Jump to content

Holy shit balls is Intel (INTC) Fucked?


Recommended Posts

26 minutes ago, Captainant said:

I've been hearing scuttlebutt that the new Intel mobile chips will have "ARM-like efficiency" so maybe their big bet on their own fabs will pay off yet

See the article I posted earlier on Intel's presentation of Lunar Lake. If there claims are accurate, it appears they are targeting Apple M3 in terms of performance efficiency, at least at the operating range for the thin and light devices these chips would be used in. It also appears to be a really big jump in GPU performance, allegedly surpassing AMD's latest and greatest integrated efforts.  

intel-lunar-lake-08-2-1440x630.jpeg

Edited by Dahobbs
Link to comment
Share on other sites

Rumors of Intel getting kicked out of the DJIA.  That would hurt it even further due to outflows from index and fund investing. It and Boeing are the 2 absolute dogs on the list in terms of financial performance but also stock and consumer sentiment.

Link to comment
Share on other sites

  • 2 weeks later...
1 hour ago, immamac said:

Among other things. They are splitting the fab and allowing outside investment in the spinco. 

Yeah, a subsidiary that can take in additional investors makes a lot more sense than the rumor of them selling off the fabs entirely. Intel retains control (presumably wouldn't let ownership get deluded below that point), but can bring in fresh investors more easily. It would also allow a separate leadership team to be more laser focused on the foundry business. 

Link to comment
Share on other sites

  • 2 months later...

Dont know what to think about this news. He went all in on 18A, so to leave this suddenly with no named successor and before the realization (good or bad)of his efforts is a bit odd. Stock is up slightly after the announcement so the market isnt spooked yet. 

Quote

Intel Corporation (NASDAQ: INTC) today announced that CEO Pat Gelsinger retired from the company after a distinguished 40-plus-year career and has stepped down from the board of directors, effective Dec. 1, 2024.

Intel has named two senior leaders, David Zinsner and Michelle (MJ) Johnston Holthaus, as interim co-chief executive officers while the board of directors conducts a search for a new CEO. Zinsner is executive vice president and chief financial officer, and Holthaus has been appointed to the newly created position of CEO of Intel Products, a group that encompasses the company’s Client Computing Group (CCG), Data Center and AI Group (DCAI) and Network and Edge Group (NEX). Frank Yeary, independent chair of the board of Intel, will become interim executive chair during the period of transition. Intel Foundry leadership structure remains unchanged.

https://www.businesswire.com/news/home/20241202016400/en/Intel-Announces-Retirement-of-CEO-Pat-Gelsinger

Link to comment
Share on other sites

9 minutes ago, immamac said:

Seems like he said fuck you guys I got chips act over the line its not my problem anymore and peaced out. 

No transition. 

He's just an MBA trying to run a complicated chip company! You cant expect him to make long term decisions for the good of the company - HES A SHAREHOLDER GODDAMNIT! They've gotta get paid!

That's why Intel bought back billions in their own shares instead of investing in their process, and had to beg for corporate welfare

Shareholder supremacy tends to rot a company away

Edited by Captainant
  • Hook 'Em 1
Link to comment
Share on other sites

3 minutes ago, Captainant said:

He's just an MBA trying to run a complicated chip company! You cant expect him to make long term decisions for the good of the company - HES A SHAREHOLDER GODDAMNIT! They've gotta get paid!

That's why Intel bought back billions in their own shares instead of investing in their process, and had to beg for corporate welfare

Shareholder supremacy tends to rot a company away

That's not gelsinger, he was a very well respected engineer and by far the most technical CEO in a long time. It's why the stock got punished. 

Link to comment
Share on other sites

6 minutes ago, immamac said:

That's not gelsinger, he was a very well respected engineer and by far the most technical CEO in a long time. It's why the stock got punished. 

Ah fair enough, he tried to stop the trend of buybacks when he took the helm in 2021 - and he got punished for it. But it's not the technical folks driving the ship at Intel, and it hasn't been for pretty much the whole 21st century 

Edit: and that he's been pushed out effective immediately tells me this was not a planned or coordinated exit. Dude got axed 

Edited by Captainant
Link to comment
Share on other sites

9 minutes ago, Captainant said:

Ah fair enough, he tried to stop the trend of buybacks when he took the helm in 2021 - and he got punished for it. But it's not the technical folks driving the ship at Intel, and it hasn't been for pretty much the whole 21st century 

Edit: and that he's been pushed out effective immediately tells me this was not a planned or coordinated exit. Dude got axed 

Yeah, he spent too much time at Intel to fuck them over, and a 7:30AM press release that the CEO is "retiring" at the age of 63 is not how Intel would do shit. I suspect we'll hear more in the coming weeks. 

