Intel CEO Pat Gelsinger is "retiring"

P

PandaBear

Joined
Aug 5, 2002
Messages
25,881
Location
Silicon Valley
https://www.marketscreener.com/quot...nounces-Executive-and-Board-Changes-48503659/

Also "Intel's problems caused by insistence on the foundry business"

https://www.msn.com/en-us/money/oth...1&cvid=a5f27b3c29964478b2f4437a5fb2ede6&ei=31

I've talked to a few people I know within Intel. According to them it seems like nobody can fix Intel's manufacturing side, and nobody with their worth will leave a well run semiconductor company to become Intel's CEO, so Pat was already the most realistic choice.

What's shocking to me is that Lip-Bu Tan was the only board member with semiconductor experience and he resigned (likely due to frustration on Intel being bloated).

https://www.bing.com/search?pglt=41...xV0gEIMzgwN2owajGoAgCwAgA&FORM=ANNAB1&PC=U531

Realistically, Intel's manufacturing / foundry side cannot be fixed if even Samsung (#2 in the industry after TSMC) cannot compete. If they split the design and manufacturing side they will likely no longer be #1 in either, and be considerably smaller. The new fab building all over the world with government / political support is likely not going to work well either with all the string attached (in semiconductor you need to move very fast and likely even the best government will not be fast enough to catch up with all the policies outside of war time).
 
Another sad corporate story like the demise of Sears? Years ago all Intel needed to worry about was AMD. What happened?

While I don't have the numbers, the IBM chip plant in East Fishkill NY was one of the largest in the world but IBM did not sell any chips, all internally used in IBM computers. So it was hard to really know how many chips they actually produced. IBM Research in Yorktown was top notch also. Then came along CMOS and everything changed.
 
Dave Hess said:
Is the Arizona fab too optimistic and too high goals ?
Click to expand...
In Taiwan, TSMC has honed a highly complex manufacturing process staffed a network of skilled engineers and specialized suppliers. Remember, Semiconductor Manufacturing is the most high tech of the high tech. How do you fab a new chip if you don't have the machine tools? We are talking 4 and 3nm nodes.

TSMC management speak Taiwanese and Mandarin; their English is overall poor. There is a culture clash. I'm sure you have heard me speak of the Silicon Valley work ethic; it's all about winning. It's cutthroat. Phoenix, and the rest of America are not accustomed to the brutality. It's a pressure cooker. Years ago I was told by our CEO that if I was not ready to jump on a plane on a moment's notice and fix a problem half way around the world then perhaps this job was not for me.

Americans are reluctant to change, we see it everywhere. High tech is all about change.

Early on the American engineers and management were sent to live and work in Taiwan for up to two years for training. It was only so effective.

Another big issue is finding skilled labor; Amercia does not produce enough engineering talent. We need to value education more.

There is much more. My understanding is TSMC is agreeing to compromises with labor in numerous ways.

Semiconductors is a tough business. Fair? You must be joking.
 
Donald said:
Another sad corporate story like the demise of Sears? Years ago all Intel needed to worry about was AMD. What happened?

While I don't have the numbers, the IBM chip plant in East Fishkill NY was one of the largest in the world but IBM did not sell any chips, all internally used in IBM computers. So it was hard to really know how many chips they actually produced. IBM Research in Yorktown was top notch also. Then came along CMOS and everything changed.
Click to expand...
My understanding is, in the last several CEOs after Andy Groove were all sales / marketing / finance guys and they couldn't really make or have the creditability to make tough decisions that investors don't like. They were a monopoly for a long time and could afford to not invest in the best, as long as their competitor in x86 CPU were also not able to compete with them. The math was that AMD cannot build a better fab making x86 CPU. However, the foundry model started working because 1) not everything is about speed and cost, power consumption of smart phone means Apple and Qualcomm can afford to pay for a more expensive chip through TSMC. Also these newer companies are not making money selling processing power, they sell mobile solutions or eco system. 2) Crypto boom and low interest rate, leads to additional customers like nVidia AMD and BitMain paying for the best to make GPUs or crypto mining chips. They too make money off the design instead of chip making efficiency. 3) Intel due to politics refuse to R&D anything that can reduce demand for x86 CPU. Their LTE modem, GPU, FPGA, etc are all not that good. Same "Kodak" problem.

