Building A Chipmaking Fab In The US Costs Twice As Much, Takes Twice As Long As In Taiwan

[fansink]

The semiconductor industry is expanding rapidly as countries race to build new fabs. While it takes around 19 months to build a fab in Taiwan, it takes a whopping 38 months to build a fab in the U.S. due to the extensive time it takes to get a permit and because fabs are not constructed 24/7, according to Exyte, a leading engineering, construction, and design company that specializes in high-tech facilities like chip production plants, reports Semiconductor Digest.

Taiwan completes fabs in around 19 months, followed by Singapore and Malaysia at 23 months. European projects take 34 months, while the U.S. is the slowest at 38 months. A key reason for this is Taiwan's streamlined permit process and round-the-clock construction, whereas the U.S. and Europe face delays in approvals and do not construct 24/7. The U.S. has enacted a law that exempts certain U.S. fabs from federal environmental assessments, but that is obviously not enough to be on par with Taiwan.

Costs also differ widely. Constructing a plant in the U.S. is about twice as expensive as in Taiwan, despite similar equipment costs, according to Exyte. This discrepancy arises from higher labor costs, extensive regulatory requirements, and inefficiencies in supply chains. Also, Taiwanese workforce is highly experienced, so Taiwanese builders require fewer detailed blueprints because they are familiar with every step of the process, which speeds up completion of fab projects, according to Herbert Blaschitz, an executive at Exyte.

To compete efficiently with Taiwan — which has well-integrated supply chain, experienced workforce, and efficient regulatory processes — the U.S. and Europe must streamline permitting, optimize construction techniques, and adopt advanced planning tools like digital twins. Blaschitz suggests adopting 'virtual commissioning,' where a digital model of the plant is created before physical construction begins. This allows potential problems to be identified early, reducing costs and environmental impact while improving speed and efficiency.

Modern semiconductor production facilities are huge, both in terms of dimensions and investments. A leading-edge fab — such as those operated by Intel, Samsung Foundry, or TSMC — requires investment exceeding $20 billion, with $4-6 billion allocated just for the structure itself, according to Blaschitz, who highlighted Taiwan's advantages at the SEMI Industry Strategy Symposium.

The construction process involves 30-40 million workhours, using 83,000 tons of steel, 5,600 miles of electrical wiring, and 785,000 cubic yards of concrete, according to Blaschitz. A typical fab may include a 40,000-square-meter cleanroom with 2,000 production tools used for lithography, deposition, etching, cleaning, and other operations. Each of the tools requires around 50 individual utility and process connections.

US fab construction costs twice as much, takes twice as long as Taiwan

U.S. and Europe need to adopt new construction techniques to be competitive.

www.tomshardware.com

 

[MKWVentures]

I will get some recent input from experts on this. But my take is that it can take twice as long but a lot of that depends on the company doing it. Supply chain and labor availability in asia is much better. I would be shocked if it actually cost 2x as much ... my spreadsheet is nowhere near that big a difference. Tools are still the largest part of capital expense. All this assumes the same subsidies. In the past, other countries gave huge subsidies and the US gave few. that has changed.

What are your thoughts @nghanayem or @Scotten Jones ?

Reminder: NEVER use the publicly stated costs to build a fab that companies publish. Those are meant to get the response they want from government. the actual spending is accurately tracked by a number of sources if you talk to the correct people.

 

[Daniel Nenni]

Does anyone know the numbers for The TSMC Japan fab? I think that was faster than even Taiwan. I asked ChatGPT:

Does Building A Chipmaking Fab In The US Costs Twice As Much, Takes Twice As Long As In Taiwan?

ChatGPT said:​

Yes, building a chipmaking fab in the U.S. generally costs significantly more and takes longer compared to Taiwan. The common estimate is that it costs about 40% to 100% more (or roughly 1.5x to 2x the cost) and can take up to twice as long to complete. Several factors contribute to this disparity:

1. Higher Construction and Labor Costs​

• The cost of materials, labor, and compliance in the U.S. is significantly higher than in Taiwan.
• Skilled labor shortages and union wages further increase costs.
• Construction efficiency is higher in Taiwan due to experience and established supply chains.

2. Regulatory and Permitting Delays​

• The U.S. has stricter environmental regulations and longer approval processes.
• In Taiwan, government support and streamlined processes allow fabs to be built in about 2 years, while in the U.S., it can take 3-4 years.

3. Supply Chain and Logistics Challenges​

• Taiwan has a well-established semiconductor supply chain, making it easier to source equipment and materials.
• In the U.S., many critical components and expertise are imported, adding costs and delays.

