INTC - Intel

[Castanza]

It's been my experience that if you pay enough money one tends to shows up.

 

Money has close to nothing to do with CEO quality.

 

2 (?) CEOs later Intel still has crappy leadership.

 

Perhaps your experience is colored by hugely successful Microsoft transition to Nadella.

 

I would say that this is more of an exception than the rule. I'd say AMD waited 10+ years for Lisa Su. IBM? Nokia?

 

We'll see how Intel fares and when - if ever - it gets a great CEO.

 

It's a fair point...."A bird in the hand is worth two in the bush."

 

In the meantime, Intel is a highly profitable company with solid growth. But your point is taken that it could turn into the WFC of the chip market.  ;D

[mwtorock]

Slightly off topic -  there is a wsj article today for Lisa Su.

 

https://www.wsj.com/articles/lisa-sus-amd-is-flying-high-as-rival-chip-maker-intel-stumbles-11596732408?mod=itp_wsj&ru=yahoo

 

Would you think Intel would take her in as CEO candidate back in 2012 or 2013?

 

*****************************

“You would think that the decisions get easier because you have more resources, but they actually don’t because you have a much broader set of opportunities,” she said. “In some sense, when you’re more constrained, it’s, ‘OK, I have two bets.’ But now I can have 10 bets.”

 

An engineer who once served as technical assistant to renowned International Business Machines Inc. CEO Louis Gerstner, the 50-year-old Ms. Su is credited with several decisions that put AMD in position to win the high-speed computing race. She shifted its strategy for outsourcing semiconductor manufacturing and embraced a radical redesign of its chips.

 

AMD was flailing when Ms. Su was appointed CEO in 2014. Poor chip-design choices and manufacturing delays had erased a strong run in the mid-2000s.

 

“I used to have conversations with distressed-asset investors about what the company was worth in a liquidation breakup,” said Stacy Rasgon, an analyst at Sanford Bernstein & Co. “That’s where a lot of the thinking was going back then.”

 

In taking over, Ms. Su faced a high-stakes choice. Conventional wisdom in the chip industry said that desktop computers, and the processors that power them, were becoming a less important market. The future, the thinking went, belonged to powering items such as mobile phones and chromebooks.

 

 

Ms. Su shifted AMD’s strategy for outsourcing semiconductor manufacturing and embraced a radical redesign of its chips.

 

AMD had started investing in chips optimized for such devices, but its strength was in making the central processing units, or CPUs, that run desktop and laptop computers and servers, as well as graphics processors that handle image displays in computers and game consoles like Microsoft Corp. ’s Xbox.

 

Ms. Su opted to double-down on those core products. Others, like chips for phones and tablets, took a back seat.

 

Her decisions paid off—especially lately. AMD’s revenue from CPUs and graphics processors rose to $4.71 billion last year, from $1.81 billion in 2015. During the pandemic, demand for servers and laptops has boomed while smartphone shipments this year are down.

 

That outcome owes a lot to another bet Ms. Su placed five years ago, when she signed off on a new design, and lined up manufacturing to embed it in chips with the tiniest and most efficient transistors in a bid to make them faster than Intel’s.

[rb]

It's been my experience that if you pay enough money one tends to shows up.

 

Money has close to nothing to do with CEO quality.

 

2 (?) CEOs later Intel still has crappy leadership.

 

Perhaps your experience is colored by hugely successful Microsoft transition to Nadella.

 

I would say that this is more of an exception than the rule. I'd say AMD waited 10+ years for Lisa Su. IBM? Nokia?

 

We'll see how Intel fares and when - if ever - it gets a great CEO.

 

It's a fair point...."A bird in the hand is worth two in the bush."

 

In the meantime, Intel is a highly profitable company with solid growth. But your point is taken that it could turn into the WFC of the chip market.  ;D

Well if money doesn't matter then why do we pay them so much?

 

Who knows maybe Warren East will want to play with chips again?

[Liberty]

Well if money doesn't matter then why do we pay them so much?

 

Who knows maybe Warren East will want to play with chips again?

