Former Intel CEO Craig Barrett on saving Intel

[hist78]

I have ONE question for Craig Barret, if you are so smart when you were CEO why did you allow Intel to go down the path that lead the company to where they are today?

"Craig Barrett served as CEO of Intel from 1998 to 2005 and as Chairman from 2005 to 2009. He was not only the captain steering the entire Intel ship, but also the steward of its long term prosperity. "

"From my analysis, Intel spent $50.32 billion on share buybacks between 2000 and 2009, when Mr. Barrett was either CEO or Chairman. I can’t help but wonder: what if he had spent only half that amount, keeping $25 billion to improve Intel’s long term competitiveness, or even to start Intel Foundry at a time when TSMC was still not as strong as Intel?"

Former Intel CEO Craig Barrett on saving Intel

A couple of teasers from the article: "6. The current Intel CEO's comments about not investing in new technology (14A) until customers sign up is a joke." "9. The FFWBMs (four former wise board members) of Intel continue to claim you have to break Intel into two pieces before any customer...

semiwiki.com

When Robert Noyce, Gordon Moore, and Andy Grove retired from Intel one by one, Craig Barrett became the person responsible for guiding the company into the 21st century and beyond. Unfortunately, he made several mistakes and questionable decisions that have had long lasting negative effects.

One example was his choice of Paul Otellini as successor, the first Intel CEO without an engineering background. Otellini is best known for rejecting Apple’s request for Intel to manufacture SoCs for the upcoming iPhone, believing the order would be too small and unprofitable. The first iPhone launched in June 2007, when Otellini was CEO and Barrett was Chairman of Intel.

Fast forward to 2024: Apple paid about $23 billion to TSMC for manufacturing services, while Intel’s total revenue for the same year was $53.1 billion. It's a stark and sobering contrast.

 

[blueone]

"Craig Barrett served as CEO of Intel from 1998 to 2005 and as Chairman from 2005 to 2009. He was not only the captain steering the entire Intel ship, but also the steward of its long term prosperity. "
"From my analysis, Intel spent $50.32 billion on share buybacks between 2000 and 2009, when Mr. Barrett was either CEO or Chairman. I can’t help but wonder: what if he had spent only half that amount, keeping $25 billion to improve Intel’s long term competitiveness, or even to start Intel Foundry at a time when TSMC was still not as strong as Intel?"

As you could read from the Fortune op-ed, Craig was always a staunch believer in being an IDM.

When Robert Noyce, Gordon Moore, and Andy Grove retired from Intel one by one, Craig Barrett became the person responsible for guiding the company into the 21st century and beyond. Unfortunately, he made several mistakes and questionable decisions that have had long lasting negative effects.

One example was his choice of Paul Otellini as successor, the first Intel CEO without an engineering background. Otellini is best known for rejecting Apple’s request for Intel to manufacture SoCs for the upcoming iPhone, believing the order would be too small and unprofitable. The first iPhone launched in June 2007, when Otellini was CEO and Barrett was Chairman of Intel.

I don't think I can stand to read this BS story another time. PSO did not want to build more N-process fabs, even just one, to build low-margin parts, from the perspective of Intel's balance sheet. It made no financial sense to him, or a lot of other people at the time. Intel was not a low-cost fab, and everything about Intel's fab process was about performance and high clock frequencies, two things you don't want to optimize for in an iPhone or an iPad. So Apple would have needed process variations, and I'm sure that would have translated into R&D resources that would have to be redeployed from high-margin CPUs. Remember, Jobs wanted a very low price. BTW, PSO didn't like Itanium either. While he wasn't an engineer, I thought he was a pretty smart and rational guy, and I got to know him well enough to come to a conclusion like that.

Fast forward to 2024: Apple paid about $23 billion to TSMC for manufacturing services, while Intel’s total revenue for the same year was $53.1 billion. It's a stark and sobering contrast.

Former Intel CEO Craig Barrett on saving Intel

A couple of teasers from the article: "6. The current Intel CEO's comments about not investing in new technology (14A) until customers sign up is a joke." "9. The FFWBMs (four former wise board members) of Intel continue to claim you have to break Intel into two pieces before any customer...

semiwiki.com

Yeah, isn't 20-20 hindsight great?

 

[hist78]

Li-Pu Tan is the first Intel CEO in recent years to have the courage to identify Intel’s deep-rooted problems and take painful but necessary steps to correct them.