Link to comment
Share on other sites

17 minutes ago, immamac said:

That's not gelsinger, he was a very well respected engineer and by far the most technical CEO in a long time. It's why the stock got punished. 

Agreed, and he was at Intel when it used to be an engineering company.  It's why I gave him a chance to turn it around albeit I knew it would take time.

This is nuts and I assume details will start to come out at some point.  I hope he's not sick.

Link to comment
Share on other sites

Looks like this was a good old-fashioned firing.

https://www.cnbc.com/2024/12/02/intel-ceo-pat-gelsinger-is-out.html

I have no idea how long the current Intel BOD has been around, but this Yeary guy has been on the board since 2009, so he had a front row seat for the downfall and likely played a role through their incompetent CEO hirings and missed executions of the past.

Spoiler

Intel ousted CEO Pat Gelsinger over the weekend, capping a tumultuous nearly four-year tenure at what was America’s leading semiconductor company before its stock price and market share collapsed.

The company announced Gelsinger’s resignation Monday morning, which a person familiar with the matter said came after a contentious board meeting last week over Gelsinger’s perceived failure to respond to Nvidia’s competitive edge and a lack of confidence in Gelsinger’s turnaround plans.

Intel CFO David Zinsner and Intel products CEO MJ Holthaus were named interim co-CEOs. Longtime board member Frank Yeary will serve as interim executive chair. Shares of Intel were down 2% Monday afternoon.

“We are working to create a leaner, simpler, more agile Intel,” Yeary said. Yeary was a key driver in Gelsinger’s ouster, said the person, who requested anonymity to discuss confidential board proceedings freely. Intel did not immediately return a request for comment on the board meeting.

Yeary, Intel’s longest-serving board member, will now have to preside over yet another CEO search process. Gelsinger, 63, had an illustrious career at Intel, rising to become the company’s first chief technical officer at the turn of the century, before he took a senior role at EMC. Gelsinger returned to the company from VMware, where he was chief executive, to stabilize Intel in 2021, succeeding then-CEO Bob Swan.

“It has been a challenging year for all of us as we have made tough but necessary decisions to position Intel for the current market dynamics,” Gelsinger said in a press release.

The board meeting that led to Gelsinger’s ouster was first reported by Bloomberg.

Gelsinger set out an audacious plan when he arrived in 2021 to transform the languishing company into a chipmaking juggernaut. He sought to achieve parity with the two leading chipmakers, Samsung and Taiwan Semiconductor Manufacturing Co. He pursued big buildouts in the U.S. and around the world, a costly endeavor that weighed heavily on Intel’s free cash flow and increased the company’s debt load.

He also wooed government investment, positioning Intel as the single-largest beneficiary of the U.S. CHIPS and Science Act. Government money has begun to flow to Intel in recent weeks and will aid the company’s chip fabs in Arizona and Ohio. Gelsinger’s retirement comes a week after Intel and the CHIPS and Science Act office finalized a $7.86 billion grant.

Gelsinger also moved to position the company as vital to U.S. national security. He won a multibillion-dollar contract with the Department of Defense to build secure chips, and in meetings with analysts and prospective customers stressed that Intel was a trusted partner of the U.S. government.

But all that was not enough to assuage investors, who increasingly began to see Intel’s aggressive spending as a folly.

 

Link to comment
Share on other sites

IMO this board seems like they're mad they missed the AI boat, and they fired Gelsinger because he wasn't able to move mountains and fix Intel's old woes around process stagnation.

They ate the seed stock a decade plus ago when they stopped improving node processes and just did 14nm+++++ to keep R&D costs low so they could continue to fund buybacks and dividends. Things were great, until they weren't. 

 

Now Intel is chasing AMD's innovations and their heavily capitalized fabs are a millstone around their necks

  • Hook 'Em 1
Link to comment
Share on other sites

Gelsinger is by far the most technical CEO Intel has had in some time.  