4. Government Incentives and Industry Expertise​

• Taiwan’s government provides generous subsidies and incentives to companies like TSMC.
• The U.S. CHIPS Act provides funding, but bureaucracy slows down distribution and effectiveness.
• Taiwan has a deep talent pool with decades of semiconductor specialization, while the U.S. faces a skills gap.

5. Cultural and Operational Differences​

• Companies like TSMC have highly optimized, lean project execution in Taiwan, while in the U.S., unfamiliarity with local business and work culture (e.g., Arizona's workforce for TSMC) creates inefficiencies.

Example: TSMC's Arizona Fab​

• TSMC's Arizona fab (N5/N4 process) was initially planned for 2024 but is delayed to 2025 due to labor and supply chain challenges.
• The cost is reportedly around $40 billion, significantly higher than comparable fabs in Taiwan.

Would you like a deeper dive into any specific cost factor?

4o

You said:​

Yes

ChatGPT said:​

Sure! Let’s break down the key cost factors in more detail:

---

1. Construction and Labor Costs​

Higher Material and Equipment Costs​

• In the U.S., building materials and equipment often cost more due to supply chain limitations and imports.
• Advanced semiconductor fab tools (e.g., EUV lithography machines) have to be imported from ASML (Netherlands) and other global suppliers.
• Specialized cleanroom materials and chemicals often cost more in the U.S. due to logistics and compliance costs.

Skilled Labor Shortages and Wages​

• In Taiwan, TSMC benefits from a highly skilled workforce with decades of experience.
• In the U.S., there is a shortage of semiconductor fab construction workers and engineers.
• U.S. labor unions demand higher wages and more benefits compared to Taiwan, increasing labor costs.
• TSMC has struggled to find enough skilled U.S. workers, leading to delays.

Example:

• Taiwan's skilled fab construction workers reportedly earn 50-60% less than their U.S. counterparts.
• TSMC had to bring in workers from Taiwan to complete its Arizona fab due to local labor inefficiencies.

---

2. Regulatory and Permitting Delays​

Stricter Environmental Regulations​

• The U.S. has more complex and time-consuming environmental and zoning regulations compared to Taiwan.
• Fabs require extensive water and energy usage, triggering lengthy environmental reviews in the U.S.
• Taiwan’s government provides fast-track approvals for semiconductor fabs due to their strategic importance.

Permitting and Bureaucratic Hurdles​

• Obtaining permits in the U.S. can take years, while Taiwan streamlines approvals within months.
• U.S. fabs must comply with federal, state, and local regulations, increasing paperwork and delays.
• The CHIPS Act aims to help, but funds have been slow to distribute due to bureaucratic processes.

Example:

• TSMC’s Arizona fab has faced multiple delays due to labor and permitting challenges, now pushing production to 2025.
• In Taiwan, fabs are often built and operational within 2 years due to strong government support.

---

3. Supply Chain and Logistics Challenges​

Limited Domestic Supply Chain​

• Taiwan has an established semiconductor ecosystem, making it easier and cheaper to source materials.
• The U.S. lacks a comparable supply chain, forcing fabs to import materials, tools, and components.
• This results in higher costs and longer lead times for critical components.

Equipment Availability and Import Costs​

• Semiconductor fab tools (e.g., lithography machines) are mostly produced outside the U.S., leading to delays and tariffs.
• Taiwan benefits from local suppliers who work closely and efficiently with TSMC, reducing costs.

Example:

• ASML’s EUV machines (needed for advanced chipmaking) cost over $150 million each and require complex shipping, installation, and training.
• Delays in importing and installing these machines have slowed progress in U.S. fabs.

---

4. Government Incentives and Industry Expertise​

Stronger Government Support in Taiwan​

• Taiwan provides generous subsidies, tax breaks, and land grants to companies like TSMC.
• The government actively coordinates semiconductor industry growth, reducing costs and delays.

U.S. CHIPS Act Challenges​

• While the CHIPS Act provides $52 billion in funding, the process has been slow and bureaucratic.
• TSMC has criticized the U.S. incentive structure, arguing that it is not competitive with Taiwan’s support.

Example:

• Intel and TSMC are still negotiating CHIPS Act subsidies, delaying investment certainty.
• Taiwan subsidizes electricity and water costs, while U.S. fabs must pay higher utility rates.