From Greg Barrett’s comments on how to “save” Intel, we can see some classic traits of Intel’s culture or Intel’s style:

1. “We are so unique, we are the chosen one!”

While TSMC, GlobalFoundries, AMD, Nvidia, Broadcom, Marvell, Qualcomm, and MediaTek fight fiercely for customers and survival, Intel has often been busy debating among itself about which high-end restaurant it should go to. Many Intel leaders, and likely many employees, seem to believe they have the luxury of choosing their customers, features, prices, and costs. Worse, they see this as the Intel way to win. In their view, it’s Intel who chooses the customers and sets the price and schedule.


2. Entitlement

Intel’s finances are in serious trouble. From 2022 to Q2 2025, the company has suffered billions in negative free cash flow, burning cash faster than it can generate. The solution? Pat Gelsinger’s approach is to ask domestic and foreign governments for “free” money. Craig Barrett’s idea went even further, asking Intel’s customers and competitors to chip in $40 billion to cover Intel’s own financial and management mistakes.

This conveniently ignores the fact that Intel spent a whopping $133.86 billion (2000~ 2021) or $83.54 billion (2010 ~ 2021) on share buybacks. Instead of investing that precious cash to position itself for the future, as many peers did, Intel used it to boost its stock price. The justification? See point #1 above.


3. “Give us some more time, Intel shall rise again!”

This mindset undermines the urgency for real, fundamental change. Intel seems to believe its customers and competitors can simply wait another two, three, or even four years while it pursues the “latest and greatest” technology. They don’t say it outright, but the underlying belief is obvious: the industry should give Intel more time to reclaim its glory. And the justification? Again, see point #1.

But Intel’s customers and competitors won’t wait. It's because they can’t afford to wait, and they don’t need to. Other solutions are readily available.

 

[EganVector]

These reasons aren't correct. While buybacks do reduce the number of outstanding shares, the motivation isn't increasing executive compensation based on EPS. I'm not aware of any company that does that.

The two most important justifications for share buybacks are:

1. To reduce dilution of share book value when a company issues stock to employees as compensation in the form of stock options or restricted stock units.

2. To return excess earnings to shareholders while avoiding the multiple-taxation of earnings delivered as dividends in the US. The earnings returned are double-taxed because the corporation pays federal income tax on the earnings, and then the recipients of the dividends pay federal income tax on them (though probably at the qualified dividend rate, which is the same as the long-term capital gains rate). And for those recipients who live in states with an income tax, they're taxed again. And if you live in a city with a local income tax the dividend funds get taxed a fourth time. Share buybacks increase the value of shares by reducing the number of shares available to trade in the stock market (called "float"), which increases the value of each share without further taxation until the shares are sold by a stockholder for a gain.

I am not aware of any company anywhere that has 100% of its executives acting 100% of the time for the benefit of the company. Every human has their own agenda, and sometimes it aligns with the agenda of the company. Most people operate on the basis of the "what's in it for me?" principle. While the justifications you gave are valid, they are still not a good use of capital and are indicative a company that has run out of ideas for future products ("hey, we have no idea what to invest in, so let's buy back shares").

 

[blueone]

I am not aware of any company anywhere that has 100% of its executives acting 100% of the time for the benefit of the company. Every human has their own agenda, and sometimes it aligns with the agenda of the company. Most people operate on the basis of the "what's in it for me?" principle.

I agree. I have run into numerous examples of what you're talking about in my career, especially with acquired groups from other companies, and especially in countries different from the acquiring company's home country. What I've observed is that the acquired teams often have their primary allegiance to the team members and the team as a whole, rather than the company's overall interest. I could list multiple examples of this I observed in Intel.

While the justifications you gave are valid, they are still not a good use of capital and are indicative a company that has run out of ideas for future products ("hey, we have no idea what to invest in, so let's buy back shares").

I don't agree. Oftentimes ideas for future products or technologies require relatively modest R&D funding to determine if they should get to the next stage of development. And the proportion of pipelined ideas that result in compelling business cases is nowhere near 100%, it might less than 20%, and there might be a long time between lab ideas and high volume manufacturing. Silicon photonics comes to mind. So does Quantum computing. (Intel had/has a promising program to do QC in CMOS chips.) To name two. Also, corporations exist for the benefit of stockholders, at least in the US, and senior executives have a fiduciary responsibility to maximize shareholder value. In the US, share buybacks are an important tool for returning excess funds to shareholders, and keeping tens or hundreds of billions of dollars in treasury bills because someone's crystal ball says the basic business model might change sometime in the future, and we should keep all of that low-performing cash around just in case, is yet another example of where 20-20 hindsight makes predicting the future look easy after the fact.

