Shared pain shared gain: the start of a monopoly in leading-edge logic chip manufacturing ?

[user nl]

so you think Arizona is large enough to have multiple fabs in same neighbor hood. a lot of work ers like it for job opportunity.
intc maybe is just dumb.
or smart, they know tsmc can't compete with them.
what you think

Arizona has been courting TSMC since 2013, so after dating for many years it seems they got engaged in 2020 and have started to produce some offspring now in 2024!
https://finance.yahoo.com/news/taiwan-arizona-ties-history-behind-093000132.html?fr=sycsrp_catchall

 

[chipsntexas]

TSMC would have invented those process flows. It was an entire semiconductor ecosystem that pushed for those innovations.

Wow! Let's dismiss decades of Intel dominance in the industry by saying that TSMC would have invented them too.
While possible, it doesn't change the fact of who did it, and when. They're hardly a yap dog - still have $50B revenue after 2 down years.
Intel is in short term crisis due to multiple poor decisions and weak execution, to be sure.
But this is a very large pool, and too complex to simply say TSMC will be the only one standing - last I checked they make no memory.
And all the AI logic chips in the world don't work without it; the chips themselves need that and a system (SoC) to be of any value.
Intel is still leading the way in some of these areas, so their lack is not so much in innovation as in execution.
That is a fixable problem, just need time and money, both of which are running in very short supply.

 

[samwilde]

This is a consequence of Rock's law.

Moore's second law - Wikipedia

en.wikipedia.org

Fab construction costs double every two process nodes. You will see them double again when High-NA comes out.

The amount of businesses with enough capital to build leading edge fabs naturally tends to decrease. Samsung and Intel are only in the race because they are IDMs.

 

[user nl]

This is a consequence of Rock's law.

Moore's second law - Wikipedia

en.wikipedia.org

Fab construction costs double every two process nodes. You will see them double again when High-NA comes out.

The amount of businesses with enough capital to build leading edge fabs naturally tends to decrease. Samsung and Intel are only in the race because they are IDMs.

Thanks, I didn't know that. So, the question is now that Moore's law may be dying, or has slowed down very much, is also Rock's law slowing down enough for IFS to keep up?

Now that IFS has revenue of about 4-4,5 B$/quarter (Q1 and Q2-2024), so about 16-18 B$/year and if IFS wants to invest some 30% in capital (assuming 50-60% gross margin) it can spend about 5-6 B$/year in capital. If they can double that by financial engineering (private capital + government subsidies) they have about 12 B$/year for capital spending.

TSMC spends now about 30-32 B$/year in capital (revenue some 20 B$/quarter in Q2-2024 with healthy 53% gross margin).

Hard to see how IFS will survive this race with TSMC in leading edge logic manufacturing.

Does this make sense?

Here a useful reference I found from the situation some 20 years ago in 2003 and Rock's law:
https://archive.ph/20130119214941/http://news.com.com/Semi+survival/2009-1001_3-981418.html

 

[hist78]

Thanks, I didn't know that. So, the question is now that Moore's law may be dying, or has slowed down very much, is also Rock's law slowing down enough for IFS to keep up?

Now that IFS has revenue of about 4-4,5 B$/quarter (Q1 and Q2-2024), so about 16-18 B$/year and if IFS wants to invest some 30% in capital (assuming 50-60% gross margin) it can spend about 5-6 B$/year in capital. If they can double that by financial engineering (private capital + government subsidies) they have about 12 B$/year for capital spending.

TSMC spends now about 30-32 B$/year in capital (revenue some 20 B$/quarter in Q2-2024 with healthy 53% gross margin).

Hard to see how IFS will survive this race with TSMC in leading edge logic manufacturing.

Does this make sense?

Here a useful reference I found from the situation some 20 years ago in 2003 and Rock's law:
https://archive.ph/20130119214941/http://news.com.com/Semi+survival/2009-1001_3-981418.html

We can use the Capital Intensity Ratio (CIR, CapEx/Revenue) to analyze it. TSMC's CIR has been high for the past several years. This means TSMC must maintain a very high net profit margin and strong revenue growth to sustain such heavy CapEx. I expect TSMC's CIR for 2024 should drop to around 40% or lower.