---

5. Cultural and Operational Differences​

Different Work Cultures and Productivity Levels​

• In Taiwan, fab construction is highly optimized, with workers trained for maximum efficiency.
• In the U.S., cultural and labor differences create friction and slower execution.
• Reports suggest that construction efficiency is lower in the U.S. due to stricter work-hour limits and safety regulations.

Example:

• TSMC’s Arizona fab faced worker disputes and safety concerns, delaying progress.
• The company had to import Taiwanese engineers to speed up production.

Final Cost Comparison Estimate​

 

[Daniel Nenni]

Here’s a detailed cost breakdown comparing the expenses of building an advanced chip fab (e.g., 3nm or 5nm) in Taiwan vs. the U.S.

---

Estimated Total Cost​

• Taiwan: $10 - $12 billion per fab
• U.S.: $15 - $20+ billion per fab

The U.S. fab costs about 40-100% more than a similar facility in Taiwan.

---

---

Key Drivers of Higher U.S. Costs​

1. Construction Costs
   • U.S. construction labor is more expensive (~$100-$150/hr vs. $30-$50/hr in Taiwan).
   • Unionized labor in the U.S. drives costs up.
   • Taiwan has faster, more efficient construction teams.
2. Equipment & Materials
   • The same ASML EUV lithography machine costs more in the U.S. due to shipping, import taxes, and regulatory fees.
   • Materials for chip production (e.g., ultra-pure chemicals, wafers, gases) are often imported into the U.S., making them 20-30% more expensive.
3. Utilities & Water Usage
   • Chip fabs consume millions of gallons of ultra-pure water daily.
   • Taiwan subsidizes water and electricity, whereas the U.S. does not.
   • Electricity in Taiwan costs about $0.10/kWh, while in the U.S., it can be $0.12 - $0.15/kWh or more.
4. Permitting & Regulatory Delays
   • Taiwan fast-tracks semiconductor approvals, while U.S. permitting can add 6-12 months of delays.
   • U.S. fabs require more environmental reviews and compliance (e.g., water usage, emissions).
5. Ongoing Operating Costs
   • Engineer salaries:
     • Taiwan: $50K - $100K per year
     • U.S.: $100K - $200K per year
   • Maintenance & operational costs are 30-50% higher in the U.S.

---

Example: TSMC Arizona vs. Taiwan Fab Costs​

• TSMC Arizona (5nm, 4nm, planned 3nm fab) → $40 billion+ for two fabs
• TSMC Taiwan (3nm Fab 18) → $15 billion
• TSMC Fab Build Time
  • Taiwan: ~2 years
  • U.S.: 3-4 years (or more, due to delays)

---

Conclusion: Why U.S. Fabs Are More Expensive​

Labor costs are 2-3x higher in the U.S.
Construction costs are significantly higher due to inefficiencies.
Regulations slow down approvals and increase compliance costs.
Utilities & supply chain constraints raise operational expenses.

Would you like me to break down potential ways to reduce U.S. fab costs?

 

[count]

I will get some recent input from experts on this. But my take is that it can take twice as long but a lot of that depends on the company doing it. Supply chain and labor availability in asia is much better. I would be shocked if it actually cost 2x as much ... my spreadsheet is nowhere near that big a difference. Tools are still the largest part of capital expense. All this assumes the same subsidies. In the past, other countries gave huge subsidies and the US gave few. that has changed.

What are your thoughts @nghanayem or @Scotten Jones ?

Reminder: NEVER use the publicly stated costs to build a fab that companies publish. Those are meant to get the response they want from government. the actual spending is accurately tracked by a number of sources if you talk to the correct people.

A lot of people will look at a spreadsheet of costs and determine that equipment is the highest capex item... however they are missing a key point.

It's equipment depreciation that's the largest expense.

Two companies can by the same piece of equipment with a useful life of 10 years. The equipment has a nameplace capacity of 1000 units a year and cost $1 million dollars.

Company 1 achieves OEE of 50% in year 1, and 95% in year two onwards
Company 2 achieves OEE of 30% in year 1, 60% in year 2, and 85% onwards

Assume straight line depreciation.

On average, over the 10 year life of the equipment, company 1 has an equipment cost of ~$115/unit of production, company 2 costs are $144/unit

Same equipment, same cost to buy the equipment, very different unit costs at the end of the day. Only difference is OEE and how long it takes to ramp to production.

