Intel Weighs Options Including Foundry Split to Stem Losses

[benb]

Strive for excellence: at TSMC, but not Intel: It's simply a fact that TSMC lets innovation in through many channels while Intel has uber-aggressive innovation-rejecting mechanisms called the Virtual Factory and Copy Exact.

TSMC operates in many ways like a startup despite being a monopoly. Intel operates like a monopoly; top down control, tough politics rather than data-driven decision-making. To survive at Intel, you need to be tough. To survive at TSMC, you need to be smart.

 

[blueone]

Strive for excellence: at TSMC, but not Intel: It's simply a fact that TSMC lets innovation in through many channels while Intel has uber-aggressive innovation-rejecting mechanisms called the Virtual Factory and Copy Exact.

TSMC operates in many ways like a startup despite being a monopoly. Intel operates like a monopoly; top down control, tough politics rather than data-driven decision-making. To survive at Intel, you need to be tough. To survive at TSMC, you need to be smart.

Have you ever worked at TSMC or Intel?

 

[BruceA]

Strive for excellence: at TSMC, but not Intel: It's simply a fact that TSMC lets innovation in through many channels while Intel has uber-aggressive innovation-rejecting mechanisms called the Virtual Factory and Copy Exact.

TSMC operates in many ways like a startup despite being a monopoly. Intel operates like a monopoly; top down control, tough politics rather than data-driven decision-making. To survive at Intel, you need to be tough. To survive at TSMC, you need to be smart.

You sound like you know the insides of both, LOL

 

[nghanayem]

Strive for excellence: at TSMC, but not Intel: It's simply a fact that TSMC lets innovation in through many channels while Intel has uber-aggressive innovation-rejecting mechanisms called the Virtual Factory and Copy Exact.

TSMC operates in many ways like a startup despite being a monopoly. Intel operates like a monopoly; top down control, tough politics rather than data-driven decision-making. To survive at Intel, you need to be tough. To survive at TSMC, you need to be smart.

TSMC introduces ‘Global Gigafab’ Program, Provides Glimpse into Revolutionary Process Replication

TSMC's Global Gigafab Manufacturing program aims to replicate its manufacturing processes across its various gigafab sites, with a focus on sustainability and social responsibility, to ensure consistent operational efficiency and production quality on a global scale.

pcge.eu

If that is truly what you believe then the era of TSMC innovation is over.

People who say CE! stifles innovation clearly don't understand how it works and why it is done. Read intel's papers on how CE! and the VF work (they aren't long) and get back to me on those views. CB also did some lovely interviews on the topic explaining his rationale for implementing it and other quality and LEAN engineering principles from Japan. After all the 1st rule of the Toyota Production System is That ALL work must have a specified content, sequence, timing, and outcome. And if there is anyone who knows how to do low cost, high quality manufacturing no matter where they are on the globe and with industry leading cycle time it is Toyota. When a new best known method is developed ALL must switch over to this new better method to avoid quality misses and creating waste. If you think that proliferating the newest BKMs to a whole factory is a bad thing, then clearly you don't understand manufacturing, because I guarantee you TSMC and everyone else who is good at manufacturing does that.

 

[Daniel Nenni]

Strive for excellence: at TSMC, but not Intel: It's simply a fact that TSMC lets innovation in through many channels while Intel has uber-aggressive innovation-rejecting mechanisms called the Virtual Factory and Copy Exact.

TSMC operates in many ways like a startup despite being a monopoly. Intel operates like a monopoly; top down control, tough politics rather than data-driven decision-making. To survive at Intel, you need to be tough. To survive at TSMC, you need to be smart.

TSMC is customer/margin driven while Samsung and Intel are technology/ego driven. What choice does Intel and Samsung have since they don't have the customers or ecosystem TSMC does? The foundry business is really rough, hundreds of billions have been spent and TSMC is the only one making a decent ROI.

TSMC has 99% market share at 3nm and 2nm will be the same. That type of customer momentum will be very difficult to compete with for the next five years. Intel or Samsung will have to hit a home run on the technology side to get back into the foundry game, my opinion. Of course there is always the NOT TSMC market segment, that will always exist.

