The Fall of Intel

[Cmaier]

Humans are humans, even engineers . Humans tend to care about symbolism, perhaps much more than we should, but we do. A huge amount of leadership isn’t just the practical decisions you make, but motivating people to carry them out.

In all my years at Intel’s competitors, I never saw any kind of petty concerns like that. We cared about the value of our options, the ineptitude of management, etc. When Jerry Sanders wore a pinstriped suit with the words “Jerry Sanders” as the pinstripes, we were bemused, but not upset about it. On the other hand, when our $42/share stock was worth $4 and, in a dressing that, he complained that he, too, was suffering because his wife couldn’t afford to furnish his second mansion, that pissed us off.

 

[dada_dave]

In all my years at Intel’s competitors, I never saw any kind of petty concerns like that. We cared about the value of our options, the ineptitude of management, etc. When Jerry Sanders wore a pinstriped suit with the words “Jerry Sanders” as the pinstripes, we were bemused, but not upset about it. On the other hand, when our $42/share stock was worth $4 and, in a dressing that, he complained that he, too, was suffering because his wife couldn’t afford to furnish his second mansion, that pissed us off.

The way I read it is:

The more important consideration is that he lacked charisma and, in the speech, a clear vision while intimating that more people would be losing their jobs. Merely the cherry on top is that his clothing choice reinforces that he is doing this as an outsider. Had he been delivering a great address that resonated with his audience, or at least had the address been delivered under better circumstances for Intel, probably that would’ve been dismissed.

It’s nowhere near as egregious as pleading poverty because his wife can’t furnish their second mansion, but it’s still a leadership faux pas under the circumstances. In your example, Jerry Sanders’ mistake was trying to be relatable, trying to say “we’re all in this together”, in a way that reinforced just how out of touch he was. In this case, Lip-Bu Tan either carelessly, or deliberately, signaled “we’re not in this together” in a small way. That’s a little different from the pin striped suit example (depending on the circumstances in the suit which I do not know) even if it doesn’t rise to the former’s height of face planting.

 

[exoticspice1]

@Cmaier , what makes Intel different from AMD?

Do you like the performance of Lisa Su and culture at AMD and how does that compare to Intel? Sorry about the long questions, its just you strong opinions regarding Intel

 

[Cmaier]

@Cmaier , what makes Intel different from AMD?

Do you like the performance of Lisa Su and culture at AMD and how does that compare to Intel? Sorry about the long questions, its just you strong opinions regarding Intel

I don’t know anything about current AMD culture. About two years ago I heard from an old colleague who told me it’s had its ups and downs since I left.

Primary, though, it was always smaller, feistier, and more entrepreneurial than Intel. Much smaller teams consisting of much more talented engineers who are more flexible in terms of understanding multiple aspects of design. I worked on the circuit design, EDA, methodology, logic design, and physical design teams at various points, sometimes more than one simultaneously. And that wasn’t too uncommon. For opteron many of us did instruction set design, logic design, and physical design of our blocks. Some also did circuit design. Others also designed the power grid or the clock network or the standard cell architecture. I worked on all that. That would never happen at Intel. And we moved from block to block depending on the chip. I did an instruction scheduler, integer ALUs, floating point unit, etc. We all had to understand the entire chip, not just keep designing an adder over and over again for our entire careers.

 

[dada_dave]

Pat Gelsinger becomes executive chairman, head of technology at church-focused platform Gloo

Any hardware projects? Probably no.

www.tomshardware.com

Pat Gelsinger, former CEO of Intel and VMWare, announced on Monday that his role at faith-focused technology company Gloo has been expanded, and he will become the company's executive chairman as well as head of technology who will be in charge of products development. After nearly 10 years as a board member and investor, he is now leading Gloo's product and engineering efforts, with a focus on building a vertical cloud platform for the faith ecosystem.

He’s found his calling?

 

[Cmaier]

Pat Gelsinger becomes executive chairman, head of technology at church-focused platform Gloo

Any hardware projects? Probably no.

www.tomshardware.com

He’s found his calling?

how many times on here did i say this guy was way too worried about converting us silicon valley pagans to his stupid religion and not worried enough about doing his damned job?

 

[dada_dave]

how many times on here did i say this guy was way too worried about converting us silicon valley pagans to his stupid religion and not worried enough about doing his damned job?

I’m just wondering what a vertical cloud platform for a faith ecosystem actually is … heaven? (I know that in the article they sort of describe some sort AI chat bot for religious people, but seriously the article writer had to write that with a smile.)

Also … Gloo?

 

[Agent47]

I’m just wondering what a vertical cloud platform for a faith ecosystem actually is … heaven?

Wouldn't that have to be horizontal?

 

[Yoused]

The thing I fail to get is the "I want people to be like me" perspective. I mean look at yourself. Really? You think we would be better off with more of that? I used to be near the end of my workday with all this stuff that needed to be done, thinking man, I could get this done faster with two or three more of me – but then thought better of the idea after imagining how much time would be wasted arguing with me, mostly about pointless things.