 

[tedc]

Li-Pu Tan is the first Intel CEO in recent years to have the courage to identify Intel’s deep-rooted problems and take painful but necessary steps to correct them.

From Greg Barrett’s comments on how to “save” Intel, we can see some classic traits of Intel’s culture or Intel’s style:

Objectively, very little is in Intel's favor right now (aside from having LBT as CEO):
• Running at a net loss with declining revenue and market share. Product challenges in HPC and AI.
• Arguably the strongest competitive landscape Intel has faced in decades.
• Tons of political and analyst noise and scrutiny.
• Limited funds to invest into leading edge processes. CHIPS Act funding tentative.
• Limited options to improve cost structure (e.g., expanding abroad).
• Have yet to build a viable foundry business. Lack of commitment and clarity around 18A and 14A (delays?).
• In the unenviable position of convincing its competitors to become customers while also asking them fund capacity expansion and R&D.
• Still financially burdened with Apollo and Brookfield JVs.

I have to commend LBT for understanding all this and still taking the job. Barrett's comments express his frustration, but fail to present a practical business path to save Intel. To your earlier point: Intel is "special" and therefore deserves a Hail Mary play by its competitors to save it (i.e., dictating the terms of their success)? What a perspective.

 

[lispwright]

I have ONE question for Craig Barret, if you are so smart when you were CEO why did you allow Intel to go down the path that lead the company to where they are today?

Perhaps we shouldn't listen to Barrett because he's infamous for not listening to others? Below quote per Bob Colwell, who was part of the P6 core team, Intel's first out of order design.

Which had enough information as it was processing data that it ended up being better than the very long instruction word (VLIW) Itanium approach which required that knowledge up front in compilers. And is said to only worked for a few HPC type workloads, so Andy Grove and company green lit it based on microbenchmarks.

Look higher in the link and you'll see him talking about the nightmare of TTL in the 1980s, which I saw happen in a company I worked at when a Fairchild FAST version of a standard ALU part failed hard, one of many reasons a lot of companies using LSI/MSI failed the transition of getting a whole CPU minus FP onto one chip:

0:26:34 BC: Grove stopped being the president in January 1998.

0:26:40 PE: Yes

0:26:41 BC: And that's when Craig Barrett took over.

0:26:43 PE: And what changed at point?

0:26:46 BC: Well Craig's not Andy, I mean he had a different way of thinking and doing things, Craig, I don't want it to sound cynical but I always sound cynical when I talk about him because I had such a bumpy relationship with him. I don't think he felt like he needed anything I could tell him, and it wasn't just me, I wasn't taking this personally. I never once got the same feeling I got with Andy that my inputs were being seriously and politely considered, and then a decision would be made that included my inputs.

0:27:21 PE: Yes.

0:27:22 BC: That never happened. Instead, for example five Intel fellows including me went to visit Craig Barrett in June of 98 with the same Itanium story, that Itanium was not going to be able to deliver what was being promised. The positioning of Itanium relative to the x86 line is wrong, because x86 is going to better than you think and Itanium is going to be worse and they're going to meet in the middle. We're being forced to put a gap in the product lines between Itanium and x86 to try to boost the prospects for Itanium. There's a gap there now that AMD is going to drive a truck through, they're going to, what do you think they're going to hit, they're going to go right after that hole" which in fact they did. It didn't take any deep insight to see all of these things, but Craig essentially got really mad at us, kicked us out of his office and said (and this is a direct quote) "I don't pay you to bring me bad news, I pay you to go make my plans work out"

0:28:22 PE: Gee.

0:28:25 BC: So.

0:28:25 PE: Yeah he's polar opposite.

0:28:26 BC: So and he, and at that point he stood up and walked out and to back of his head. I said, "Well that's just great Craig. You ignored the message and shot the messengers. I'll never be back no matter how strong of a message I've got that you need to hear, I'll never bring it to you now.”

0:28:38 PE: Yeah.

0:28:40 BC: It's not rewardable behavior. It was sad, a culture change in the company that was not a good one and there was no way I could fix it. If it had been Andy doing something that I thought was dumb, I'd go and see him and say "Andy what you're doing is dumb", and maybe I'd convince, maybe I wouldn't. But as soon as you close that door, it is a dictatorship. You can't vote the guy out of office anymore, you can't reach him. There's no communication channel.