Very few people, even in manufacturing, really understand this point. When your biggest cost is equipment you need to squeeze every bit of production out of every piece of equipment, which means getting the equipment in service as quickly as possible, running 24/7, minimizing downtime, and maximizing yield. To do this you need people who are absolutely dedicated to solving problems quickly (at 2AM if needed), have a mindset for quality and continuous improvement, and understand the process and the equipment inside and out. This is what I mean when I often talk about "manufacturing culture", that is missing in the US.

 

[MKWVentures]

A lot of people will look at a spreadsheet of costs and determine that equipment is the highest capex item... however they are missing a key point.

It's equipment depreciation that's the largest expense.

Two companies can by the same piece of equipment with a useful life of 10 years. The equipment has a nameplace capacity of 1000 units a year and cost $1 million dollars.

Company 1 achieves OEE of 50% in year 1, and 95% in year two onwards
Company 2 achieves OEE of 30% in year 1, 60% in year 2, and 85% onwards

Assume straight line depreciation.

On average, over the 10 year life of the equipment, company 1 has an equipment cost of ~$115/unit of production, company 2 costs are $144/unit

Same equipment, same cost to buy the equipment, very different unit costs at the end of the day. Only difference is OEE and how long it takes to ramp to production.

Very few people, even in manufacturing, really understand this point. When your biggest cost is equipment you need to squeeze every bit of production out of every piece of equipment, which means getting the equipment in service as quickly as possible, running 24/7, minimizing downtime, and maximizing yield. To do this you need people who are absolutely dedicated to solving problems quickly (at 2AM if needed), have a mindset for quality and continuous improvement, and understand the process and the equipment inside and out. This is what I mean when I often talk about "manufacturing culture", that is missing in the US.

100% agree on the efficiency and impact. you are preaching to the converted. My experience is that this is biggest wafer cost and unit cost differentiator

they were talking about capex cost to build the factory I believe.... correct me if I am wrong.

 

[MKWVentures]

Would you like me to break down potential ways to reduce U.S. fab costs?

Chat GPT? SMH.

we should see how much TSMC actually spent and how much other companies actually spend vs the claims. I do not believe TSMC will spend 40B in Arizona by 2026. I might be wrong.

 

[count]

Chat GPT? SMH.

we should see how much TSMC actually spent and how much other companies actually spend vs the claims. I do not believe TSMC will spend 40B in Arizona by 2026. I might be wrong.

You aren’t wrong, but the underlying point of the article is competitiveness, and capex only tells part of the story when it comes to competitiveness.

 

[Artificer60]

A lot of people will look at a spreadsheet of costs and determine that equipment is the highest capex item... however they are missing a key point.

It's equipment depreciation that's the largest expense.

Two companies can by the same piece of equipment with a useful life of 10 years. The equipment has a nameplace capacity of 1000 units a year and cost $1 million dollars.

Company 1 achieves OEE of 50% in year 1, and 95% in year two onwards
Company 2 achieves OEE of 30% in year 1, 60% in year 2, and 85% onwards

Not disagreeing with your point, but OEE is one of those metrics that looks great on paper, but when you are looking at a wafer fab with multiple passes through re-entrant tools it can be a real challenge to evaluate that quality metric unless you are willing to pay to have a lot of inline metrology. Not every fab can afford that expense.

To do this you need people who are absolutely dedicated to solving problems quickly (at 2AM if needed), have a mindset for quality and continuous improvement, and understand the process and the equipment inside and out. This is what I mean when I often talk about "manufacturing culture", that is missing in the US.

As to that 2AM comment, in my 15+ years as a process engineer I never noticed any hesitancy for people to page me at all hours of the day or night when tools needed attention. And I did what was needed to get the tools restored to a state to produce high quality output as quickly as I would have if it had happened at 10:00 am on a weekday. I have worked with and continue to work with a lot of people in the US who embody than "manufacturing culture". Sorry if this comes across as a bit snippy, but I take offense at the insinuation that I and most of my hard-working colleagues lack skill and dedication.

It never ceases to amaze me that people forget that Intel held a technology lead over the whole industry for a very long time. That didn't happen due to a poor "manufacturing culture" and they didn't lose it because their engineers and technicians suddenly became stupid and lazy overnight. They lost it because management at the highest level of the company decided to drive the bus off a cliff.

 

[osnium]

Not disagreeing with your point, but OEE is one of those metrics that looks great on paper, but when you are looking at a wafer fab with multiple passes through re-entrant tools it can be a real challenge to evaluate that quality metric unless you are willing to pay to have a lot of inline metrology. Not every fab can afford that expense.