The ecosystem (IP/EDA/Etc.) is push and pull. Every company in the ecosystem is pushing TSMC while Intel, Samsung, and Rapidus are pulling. A very different dynamic, absolutely.

 

[BruceA]

TSMC is customer/margin driven while Samsung and Intel are technology/ego driven. What choice does Intel and Samsung have since they don't have the customers or ecosystem TSMC does? The foundry business is really rough, hundreds of billions have been spent and TSMC is the only one making a decent ROI.

TSMC has 99% market share at 3nm and 2nm will be the same. That type of customer momentum will be very difficult to compete with for the next five years. Intel or Samsung will have to hit a home run on the technology side to get back into the foundry game, my opinion. Of course there is always the NOT TSMC market segment, that will always exist.

The ecosystem (IP/EDA/Etc.) is push and pull. Every company in the ecosystem is pushing TSMC while Intel, Samsung, and Rapidus are pulling. A very different dynamic, absolutely.

If you count and assume Panther Lake and CCG is a IFS captive foundry customer Intel will have the basis of volume and start of IP to have the beginning of a real alternative foundry offering other than TSMC, the world needs it.

The question is does Intel have the stomach as well as necessary leadership to make the culture pivot as well as the cash to carry them over to 2027 or so.

 

[kopper]

I can't believe you've repeated this complete nonsense about US work culture. I let it go last time (this being a professional semiconductor discussion site and not wishing to get sucked into your tirades of personal abuse).

I seriously wonder exactly where you're getting all this from. My experience over more than three decades around the semiconductor industry is the exact opposite. It is the US companies - certainly the start-ups - who are the most innovative and place the least emphasis on traditional values and social coshesion and stability. And have the least attachments to respecting people purely for their seniority. Of course, you will always find exceptions - and large companies everywhere tend to sclerosis unless well managed.

How on earth do you think Silicon Valley happened, thrived and continues to grow if US work values are so pathetic and anti-innovative ?

Why would you assume that TI work culture is the same as US work culture in general ? Silicon Valley work culture is vastly different from that in Texas (or certainly was when I was at TI). Yes, TI had some specific things at that time that we might consider failings - rarely hired externally after graduate level, top management all seemed to have gone through SMU, etc. . I imagine some of those have changed since then. However, complacency and mediocrity were not amongst them. They put a lot of emphasis on quality, customers and continuous improvement.

I came across an interesting phrase yesterday watching a video about Korea: "creative innovation". It strikes me that this is what Asia (Japan, Korea, Taiwan, China) is really good at - getting into existing industries after the initial innovation and growth phase and then mastering the incremental innovations in the following phases. It's also what the UK is spectacularly bad at, but that's another discussion. But we risk confusing that sort of innovation with the initial "breakout innovation" where completely new products and concepts are developed.

As for your repeated assertion that the US is somehow prejudiced against promoting foreigners: I can only assume you've never worked in countries like France or Korea ! Nowhere is a perfect meritocracy. But the US is at the top end of the league table. Note: I'm saying this as a Brit, so no dog in this fight.

I, too, was avoiding this one as I have spoken about it enough within the fab-building topic. The levels of labor inefficiency I've seen at the AZ build remind me of my time living in TW. Why do 5 people from 4 companies touch the same data, manually moving it from one document to another? I suspect the answer is, "security," and complete BS. The innovative, efficient method is a shared file where we all update our parts individually with no duplication of data entry (eliminating the extra movement, a LEAN waste). So me, the US guy who has an incredibly constrained labor supply of skilled tradespeople, is applying Japanese manufacturing methodology (Toyota Production System) to maximize output through efficiency. Meanwhile, my customer wants me to waste resources doing things out of sequence so that his boss can see "progress" only for us to remove the work since it physically blocks the work that is first in sequence but does not provide a visual sense of accomplishment. That is 4x the labor consumption compared to remaining in sequence and being able to show the boss how the progress unfolds.

That entire post read like a projection. Going against a formal hierarchy is part of our business ethos. How many times, just within Silicon Valley, was a pioneering company formed from the innovators being stifled at their current employer and striking out on their own? While we have a segment of the workforce that will kowtow to senior management, this behavior is generally out of personal survival needs and, simultaneously, viewed with disdain by the non-butt-kissing employees. Corporate giants can get stuck in the status quo when they've allowed the layers of management to grow too thick and create survival behaviors - this can be fixed through a re-organization with the fat trimmed out - and this is part of the current Intel problem (from personal observation).

