Tbh, I'm not convinced Intel's problem is the CEO. I mean, sure he oversees everything and he's the scapegoat when things go south. But Intel's woes seem to be more in the talent/overconfidence department. And I'm not that is for the CEO to fix.

bugTbh, I'm not convinced Intel's problem is the CEO. I mean, sure he oversees everything and he's the scapegoat when things go south. But Intel's woes seem to be more in the talent/overconfidence department. And I'm not that is for the CEO to fix.

From my understanding, a lot of it comes down to shitty company culture, but that is reading posts from disgruntled ex employees.

Intel should become a foundry for hire only so this would be a good CEO for that. I don't see a future for Intel in making their own integrated circuits. But I'm only basing that on decreasing revenue and market share with no near future roadmap that can reverse the trend. So don't let those pesky trends get in your way of bashing me for saying it.

DavenIntel should become a foundry for hire only so this would be a good CEO for that. I don't see a future for Intel in making their own integrated circuits. But I'm only basing that on decreasing revenue and market share with no near future roadmap that can reverse the trend. So don't let those pesky trends get in your way of bashing me for saying it.

I'll just leave this here: AMD was on the same trend back in Bulldozer days.

Pat Gelsinger became Intel CEO on 15 February, 2021. Intel now only has a few days left before they miss on the grand plan, "5C4Y".

bugI'll just leave this here: AMD was on the same trend back in Bulldozer days.

Not this again. Look, here at TPU we all have tons of knowledge regarding building computers but near zero knowledge about the market. This includes me. But my years of following tech has shown me at least one simple fact: there are MORE players now. Intel is NOT losing market share because of AMD. They are losing market share because of Apple, Nvidia, AMD, Qualcomm, Mediatek, Huawei, Fujitsu, Amazon, Ampere, etc. etc. This is DIFFERENT than the Bulldozer days and therefore NOT comparable.

TheLostSwedeFrom my understanding, a lot of it comes down to shitty company culture, but that is reading posts from disgruntled ex employees.

This. As I've said in other threads, Intel's problem is the same one that Boeing has: a company culture of engineering has been replaced by one of money-making, and after Gelsinger's ousting I'm doubtful it's possible for that culture to be fixed. Bad managers are incredibly good at making themselves difficult to fire, when you get enough of them they metastasise into a bloc that exists simply to ensure each member keeps its job, and once that bloc builds up enough political capital it becomes incredibly difficult to break.

Then from the actual engineering side, word has already got out among the job market that Intel is no longer a place to work at if you want to be able to build things, so the best engineers either don't apply to work there at all, or only stay a couple of years just to get the Intel name on their CV, and then move on.

Is the culture at TSMC better? It's only the inertia that keeps TSMC still looking good, but because you can't shrink the transistors indefinitely, and the Moore's law is dead, that same thing will strike all of them. It's matter of time. The sooner, the better.

3valatzyIs the culture at TSMC better? It's only the inertia that keeps TSMC still looking good, but because you can't shrink the transistors indefinitely, and the Moore's law is dead, that same thing will strike all of them. It's matter of time. The sooner, the better.

You can overcome this by stacking. Also there's photonic chips coming. Quantum computers is a thing. TSMC as a chip manufacturer has years and years of new ways to build chips ahead of them. This is why this is a good CEO pick for Intel and why Intel needs to move to chip fabbing only.

DavenYou can overcome this by stacking. Also there's photonic chips coming. Quantum computers is a thing. TSMC as a chip manufacturer has years and years of new ways to build chips ahead of them. This is why this is a good CEO pick for Intel and why Intel needs to move to chip fabbing only.

I don't agree. You can stack, but you can't add power requirements. I won't put a 300-watt furnace chip in my case. If you want to make 1000-watt, 2000-watt chips, good luck !

AssimilatorThis. As I've said in other threads, Intel's problem is the same one that Boeing has: a company culture of engineering has been replaced by one of money-making, and after Gelsinger's ousting I'm doubtful it's possible for that culture to be fixed. Bad managers are incredibly good at making themselves difficult to fire, when you get enough of them they metastasise into a bloc that exists simply to ensure each member keeps its job, and once that bloc builds up enough political capital it becomes incredibly difficult to break.