Then the chip act forcing them not to split up, and money not released while demanding them to build new fab first (they don't have money), and the crazy goal to build many new fabs all over the world before they sort out manufacturing problem, and them skipping a node they are late and just jump into the next node instead (happens a lot with incompetent management, I've worked in a few of these companies)...

Finally the board members aren't really engineering background, most of them are finance guys. You don't want a finance guy running an engineering company.
 
PandaBear said:
My understanding is, in the last several CEOs after Andy Groove were all sales / marketing / finance guys and they couldn't really make or have the creditability to make tough decisions that investors don't like. They were a monopoly for a long time and could afford to not invest in the best, as long as their competitor in x86 CPU were also not able to compete with them. The math was that AMD cannot build a better fab making x86 CPU. However, the foundry model started working because 1) not everything is about speed and cost, power consumption of smart phone means Apple and Qualcomm can afford to pay for a more expensive chip through TSMC. Also these newer companies are not making money selling processing power, they sell mobile solutions or eco system. 2) Crypto boom and low interest rate, leads to additional customers like nVidia AMD and BitMain paying for the best to make GPUs or crypto mining chips. They too make money off the design instead of chip making efficiency. 3) Intel due to politics refuse to R&D anything that can reduce demand for x86 CPU. Their LTE modem, GPU, FPGA, etc are all not that good. Same "Kodak" problem.

Then the chip act forcing them not to split up, and money not released while demanding them to build new fab first (they don't have money), and the crazy goal to build many new fabs all over the world before they sort out manufacturing problem, and them skipping a node they are late and just jump into the next node instead (happens a lot with incompetent management, I've worked in a few of these companies)...

Finally the board members aren't really engineering background, most of them are finance guys. You don't want a finance guy running an engineering company.
Click to expand...
Everything in your first paragraph describes why management needs to change. You can't change a giant business with a guy thats been there 40 years. Too much baggage. Doesn't automatically mean the next guy might not be worse, but you have to try. They might need to cycle through a few? I have sat in similar meetings as an outsider. The long timers all have the excuses that they tried something like that 20 years ago and it won't work now. You can't change their opinion. Your description of their x86 affinity explains it perfectly.

There building new fabs, not just here, but in Malaysia, Germany, and other places. A lot of it is new tech, new automation processes. It will be frustrating and take longer than expected - it always does.

Globalization was the most unique time in human history. Were reverting to the norm - imperialistic structured agreements and conflict everywhere else. People are not going to sole source their IC's from an island which is most likely to be in the middle of some sort of conflict for the foreseeable future. And not every application needs 1.8nm. I think the foundry business is a reasonable bet for someone. Guess we find out if there the right someone.

Full disclosure, I own a little bit of INTC stock. Its a speculative bet. It may go to zero - who knows, I don't have much money in it. Succeed or fail at least there swinging at the pitch, unlike a lot of places which likely will fail by doing nothing.
 
PandaBear said:
My understanding is, in the last several CEOs after Andy Groove were all sales / marketing / finance guys and they couldn't really make or have the creditability to make tough decisions that investors don't like. They were a monopoly for a long time and could afford to not invest in the best, as long as their competitor in x86 CPU were also not able to compete with them. The math was that AMD cannot build a better fab making x86 CPU. However, the foundry model started working because 1) not everything is about speed and cost, power consumption of smart phone means Apple and Qualcomm can afford to pay for a more expensive chip through TSMC. Also these newer companies are not making money selling processing power, they sell mobile solutions or eco system. 2) Crypto boom and low interest rate, leads to additional customers like nVidia AMD and BitMain paying for the best to make GPUs or crypto mining chips. They too make money off the design instead of chip making efficiency. 3) Intel due to politics refuse to R&D anything that can reduce demand for x86 CPU. Their LTE modem, GPU, FPGA, etc are all not that good. Same "Kodak" problem.

Then the chip act forcing them not to split up, and money not released while demanding them to build new fab first (they don't have money), and the crazy goal to build many new fabs all over the world before they sort out manufacturing problem, and them skipping a node they are late and just jump into the next node instead (happens a lot with incompetent management, I've worked in a few of these companies)...

Finally the board members aren't really engineering background, most of them are finance guys. You don't want a finance guy running an engineering company.
Click to expand...
I think it was finance guys and not retail guys that brought down Sears.
 