The number of times I heard, "That's not how we do it in TW," demonstrates the level of arrogance that there is nothing to be learned from others. Arrogance bred by being the stalwart of an entire nation. Hearing that the expectation for US workers to be honored to work there is laughable. We don't care. We just want to build your factory and products and go home to our families. A happy employee is a more productive employee. Extended overtime leads to mistakes and burnout. Enough books have already been written on the topic.

I don't see enough mention in the Intel split concept regarding the military and government contracts that would need to be re-homed to a TI, Micron, Microchip, or similar. TSMC doesn't qualify as it is not US-headquartered and does not have back-end processing in the US.

 

[BruceA]

I, too, was avoiding this one as I have spoken about it enough within the fab-building topic. The levels of labor inefficiency I've seen at the AZ build remind me of my time living in TW. Why do 5 people from 4 companies touch the same data, manually moving it from one document to another? I suspect the answer is, "security," and complete BS. The innovative, efficient method is a shared file where we all update our parts individually with no duplication of data entry (eliminating the extra movement, a LEAN waste). So me, the US guy who has an incredibly constrained labor supply of skilled tradespeople, is applying Japanese manufacturing methodology (Toyota Production System) to maximize output through efficiency. Meanwhile, my customer wants me to waste resources doing things out of sequence so that his boss can see "progress" only for us to remove the work since it physically blocks the work that is first in sequence but does not provide a visual sense of accomplishment. That is 4x the labor consumption compared to remaining in sequence and being able to show the boss how the progress unfolds.

That entire post read like a projection. Going against a formal hierarchy is part of our business ethos. How many times, just within Silicon Valley, was a pioneering company formed from the innovators being stifled at their current employer and striking out on their own? While we have a segment of the workforce that will kowtow to senior management, this behavior is generally out of personal survival needs and, simultaneously, viewed with disdain by the non-butt-kissing employees. Corporate giants can get stuck in the status quo when they've allowed the layers of management to grow too thick and create survival behaviors - this can be fixed through a re-organization with the fat trimmed out - and this is part of the current Intel problem (from personal observation).

The number of times I heard, "That's not how we do it in TW," demonstrates the level of arrogance that there is nothing to be learned from others. Arrogance bred by being the stalwart of an entire nation. Hearing that the expectation for US workers to be honored to work there is laughable. We don't care. We just want to build your factory and products and go home to our families. A happy employee is a more productive employee. Extended overtime leads to mistakes and burnout. Enough books have already been written on the topic.

I don't see enough mention in the Intel split concept regarding the military and government contracts that would need to be re-homed to a TI, Micron, Microchip, or similar. TSMC doesn't qualify as it is not US-headquartered and does not have back-end processing in the US.

It was indeed amazing at the arrogance and closed mind of the TSMC assignes and leadership, the bridges they burned and credibility lost with their arrogance. What is more strange they got lots of feedback bit internally and from consultants, all ignored. Can’t wait to see what happens in Germany, LOL

 

[kopper]

It was indeed amazing at the arrogance and closed mind of the TSMC assignes and leadership, the bridges they burned and credibility lost with their arrogance. What is more strange they got lots of feedback bit internally and from consultants, all ignored. Can’t wait to see what happens in Germany, LOL

No kidding. If you think Arizona labor is challenging (and we are an open-shop state), wait and see what it is like with the uber-rigid German mindset. Perhaps the reason we see Bosch involved with the factory build.

 

[delong.height]

It seems Pat is going all in on IFS, sell non-core units. No split yet

Exclusive-Intel CEO to pitch board on plans to shed assets, cut costs, source says

Intel CEO Pat Gelsinger and key executives are expected to present a plan later this month to the company’s board of directors to slice off unnecessary businesses and revamp capital spending, according to a source familiar with the matter, as they try to revive the once-dominant chipmaker's...

finance.yahoo.com

Will the BOD accept the plan or find another CEO?