Then from the actual engineering side, word has already got out among the job market that Intel is no longer a place to work at if you want to be able to build things, so the best engineers either don't apply to work there at all, or only stay a couple of years just to get the Intel name on their CV, and then move on.

Money making is always good. Makes sure you have a job to come back to tomorrow.
The problem is when the money making ability gets disconnected from the technical ability*. Though I'm not sure that's what happens at Intel. It could be the management is disconnected, it could be they're connected, but fed bad data. But that's something we'll probably never find out.

*I'm experiencing this (on a much, much smaller scale) at my current job. A PO that has zero tech inclination. The moment you ask something remotely technical you know the next thing you'll hear is: "I don't understand, could you explain it a bit better".

3valatzyI don't agree. You can stack, but you can't add power requirements. I won't put a 300-watt furnace chip in my case. If you want to make 1000-watt, 2000-watt chips, good luck !

I'm more thinking that you can remove things from the core like L3 cache to make room for more cores or other circuitry. You can then stack the cache. Same can go for other parts of a chip like memory controllers, NPUs, etc. Stacking allows more room for something else on the same process node that doesn't necessarily take a ton more power.

AssimilatorThis. As I've said in other threads, Intel's problem is the same one that Boeing has: a company culture of engineering has been replaced by one of money-making, and after Gelsinger's ousting I'm doubtful it's possible for that culture to be fixed. Bad managers are incredibly good at making themselves difficult to fire, when you get enough of them they metastasise into a bloc that exists simply to ensure each member keeps its job, and once that bloc builds up enough political capital it becomes incredibly difficult to break.

Then from the actual engineering side, word has already got out among the job market that Intel is no longer a place to work at if you want to be able to build things, so the best engineers either don't apply to work there at all, or only stay a couple of years just to get the Intel name on their CV, and then move on.

Pat was hired on January 13, 2021. Here is Intel's revenue over that time period:


I'm sorry but Pat was a horrible CEO just from his failure of the basic duty of a CEO...make the company money.

3valatzyIs the culture at TSMC better? It's only the inertia that keeps TSMC still looking good, but because you can't shrink the transistors indefinitely, and the Moore's law is dead, that same thing will strike all of them. It's matter of time. The sooner, the better.

You can't compare TSMC and Intel. TSMC's "product" is its ability to manufacture complex chips for others, whereas Intel actually designs and makes their own chips.
No-one else have been able to do what TSMC does and we may never see another company like them, even if chips end up being made in an entirely different way in the future.
The company culture is also very different, as it's not a US company and the top level management doesn't have 100 millions dollar plus salaries (I guess this didn't quite apply to Pat) and massive stock incentives.
Are they still getting paid obscene money? Of course, but nowhere near what some US company CEO's earn.
This is a couple of years old, but both the CEO and the co-CEO of TSMC made less than US$20 million a year, after almost trebling their salaries. That said, salaries in Taiwan sucks, I know, as I've lived and worked there for 15 years.
www.taiwannews.com.tw/news/4883687

DavenI'm more thinking that you can remove things from the core like L3 cache to make room for more cores or other circuitry. You can then stack the cache. Same can go for other parts of a chip like memory controllers, NPUs, etc. Stacking allows more room for something else on the same process node that doesn't necessarily take a ton more power.


Pat was hired on January 13, 2021. Here is Intel's revenue over that time period:


I'm sorry but Pat was a horrible CEO just from his failure of the basic duty of a CEO...make the company money.

Short sigthness.
In chip industry, the impact of someone is counted >5years.

There has been 10 years of mismanagement at Intel. That missmanagement resulted in the awful 2024 years.

Pat invested big in the foundry to get back on track. Yes it cost money. Yes 2024 was abysmall, because of the invstement..

Imagine if 2025 turns well and 2026 even better, with a good 18A, similar (probably a bit behind) TSMN 2nm, and a good panther lake (on intel 18A)?

Will you think this will be due to the new CEO not yet around?

In a couple of years, we will be able to tell if Pat was a good CEO.

My take is that we will miss him. He was a true engineer, not a MBA guy who ruined Boeing.
The MBA guy makes money short term, and tank the company long term.
The engineer invest, cost money short term, but long term it s positive. That's what I am expected; I can be wrong. But certainly, we cannot know today.

bgxShort sigthness.
In chip industry, the impact of someone is counted >5years.