  • Key architect of the 386 - which bailed out the 432 failure, post 286
  • Chief architect of the 486
  • Key technical leader of the CPU group, then company CTO, until they pushed him out during the Larrabee program

When he returned, he wanted to recreate a Grove-ian culture, which is good prima facie, but ignores many past realities that no longer exist. 

  1. Intel was one of about 4 groups developing CPUs in the 80s and 90s: Intel, IBM, DEC, Motorola.  Now, Apple, Arm, Amazon, AMD, nVidia, Google, MIPs, and a bevy of RISC-V startups all compete for CPU design talent.  Each are "sexier" than Intel in some form or fashion,
  2. In the late 1990s to around 2014, Intel was 1.5-2 generations ahead of all other companies in process technology.  That allowed them to develop higher performing, lower power, lower area CPUs than any of the competitors.
  3. Intel had superior microarchitectures to all competitors - and this is largely due to #1 and #2.
  4. Intel played hardball with the box movers (Dell, HP, etc.) when those companies entertained alternatives (AMD, VIA, Transmeta, Cyrix). 

None of those advantages exist today.  I suggest that Intel succeeded despite Grove's approaches and processes, not because of them.  Or his approaches helped manage a complex and growing organization that benefitted from competitive advantages.  Bringing back Grove's OKR approach to program and people management was not going to solve the problems, but it sounded good.

IMO, focusing on fixing the FABs/process was the right approach.  Client market share was eroding to AMD and Apple.  Server share was eroding to AMD, Arm, and the vertical integrators.  GPU has always been second and third tier (to nVidia, AMD/ATI, imagination, S3).  Some of that market erosion is b/c of FAB problems.  Get that fixed, get external customers to the fab, AND get subsidies for on-shoring of semiconductor supply chain.  Profit.  I think they planned as if 2021 levels of revenue would persist through 2024 - and the CFO often noted "this was not in our models" in quarterly earnings discussions - not good.

I believe Gelsinger was pushed out once the CHIPs check cleared.  The market has punished this decision to the tune of $10B since an intraday high of $25 yesterday, closing at $22.47 today.

  • Hook 'Em 4
Link to comment
Share on other sites

12 minutes ago, TwiceHorn said:

Templeton is a EE, but in my opinion, did a fairly incredible job of steering TI from a ponderous, diverse electronics/SC manufacturer into a fairly lean and targeted company that's still a technical leader with some commodity products.

I have not followed TXN very closely since my father worked there when I was in MS and HS, which was fairly long ago.  I recall the 99/4 and Professional Computer from that era :)

What did Templeton do that made them more lean and targeted?

Also, when Intel announced 15k layoffs in August, several folks in my network wondered if Gelsinger wasn't significantly undercalling the required streamlining.  Many large companies have cultures that prefer # of people over quality of people and everyone believes they need more people - not more highly skilled people.

Link to comment
Share on other sites

On 12/2/2024 at 3:47 PM, Captainant said:

IMO this board seems like they're mad they missed the AI boat, and they fired Gelsinger because he wasn't able to move mountains and fix Intel's old woes around process stagnation.

They ate the seed stock a decade plus ago when they stopped improving node processes and just did 14nm+++++ to keep R&D costs low so they could continue to fund buybacks and dividends. Things were great, until they weren't. 

 

Now Intel is chasing AMD's innovations and their heavily capitalized fabs are a millstone around their necks

 

2 hours ago, boilerhorn said:

Gelsinger is by far the most technical CEO Intel has had in some time.  

  • Key architect of the 386 - which bailed out the 432 failure, post 286
  • Chief architect of the 486
  • Key technical leader of the CPU group, then company CTO, until they pushed him out during the Larrabee program

When he returned, he wanted to recreate a Grove-ian culture, which is good prima facie, but ignores many past realities that no longer exist. 

  1. Intel was one of about 4 groups developing CPUs in the 80s and 90s: Intel, IBM, DEC, Motorola.  Now, Apple, Arm, Amazon, AMD, nVidia, Google, MIPs, and a bevy of RISC-V startups all compete for CPU design talent.  Each are "sexier" than Intel in some form or fashion,
  2. In the late 1990s to around 2014, Intel was 1.5-2 generations ahead of all other companies in process technology.  That allowed them to develop higher performing, lower power, lower area CPUs than any of the competitors.
  3. Intel had superior microarchitectures to all competitors - and this is largely due to #1 and #2.
  4. Intel played hardball with the box movers (Dell, HP, etc.) when those companies entertained alternatives (AMD, VIA, Transmeta, Cyrix). 