I'm a shareholder, I don't agree on selling non-core units (at this point IFS) for cash. Pat G is a double-edged sword, he brought 18A to life which is great, yet tank shareholders' return in doing so. I don't think getting more cash will change any of the picture. I would say to Pat G. If you want to build fab, get customers now (you have good technologies, but if you can't get customer,; how can i trust you with your sayings). And, no money for new fab shell that won't be in life until 4 years from now.

I would like them to spin off companies (mobileye, altera, x86 design teams for Core and Xeon). Keep IFS under intel's name. Return those companies' shares to shareholder. Or sell those companies to Nvidia, Broadcom, and Qualcomm, they need to make wafer agreement with those companies if they want to get it. Move their designs from TSMC to Intel, while Intel making sure that their supply is good and solid, and gorwing from now on. Otherwise, I will be selling my shares (if they are getting cash but not 1). return stocks of spun-off companies to shareholders, 2). make wafer agreement with buyer companies(Nvidia, Broadcom, and Qualcomm)).

 

[kopper]

TSMC introduces ‘Global Gigafab’ Program, Provides Glimpse into Revolutionary Process Replication

TSMC's Global Gigafab Manufacturing program aims to replicate its manufacturing processes across its various gigafab sites, with a focus on sustainability and social responsibility, to ensure consistent operational efficiency and production quality on a global scale.

pcge.eu

If that is truly what you believe then the era of TSMC innovation is over.

People who say CE! stifles innovation clearly don't understand how it works and why it is done. Read intel's papers on how CE! and the VF work (they aren't long) and get back to me on those views. CB also did some lovely interviews on the topic explaining his rationale for implementing it and other quality and LEAN engineering principles from Japan. After all the 1st rule of the Toyota Production System is That ALL work must have a specified content, sequence, timing, and outcome. And if there is anyone who knows how to do low cost, high quality manufacturing no matter where they are on the globe and with industry leading cycle time it is Toyota. When a new best known method is developed ALL must switch over to this new better method to avoid quality misses and creating waste. If you think that proliferating the newest BKMs to a whole factory is a bad thing, then clearly you don't understand manufacturing, because I guarantee you TSMC and everyone else who is good at manufacturing does that.

And yet, if you have been in multiple copies of the same fab, you quickly see how things were changed on each project due to cost savings directives or schedule issues (procurement challenges drive many of the deviations). The VF will keep the operating parameters consistent, but for construction, things get changed at the project level (moving countries, some of the changes are required by differences in utilities, codes, and standard construction practices). Even adopting TICE cannot prevent these changes as we must meet the milestone dates.

 

[kopper]

I'm a shareholder, I don't agree on selling non-core units (at this point IFS) for cash. Pat G is a double-edged sword, he brought 18A to life which is great, yet tank shareholders' return in doing so. I don't think getting more cash will change any of the picture. I would say to Pat G. If you want to build fab, get customers now (you have good technologies, but if you can't get customer,; how can i trust you with your sayings). And, no money for new fab shell that won't be in life until 4 years from now.

I would like them to spin off companies (mobileye, altera, x86 design teams for Core and Xeon). Keep IFS under intel's name. Return those companies' shares to shareholder. Or sell those companies to Nvidia, Broadcom, and Qualcomm, they need to make wafer agreement with those companies if they want to get it. Move their designs from TSMC to Intel, while Intel making sure that their supply is good and solid, and gorwing from now on. Otherwise, I will be selling my shares (if they are getting cash but not 1). return stocks of spun-off companies to shareholders, 2). make wafer agreement with buyer companies(Nvidia, Broadcom, and Qualcomm)).

History supports your position. More cash is a band-aide. Growing up, I watched an industry giant self-destruct. Using "non-core" business units as cash generators via sell-off ultimately hurt more through the reduction of diversified revenue, resulting margin, and loss of leverageable PPE. Meanwhile, the core business continued to lose sales as the consumer market shifted from analog to digital (their own boat they declined to board, but that is another book to write). Thankfully, dad survived all the rounds of lay-off until he was ready to take an early retirement option and change companies.