There has been 10 years of mismanagement at Intel. That missmanagement resulted in the awful 2024 years.

Pat invested big in the foundry to get back on track. Yes it cost money. Yes 2024 was abysmall, because of the invstement..

Imagine if 2025 turns well and 2026 even better, with a good 18A, similar (probably a bit behind) TSMN 2nm, and a good panther lake (on intel 18A)?

Will you think this will be due to the new CEO not yet around?

In a couple of years, we will be able to tell if Pat was a good CEO.

My take is that we will miss him. He was a true engineer, not a MBA guy who ruined Boeing.
The MBA guy makes money short term, and tank the company long term.
The engineer invest, cost money short term, but long term it s positive. That's what I am expected; I can be wrong. But certainly, we cannot know today.

Pat insulted suppliers and their countries, said Nvidia's success was due to luck, oversaw cancellation after cancellation of process nodes, allowed the abysmal failure of the 13th and 14th gen series, missed completely out on the GPU market spikes (AI and cryptocurrency), oversaw delay after delay of their GPUs with final launch marred by extremely poor driver execution and much, much more that I'm missing.

I know many here wanted an 'engineer' instead of an 'MBA' in charge of Intel. The engineer was worse! Way, way worse. That is reflected in the financials and the fact that HE WAS FIRED!

Edit: Oh and I forget the horrible failure that was Aurora and Ponte Vecchio.

Edit2: I posted this in the article about him being fired:
Intel CEO Pat Gelsinger Retires, Company Appoints two Interim co-CEOs | TechPowerUp
"A great CEO increases market share in existing markets, successfully enters markets with no previous market presence and creates new markets with paradigm shifting innovation. A good CEO does at least one of these. Pat tried and failed in ALL three. If you don't like Intel's outcome going into the second half of the current decade, please do not blame those of us who noticed it's not going well."

I sincerely hope Intel gets a better CEO than Pat. He was abyssmal.

Edit 3: Oh and I forget about the lackluster launch of Arrow Lake and Meteor Lake. Those were all Pat.

DavenI'm more thinking that you can remove things from the core like L3 cache to make room for more cores or other circuitry. You can then stack the cache. Same can go for other parts of a chip like memory controllers, NPUs, etc. Stacking allows more room for something else on the same process node that doesn't necessarily take a ton more power.

Yes and No, you still need real estate for the the TSVs and that is not a small amount of die space. Think the difference between Zen4 and Zen4c.

Intel are 50% of the way to doing this with the Foveros Interpose Die already on Meteor and Arrow lake but its on a really old node (22nm) so not sure how suitable it would be to adding cache onto it.


What I could possibly see coming from intel is a design in the near future where there is multiple Compute tiles on the high end processor parts for us gamers but being able to swap one or multiple out for NPUs/Graphic tiles for AI workloads/Handheld designs/General Purpose compute and then in their server parts its just lots of compute tiles meaning they can start to achieve the benefits of scale similar to how AMD have done with their chiplet designs.

IMO what they need to achieve to really gain dominance in the market would be to get their Foeveros tech to be able to have the interposer act as the IO die and then being able to plug in the different tiles that are required for the different SKUs out there. Also development in their software stack to take advantage of their NPUs in different areas. One thing I could see being a key area for this is in networking/security devices. Firewalls that utilise NPUs for IDS/VPN/NG Firewall aspects while still having the CPU horsepower to route packets at line speed.

bgxShort sigthness.
In chip industry, the impact of someone is counted >5years.

There has been 10 years of mismanagement at Intel. That missmanagement resulted in the awful 2024 years.

Pat invested big in the foundry to get back on track. Yes it cost money. Yes 2024 was abysmall, because of the invstement..

Imagine if 2025 turns well and 2026 even better, with a good 18A, similar (probably a bit behind) TSMN 2nm, and a good panther lake (on intel 18A)?

Will you think this will be due to the new CEO not yet around?

In a couple of years, we will be able to tell if Pat was a good CEO.

My take is that we will miss him. He was a true engineer, not a MBA guy who ruined Boeing.
The MBA guy makes money short term, and tank the company long term.
The engineer invest, cost money short term, but long term it s positive. That's what I am expected; I can be wrong. But certainly, we cannot know today.