TSMC Capital Intensity Ratio (CIR)

 

[Daniel Nenni]

What about the EU, should they provide Intel a separate bowl too? But perhaps Intel will find out that before approaching the EU-bowl it got terribly sick from all this extra food and needs to go on a long-needed diet to become healthy again?
https://bits-chips.nl/article/report-intels-magdeburg-fab-on-the-chopping-board/

Intel already has fabs in the EU. Personally I think Intel needs to focus on expanding manufacturing in the US. Clearly Intel bit off more than they can chew so smaller bites and chew thoroughly before swallowing please.

 

[jim_22222]

Arizona has been courting TSMC since 2013, so after dating for many years it seems they got engaged in 2020 and have started to produce some offspring now in 2024!
https://finance.yahoo.com/news/taiwan-arizona-ties-history-behind-093000132.html?fr=sycsrp_catchall

still, compared with the tw side, US might just been too optimistic and lack of consideration.

TSMC's investment in a new fab in Arizona could have several implications. Firstly, there's the potential for a talent drain from Intel's existing facilities in the state, as skilled workers may be lured by new opportunities at TSMC. This could intensify competition for semiconductor professionals in the region and potentially disrupt Intel's operations.
Secondly, Taiwan's policy regarding AMD's investment in the country, which stipulates sourcing 50% of talent from abroad, indicates a proactive approach to addressing potential workforce shortages and diversifying talent. This could serve as a model for TSMC's Arizona venture, helping to mitigate local talent scarcity.
Lastly, from the U.S. perspective, there may be concerns about underestimating the complexity of ramping up a highly skilled semiconductor workforce quickly enough to meet the demands of these new fabs. The U.S. will need to consider long-term strategies for developing domestic talent, including education and training programs, to support the growing semiconductor industry and ensure the success of investments like TSMC's.

the prospect of a talent drain from Intel's fabs, particularly with TSMC's new Arizona facility entering the scene, raises several concerns. Firstly, Intel may find itself in a position where it needs to substitute departing experienced workers with less qualified or inexperienced talent. This could lead to a knowledge gap within Intel's workforce and potentially impact productivity and innovation.
Secondly, the impact of such a talent drain could be more subtle yet profound. Experienced employees carry with them not just skills, but also institutional knowledge and a nuanced understanding of the company's processes and culture. Their departure can affect team dynamics and the transfer of tacit knowledge that is critical for complex manufacturing operations.
Lastly, there are long-term implications for quality and safety. Semiconductor manufacturing is a highly sophisticated process where expertise is crucial to maintaining the high standards required for production. A less experienced workforce could lead to a higher incidence of errors, affecting the reliability and performance of the chips produced. This, in turn, could have repercussions for the safety and functionality of the end products that rely on these semiconductors. Intel will need to address these challenges strategically to maintain its reputation for quality and safeguard its long-term operational integrity.

 

[osnium]

still, compared with the tw side, US might just been too optimistic and lack of consideration.

TSMC's investment in a new fab in Arizona could have several implications. Firstly, there's the potential for a talent drain from Intel's existing facilities in the state, as skilled workers may be lured by new opportunities at TSMC. This could intensify competition for semiconductor professionals in the region and potentially disrupt Intel's operations.
Secondly, Taiwan's policy regarding AMD's investment in the country, which stipulates sourcing 50% of talent from abroad, indicates a proactive approach to addressing potential workforce shortages and diversifying talent. This could serve as a model for TSMC's Arizona venture, helping to mitigate local talent scarcity.
Lastly, from the U.S. perspective, there may be concerns about underestimating the complexity of ramping up a highly skilled semiconductor workforce quickly enough to meet the demands of these new fabs. The U.S. will need to consider long-term strategies for developing domestic talent, including education and training programs, to support the growing semiconductor industry and ensure the success of investments like TSMC's.

the prospect of a talent drain from Intel's fabs, particularly with TSMC's new Arizona facility entering the scene, raises several concerns. Firstly, Intel may find itself in a position where it needs to substitute departing experienced workers with less qualified or inexperienced talent. This could lead to a knowledge gap within Intel's workforce and potentially impact productivity and innovation.
Secondly, the impact of such a talent drain could be more subtle yet profound. Experienced employees carry with them not just skills, but also institutional knowledge and a nuanced understanding of the company's processes and culture. Their departure can affect team dynamics and the transfer of tacit knowledge that is critical for complex manufacturing operations.
Lastly, there are long-term implications for quality and safety. Semiconductor manufacturing is a highly sophisticated process where expertise is crucial to maintaining the high standards required for production. A less experienced workforce could lead to a higher incidence of errors, affecting the reliability and performance of the chips produced. This, in turn, could have repercussions for the safety and functionality of the end products that rely on these semiconductors. Intel will need to address these challenges strategically to maintain its reputation for quality and safeguard its long-term operational integrity.