 

Money matters to attract and retain the good leaders.

 

But paying someone a lot doesn't make them a good leader. There's a lot more high-paid CEOs out there than there are good CEOs worth their salaries...

 

Necessary condition, but not sufficient.

[Jurgis]

Anybody knows first-hand or second-hand how Altera and/or Mobileye are doing after Intel acquired them? Growing? Shrinking? Holding steady? How are their products faring? What's the morale?

[zippy1]

Not sure whether this has been mentioned or not. Dr. Lisa Su is the niece of Jensen Huang, the CEO of Nvidia......

[Castanza]

https://finance.yahoo.com/news/intel-says-transistor-technology-could-130846233.html

 

"It is 20%, the largest intra-node jump ever in our history," Raja Koduri, Intel's chief architect, said of the performance gain in an interview with Reuters. "It's actually same as what you would get with one full Moore's Law node of performance."

[scorpioncapital]

Isn't Moore's law a doubling? Or is a moore's node different?

 

 

[rkbabang]

Isn't Moore's law a doubling? Or is a moore's node different?

 

Moore's law was a doubling of the number of transistors on a chip every 18-24 months.  It wasn't directly about performance (although increased performance was the result).

 

[Liberty]

Great essay on Intel problems and larger implications:

 

https://medium.com/@gavin_baker/investing-mistakes-chapter-1001-9f0b7dbc6637

[Castanza]

Great essay on Intel problems and larger implications:

 

https://medium.com/@gavin_baker/investing-mistakes-chapter-1001-9f0b7dbc6637

 

There is a lot of conjecture in that article. First the author says Intel’s chips are better in real world scenarios and their advantage is that the software is written for their x86 chipset. They go on to say that it would take at least four years of dominance from AMD (what dominance?) for coders to begin to write programs optimized for their architecture.

 

Then they contradict themselves and say...”once a company stops trying to compete at the leading edge they can never get it back.” Alluding that Intel’s chipset delay was a permanent impairment and they will never again catchup. Even though Intels 10nm chip is equivalent to AMD 7nm in terms of performance.

 

So is Intel ahead or behind? Do real world results matter or not? If AMD was behind for the past two decades how can they be on the cutting edge now if by the authors own admission this is not possible.

 

Then the author says that software will begin to be written for AMD because of Intels 7nm delay. What? I thought they said it could take 4 years for this shift. So far Intel said a 6-12 month delay could be expected for a 7nm rollout. And that their options are open to outsourcing manufacturing which could significantly cut down this delay.

 

The author also said that TSMC gave AMD a large advantage, but why would this not also be an advantage for Intel if they used their 7nm products as well?

 

There is also zero mention of any other market segment. AMD competes with Intel in what 2 of 10 segments Or something like that?

 

Again, I’m not trying to pump Intel....this just comes across more as a hit piece than anything else. Time will tell. I guess AMD could pull it off in some way. They have good management which Intel lacks. From an investment standpoint to me it’s either Invest in Intel or don’t. AMD is way too hot to touch right now. Perhaps TWSC is worth a look in the short term. Not sure.

 

 

[lnofeisone]

Great essay on Intel problems and larger implications:

 

https://medium.com/@gavin_baker/investing-mistakes-chapter-1001-9f0b7dbc6637

 

There is a lot of conjecture in that article. First the author says Intel’s chips are better in real world scenarios and their advantage is that the software is written for their x86 chipset. They go on to say that it would take at least four years of dominance from AMD (what dominance?) for coders to begin to write programs optimized for their architecture.

 

The. They contradict themselves and say...”once a company stops trying to compete at the leading edge they can never get it back.” Alluding that Intel’s chipset delay was a permanent impairment and they will never again catchup. Even though Intels 10nm chip is equivalent to AMD 7nm in terms of performance.

 

So is Intel ahead or behind? Do real world results matter or not? If AMD was behind for the past two decades how can they be on the cutting edge now if by the authors own secret this is not possible.