 

[lispwright]

I'm gaining increased respect for Intel's 2 "non-technical" CEOs, although Paul Otellini was also a programmer. He's got a famous 20/20 hindsight quote about turning down Apple for the iPhone CPU:

We ended up not winning it or passing on it, depending on how you want to view it. And the world would have been a lot different if we'd done it. The thing you have to remember is that this was before the iPhone was introduced and no one knew what the iPhone would do... At the end of the day, there was a chip that they were interested in that they wanted to pay a certain price for and not a nickel more and that price was below our forecasted cost. I couldn't see it. It wasn't one of these things you can make up on volume. And in hindsight, the forecasted cost was wrong and the volume was 100x what anyone thought.

Bob Swan at least realized Intel was failing hard with 10 nm and booked enough TSMC capacity to tide Intel over. Which also put them on the path to using industry tools, a prerequisite to becoming a real foundry.

So I don't score Lion Cove parts like Arrow Lake and Lunar Lake as harshly as gamers do, because the core is sort of a clean sheet redesign. Tweak it and it should go from adequate to good.

 

[MKWVentures]

Bob Swan decided Intel Manufacturing was not working out. All the product groups agreed with him. Apple, AMD, Nvidia, etc agree with him.

Intel was great during a very different era of semiconductors. Those ways don't work any more. They tried lots of options to fix it .... but Intel is Intel. Companies are not entitled to be successful forever

.... and then PG went all in to expand Intels worst performing group.... Now Pat's gone and the product group is back to "Please do not make us use IFS"

Intel Still has very very smart people and a smaller, more agile, more customer centric, Intel Product group will be successful.

 

[blueone]

Perhaps we shouldn't listen to Barrett because he's infamous about not listening to others? Below quote per Bob Colwell, who was part of the P6 core team, Intel's first out of order design.

Which had enough information as it was processing that it ended up being better than the very long instruction word (VLIW) Itanium approach that required that knowledge up front in compilers. Which is said to only worked for a few HPC type workloads, so Andy Grove and company green lit it based on microbenchmarks.

Look higher in the link and you'll see him talking about the nightmare of TTL in the 1980s, which I saw happen in a company I worked at when a Fairchild FAST version of a standard ALU part failed hard, one of many reasons a lot of companies using LSI/MSI failed the transition of getting a whole CPU minus FP onto one chip:

Excellent reference. I didn't know Bob Colwell well, but I did spend a lot of time talking to him once at a party, and it was enjoyable and enlightening. I fondly remember his columns in IEEE Computer. I got in trouble too when I was assigned to Itanium, unwillingly, because I thought VLIW was suitable for processors with constrained workloads, but would be a really bad idea for general purpose computing. (I could argue that much of my career was defined by being assigned to truly bad ideas and tasked with making them leadership products anyway. I think I could write a book on the topic.)

Barrett also got very annoyed with me once and let me know it. He's very intimating in person, at least back then. Nothing bad came of the incident, but I did not look forward to any future interactions.

 

[DanX]

I have ONE question for Craig Barret, if you are so smart when you were CEO why did you allow Intel to go down the path that lead the company to where they are today?

BOD

 

[tomatoma]

Bob Swan decided Intel Manufacturing was not working out. All the product groups agreed with him. Apple, AMD, Nvidia, etc agree with him.

Intel was great during a very different era of semiconductors. Those ways don't work any more. They tried lots of options to fix it .... but Intel is Intel. Companies are not entitled to be successful forever

.... and then PG went all in to expand Intels worst performing group.... Now Pat's gone and the product group is back to "Please do not make us use IFS"

Intel Still has very very smart people and a smaller, more agile, more customer centric, Intel Product group will be successful.

No, even if it's downsized, Intel has no future.
Intel will always fall
That's destiny

 

[hist78]

I'm gaining increased respect for Intel's 2 "non-technical" CEOs, although Paul Otellini was also a programmer. He's got a famous 20/20 hindsight quote about turning down Apple for the iPhone CPU:
Bob Swan at least realized Intel was failing hard with 10 nm and booked enough TSMC capacity to tide Intel over. Which also put them on the path to using industry tools, a prerequisite to becoming a real foundry.

So I don't score Lion Cove parts like Arrow Lake and Lunar Lake as harshly as gamers do, because the core is sort of a clean sheet redesign. Tweak it and it should go from adequate to good.