Donald said:
I think it was finance guys and not retail guys that brought down Sears.
Click to expand...
Sears died because the "everything" department store died. There product was way to expensive comparing to a sheet at walmart and a fridge at home depot. They got rid of their catalog 10 years early - they could have kept it and transitioned to being a powerhouse in online, but couldn't see the future. They wanted to sell Kenmore not maytag and keep the margin for themselves. There stores were anchors in malls that were also dyeing.

So no, retailers were in fact the ones that brought down sears, they wanted the future to be like the past. Finance guys drove the final nail in the coffin.

kmart - same.
 
SC Maintenance said:
Sears died because the "everything" department store died. There product was way to expensive comparing to a sheet at walmart and a fridge at home depot. They got rid of their catalog 10 years early - they could have kept it and transitioned to being a powerhouse in online, but couldn't see the future. They wanted to sell Kenmore not maytag and keep the margin for themselves. There stores were anchors in malls that were also dyeing.

So no, retailers were in fact the ones that brought down sears, they wanted the future to be like the past. Finance guys drove the final nail in the coffin.

kmart - same.
Click to expand...
Every big Sears I used to go to = gangsta areas …
You feel like that old building moved to another country …
 
If one hasn't been written already, there could be a book written about the downfall of Sears.

What many don't realize is that the road to the most advanced chip making like TSMC does passes through…a Dutch company.
 
SC Maintenance said:
Everything in your first paragraph describes why management needs to change. You can't change a giant business with a guy thats been there 40 years. Too much baggage. Doesn't automatically mean the next guy might not be worse, but you have to try. They might need to cycle through a few? I have sat in similar meetings as an outsider. The long timers all have the excuses that they tried something like that 20 years ago and it won't work now. You can't change their opinion. Your description of their x86 affinity explains it perfectly.

There building new fabs, not just here, but in Malaysia, Germany, and other places. A lot of it is new tech, new automation processes. It will be frustrating and take longer than expected - it always does.

Globalization was the most unique time in human history. Were reverting to the norm - imperialistic structured agreements and conflict everywhere else. People are not going to sole source their IC's from an island which is most likely to be in the middle of some sort of conflict for the foreseeable future. And not every application needs 1.8nm. I think the foundry business is a reasonable bet for someone. Guess we find out if there the right someone.

Full disclosure, I own a little bit of INTC stock. Its a speculative bet. It may go to zero - who knows, I don't have much money in it. Succeed or fail at least there swinging at the pitch, unlike a lot of places which likely will fail by doing nothing.
Click to expand...

People do source critical technologies from front line of potential warzone a lot: telecom equipment from Finland (Erikson / Nokia), precision equipment from Germany and Japan, Taiwan is just one of them. On the contrary, I would imagine in a cold war era (nobody with nuke will fight a hot war with each other these days), keeping highly valuable and profitable industry near front line is to help keep them wealthy and sustain the military bases either they or US have there. S Korea would not be able to support their massive (I think within the top 5 if not top 10 of the world) for such a small country and population had they not have Samsung, Hyundai, or SK.

I'm starting to believe TSMC and nVidia have a lot more to do with leadership discipline and is the outlier these days so it would be hard to replicate their successes with just a 'I can do it too' non founder CEO.
 
Dave Hess said:
Is the Arizona fab too optimistic and too high goals ?
Click to expand...

Part of it is the fab. Maybe Intel can pull some political string to get the equipment allocation prioritized but the people, you can't build a team out of nowhere after squeezing their pays for 2 decades and all your best talents already left for better paying competitors like Apple, nVidia, Google, Meta / Facebook, etc. They are competing with people in industries also making chips but make most of their money outside of chips in ecosystems.

TSMC is the best paying employer in Taiwan, Intel is one of the worst paying employer in the semiconductor industry today (at least in Santa Clara). Arizona is probably ok but if TSMC starts hiring there the best of Intel in Arizona will definitely jump ship.

https://www.msn.com/en-us/news/tech...1&cvid=9323b57746d3414a920d837bef6b7894&ei=11

As an example. From my sources I was told for example an nVidia H100 would cost like $150 to make in TSMC just for the silicon, then about $300 more for the high bandwidth memory for it, that's a total of $450. nVidia sells it for $30k each. Intel is focusing on the foundry side in its revitalization plan, making that equivalent of $150 instead of the $29550 on the design and ecosystem side.
 
Carmudgeon said:
If one hasn't been written already, there could be a book written about the downfall of Sears.