DavenPat insulted suppliers and their countries, said Nvidia's success was due to luck, oversaw cancellation after cancellation of process nodes, allowed the abysmal failure of the 13th and 14th gen series, missed completely out on the GPU market spikes (AI and cryptocurrency), oversaw delay after delay of their GPUs with final launch marred by extremely poor driver execution and much, much more that I'm missing.

I know many here wanted an 'engineer' instead of an 'MBA' in charge of Intel. The engineer was worse! Way, way worse. That is reflected in the financials and the fact that HE WAS FIRED!

Edit: Oh and I forget the horrible failure that was Aurora and Ponte Vecchio.

@bgx See? It went right over his head.

Even AMD has bad company culture apparently hmm...

DavenPat insulted suppliers and their countries, said Nvidia's success was due to luck,

This is entirely valid criticism.

Davenoversaw cancellation after cancellation of process nodes, allowed the abysmal failure of the 13th and 14th gen series, missed completely out on the GPU market spikes (AI and cryptocurrency), oversaw delay after delay of their GPUs with final launch marred by extremely poor driver execution and much, much more that I'm missing.

This is not.

• The sunk-cost fallacy in trying to make obviously failed nodes work (refusal to pursue EUV), is one of the primary reasons Intel's foundry business - and the company as a whole - is in the s**t. Killing those projects was the right thing to do and might very well be the thing that saves the company. Yeah, nobody enjoys eating humble pie, but engineers are a lot better at it than managers.
• 13th and 14th gen are Skylake++++++ continuations that the company was forced into producing to keep its head above water, since they were waiting and hoping for the foundry side to come good so that they could fab newer and better designs. The alternative was not releasing a (perceived) new product and that is unacceptable to shareholders.
• Intel's dGPU effort was started long before Pat was CEO, he inherited Raja's mess and one of the things he did to try fixing it was fire Raja. I do suspect however that the dGPU division will be wound down by the next CEO, it's just not produced the results the company needs and it's a dangerous distraction at a time when they need to be focusing on their core competencies.

DavenI'm sorry but Pat was a horrible CEO just from his failure of the basic duty of a CEO...make the company money.

Not to speak about the clown Gelsinger being personally responsible for outright invalidating a 40% rebate over an actual amount of $15 billion USD!

HardwareTimes.com – Intel May Outsource Processor Chiplet Production Worth $15 Billion to TSMC Between 2024 and 2025

DavenEdit: Oh and I forget the horrible failure that was Aurora and Ponte Vecchio.

Yup, the years-long and over and over postponed Aurora NEXT-disaster, is fully to blame on Pat Gelsinger alone,
including the $600 million contractual fine as a penalty for breach of contract, for the overall damages caused by repeated delays for literal y-e-a-r-s!

Also the resulting multi-billion of costs (and heavy losses!) for Ponte Vecchio and for bringing PV to market knifing it immediately afterwards (the moment Aurora NEXT was delivered as per contract, even if it was late), when contractually promise a system consisting of devices, Intel by that time didn't even had sported any greater thought about on how to eventually realise in the first place – It's not that this already back-fired more than once in any past for Intel, like with the first Aurora from 2015–2018, or the first joke of Larrabee and its rehash Xeon Phi …

Their now complete line-up of everything HPC-accelerators being eventually knifed after being also repeatedly delayed, is also a direct result of that.

DavenPat insulted suppliers and their countries, said Nvidia's success was due to luck, oversaw cancellation after cancellation of process nodes, allowed the abysmal failure of the 13th and 14th gen series, missed completely out on the GPU market spikes (AI and cryptocurrency), oversaw delay after delay of their GPUs with final launch marred by extremely poor driver execution and much, much more that I'm missing.

I know many here wanted an 'engineer' instead of an 'MBA' in charge of Intel. The engineer was worse! Way, way worse. That is reflected in the financials and the fact that HE WAS FIRED!

Edit: Oh and I forget the horrible failure that was Aurora and Ponte Vecchio.

Edit2: I posted this in the article about him being fired:
Intel CEO Pat Gelsinger Retires, Company Appoints two Interim co-CEOs | TechPowerUp
"A great CEO increases market share in existing markets, successfully enters markets with no previous market presence and creates new markets with paradigm shifting innovation. A good CEO does at least one of these. Pat tried and failed in ALL three. If you don't like Intel's outcome going into the second half of the current decade, please do not blame those of us who noticed it's not going well."