None of those advantages exist today.  I suggest that Intel succeeded despite Grove's approaches and processes, not because of them.  Or his approaches helped manage a complex and growing organization that benefitted from competitive advantages.  Bringing back Grove's OKR approach to program and people management was not going to solve the problems, but it sounded good.

IMO, focusing on fixing the FABs/process was the right approach.  Client market share was eroding to AMD and Apple.  Server share was eroding to AMD, Arm, and the vertical integrators.  GPU has always been second and third tier (to nVidia, AMD/ATI, imagination, S3).  Some of that market erosion is b/c of FAB problems.  Get that fixed, get external customers to the fab, AND get subsidies for on-shoring of semiconductor supply chain.  Profit.  I think they planned as if 2021 levels of revenue would persist through 2024 - and the CFO often noted "this was not in our models" in quarterly earnings discussions - not good.

I believe Gelsinger was pushed out once the CHIPs check cleared.  The market has punished this decision to the tune of $10B since an intraday high of $25 yesterday, closing at $22.47 today.

Yeah, this seems like a short sighted decision by the board. The replacement will tell us a lot about who is really to blame for Intel's problems. My guess is they hire another finance guy to try and work some short term magic while neglecting the core business assets. 

Link to comment
Share on other sites

2 hours ago, boilerhorn said:

I have not followed TXN very closely since my father worked there when I was in MS and HS, which was fairly long ago.  I recall the 99/4 and Professional Computer from that era :)

What did Templeton do that made them more lean and targeted?

Also, when Intel announced 15k layoffs in August, several folks in my network wondered if Gelsinger wasn't significantly undercalling the required streamlining.  Many large companies have cultures that prefer # of people over quality of people and everyone believes they need more people - not more highly skilled people.

Well, perhaps you recall (my Dad worked there from 57-86), their business included SC manufacturing, mostly DRAM and SRAM, defense systems, calculators, analog semiconductor devices, clad metals and kind of a wide range of shit.

They started spinning stuff off, namely defense systems to Raytheon, then memory operations to Micron (they were only competing by extracting licenses from Asia, Inc.) and so on.  I don't believe Templeton presided over most of that, but he was there to refocus what was left.

Templeton reduced them to educational products/calculators, analog SC devices, and embedded processors/ASICs/SOCs.

Their stock has done very, very well simply on the strength of a well-run company that's a leader in its several fields.

Edited by TwiceHorn
  • Hook 'Em 2
Link to comment
Share on other sites

50 minutes ago, TwiceHorn said:

They started spinning stuff off, namely defense systems to Raytheon, then memory operations to Micron (they were only competing by extracting licenses from Asia, Inc.) and so on.  I don't believe Templeton presided over most of that, but he was there to refocus what was left.

Templeton reduced them to educational products/calculators, analog SC devices, and embedded processors/ASICs/SOCs.

Their stock has done very, very well simply on the strength of a well-run company that's a leader in its several fields.

I worked at TI during that era- in the  Semiconductor group in Houston (DSPs specifically).

The CEO (Jerry Junkins) died the same week I started, and Rich Templeton took over the Semiconductor group at that point... selling off everything non-SC. I'd graduated from college 3 weeks earlier so I had no idea what good leadership/management looked like at the time but he seemed sharp.

  • Hook 'Em 2
Link to comment
Share on other sites

1 hour ago, TwiceHorn said:

Well, perhaps you recall (my Dad worked there from 57-86), their business included SC manufacturing, mostly DRAM and SRAM, defense systems, calculators, analog semiconductor devices, clad metals and kind of a wide range of shit.

They started spinning stuff off, namely defense systems to Raytheon, then memory operations to Micron (they were only competing by extracting licenses from Asia, Inc.) and so on.  I don't believe Templeton presided over most of that, but he was there to refocus what was left.

Templeton reduced them to educational products/calculators, analog SC devices, and embedded processors/ASICs/SOCs.

