to an activist investor. He's just agreeing with stuff he likes.

That Oregon can't have a loyal team of 5000 in production says something. This wouldn't be happening if .....
 
Intel will start at 3nm otherwise they will be behind AMD. Intel has already started hiring foundry teams in Taiwan and Korea so outsourcing will happen, absolutely. I would bet it will be a fab-lite situation versus fabless for Intel. Especially with a corporate raider in the mix. TSMC is upping CAPEX to $20B and Intel cannot keep up with that pace. Samsung can but they can't yield so it's TSMC's business to lose.
Intel is going to start from behind AMD, that's the situation they have put themselves in by not being decisive in the past. A fab lite model would not be decisive and would leave them even further behind in the future. In my opinion, Intel needs to go all in to fabless, knowing that it means the next 2-3 years will look really bad while they catch up with AMD, but at least they have the balance sheet right now that they will be able to ride out the transition. If they continue to be indecisive on this there will come a point where it'll be too late. It may already be too late - AMD is now a $120b company that's unburdened by foundries so they can put all their capex into making their chips better. Intel's Capex is spread apart between too many businesses to keep track of.
 
Putting everything in one place (Taiwan) is a bad idea.

It was mentioned that 7nm Intel is an old generation for tsmc.

Israel has a team of 5000-5500 that's dependable. Oregon doesn't have that. To be honest silicon forest always relied on importing talent from Israel and now that they don't have that at best there are only partially completed projects.
 
Last edited by a moderator:
Putting everything in one place (Taiwan) is a bad idea.
but who can beat TSMC? I've listened to Daniel interview on Moores laws is dead podcast and someone ask him about "Big Red". Daniel said he is disappointed or something in that regard that China can't get their semi industry up and they have been doing it way before Trump.

this interview with Powerchip Technology founder (its in Chinese), the founder said Taiwan have the lowest cost compare to other country like South Korea and even China! Now, i don't know if that's true or not but since he say it and his company own fabs. i'll take his words for it.

so far i got two sources (Daniel and Frank Huang) said "Big Red" semi industry just can't compete. Can you tell me who can beat TSMC? Samsung? Intel?

Isn't Trump asking TSMC to setup shop in US is to address this problem?

 
Last edited:
Putting everything in one place (Taiwan) is a bad idea.

It was mentioned that 7nm Intel is an old generation for tsmc.

Israel has a team of 5000-5500 that's dependable. Oregon doesn't have that. To be honest silicon forest always relied on importing talent from Israel and now that they don't have that at best there are only partially completed projects.

I think Intel made a strategical mistake by not establishing a major R&D and/or a major manufacturing facility in Taiwan. TI, Micron, and Qualcomm all have important Taiwan based operations. With so many experienced and talented semiconductor engineers and robust semiconductor supply chain in Taiwan, Intel didn't take full advantage on it.
 
So far island countries like Cuba, new Zealand, and taiwan haven't been effected but I'm assuming covid will shut those countries down too and will cause havoc.
 
We've all seen this story before. Look for Bob Swan to arrange a golden parachute for himself, allow third point to dismantle the company and collect a bonus for sticking around through the carnage, and then float off into either retirement or politics as a very rich man.
He must be reading this forum.
 
The first thing Intel has to do is recognize the PC business has become like the DRAM business back in the late 1970s; it is becoming commoditized, too competitive, and will not support their traditional margins. ARM ecosystem mobile chips will be "good enough" to replace x86 chips in 1-5 years, at a decisively lower price point. Intel may be able to retain the server business longer than this.

Intel has to start over with a new CEO and a new business model. A tiny spinoff from the legacy Intel. Like Intel in the late 1970s.
Hey there 2015 self. Pandemic’s a surprise, huh?

”PC business...it is becoming commoditized, too competitive, and will not support their traditional margins”.
Commoditized, yes, competitive, yes, but it has supported their traditional margins. Surprisingly.
”Intel has to start over with a new CEO”
Yep. And then keep starting over again. Brian K. was CEO in 2015, then Bob S., now Pat G.
”...and a new business model”
There hasn’t been much change: Still x86, still IDM, still stagnant in all respects.

So, let’s make some new predictions about Intel’s future.

- They will develop advanced nodes for a smaller VF
- They will keep F28 and F42 going in some capacity, but it won’t be pretty
- They won’t regain process development power
- They will gain for the first time, cost/chip competitiveness which will justify keeping some fabs, to produce some products
- Intel designs on foundry nodes will be surprisingly good and profitable, like AMD’s have been
 
Hey there 2015 self. Pandemic’s a surprise, huh?

”PC business...it is becoming commoditized, too competitive, and will not support their traditional margins”.
Commoditized, yes, competitive, yes, but it has supported their traditional margins. Surprisingly.
”Intel has to start over with a new CEO”
Yep. And then keep starting over again. Brian K. was CEO in 2015, then Bob S., now Pat G.
”...and a new business model”
There hasn’t been much change: Still x86, still IDM, still stagnant in all respects.

So, let’s make some new predictions about Intel’s future.

- They will develop advanced nodes for a smaller VF
- They will keep F28 and F42 going in some capacity, but it won’t be pretty
- They won’t regain process development power
- They will gain for the first time, cost/chip competitiveness which will justify keeping some fabs, to produce some products
- Intel designs on foundry nodes will be surprisingly good and profitable, like AMD’s have been
One thing for sure, as early as 2015, many people on this Semieiki already sensed Intel was on a trouble and dangerous path.
 
It’s fun to look back 5 years, good thread redux.

Before we bury Intel too much, just wanted to point out...
How amazing is it that from 2015 to 2021 Intel has been on the same node, has fallen behind by 3 nodes (TSMC is on 5nm), but their aren’t tumbleweeds at all the Intel sites like they’re Kmarts? Something is wrong there, but also right.
I think the surprising rightness of Intel is found in the way they cost reduced and optimized 14nm so that the gap with 5nm isn’t that great. The whole organization, from chip designers to manufacturing sites, participated in extending the life of a node to an extraordinary degree. It’s not something foundries could do.
 
Back
Top