"He's got a famous 20/20 hindsight quote about turning down Apple for the iPhone CPU".

Was it truly a case of 20/20 hindsight, or was it a leader who was disconnected from market trends and reality? This is a good topic for discussion.

But one thing is very clear: in 2012, five years after Apple released the first iPhone in 2007, Mr. Paul Otellini was still very much stuck in an Intel-centric world (aka “Intel Bubble”). While the rest of the world had been moving in different directions for at least five years, he still failed to recognize the shift and did not adapt Intel accordingly. Many of Intel’s problems today can be traced back to his decisions or lack of action. Was he arrogant, or simply naive?

The following interview was conducted in October 2012 at the University of California, Berkeley. Mr. Paul Otellini served as Intel’s fifth CEO from 2005 to 2013.

Paul Otellini, CEO of Intel, in conversation with Berkeley-Haas Dean Rich Lyons

Paul Otellini, CEO of Intel and a Haas MBA graduate, spoke to Haas Dean Rich Lyons on October 3rd, 2012 as part of the Dean's Speaker Series Paul Otellini: CEO of Intel from 2005 to 2013.

semiwiki.com

 

[Anderson style]

is intel going bankruptcy?

 

[DanX]

"He's got a famous 20/20 hindsight quote about turning down Apple for the iPhone CPU".

Was it truly a case of 20/20 hindsight, or was it a leader who was disconnected from market trends and reality? This is a good topic for discussion.

I agree with you. This is an opportunity , not garrantee a success.
Samsung had the opportunity but did not transfer it to a success.

 

[tomatoma]

"He's got a famous 20/20 hindsight quote about turning down Apple for the iPhone CPU".

Was it truly a case of 20/20 hindsight, or was it a leader who was disconnected from market trends and reality? This is a good topic for discussion.

But one thing is very clear: in 2012, five years after Apple released the first iPhone in 2007, Mr. Paul Otellini was still very much stuck in an Intel-centric world (aka “Intel Bubble”). While the rest of the world had been moving in different directions for at least five years, he still failed to recognize the shift and did not adapt Intel accordingly. Many of Intel’s problems today can be traced back to his decisions or lack of action. Was he arrogant, or simply naive?

The following interview was conducted in October 2012 at the University of California, Berkeley. Mr. Paul Otellini served as Intel’s fifth CEO from 2005 to 2013.

Paul Otellini, CEO of Intel, in conversation with Berkeley-Haas Dean Rich Lyons

Paul Otellini, CEO of Intel and a Haas MBA graduate, spoke to Haas Dean Rich Lyons on October 3rd, 2012 as part of the Dean's Speaker Series Paul Otellini: CEO of Intel from 2005 to 2013.

semiwiki.com

He's just incompetent

 

[Xebec]

So I don't score Lion Cove parts like Arrow Lake and Lunar Lake as harshly as gamers do, because the core is sort of a clean sheet redesign. Tweak it and it should go from adequate to good.

Is ARL more expensive to manufactured than RPL given the amount of packaging required?

I recognize it's using older nodes for much of the silicon, and yields are higher with smaller dice, but you also 'waste' a lot of transistors communicating back and forth between those dice. Not to mention reduced ASP potential from lower performance of not being monolithic.

 

[BruceA]

A good read about Intel's iPhone moment

How Intel Missed the iPhone : The XScale Era

Challenging the conventional narrative on Intel's smartphone miss

thechipletter.substack.com

 

[BruceA]

Everybody can have easy 20:20 hindsight from the retirement corner pontificate about this or that. Once you've walked a ton of miles in any of the players shoes is it really that easy to separate what got you to where you are and what you need to do next that is so obvious now? Unless you've lived it deeply first hand to appreciate what drove the culture creation, the personalities and the decisions that enabled people to rise to their position of power to make those decisions, to easy to just state the obvious mistake.

The look back is always obvious, looking forward very hard sometimes to see what could be the real pivot changing direction.

 

[KevinK]

The look back is always obvious, looking forward very hard sometimes to see what could be the real pivot changing direction.

The interesting question looking forward is what happens beyond 14A - what is the the main driver as the economics of the next processes fall solely on TSMC, as Intel and Samsung fall away ? Seems to me the real drivers for greater density from future nodes are client-side inference and data center inference / training, meaning general processing is good enough as a supporting function to AI, today. Both of those are going to be driven by energy efficiency and memory requirements of the next generations of models.