I sincerely hope Intel gets a better CEO than Pat. He was abyssmal.

Edit 3: Oh and I forget about the lackluster launch of Arrow Lake and Meteor Lake. Those were all Pat.

One more edit for you: the last time Pat was in charge, we got the glorious NetBurst era of intel, along with the infamous Bribes they got hammered for.

One thing is happenstance, two is a coincidence, but for Pat, he's got a long track record of bad decisions surrounding him. Investing in fabs is good, but his leadership has been.....almost juvenile? Picking fights with TSMC, insulting your competition with nothing to show, and crucially a total loss of profit, actually LOSING money, with nothing to show.

Let's also not forget, Pat has been around long enough to see one of his fruits: a new CPU design, in arrow lake. And? Underwhelming at best. Sure, its made some improvements over the alderlake design and it's more efficient, but it sacrifices performance to do so, and much of that comes from being on TSMC instead of just intel. It's still losing to AMD and AMD is much better positioned to continue improvements.

Sure, Pat isnt the only problem. The entire management structure of Intel is extremely poor, and needed purged a decade ago. The current culture is not conductive to adapting to modern business. Frankly, that should be the CEO's real job, leadership. And in that regard, Pat has failed spectacularly. I mean, compare him to what Lisa su did in her first 4 years at AMD, the sheer difference in attitude and management and the delivery of product.

TheinsanegamerNOne more edit for you: the last time Pat was in charge, we got the glorious NetBurst era of intel, along with the infamous Bribes they got hammered for.

One thing is happenstance, two is a coincidence, but for Pat, he's got a long track record of bad decisions surrounding him. Investing in fabs is good, but his leadership has been.....almost juvenile? Picking fights with TSMC, insulting your competition with nothing to show, and crucially a total loss of profit, actually LOSING money, with nothing to show.

Let's also not forget, Pat has been around long enough to see one of his fruits: a new CPU design, in arrow lake. And? Underwhelming at best. Sure, its made some improvements over the alderlake design and it's more efficient, but it sacrifices performance to do so, and much of that comes from being on TSMC instead of just intel. It's still losing to AMD and AMD is much better positioned to continue improvements.

Sure, Pat isnt the only problem. The entire management structure of Intel is extremely poor, and needed purged a decade ago. The current culture is not conductive to adapting to modern business. Frankly, that should be the CEO's real job, leadership. And in that regard, Pat has failed spectacularly. I mean, compare him to what Lisa su did in her first 4 years at AMD, the sheer difference in attitude and management and the delivery of product.

Pat was also around when the Core 2 Duo did a number on AMD and won Apple over.
There's literally a picture of him holding a frame with a core 2 duo CPU. He also presented the arch at IDF 2006. Man, sometimes tech forums are really being too hardcore on the hate. As it's been said above, a lot of Intel trouble started way before Pat took the lead. He's a CEO, not the second coming of Jesus who could single handedly fix chips arch issues, software issues, fabs issues both CPU and GPU side. Reminder that MTL design started before he was CEO, various issued forced Intel to scramble together Raptor lake while they fix what prevented them to launch MTL. And arrow lake is an evolution of the MTL design...design that was initially planned to come after Alder lake in 2022. And alder lake itself was also delayed while Intel was in foundry hell, and refused to go external.
www.anandtech.com/show/1962

dyonoctisMan, sometimes tech forums are really being too hardcore on the hate.

Intel started it by threatening any company for over two decades that sold a non-Intel x86 processor based product. The threats ranged from total lock out of Intel processors to literal threats of violence against employees. Sorry, not sorry for the hate. Although, Pat does look cute in that photo. Who could be mad at that face?

So the person who couldn't keep GF competitive in leading edge, is going to lead intel. Excellent!

DavenIntel started it by threatening any company for over two decades that sold a non-Intel x86 processor based product. The threats ranged from total lock out of Intel processors to literal threats of violence against employees. Sorry, not sorry for the hate. Although, Pat does look cute in that photo. Who could be mad at that face?

I don't recall any stories about threats of violence. Do you have any sources?