As to that 2AM comment, in my 15+ years as a process engineer I never noticed any hesitancy for people to page me at all hours of the day or night when tools needed attention. And I did what was needed to get the tools restored to a state to produce high quality output as quickly as I would have if it had happened at 10:00 am on a weekday. I have worked with and continue to work with a lot of people in the US who embody than "manufacturing culture". Sorry if this comes across as a bit snippy, but I take offense at the insinuation that I and most of my hard-working colleagues lack skill and dedication.

It never ceases to amaze me that people forget that Intel held a technology lead over the whole industry for a very long time. That didn't happen due to a poor "manufacturing culture: and they didn't lose it because their engineers and technicians suddenly became stupid and lazy overnight. They lost it because management at the highest level of the company decided to drive the bus off a cliff.

Does the hard core Intel culture that existed pre 2014 still exist today? I’d imagine many of the qualified and experienced people were let go as part of the 2016 and 2024 layoffs in the name of cost cutting.

 

[Artificer60]

Does the hard core Intel culture that existed pre 2014 still exist today? I’d imagine many of the qualified and experienced people were let go as part of the 2016 and 2024 layoffs in the name of cost cutting.

If they are able to successfully launch 18A and Panther Lake this year and they live up to expectations, I would say yes. You don't close that kind of technology gap in 5 years without a culture that values skill and dedication.

Though it will be interesting to see the longer term impact of the last round of layoffs given that the use of voluntary retirement packages to reduce headcount encouraged the most experienced people to leave.

 

[blueone]

Does the hard core Intel culture that existed pre 2014 still exist today? I’d imagine many of the qualified and experienced people were let go as part of the 2016 and 2024 layoffs in the name of cost cutting.

What a silly post.

 

[osnium]

What a silly post.

It was a question inspired by articles in the Oregonian about the number of Intel Fellows who were laid off, LinkedIn posts about senior Intel engineers taking the severance packages and the general downward trend of the fab’s outcomes. Also, if you have a PhD in EE and have the option to join Intel right now to do process development vs Apple to work on displays vs TI to do process development on new material systems like Gallium Nitride, would you choose to join Intel?

Intel’s Oregon layoffs by the numbers: VPs, fellows and engineering all took hits

Executives, scientists, engineers and administrative assistants all took a hit.

www.oregonlive.com

 

[blueone]

It was a question inspired by articles in the Oregonian about the number of Intel Fellows who were laid off, LinkedIn posts about senior Intel engineers taking the severance packages and the general downward trend of the fab’s outcomes. Also, if you have a PhD in EE and have the option to join Intel right now to do process development vs Apple to work on displays vs TI to do process development on new material systems like Gallium Nitride, would you choose to join Intel?

Intel’s Oregon layoffs by the numbers: VPs, fellows and engineering all took hits

Executives, scientists, engineers and administrative assistants all took a hit.

www.oregonlive.com

Fellows were not laid off in 2014 and 2016, unfortunately in a few cases I can think of. If anything, Intel still has too many Fellows and Senior Fellows. Intel still has too many employees in general - they still have about 109,000. As for VPs, I think they should have the VPs count off by threes, and lay off all of the threes.

And getting information about Intel, any other semi company, or the semi industry from The Oregonian is not a good idea. Their reporter, Mike Rogoway, is mostly about rumors and discussions with unnamed sources, and has never worked in the semi industry.

 

[DanX]

Not disagreeing with your point, but OEE is one of those metrics that looks great on paper, but when you are looking at a wafer fab with multiple passes through re-entrant tools it can be a real challenge to evaluate that quality metric unless you are willing to pay to have a lot of inline metrology. Not every fab can afford that expense.



As to that 2AM comment, in my 15+ years as a process engineer I never noticed any hesitancy for people to page me at all hours of the day or night when tools needed attention. And I did what was needed to get the tools restored to a state to produce high quality output as quickly as I would have if it had happened at 10:00 am on a weekday. I have worked with and continue to work with a lot of people in the US who embody than "manufacturing culture". Sorry if this comes across as a bit snippy, but I take offense at the insinuation that I and most of my hard-working colleagues lack skill and dedication.

It never ceases to amaze me that people forget that Intel held a technology lead over the whole industry for a very long time. That didn't happen due to a poor "manufacturing culture" and they didn't lose it because their engineers and technicians suddenly became stupid and lazy overnight. They lost it because management at the highest level of the company decided to drive the bus off a cliff.

This has nothing to do with the engineers or the management. It has everything to do with the competitor.
With the competitor catching up, it is almost impossible for Intel to compete again without government help .
In 2014, no one was near Intel's technology.