Their stock has done very, very well simply on the strength of a well-run company that's a leader in its several fields.

My dad was based in Austin from 1978-1985 in the business/planning domains.  I had forgotten about all of the ancillary businesses they had at the time.  Templeton's decision to focus was a good idea.

99/4 and 99/4a were fantastic for me to play with.  I saved my money to buy a patch cord to allow me to save programs to a cassette tape.  A friend's father was a SW developer and was able to procure a floppy drive.  The speedup was amazing.

And who can forget the Little Professor and Speak and Spell? :)

Non sequitur, somewhat...
In the mid-late 80s, UT ECE hired a few industry luminaries into its department.  Harvey Cragon left TI to join the UT faculty; he was a fully-tenured, chaired professor with only a bachelor's degree.  Cragon worked with Jack Kilby on an early TI computer for the air force, on TI's supercomputer, and later its DSPs.  I took several of his classes and kept in touch with him until around 2014.  He passed away in 2018.

  • Hook 'Em 1
Link to comment
Share on other sites

4 hours ago, boilerhorn said:

Gelsinger is by far the most technical CEO Intel has had in some time.  

  • Key architect of the 386 - which bailed out the 432 failure, post 286
  • Chief architect of the 486
  • Key technical leader of the CPU group, then company CTO, until they pushed him out during the Larrabee program

When he returned, he wanted to recreate a Grove-ian culture, which is good prima facie, but ignores many past realities that no longer exist. 

  1. Intel was one of about 4 groups developing CPUs in the 80s and 90s: Intel, IBM, DEC, Motorola.  Now, Apple, Arm, Amazon, AMD, nVidia, Google, MIPs, and a bevy of RISC-V startups all compete for CPU design talent.  Each are "sexier" than Intel in some form or fashion,
  2. In the late 1990s to around 2014, Intel was 1.5-2 generations ahead of all other companies in process technology.  That allowed them to develop higher performing, lower power, lower area CPUs than any of the competitors.
  3. Intel had superior microarchitectures to all competitors - and this is largely due to #1 and #2.
  4. Intel played hardball with the box movers (Dell, HP, etc.) when those companies entertained alternatives (AMD, VIA, Transmeta, Cyrix). 

None of those advantages exist today.  I suggest that Intel succeeded despite Grove's approaches and processes, not because of them.  Or his approaches helped manage a complex and growing organization that benefitted from competitive advantages.  Bringing back Grove's OKR approach to program and people management was not going to solve the problems, but it sounded good.

IMO, focusing on fixing the FABs/process was the right approach.  Client market share was eroding to AMD and Apple.  Server share was eroding to AMD, Arm, and the vertical integrators.  GPU has always been second and third tier (to nVidia, AMD/ATI, imagination, S3).  Some of that market erosion is b/c of FAB problems.  Get that fixed, get external customers to the fab, AND get subsidies for on-shoring of semiconductor supply chain.  Profit.  I think they planned as if 2021 levels of revenue would persist through 2024 - and the CFO often noted "this was not in our models" in quarterly earnings discussions - not good.

I believe Gelsinger was pushed out once the CHIPs check cleared.  The market has punished this decision to the tune of $10B since an intraday high of $25 yesterday, closing at $22.47 today.

Good summary. Key emphasis on their fab lead and "hardball" as you put it. The process technology was their crown jewel.  As one ex-Intel exec told me "Intel is a manufacturing company that makes products to fill their fabs".  When TSMC caught up and surpassed, it allowed companies with better designs to shine. Their designs weren't all that special. They did a great job of finding ways to lock customers into their technology.

And "hardball" is why this has taken so long to come to fruition. The board and execs whistled past the graveyard for a decade taking their short term profits while threatening and bribing everyone in their way, completely ignoring the bigger picture. 

I thought Gelsinger had a chance to get them back by focusing on the Fabs, but it seems like the board wants short term returns. Good luck with that. 

  • Hook 'Em 1
Link to comment
Share on other sites

4 hours ago, boilerhorn said:

My dad was based in Austin from 1978-1985 in the business/planning domains.  I had forgotten about all of the ancillary businesses they had at the time.  Templeton's decision to focus was a good idea.

99/4 and 99/4a were fantastic for me to play with.  I saved my money to buy a patch cord to allow me to save programs to a cassette tape.  A friend's father was a SW developer and was able to procure a floppy drive.  The speedup was amazing.