 

Then the author says that software will begin to be written for AMD because of Intels 7nm delay. What? I thought they said it could take 4 years for this shift. So far Intel said a 6-12 month delay could be expected for a 7nm rollout. And that their options are open to outsourcing manufacturing.

 

The author also said that TSMC gave AMD a large advantage, but why would this not also be an advantage for I tel if they used their 7nm products?

 

There is also zero mention of any other market segment. AMD competes with Intel in what 2 of 10 segments Or something like that?

 

Again, I’m not trying to pump Intel....this just comes across more as a hit piece than anything else. Time will tell. I guess AMD could pull it off in some way. They have good management which Intel lacks. From an investment standpoint to me it’s either Invest in Intel or don’t. AMD is way too hot to touch right now. Perhaps TWSC is worth a look in the short term. Not sure.

 

I thought this was phenomenal essay though it goes in and out on some topics that might cause confusion. Intel has historically been ahead in manufacturing (harder of the two to keep up at due to physics) and software (Intel's own and 3rd party). Intel lost its advantage wrt manufacturing, however its years of entrenchment in software allow it to be comparable or ahead on single thread tests. At multi-thread tests, Intel isn't doing as well. AMD's successes will start chipping away at Intel's success with software. The 4 year shift that's being referenced is likely a complete shift. Doesn't mean there aren't lower hanging fruit that can be tackled immediately further cementing AMD's lead.

 

As far as investments, AMD is very hot. Intel is a turnaround story with management way over their head and depleted talent ranks. 

[Castanza]

Great essay on Intel problems and larger implications:

 

https://medium.com/@gavin_baker/investing-mistakes-chapter-1001-9f0b7dbc6637

 

There is a lot of conjecture in that article. First the author says Intel’s chips are better in real world scenarios and their advantage is that the software is written for their x86 chipset. They go on to say that it would take at least four years of dominance from AMD (what dominance?) for coders to begin to write programs optimized for their architecture.

 

The. They contradict themselves and say...”once a company stops trying to compete at the leading edge they can never get it back.” Alluding that Intel’s chipset delay was a permanent impairment and they will never again catchup. Even though Intels 10nm chip is equivalent to AMD 7nm in terms of performance.

 

So is Intel ahead or behind? Do real world results matter or not? If AMD was behind for the past two decades how can they be on the cutting edge now if by the authors own secret this is not possible.

 

Then the author says that software will begin to be written for AMD because of Intels 7nm delay. What? I thought they said it could take 4 years for this shift. So far Intel said a 6-12 month delay could be expected for a 7nm rollout. And that their options are open to outsourcing manufacturing.

 

The author also said that TSMC gave AMD a large advantage, but why would this not also be an advantage for I tel if they used their 7nm products?

 

There is also zero mention of any other market segment. AMD competes with Intel in what 2 of 10 segments Or something like that?

 

Again, I’m not trying to pump Intel....this just comes across more as a hit piece than anything else. Time will tell. I guess AMD could pull it off in some way. They have good management which Intel lacks. From an investment standpoint to me it’s either Invest in Intel or don’t. AMD is way too hot to touch right now. Perhaps TWSC is worth a look in the short term. Not sure.

 

I thought this was phenomenal essay though it goes in and out on some topics that might cause confusion. Intel has historically been ahead in manufacturing (harder of the two to keep up at due to physics) and software (Intel's own and 3rd party). Intel lost its advantage wrt manufacturing, however its years of entrenchment in software allow it to be comparable or ahead on single thread tests. At multi-thread tests, Intel isn't doing as well. AMD's successes will start chipping away at Intel's success with software. The 4 year shift that's being referenced is likely a complete shift. Doesn't mean there aren't lower hanging fruit that can be tackled immediately further cementing AMD's lead.

 

As far as investments, AMD is very hot. Intel is a turnaround story with management way over their head and depleted talent ranks.

 

I read it a second time and there are solid point made. Generally I like Gavins writing a lot. I just think there could have been more analysis done on the two companies. Investing in Intel or AMD should consider the entire scope .