I suspect the layoffs that Intel has announced will accelerate the decline in the quality of Intel’s workforce.

 

[jim_22222]

I suspect the layoffs that Intel has announced will accelerate the decline in the quality of Intel’s workforce.

good companies send out low performers.
bad ones do the opposite.
which category you think intc belongs to?

 

[osnium]

good companies send out low performers.
bad ones do the opposite.
which category you think intc belongs to?

Intel severance: Chipmaker sets terms for buyouts, early retirement and layoffs

The company wants to cut 15,000 jobs and save $10 billion.

www.oregonlive.com

According to the Oregonian, they are starting with voluntary severance and retirement packages. Generous retirement packages will incentivize employees with key institutional knowledge to leave. Folks who take voluntary severance are also most likely more confident about landing jobs later on. It seems to me Intel is going to lose a lot of good talent. If you are in AZ and are confident that TSMC will hire you, why wouldn’t you take the package and then jump ship?

 

[hist78]

still, compared with the tw side, US might just been too optimistic and lack of consideration.

TSMC's investment in a new fab in Arizona could have several implications. Firstly, there's the potential for a talent drain from Intel's existing facilities in the state, as skilled workers may be lured by new opportunities at TSMC. This could intensify competition for semiconductor professionals in the region and potentially disrupt Intel's operations.
Secondly, Taiwan's policy regarding AMD's investment in the country, which stipulates sourcing 50% of talent from abroad, indicates a proactive approach to addressing potential workforce shortages and diversifying talent. This could serve as a model for TSMC's Arizona venture, helping to mitigate local talent scarcity.
Lastly, from the U.S. perspective, there may be concerns about underestimating the complexity of ramping up a highly skilled semiconductor workforce quickly enough to meet the demands of these new fabs. The U.S. will need to consider long-term strategies for developing domestic talent, including education and training programs, to support the growing semiconductor industry and ensure the success of investments like TSMC's.

the prospect of a talent drain from Intel's fabs, particularly with TSMC's new Arizona facility entering the scene, raises several concerns. Firstly, Intel may find itself in a position where it needs to substitute departing experienced workers with less qualified or inexperienced talent. This could lead to a knowledge gap within Intel's workforce and potentially impact productivity and innovation.
Secondly, the impact of such a talent drain could be more subtle yet profound. Experienced employees carry with them not just skills, but also institutional knowledge and a nuanced understanding of the company's processes and culture. Their departure can affect team dynamics and the transfer of tacit knowledge that is critical for complex manufacturing operations.
Lastly, there are long-term implications for quality and safety. Semiconductor manufacturing is a highly sophisticated process where expertise is crucial to maintaining the high standards required for production. A less experienced workforce could lead to a higher incidence of errors, affecting the reliability and performance of the chips produced. This, in turn, could have repercussions for the safety and functionality of the end products that rely on these semiconductors. Intel will need to address these challenges strategically to maintain its reputation for quality and safeguard its long-term operational integrity.

Among the 2,000 TSMC employees in Arizona, half of them are from Taiwan. I believe this is the original plan agreed between US government and TSMC. It addressed several issues you mentioned.

 

[jim_22222]

Among the 2,000 TSMC employees in Arizona, half of them are from Taiwan. I believe this is the original plan agreed between US government and TSMC. It addressed several issues you mentioned.

fake news. it won't delayed that much if this is true.

As TSMC steps up its global expansion, the company said production at its first plant in Arizona will be delayed until 2025 due to a shortage of specialist workers.

“While we are working to improve the situation, including sending experienced technicians from Taiwan to train the local skilled workers for a short period of time, we expect the production schedule of N4 process technology to be pushed out to 2025,” TSMC chairman Mark Liu said Thursday.

https://edition.cnn.com/2023/07/21/tech/tsmc-arizona-production-delay/index.html

 

[DepthFirst]

still, compared with the tw side, US might just been too optimistic and lack of consideration.