What many don't realize is that the road to the most advanced chip making like TSMC does passes through…a Dutch company.
Click to expand...
Yes. And the Dutch company ASML gets there neon from Ukraine. And 90% of the highly pure sand for worldwide IC's comes from a small town in North Carolina that just about got washed away during the last storm. The IC supply chain is the longest and most frail in the world.
 
JeffKeryk said:
CEOs come and go. My problem is with TSMC and the China-Taiwan conflict. The Chips Act fostered the Arizona fab, but there is so much more complexity.
Click to expand...
We're supposed to be getting some of that here in Central NY state as well. I'm hoping it does happen. Getting chip manufacturing domestic is a matter of national security at this point. Plus , who knows what China will do to the chip market if they get ahold of the ability to make these chips.
 
PandaBear said:
As an example. From my sources I was told for example an nVidia H100 would cost like $150 to make in TSMC just for the silicon, then about $300 more for the high bandwidth memory for it, that's a total of $450. nVidia sells it for $30k each. Intel is focusing on the foundry side in its revitalization plan, making that equivalent of $150 instead of the $29550 on the design and ecosystem side.
Click to expand...
All those numbers are meaningless unless you know also the R&D and overhead cost NVDA has. Your ignoring the decades of R&D it took to get there.

No doubt the design side is more profitable - assuming people want your chip. Personally I think the NVDA guys are extremely lucky. They didn't start out making a AI chip. They were making graphics chips that just happened to be the best fit for AI. Super smart guys no doubt, but also right place right time.

Manufacturing is a different business for sure - high volume, thin margins, but more stable. Your comparing apples to peanuts. INTC - market cap $80B, to NVDA - market cap 3350B - 42 times higher. I am comparing INTC to TSM - and I think there might be a opportunity there - primarily because of the de-globalization efforts. The fact they have a design business at all is gravy.

Also, my long experience across hundreds of companies is the margin all resides in understanding the application. Ie the people that figure out how to use AI like Microsoft, google, apple or some completely new company will make most of the profit. Those types of companies were the winners in the internet revolution, not Lucent. Thats why NVDA is actually working on applications not focusing on just chips. And there customers are already working on their own chips:

https://www.cnbc.com/2024/12/03/apple-says-it-uses-amazons-custom-ai-chips-.html
 
SwampSurvivor said:
We're supposed to be getting some of that here in Central NY state as well. I'm hoping it does happen. Getting chip manufacturing domestic is a matter of national security at this point. Plus , who knows what China will do to the chip market if they get ahold of the ability to make these chips.
Click to expand...
China has been a major player for many years. It is a huge region in the global Semiconductor market.
 
JeffKeryk said:
China has been a major player for many years. It is a huge region in the global Semiconductor market.
Click to expand...
For now the export ban is limiting their capabilities. If any indication of what they did in the past the market would be flooded and oversupply in their own market as well as outside of their own market. This happened in solar panel, LCD, EVs, cotton, film camera film, steel, etc. This is a major problem with any central planning economy that China is still partly in, they will have no way to micro manage it efficiently until it is too late.
 
PandaBear said:
For now the export ban is limiting their capabilities. If any indication of what they did in the past the market would be flooded and oversupply in their own market as well as outside of their own market. This happened in solar panel, LCD, EVs, cotton, film camera film, steel, etc. This is a major problem with any central planning economy that China is still partly in, they will have no way to micro manage it efficiently until it is too late.
Click to expand...
Its not a bug, its a feature.

China uses control of certain markets as a hegemon of sorts - much like we use the US dollar. Control certain critical markets, you control a country or partners economy. We saw this in the pandemic for example - all the PPE had to come from China.

They have control of a significant portion of our prescription drugs. Textiles, shoes, electronics. The next planned attack was EV's. They don't care about making money - they don't have to. They print as much money internally as they need to. Its estimated China printed 4 to 5X more money than the US has over the last decade. You never know, because they control the flow of data out also.
 
PandaBear said:
For now the export ban is limiting their capabilities. If any indication of what they did in the past the market would be flooded and oversupply in their own market as well as outside of their own market. This happened in solar panel, LCD, EVs, cotton, film camera film, steel, etc. This is a major problem with any central planning economy that China is still partly in, they will have no way to micro manage it efficiently until it is too late.
Click to expand...
There is another issue; an upstream issue. If sanctions against China continue/increase, the Semiconductoe Mfg Equipment companies will be greatly impacted.
 
You must log in or register to reply here.
Back
Top Bottom