siliconbruh999
Well-known member
Found A short of Pat saying how they missed AI and mobile will Intel make the same mistakes again ?
Intel's biggest Misses AI and Mobile?
- Thread starter siliconbruh999
- Start date
Not really convincing.
siliconbruh999
Well-known member
Yeah after missing so many things it's hard to believe them
hist78
Well-known member
I believe Intel and Pat Gelsinger need to seriously reflect on why Intel keeps making mistakes and poor decisions. They can't simply blame previous Intel CEOs, CFOs, the Board of Directors, or division leaders.
In my view, the root cause lies in Intel's IDM business model. This model drives all of its strategic directions, profit goals, technology choices, resource allocations, product decisions, and the hiring and compensation of executives. Without changing the IDM business model, Intel's CEOs and leadership team will continue to make critical errors. Any new branding, such as IDM 2.0 or IDM 3.0, will not address Intel's fundamental problems and will not save the company.
In my view, the root cause lies in Intel's IDM business model. This model drives all of its strategic directions, profit goals, technology choices, resource allocations, product decisions, and the hiring and compensation of executives. Without changing the IDM business model, Intel's CEOs and leadership team will continue to make critical errors. Any new branding, such as IDM 2.0 or IDM 3.0, will not address Intel's fundamental problems and will not save the company.
I grew up with Intel and truthfully it is hard to watch. Sometimes your bravado hits and sometimes it misses. With Intel there have been too many misses of late which is why I feel a pivot is in order.
The IDM business model could be the root cause as you suggest but it will be very hard to change after the big IDM 2.0 push Pat Gelsinger brought in without losing Pat.
Last edited:
Intel didn't get "lucky"
I agree, IDM makes the business more inward looking, so when they are asked to design a new mobile or AI chip they look at it and say “oh the volumes aren’t that big and itll cost too much to make”. That’s because they are looking at their internal costs and bloated overheads. The overhead is bloated because it’s a monopoly business in its core markets. Bets like mobile and AI don’t compete against the core business when they are small, and by the time they are big enough to compete the winner has already been crowned.
Well, that was the history of Intel declining to make chips for Apple, wasn't it? I think Intel has learned that lesson. And it may split itself into several companies in the future when it is in a position of strength. So we will see...
What model change exactly are you thinking would help the pivot? How would not being IDM have helped them decide to go for Apple's iphone business when they already were manufacturing StrongARM at Intel's fabs?
siliconbruh999
Well-known member
I don't think IDM model is at fault it is their inefficiency and mindset that is holding them back back when they showed Inefficiency with foundry data that was a key part it needed to be corrected they have a very bad mindset of Margins/Market they don't take risks like they should
I look at Intel's inability to pivot as emblematic of the lessons of Christensen's The Innovators Dilemma; the same things that make a business great in one environment locks them into failure later. Focus on high-end margin over volume; focus on wide moat x86 over narrow-moat ARM; these are all "good" decisions that led inevitably to the bad outcome of today.
The more successful the business, the more likely the seeds of failure have already been laid. This is something for TSMC and Nvidia to consider, in the manner of Only The Paranoid Survive, ironically written by Intel's former CEO, and somewhat of a counterpoint to Christensen.
The more successful the business, the more likely the seeds of failure have already been laid. This is something for TSMC and Nvidia to consider, in the manner of Only The Paranoid Survive, ironically written by Intel's former CEO, and somewhat of a counterpoint to Christensen.
I don't think I agree. Intel have to be good, or at least decent, at both design and manufacturing due to their tick-tock model.
Nvidia (fabless) and TSMC (foundry) can focus on what they do best since there's only one thing they have do. if anything, TSMC has become more relying on tried-and-true methods before making a leap to anything new. This is in contrast to Samsung, whose jump into GAA and it backfired
siliconbruh999
Well-known member
TSMC is not a risk taker due to it's customer has there anything risky they have done before ? High-K strained silicon Finfet Etc were all done by Intel hey have to carry multiple customer so they take low risk approach shifting of BSPD from N2 -> A16 is such a move
Apple used to be a big influence on TSMC process technology but now the list of influencers is much longer (AMD, QCOM, Nvidia, Etc...) Still, with the volumes these company ship, the low risk / high yield path is all the more important.
Would TSMC be doing BSPD if Intel had not announced it? I do not know.
That challenge is actually easy, because there are completely separate engineering organizations for design and manufacturing. There is no overlap. There is a funding challenge for both, but Intel has been overspending for so long that even with big headcount reductions it is still a viable competitor. The real challenge is at the CEO level. It is nearly impossible to find someone to manage the overall company who can be a deep expert in both, so that strategies and decisions can be judged in the CEO office. And Intel's real problem is that the current CEO has a decades-long agenda which is highly biased WRT design, and that agenda has proven to be less successful than the competitions'.
I think Samsung's early jump into GAA smells a lot like an ego trip.
Samsung prides itself on being first in the semiconductor business. Unfortunately profits suffer due to low yield and that loses customers. It is a cultural thing, the same with TSMC being customer centric rather than racing to the finish line leaving customers behind. It reminds me of the saying "build it and they will come" only semiconductor companies want a say in what's being built which does not fit Intel's or Samsung's IDM business model.
hist78
Well-known member
@siliconbruh999
"their inefficiency and mindset that is holding them back"
If these kinds of inefficiencies and mindsets have been recurring at Intel over the past 10 to 20 years, then we must point to a deeper issue: Intel's IDM business model. This is precisely one of the problems highlighted in "The Innovator's Dilemma," as mentioned by @benb.
Due to this business model, no matter how many brilliant engineers or executives come and go, they inevitably follow the same path and fall into the same trap.
tooLongInEDA
Moderator
Perhaps it's just nuance, but do we really think that the CEO needs to be a "deep expert" in both design and manufacturing ? Doesn't feel quite right to me. I'd rather say he/she needs to have some instinctive grasp and feel (and some experience) for both disciplines - to know in advance what definitely won't work, what might and what the risks and opportunities are and how to balance these. It's hard enough to find that in one person. I agree that if someone's too biased to one discipline, that's not ideal.
tooLongInEDA
Moderator
You say it must be the IDM model. But if it's been recurring for 10-20 years, couldn't it equally well be culture (or both) ?
Without any real personal experience of Intel, I can only "characterise" from external observations. One of my strongest impressions is that it grew up with a very strong NIH element to its culture. TI sometimes felt like that (note: this is decades old data) - very large sites, somewhat geographically isolated from other companies, resistance to external hiring for experienced staff, legacy of being a leader breeding over-confidence. But TI had to confront the reality of being one of many companies rather sooner.