 

[dada_dave]

Further possible evidence of yield problems with Intel 18A:

Intel reportedly sticking with TSMC as part of new long-term partnership strategy amid underwhelming 18A yields

Intel is seemingly shifting its strategy of no longer outsourcing wafers to third parties to instead embracing a longer partnership with TSMC. This could be due to the unfavourable yields from Intel’s own 18A process.

www.notebookcheck.net

I wouldn't say any of this is definite, but the tea leaves are not great.

Intel says 18A entering risk production:

Intel announces 18A process node has entered risk production — crucial milestone comes as company ramps to Panther Lake chips

Five nodes in four years is nearly accomplished.

www.tomshardware.com

 

[dada_dave]

Intel says 18A entering risk production:

Intel announces 18A process node has entered risk production — crucial milestone comes as company ramps to Panther Lake chips

Five nodes in four years is nearly accomplished.

www.tomshardware.com

So combining everything together it appears Panther Lake will ship in a “limited capacity” in H2 2025 so they can claim they met their deadline but in reality most Panther Lake parts will ship in 2026:

Intel Vision Presentation Labels Core Ultra 300 "Panther Lake" CPU Series as 2026 Products

Intel's freshly concluded Vision 2025 "Products Update and GTM" showcase included a segment dedicated to forthcoming Core Ultra 300 "Panther Lake" client processors. Industry watchdogs have grabbed a select few screenshots from Team Blue's broadcast from Las Vegas, Nevada—one backdropped slide...

www.techpowerup.com

 

[casperes1996]

The thing I fail to get is the "I want people to be like me" perspective. I mean look at yourself. Really? You think we would be better off with more of that? I used to be near the end of my workday with all this stuff that needed to be done, thinking man, I could get this done faster with two or three more of me – but then thought better of the idea after imagining how much time would be wasted arguing with me, mostly about pointless things.

If you hire me five more developers in my department to get our projects done. I could probably readjust the estimated time to finish by no less than 6 months slower. Maybe even a year more into the future.

 

[throAU]

One of the most dishonest - though I don't think uniformly with that intent so much as ignorant - retcons of Intel is that the fall was mostly about fabrication and EUV. That was the biggest dip they've ever had and made various other issues more salient, yes, particularly given AMD's foray into the market and Apple's in 2020.

But Intel's design unit is a mess, their corporate leadership and organization essentially shunted diversification multiple times over from mobile to GPUs and AI or actively harmed new acquisitions. Further their design even in their core market was increasingly sloppy and thus the gaps just grew and grew with competitors of a similar disposition (AMD) or others with better talent/IP/scale (as we now see with Arm/QC vs Intel comparisons - or even the fact that it took Apple half a decade to get their modem team up to relative par).

I'm a broken record about this but it's just remarkably obvious if you look through their products, the metrics on power/IPC/performance/area, the trends, etc. Now you don't even have to do that. Lunar Lake and Arrow Lake are on TSMC N3 and multiple headlines and reviews point out it was hardly a panacea, almost alarmingly far from that.

I think it’s both

Since the 80s intel has ridden on having the best fab tech and their design department always relied on that.

They’ve been getting away with that for years (because they DID have the best fab) and since skylake in particular Intel has been phoning it in design wise. At least until their little.big stuff and even that was a parts bin special.

They bet the farm on being a tier 1 foundry based on their historical leadership and TSMC just ate their lunch.

Now they have no fab advantage and their design department has been coasting and bleeding talent, they’re fucked.

At least as I see it as a reasonably techy consumer who’s been buying computers for ~30 years

 

[mr_roboto]

I think it’s both

Since the 80s intel has ridden on having the best fab tech and their design department always relied on that.

They’ve been getting away with that for years (because they DID have the best fab) and since skylake in particular Intel has been phoning it in design wise. At least until their little.big stuff and even that was a parts bin special.

A little more nuance on this one... it plays into why their design side has been bleeding talent.

Everyone knows about tick-tock, right? Haswell was Intel's 22nm-optimized product family, followed by a shrink for 14nm (Broadwell), followed by Skylake which was the 14nm-optimized family. Skylake should have been immediately followed by a 10nm Skylake shrink (I forget what it was supposed to be called), but 10nm was commercially useless for many years, breaking the tick-tock cadence.

Intel upper management kept buying into promises from the manufacturing side of the company that 10nm would be fixed real soon. Therefore, all they allowed the design side of the company to work on for 14nm (and 14+, ++, +++ etc) was tweaks of Skylake. There were teams working on new CPU core (and other) designs the whole time, but none of it could ship for revenue because the node(s) it was designed for were unusable.

If you were a design engineer watching your work rot because management refused to concede to reality, you'd want to get out. And lots of them did.