 

[BruceA]

Fellows were not laid off in 2014 and 2016, unfortunately in a few cases I can think of. If anything, Intel still has too many Fellows and Senior Fellows. Intel still has too many employees in general - they still have about 109,000. As for VPs, I think they should have the VPs count off by threes, and lay off all of the threes.

And getting information about Intel, any other semi company, or the semi industry from The Oregonian is not a good idea. Their reporter, Mike Rogoway, is mostly about rumors and discussions with unnamed sources, and has never worked in the semi industry.

Mike is a shril…. lol

Intel is a lot different than the Yousseef, Bill, Sohail, Mike, Ann and now Navid…. Given how different the personalities, background, expertise and skills were, but I’d say Sunlin and Yousseef could stand toe to toe with any of the past or current TSMC leaders, the new ones well that is laughable and core to the problem. Same can be said about CEOS Paul, BK ( the worst) Bob, and lastly Pat

 

[BruceA]

This has nothing to do with the engineers or the management. It has everything to do with the competitor.
With the competitor catching up, it is almost impossible for Intel to compete again without government help .
In 2014, no one was near Intel's technology.

There is a lot more than competitor catching up. Under Sohail and BK the whole technology engine grinded and really grinded completely like a hamster spinning and going nowhere.

Not clear with Navid at the wheel it has the DNA at the top to drive, as to the talent, it’s there. As to if they got enough money and scale without a couple whales is the question. We will see if Panther launches of it does anything is possible with Trump

 

[BruceA]

A

A lot of people will look at a spreadsheet of costs and determine that equipment is the highest capex item... however they are missing a key point.

It's equipment depreciation that's the largest expense.

Two companies can by the same piece of equipment with a useful life of 10 years. The equipment has a nameplace capacity of 1000 units a year and cost $1 million dollars.

Company 1 achieves OEE of 50% in year 1, and 95% in year two onwards
Company 2 achieves OEE of 30% in year 1, 60% in year 2, and 85% onwards

Assume straight line depreciation.

On average, over the 10 year life of the equipment, company 1 has an equipment cost of ~$115/unit of production, company 2 costs are $144/unit

Same equipment, same cost to buy the equipment, very different unit costs at the end of the day. Only difference is OEE and how long it takes to ramp to production.

Very few people, even in manufacturing, really understand this point. When your biggest cost is equipment you need to squeeze every bit of production out of every piece of equipment, which means getting the equipment in service as quickly as possible, running 24/7, minimizing downtime, and maximizing yield. To do this you need people who are absolutely dedicated to solving problems quickly (at 2AM if needed), have a mindset for quality and continuous improvement, and understand the process and the equipment inside and out. This is what I mean when I often talk about "manufacturing culture", that is missing in the US.

And why without tariffs manufacturing in US for the likes of semiconductors, tech will be uncompetitive.

For a decade Intel had a unique situation with their crazy competitive moat, today that is all gone

 

[osnium]

There is a lot more than competitor catching up. Under Sohail and BK the whole technology engine grinded and really grinded completely like a hamster spinning and going nowhere.

Not clear with Navid at the wheel it has the DNA at the top to drive, as to the talent, it’s there. As to if they got enough money and scale without a couple whales is the question. We will see if Panther launches of it does anything is possible with Trump

Didn’t Sohail grow up in the old school hard core Intel culture that Youssef El Mansy and Sunlin Chou promoted? Why do you think he failed and killed the innovation engine?

 

[Artificer60]

After asking some long time Intel folks I know I can offer three reasons that Intel R&D died under Sohail.

First, he did not demand robust processes. He was quite happy to accept a process that was held together with human glue provided by constant attention of process engineers. All that mattered is that it worked, the cost in human capital was irrelevant. This resulted in tight process windows and little time to spend on process improvement.

Second, he became a bottleneck in the development process. The story I was told is that there were so many process changes going on at once that there was very little baseline material in the factory to compare the results of the experiments against. Sohail's solution was to limit the number of changes that were allowed to run in the factory at any one time. He did this by keeping a list and reviewing each suggestion personally. If an idea didn't make the cut it didn't get on the list and didn't get run. I don't care who you are no one can be an expert in all areas of the fab. It is just too complex.

Third and I think the most important. He created an environment of fear in which people were afraid to share bad news. Everything got sugarcoated and issues were hidden. That is the kiss of death in any development process.

I was also told that Bill Holt reigned in many of Sohail's worst impulses. When Bill left Sohail was left to his own devices.