And who can forget the Little Professor and Speak and Spell? :)

Non sequitur, somewhat...
In the mid-late 80s, UT ECE hired a few industry luminaries into its department.  Harvey Cragon left TI to join the UT faculty; he was a fully-tenured, chaired professor with only a bachelor's degree.  Cragon worked with Jack Kilby on an early TI computer for the air force, on TI's supercomputer, and later its DSPs.  I took several of his classes and kept in touch with him until around 2014.  He passed away in 2018.

Also Willis Adcock.  Yeah, I had forgotten, but I guess Austin was the "PC" operation, IIRC.  Or maybe that was Houston, Compaq sprang from it.

TI took an early license to the transistor from Bell Labs and wound up "making a market" for it by helping Sony and others design transistor radios.  For better or worse, they saw that raised their public profile, and so was always dabbling in consumer products that made use of their technologies.  With the demise of HP and the use of graphing calculators in education, they've actually got a pretty thriving market there and some dominance.

  • Hook 'Em 1
Link to comment
Share on other sites

On 12/2/2024 at 3:17 PM, Captainant said:

He's just an MBA trying to run a complicated chip company! You cant expect him to make long term decisions for the good of the company - HES A SHAREHOLDER GODDAMNIT! They've gotta get paid!

Shareholder supremacy tends to rot a company away

 

12 hours ago, boilerhorn said:

Gelsinger is by far the most technical CEO Intel has had in some time.  

 

Another knee-jerk anti-capitalism screed turns out to be just shooting from the hips.  Many such examples.

Link to comment
Share on other sites

2 hours ago, 52-80 said:

Another knee-jerk anti-capitalism screed turns out to be just shooting from the hips.  Many such examples.

Well the strategy of eating their seed stock is really working out great for them, scout. I'm sure you can find a FRED chart to explain it. Sorry I'm not perfect to every detail, please forgive my trespass into your domain oh great and wealthy and mighty seer of the finance bro 

Link to comment
Share on other sites

5 hours ago, 52-80 said:

Another knee-jerk anti-capitalism screed turns out to be just shooting from the hips.  Many such examples.

Not sure how my comments were either anti-capitalist or a screed.  Please elaborate.

Re: AI.
Around 2015-2016, there were dozens, if not hundreds, of AI startups in the marketplace.  Many of them were focusing on specialized domain-specific architectures for accelerating AI workloads.  Some focused on inference, some on training, some tried to do both.  For most, the software problem was secondary.  "We have this great new architecture with peak FLOPs of X."  The assumption was that SW developers would flock to the best HW.  However, for most of these firms, the SW was indeed the constraint.  New models were being developed at breakneck speed; internal SW teams could not keep up, as the "we'll just deploy an army of SW folks to optimize to our HW" mindset failed.  On top, nVidia had CUDA.  While not perfect, it was low friction.  It remains low friction today.

Intel had a multi-pronged approach to AI.  First, expand the x86 ISA - AVX, AVX512, AMX, etc.  To the CPU architect every problem has a CPU/ISA solution - zero thought of adoption in the SW community and the inherent limitations of CPU vs. GPU in peak/effective performance.  "If we build it, they will come."  Second, Intel acquired several AI firms: Habana, Nervana, Movidius, Mobileye - for around $18B (Mobileye was $15B of that).  Each of them - in some way - led the field in performance.  For example, Habana had fantastic training performance numbers in MLPerf.  Nervana is (more or less gone), Movidius is used in Intel's vision processing units, Mobileye was spun off, and trades at $13.9B.  Habana created Gaudi, but the market acceptance has been low.  Moreover, most key parts of the Habana team - innovators, developers, etc. - have left Intel.

It seemed as though the strategy was "buy all the big names and see how it goes."  

  • Hook 'Em 2
Link to comment
Share on other sites

8 minutes ago, boilerhorn said:

Around 2015-2016, there were dozens, if not hundreds, of AI startups in the marketplace.  Many of them were focusing on specialized domain-specific architectures for accelerating AI workloads.  Some focused on inference, some on training, some tried to do both.  For most, the software problem was secondary.  "We have this great new architecture with peak FLOPs of X."  The assumption was that SW developers would flock to the best HW.  However, for most of these firms, the SW was indeed the constraint.  New models were being developed at breakneck speed; internal SW teams could not keep up, as the "we'll just deploy an army of SW folks to optimize to our HW" mindset failed.  On top, nVidia had CUDA.  While not perfect, it was low friction.  It remains low friction today.