TSMC's investment in a new fab in Arizona could have several implications. Firstly, there's the potential for a talent drain from Intel's existing facilities in the state, as skilled workers may be lured by new opportunities at TSMC. This could intensify competition for semiconductor professionals in the region and potentially disrupt Intel's operations.
Secondly, Taiwan's policy regarding AMD's investment in the country, which stipulates sourcing 50% of talent from abroad, indicates a proactive approach to addressing potential workforce shortages and diversifying talent. This could serve as a model for TSMC's Arizona venture, helping to mitigate local talent scarcity.
Lastly, from the U.S. perspective, there may be concerns about underestimating the complexity of ramping up a highly skilled semiconductor workforce quickly enough to meet the demands of these new fabs. The U.S. will need to consider long-term strategies for developing domestic talent, including education and training programs, to support the growing semiconductor industry and ensure the success of investments like TSMC's.

the prospect of a talent drain from Intel's fabs, particularly with TSMC's new Arizona facility entering the scene, raises several concerns. Firstly, Intel may find itself in a position where it needs to substitute departing experienced workers with less qualified or inexperienced talent. This could lead to a knowledge gap within Intel's workforce and potentially impact productivity and innovation.
Secondly, the impact of such a talent drain could be more subtle yet profound. Experienced employees carry with them not just skills, but also institutional knowledge and a nuanced understanding of the company's processes and culture. Their departure can affect team dynamics and the transfer of tacit knowledge that is critical for complex manufacturing operations.
Lastly, there are long-term implications for quality and safety. Semiconductor manufacturing is a highly sophisticated process where expertise is crucial to maintaining the high standards required for production. A less experienced workforce could lead to a higher incidence of errors, affecting the reliability and performance of the chips produced. This, in turn, could have repercussions for the safety and functionality of the end products that rely on these semiconductors. Intel will need to address these challenges strategically to maintain its reputation for quality and safeguard its long-term operational integrity.

There have been a lot of grumbling online about TSMC work culture among American employees. Unless TSMC is willing to open their purse to offer very attractive comps it seems more likely that any loss of employees will go both ways at the minimum. Which considering who is currently ahead seems more advantageous to Intel.

 

[jim_22222]

There have been a lot of grumbling online about TSMC work culture among American employees. Unless TSMC is willing to open their purse to offer very attractive comps it seems more likely that any loss of employees will go both ways at the minimum. Which considering who is currently ahead seems more advantageous to Intel.

giving the prospect of getting a lay off package any time in intc, they might have to take the other direction. this is real life.

 

[BruceA]

A monopoly in general is never good. When anything ends up with a single supplier the motivation to relentlessly motivate or not to maximize profits becomes less.

Imagine a world with only Boeing or only TSMC. Of course they have the mantra of not competing and when their customers succeed they succeed may apply at the beginning but in the end any company services its stockholders first, employees second and customers third in that order in a monopoly.

In a high demand single source supplier of Samsung and Intel was to disappears, imagine a Boeing without Airbus what motivation is there to innovate?

Add in that semiconductors drive national compute leadership even harder to fathom that the world longterm can tolerate all leadership silicon coming from Taiwan.

If this was pre COVID and Pooh Bear hadn’t gotten his latest term, think no chips act and end of intel fabs would have been sold off by Bob to TSMC, now no way.

It’s all in and Intel will either limp forward with a successful I18A and larger I14A. Let’s be honest this with Panther and Clearwater all tied to I18A, all of Intel will go with IFS success or completely end. AMD, Nvidia, Apple, Qualcomm and others are in a far different place as is TSMC then the 10nm days.

The world needs Intel to be successful as a not TSMC as Samsung most certainly isn’t an alternative currently

 

[BruceA]

There have been a lot of grumbling online about TSMC work culture among American employees. Unless TSMC is willing to open their purse to offer very attractive comps it seems more likely that any loss of employees will go both ways at the minimum. Which considering who is currently ahead seems more advantageous to Intel.

The issue in Az is not money but how the non assignees are treated and expected to conform and the total failure on both construction and fab side to train, and appreciate the local work norms. Amazing given TSMC success as a customer oriented manufacturing, but arrogance is rampant too to bottoms across the site and in HQ.

With the new packages announced to the assignees it will be interesting to see if they keep all the assignees that came as well as get the needed assignees to do P2.

In Japan and Germany given the node and partnerships they likely won’t have such issues. German construction will be interesting to watch.

 

[Daniel Nenni]

I suspect the layoffs that Intel has announced will accelerate the decline in the quality of Intel’s workforce.