 

[hist78]

I, too, was avoiding this one as I have spoken about it enough within the fab-building topic. The levels of labor inefficiency I've seen at the AZ build remind me of my time living in TW. Why do 5 people from 4 companies touch the same data, manually moving it from one document to another? I suspect the answer is, "security," and complete BS. The innovative, efficient method is a shared file where we all update our parts individually with no duplication of data entry (eliminating the extra movement, a LEAN waste). So me, the US guy who has an incredibly constrained labor supply of skilled tradespeople, is applying Japanese manufacturing methodology (Toyota Production System) to maximize output through efficiency. Meanwhile, my customer wants me to waste resources doing things out of sequence so that his boss can see "progress" only for us to remove the work since it physically blocks the work that is first in sequence but does not provide a visual sense of accomplishment. That is 4x the labor consumption compared to remaining in sequence and being able to show the boss how the progress unfolds.

That entire post read like a projection. Going against a formal hierarchy is part of our business ethos. How many times, just within Silicon Valley, was a pioneering company formed from the innovators being stifled at their current employer and striking out on their own? While we have a segment of the workforce that will kowtow to senior management, this behavior is generally out of personal survival needs and, simultaneously, viewed with disdain by the non-butt-kissing employees. Corporate giants can get stuck in the status quo when they've allowed the layers of management to grow too thick and create survival behaviors - this can be fixed through a re-organization with the fat trimmed out - and this is part of the current Intel problem (from personal observation).

The number of times I heard, "That's not how we do it in TW," demonstrates the level of arrogance that there is nothing to be learned from others. Arrogance bred by being the stalwart of an entire nation. Hearing that the expectation for US workers to be honored to work there is laughable. We don't care. We just want to build your factory and products and go home to our families. A happy employee is a more productive employee. Extended overtime leads to mistakes and burnout. Enough books have already been written on the topic.

I don't see enough mention in the Intel split concept regarding the military and government contracts that would need to be re-homed to a TI, Micron, Microchip, or similar. TSMC doesn't qualify as it is not US-headquartered and does not have back-end processing in the US.

"I don't see enough mention in the Intel split concept regarding the military and government contracts that would need to be re-homed to a TI, Micron, Microchip, or similar. TSMC doesn't qualify as it is not US-headquartered and does not have back-end processing in the US."

Many semiconductor products needed by the U.S. defense industry are not produced by Intel at all. That's why the DoD declined to endorse the so-called "secure enclave" fab facility promoted by Intel. Please refer to the DoD's Trusted Foundry Program document listed below. On the other hand, the DoD, DARPA, and DoE have been using and collaborating with TSMC directly or indirectly through the defense industry supply chain for many years, even though TSMC is not part of the DoD Trusted Foundry Program. In fact, Intel is not in the DoD Trusted Foundry Program either.

One of the leading agencies advocating for TSMC to bring its manufacturing capabilities to the U.S. is the Department of Defense, for obvious reasons. In many cases, the DoD and DoE are more practical than people might think. They want real products delivered on time and within budget. A DoD contractor told me a while back that the DoD doesn't have much interest in the "technology" itself; it wants "real" products that can help weapon systems work. Technology or a roadmap on paper can't help a fighter jet fly too far.


DoD Trusted Foundry Program:

https://www.acq.osd.mil/asds/dmea/tapo/docs/tp/Accredited-Supplier-22May2024.pdf

 

[benb]

TSMC improves their facility design specifications with each fab, several fabs a year for the last decade, guided partly by TSMC (in the form of a spec) but more by the turn-key contract holder. Turn-key is a key enabler for innovation, and efficiency, in my opinion. But it only works in TW, where contractors move quickly from contract to contract, and can live with getting paid from a fixed pot of money they know in advance. Like when you go to a lawyer and they ask for a retainer. That's not how US contractors operate.

 

[jim_22222]

I can't believe you've repeated this complete nonsense about US work culture. I let it go last time (this being a professional semiconductor discussion site and not wishing to get sucked into your tirades of personal abuse).

I seriously wonder exactly where you're getting all this from. My experience over more than three decades around the semiconductor industry is the exact opposite. It is the US companies - certainly the start-ups - who are the most innovative and place the least emphasis on traditional values and social coshesion and stability. And have the least attachments to respecting people purely for their seniority. Of course, you will always find exceptions - and large companies everywhere tend to sclerosis unless well managed.

How on earth do you think Silicon Valley happened, thrived and continues to grow if US work values are so pathetic and anti-innovative ?