[Spekulatius]

I also like Gavin Baker, but you always wonder how much these writers know. He doesn’t really seem to have a lot of semiconductor experience, but he has been an investor in tech for a long time, so there is that. He probably also has inside sources.

 

The Cruz of the problem for Intel is that the Lithography at 193nm went towards their limits and it seems that ASML (and their lead customer TSM) decided to go for a quantum leap to 13.5nm (which is almost X-RAY) while Canon (and their lead customer Intel) tweaked the existing 193nm tech once more and that didn’t seem to work well.

 

It is also noteworthy that Canons next gen tech is not based not EUV, but a contact tech that Canon calls nanoimprint which works totally different and we don’t know if it works well at all yet.

https://global.canon/en/technology/interview/nanoimprint/index.html

 

 

Seems pretty bullish for ASML (but well reflected in the shareprice already ) because it could be that ASML becomes a monopolist for the most advanced lithography equipment for a while (and possibly a long while).

 

In a way that’s bullish for INTC though because if betting on the wrong horse in lithography is the main issue, they should be able to correct that with their next gen process by going along with ASML equipment. It’s likely more complex than that, but then again, as outsides, we only see the tip of an iceberg in terms of what’s really going on.

[Liberty]

I read it a second time and there are solid point made. Generally I like Gavins writing a lot. I just think there could have been more analysis done on the two companies. Investing in Intel or AMD should consider the entire scope .

 

It's not a stock pitch. It's not the point, or what he was trying to accomplish writing this. He's sharing an investment mistake, and why he made it. Doesn't have to go into all the details of why streaming SIMD extensions that are Intel-only provide an advantage because software makers will optimize for them more than AMD's because they have such a lower market share and such... At some point you can't go back to the big bang in every essay.

[scorpioncapital]

I'm not really sure he made a mistake. He's just upset he bought before a 16% drop after watching it for 20 years supposedly. Hardly known yet if it's a mistake... Also he has quite a few investor biases. One could argue his entire article is trying to justify his failure to jump in a few weeks later and pick the bottom? He thought his purchase would instantly zoom up?  On a side note, I'm a little confused about who is ahead. If Intel 10nm is equivalent to AMD 7nm and Intel 7nm is delayed but will come out eventually, is 7nm Intel equivalent to like 5nm AMD?

 

 

 

[lnofeisone]

I'm not really sure he made a mistake. He's just upset he bought before a 16% drop after watching it for 20 years supposedly. Hardly known yet if it's a mistake... Also he has quite a few investor biases. One could argue his entire article is trying to justify his failure to jump in a few weeks later and pick the bottom? He thought his purchase would instantly zoom up?  On a side note, I'm a little confused about who is ahead. If Intel 10nm is equivalent to AMD 7nm and Intel 7nm is delayed but will come out eventually, is 7nm Intel equivalent to like 5nm AMD?

 

This is getting really in the weeds but basically the whole 7nm vs. 10nm is a marketing shtick done well. If you look at the actual density, AMD's (TSMC's I suppose*) 7nm has density of about 66Mtr/mm^2 ( megatransitors/ squared millimiter) while Intel is trying to go for about 90 Mtr/mm^2 on its 10nm. But, density alone is also not a good indicator of an actual performance.

 

The actual performance of a chip will be be determined by clock speed (how many cycles a cpu can complete in a second) and instructions per clock (how many tasks a cpu can do in a cycle). So, there are generally two primary sources of improvements in processors: increase clock rates or improve instructions per second (IPC). You can improve clock speed by reducing the number of logic levels per cycle. Improving IPC can be done through increasing number of transistors or improving compiler technology.

 

Clock speeds are about the same for both. Intel's 10nm will have higher density than AMD's 7nm. And, because of its dominance, Intel's IPC is better (more optimized software). Hence, Intel's 10nm is better than AMD's 7nm. There are other consideration (e.g., security) but let's ignore them for now. Where AMD really shines is in its performance/$. In low and mid performance bands it's just not worth getting an Intel chip.

 

*I think this is important - TSMC 7nm is used by Apple, Qualcomm, etc. This means there is a broad consortium that stands behind TSMC. Intel is sort of on its own.