I doubt AZ will be affected by the layoffs, except for the people they don't want anymore. Fab work is pretty static but I do agree TSMC in AZ is really stirring things up. Take a look at the current job openings:

AZ - TSMC Jobs

Find at TSMC

ro.careers.tsmc.com

TSMC AZ jobs get the most views on the SemiWiki job board by a long shot.

Intel severance: Chipmaker sets terms for buyouts, early retirement and layoffs

The company wants to cut 15,000 jobs and save $10 billion.

www.oregonlive.com

According to the Oregonian, they are starting with voluntary severance and retirement packages. Generous retirement packages will incentivize employees with key institutional knowledge to leave. Folks who take voluntary severance are also most likely more confident about landing jobs later on. It seems to me Intel is going to lose a lot of good talent. If you are in AZ and are confident that TSMC will hire you, why wouldn’t you take the package and then jump ship?

Generous retirement programs will get rid of people who are unhappy or planning on retiring anyway. Great strategy.

 

[user nl]

Thanks to all for their continuing contributions/views/ideas in this thread!

For those interested in some of the (recent) scientific literature on geopolitical/economical/views on the strategic aspects of this topic around (foundry) chip manufacturing, see here some interesting (open access) papers providing deeper understanding of important driving issues in the decades long complex and global dynamics in (semiconductor) technology:

1) Chris Miller (Tufts University) in Nature Reviews (Jan 2024): Global chip war for strategic semiconductors: https://rdcu.be/dTXaM

2) Cecilia Rikap (University College London) in Industrial and Corporate Change (Oct 2024): Intellectual monopolies as a new pattern of innovation and technological regime: https://academic.oup.com/icc/article/33/5/1037/7462137

"7. Final remarks All in all, by introducing the IM pattern of innovation and technological regime, in which benefits accrued from innovation developed by corporate innovation systems are disproportionately harvested by intellectual monopolies, this paper has contributed to our understanding of patterns of innovation and innovation systems. The focus on systems was crucial for identifying and studying the interplay between different heterogeneous firms co-producing innovation. As evidenced empirically, this pattern of innovation is characterized by a stable core and turbulent and dynamic peripheries among which many innovating companies subordinate to intellectual monopolies, as in the cases of AWS partners and innovating firms that are heavily reliant on funding -and technology- provided by an intellectual monopoly. The second general contribution of this paper was its historical approach to explain the IM pattern of innovation. The confluence of political, institutional, technological, and economic transformations fostered the perpetuation of intellectual monopolies but also encouraged knowledge modularization and the spread of start-ups in certain sectors—pharmaceuticals and information technologies—leading to the emergence of a distinct pattern. Among others, the role of the US hidden industrial policy as well as China’s protectionism, the weakening of antitrust, a more stringent and extensive IPRs regime, new technologies—that accelerate but also make R&D more expensive—and globalization have contributed to perpetuating intellectual monopolies........"



Good weekend reading to all!

 

[Fred Chen]

8) it appears to me that TSMC is using the same open-innovation systems approach in large-scale (now EUV-based) leading edge logic chip manufacturing. Since 2008 TSMC has developed this open-innovation approach: see https://www.tsmc.com/english/dedicatedFoundry/oip
At the same time TSMC has developed some hugely beneficial internal proprietary EUV tooling expertise (see eg https://pcge.eu/2024/05/tsmcs-euv-triumph-amplifying-tools-boosting-wafers-and-unmatched-pellicles/)

As you may know, I am particularly picky about accuracy in portraying EUV technology; bringing up TSMC's own developments reminded me of finding this disturbing patent https://patents.google.com/patent/US12025922B2/. The issue being addressed, hydrogen blistering, should be well-known but is not the only concern here; more disturbing is the patent's description of how to deal with it, which indicates that a correct fundamental understanding of the EUV system and its workings is apparently not generally present at TSMC.

 

[fansink]

As you may know, I am particularly picky about accuracy in portraying EUV technology; bringing up TSMC's own developments reminded me of finding this disturbing patent https://patents.google.com/patent/US12025922B2/. The issue being addressed, hydrogen blistering, should be well-known but is not the only concern here; more disturbing is the patent's description of how to deal with it, which indicates that a correct fundamental understanding of the EUV system and its workings is apparently not generally present at TSMC.

What specifically is disturbing in this patent, and why?