Why would you assume that TI work culture is the same as US work culture in general ? Silicon Valley work culture is vastly different from that in Texas (or certainly was when I was at TI). Yes, TI had some specific things at that time that we might consider failings - rarely hired externally after graduate level, top management all seemed to have gone through SMU, etc. . I imagine some of those have changed since then. However, complacency and mediocrity were not amongst them. They put a lot of emphasis on quality, customers and continuous improvement.

I came across an interesting phrase yesterday watching a video about Korea: "creative innovation". It strikes me that this is what Asia (Japan, Korea, Taiwan, China) is really good at - getting into existing industries after the initial innovation and growth phase and then mastering the incremental innovations in the following phases. It's also what the UK is spectacularly bad at, but that's another discussion. But we risk confusing that sort of innovation with the initial "breakout innovation" where completely new products and concepts are developed.

As for your repeated assertion that the US is somehow prejudiced against promoting foreigners: I can only assume you've never worked in countries like France or Korea ! Nowhere is a perfect meritocracy. But the US is at the top end of the league table. Note: I'm saying this as a Brit, so no dog in this fight.

One example is that TSMC recently promoted CC Wei as chairman, and Chang considered it a difficult decision since he got his PhD from Yale instead of one of the top engineering institutions. In contrast, Rich Templeton did not face any such problems from the board, and the same applies to BK at Intel. If you don't see the difference in attitude here, it's hard to continue the discussion.

 

[Daniel Nenni]

If you count and assume Panther Lake and CCG is a IFS captive foundry customer Intel will have the basis of volume and start of IP to have the beginning of a real alternative foundry offering other than TSMC, the world needs it.

The question is does Intel have the stomach as well as necessary leadership to make the culture pivot as well as the cash to carry them over to 2027 or so.

Are you assuming that Intel internal design groups use the Intel Foundry PDKs and commercial IP like external customers?

 

[Daniel Nenni]

One example is that TSMC recently promoted CC Wei as CEO, and Chang considered it a difficult decision since he got his PhD from Yale instead of one of the top engineering institutions. In contrast, Rich Templeton did not face any such problems from the board, and the same applies to BK at Intel. If you don't see the difference in attitude here, it's hard to continue the discussion.

BK had a bachelors degree in chemistry from San Jose State. Hardly fortune 500 CEO material. Paul Otellini had an MBA and BA in economics. The fall of Intel started with Paul, my opinion.

The Updated Legacy of Intel CEOs - Semiwiki

(First published December 24, 2014) A list of the best and worst CEOs in 2014 was recently published. The good news is that none of our semiconductor CEOs were on the worst list. The bad news is that none of our semiconductor CEOs were on the best list either. I will be writing about the…

semiwiki.com

 

[BruceA]

Are you assuming that Intel internal design groups use the Intel Foundry PDKs and commercial IP like external customers?

LOL

 

[hist78]

One example is that TSMC recently promoted CC Wei as CEO, and Chang considered it a difficult decision since he got his PhD from Yale instead of one of the top engineering institutions. In contrast, Rich Templeton did not face any such problems from the board, and the same applies to BK at Intel. If you don't see the difference in attitude here, it's hard to continue the discussion.

"TSMC recently promoted CC Wei as CEO"


C.C. Wei has been TSMC's CEO since June 2018 and took the additional role as the Chairman of TSMC since June 2024 after Mark Liu's retirement.

 

[BruceA]

BK had a bachelors degree in chemistry from San Jose State. Hardly fortune 500 CEO material. Paul Otellini had an MBA and BA in economics. The fall of Intel started with Paul, my opinion.

The Updated Legacy of Intel CEOs - Semiwiki

(First published December 24, 2014) A list of the best and worst CEOs in 2014 was recently published. The good news is that none of our semiconductor CEOs were on the worst list. The bad news is that none of our semiconductor CEOs were on the best list either. I will be writing about the…

semiwiki.com

BK and his multiple wives and infamous closed meeting rants! PSO and his lack of any technical competence.

Than afterwards we had Bob…. Failure started at the BoD and still a fundamental problem there. When a company is rotting the BoD is the last chance. When you have a highly competent and performing leadership the BoD is less important.

Intel’s failure starts with Andy, Omar and their inability to provide the necessary governance of a sinking company.