 

[scorpioncapital]

So a big issue is price? Is that why many switched to TSMC? I know people use ARM due to power/cost tradeoff. It seems a little like Fedex vs UPS or DHL. Premium service at a higher cost. And it better be premium. Is there a way for Intel to make better chips cheaper?

[lnofeisone]

So a big issue is price? Is that why many switched to TSMC? I know people use ARM due to power/cost tradeoff. It seems a little like Fedex vs UPS or DHL. Premium service at a higher cost. And it better be premium. Is there a way for Intel to make better chips cheaper?

 

Personally, I think Intel will continue losing market share. Intel is trying to keep up with features while coming in at a higher price. Intel's talent is leaving (as mentioned in the essay) and, this is my personal view, it is top ranks are lacking engineering firepower/background. This isn't a type of company/industry where financial engineering will win the day.

 

Where Intel stands to benefit is if there is a disruption in AMD's supply chain or a big misstep by AMD. Neither of these scenarios seem likely.

[cameronfen]

I don’t know much about intel. But the RISC vs CISC debate seems relevant if not already mentioned:

 

https://cs.stanford.edu/people/eroberts/courses/soco/projects/risc/risccisc/

 

https://medium.com/macoclock/interesting-remarks-on-risc-vs-cisc-microprocessors-1f034dca16ff

 

 

[cameronfen]

I don’t know much about intel. But the RISC vs CISC debate seems relevant if not already mentioned:

 

https://cs.stanford.edu/people/eroberts/courses/soco/projects/risc/risccisc/

 

https://medium.com/macoclock/interesting-remarks-on-risc-vs-cisc-microprocessors-1f034dca16ff

 

 

[Spekulatius]

I don’t know much about intel. But the RISC vs CISC debate seems relevant if not already mentioned:

 

https://cs.stanford.edu/people/eroberts/courses/soco/projects/risc/risccisc/

 

https://medium.com/macoclock/interesting-remarks-on-risc-vs-cisc-microprocessors-1f034dca16ff

 

The RISC vs CISC debate is probably at least 20 years old. I remember hearing about this when the Power PC processors came out. I think the lines between the two are too muddy anyways now to be meaningful, since Intel Chips have elements of both for example.

[cameronfen]

Yea but i think Intel’s competitive advantage is in CISC chips which are ill-suited for the current market.  I think that’s one reason other players damage to Intel’s moat have staying power.  Everyone is trying to move to a RISC architecture and that does not benefit Intel. 

[Liberty]

Yea but i think Intel’s competitive advantage is in CISC chips which are ill-suited for the current market.  I think that’s one reason other players damage to Intel’s moat have staying power.  Everyone is trying to move to a RISC architecture and that does not benefit Intel.

 

Intel is already on RISC and has been for a long time, though as a kind of hybrid (but that's not the reason why they're having problems):

 

"since the Pentium Pro (P6), Intel x86 processors have internally translated x86 CISC instructions into one or more RISC-like micro-operations, scheduling and executing the micro-operations separately"

[lnofeisone]

Yea but i think Intel’s competitive advantage is in CISC chips which are ill-suited for the current market.  I think that’s one reason other players damage to Intel’s moat have staying power.  Everyone is trying to move to a RISC architecture and that does not benefit Intel.

 

Intel is already on RISC and has been for a long time, though as a kind of hybrid (but that's not the reason why they're having problem):

 

"since the Pentium Pro (P6), Intel x86 processors have internally translated x86 CISC instructions into one or more RISC-like micro-operations, scheduling and executing the micro-operations separately"

 

^^This. CISC/RISC is largely an academic feud more than a market feud. Memory access is the real limit wrt CPUs which makes CISC/RISC a moot point. Also, Intel has switched to RISC (i860, i960, etc.) but exposed CISC to outside world for compatibility purposes. Even itanium (EPIC), I'd consider it more RISC-like. The switch wasn't particularly successful because it requires a really sophisticated compiler.

 

To echo Liberty - CISC/RISC is the least of their problems.