Intel had a multi-pronged approach to AI.  First, expand the x86 ISA - AVX, AVX512, AMX, etc.  To the CPU architect every problem has a CPU/ISA solution - zero thought of adoption in the SW community and the inherent limitations of CPU vs. GPU in peak/effective performance.  "If we build it, they will come."  Second, Intel acquired several AI firms: Habana, Nervana, Movidius, Mobileye - for around $18B (Mobileye was $15B of that).  Each of them - in some way - led the field in performance.  For example, Habana had fantastic training performance numbers in MLPerf.  Nervana is (more or less gone), Movidius is used in Intel's vision processing units, Mobileye was spun off, and trades at $13.9B.  Habana created Gaudi, but the market acceptance has been low.  Moreover, most key parts of the Habana team - innovators, developers, etc. - have left Intel.

Yeah honestly the biggest differentiator between chip firms has been the software - even AMD has recently announced that their focus will be on software rather than hardware. Nvidia's work on not just researching new fields of mathematics but building in hardware implementations to solve those systems is the biggest part of why they OWN the AI space. And then with the software layer it becomes very very difficult to move off of a NV platform once you've utilized some of their built in optimizations.

11 minutes ago, boilerhorn said:

It seemed as though the strategy was "buy all the big names and see how it goes."  

Hey man, it's what their highly paid consultants suggested they do! It's not the board's fault the strategy didn't work

  • Hook 'Em 1
Link to comment
Share on other sites

2 hours ago, boilerhorn said:

Not sure how my comments were either anti-capitalist or a screed.  Please elaborate.

Its the other guy, who thinks gelsinger is an MBA because anything bad that happens is because of business or public markets. 

Link to comment
Share on other sites

1 hour ago, 52-80 said:

Its the other guy, who thinks gelsinger is an MBA because anything bad that happens is because of business or public markets. 

My comment was broadly about the leadership and decision making that brought them to their current state - but your obsession is noted. By the way, congrats on posting about me multiple times without using a slur!

Link to comment
Share on other sites

This is a really solid write up on where Intel is at, product wise. They've got abandoned products from 2017 that are still offers of magnitude better than the completion today, but lack the leadership and long term vision to do anything more than stock buybacks

https://blog.ameri.coffee/intel-becoming-known-for-past-glories

The news yesterday that Pat Gelsinger has been forced out of Intel can only mean one of two things:

Intel's upcoming 18A node (which Intel promised would leapfrog TSMC) is having problems, is delayed, or is not meeting performance or yield expectations; or

Intel's board has given in to pressure from short-term-focused investors and is going to stop investing in R&D and in cutting-edge nodes to meet short-term profitability goals.

Neither of these is good.

Link to comment
Share on other sites

Dear Intel Board,
Great move pushing out Gelsinger without a plan.  Since you announced the change of leadership, Intel has shed roughly 15% of its market cap.
Congrats,
The Shareholder

That aside, part of Intel's winning culture involved eschewing lower margin, industry leading technologies in favor of higher margin domains.  Grove's famous pivot from DRAM to CPU is why Intel was what it was in the 80s and 90s.   I suspect that's why the SSD technologies, etc. were abandoned.  AND, they can have cyclical major downturns.

For what it's worth, the 18A node is yielding at 10%.  That alone could have pushed Gelsinger out, esp. since there were earlier quotes about it being at 50%-60%.  That said, yield is a function of area, custom/stdcell/SRAM, but if he's been claiming 50% to the board and someone has a legitimate report of 10%, he's gone.

As @Captainant article notes, CEO BK was an abject disaster.  Everything he did was about efficiency, etc. and he gutted/reduced/etc, several strong R/D teams in his time.  For example, the entirety of the beginning of MSFT's SOC/silicon team came from Intel (2016).  Ampere was started - and largely staffed - by a former Intel exec in 2017.  Google, AMD, nVidia, etc. have built (or are building) design centers near key Intel development centers.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.



×
×
  • Create New...