Intel's head of manufacturing announces departure

[oak8292]

Opinions are a dime a dozen but I think BK as a manufacturing guy was over confident in his 'lead' and tried to manage margin with the node ramp. Was the 14nm process really that bad or was it somewhat worse than 22nm and margins might have slipped below 60%. At the foundries schedule is often more important than the yield specifically in the mobile 'race'. Volume manufacturing is being ramped to meet release timelines and margins will suffer if yields are 'optimal' but that is just a price to pay. Intel's margins have never slipped below 60% are they requiring a much higher yield than the foundries before they really ramp or have their manufacturing processes really slipped that badly?

Doesn't wafer volume really help in the yield ramp? I thought this was an Intel 'advantage' with their huge volume in leading edge wafers?

 

[jpiniero]

Intel's margins have never slipped below 60% are they requiring a much higher yield than the foundries before they really ramp or have their manufacturing processes really slipped that badly?

As I mentioned I think Intel is cutting cost to maintain margins as long as they can. You can see why people would want to leave because there will be more layoffs to try to make the numbers look better. IoT, etc, isn't going to cut it.

 

[oak8292]

Intel has a long way to go in cutting costs. They are really disadvantaged compared to TSMC in R&D and SG&A according to annual reports. Intel went from making a product that fit into a market to a company looking to develop and manage all aspects of the computing market. Their strategy on 'adjacencies' has started to backfire. The market is 'specializing' faster than they can get in front of every 'innovation'. It is classic innovators dilemma with a dominant solution that is being bypassed. They aren't willing to give up margin to disrupt themselves.

 

[witeken]

I have sometimes wondered this as well. The one time that I met Pat Gelsinger, I was unbelievably impressed with how sharp he was and how much of a broad depth of engineering knowledge that he had. In my 25 year career as an engineer, I never have met anyone at that level of management who knew so much about low-level electrical engineering, design and testing. The other Intel executive that has always impressed me in when I've spoken with her is Diane Bryant. I miss both of them at Intel.

Back to the opening post though:
Francois was a principal engineer at Intel. The title "principal engineer" is approximately equivalent to a second tier manager - except on the technical side - and it's definitely not "Intel's head engineer". Above a principal engineer is a "senior principal engineer", above that is "Intel fellow" and above that is "Intel senior fellow". The title "principal engineer" is the first named promotion for an engineer at Intel and you would expect approximately one in every 30 engineers or so to be a principal engineer. If I had to pick one specific "head engineer" at Intel from my perspective it would be Mark Bohr . There's a list of Intel's senior fellows here: https://newsroom.intel.com/biographies/senior-fellows/

I logged in just to like this post. Informative and on point.

 

[VirtualLarry]

BK is leading the shift from a CPU centric company to a "almost everything else" company. Forget about Intel as you knew it, Intel of the future might be a slightly smaller but leaner company, which intents to put their chips in all electrical appliances, cars, buildings, etc'. ChipZilla is no more.

If the PC CPU (and chipset) market continues to shrink, do you feel it's possible that Intel might actually cede the majority of the PC CPU market to AMD (and a fraction of a percentage point to Via), and go off to explore "greener pastures", in terms of cell phone modems, and IoT devices?

 

[Arachnotronic]

If the PC CPU (and chipset) market continues to shrink, do you feel it's possible that Intel might actually cede the majority of the PC CPU market to AMD (and a fraction of a percentage point to Via), and go off to explore "greener pastures", in terms of cell phone modems, and IoT devices?

Uh, no. PC CPU business is >$30B/yr in revenue, about 2X its second largest business, the data center group.

PC CPU biz provides Intel with fab scale that it simply couldn't achieve with any other business (or its other businesses combined).

There's also the fact that Intel couldn't really afford to spend anywhere close to what it does on product R&D if >50% of its revenue just went away

 

[VirtualLarry]

Thanks for putting it all into perspective. So, even though the market for PC CPUs and chipsets is on the decline, it's still a HUGE revenue-generator.

 

[Arachnotronic]

Thanks for putting it all into perspective. So, even though the market for PC CPUs and chipsets is on the decline, it's still a HUGE revenue-generator.

Correct. DIY/gaming/enthusiast is a small part of that $30B, but it's a really huge market. People still need laptops, businesses still need desktops/workstations, etc.

 

[WhoBeDaPlaya]

Smith probably left because he wants a CEO position somewhere and that ain't happening at Intel anytime soon.

It's exactly what happened with my company (Maxim Integrated).
Matt Murphy stirred up a bunch of sh*t (layoffs, reorgs, etc.), then jumped ship to become Marvell's new CEO.

 

[CHADBOGA]

BK is leading the shift from a CPU centric company to a "almost everything else" company. Forget about Intel as you knew it, Intel of the future might be a slightly smaller but leaner company, which intents to put their chips in all electrical appliances, cars, buildings, etc'. ChipZilla is no more.

Is a company that can't compete making tablet and phone chips, really likely to succeed in all those other areas?

I would have thought that ARM would be even better placed in those areas.

 

[DrMrLordX]

I was under the impression that Intel is pulling away from IoT. Am I wrong? They are definitely pulling in the 'maker' products.

 

[WhoBeDaPlaya]

I've always thought of Intel as a fab company that happened to make CPUs, you know, to keep those silicon ingots from becoming frisbees.

Speaking as a physical design engineer, it would certainly be a heckuva lot easier to meet timing, DRV, DRC, ANT, etc. if that were the case

 

[2is]

Intel. The difficulties and delays with 10nm indicate (IMO)that they are having problems getting the requisite performance out of the new node, which doesn't bode well for taking down AMD and 2nd gen Ryzen.

You're playing the doom and gloom card entirely WAY too early. Every one of AMD's processors released after Intel came out with Core 2 Duo were sub par. The bled money, people/talent were coming and going left and right and they still managed to finally release a decent product with Ryzen, which only puts them at best, at equilibrium with Intel in terms of mainstream performance.

This has happened before. AMD hit equilibrium with Intel with the release of the Athlon64, then they took the performance crown with the X2's. They held that performance crown for just over 1 year before Intel took it back for the next 12

 

[itsmydamnation]

By 1 you mean 3 years and 5 in the datacentre. Even then products like the x6 where quite competitive.

Even with bulldozer if you look at the components of the uarch, they where just as advanced, they just completely miss judged the target/ had development hell.

But Zen 2 is looking really good simply because IBM's 7 nm looks the goods. 14nm lpp is 3rd best by quite some distance.

What a Zepplin refresh looks like in 4-6 months time we will have to wait and see, bd to pd delivered around 400-500mhz, hopefully this refresh can to.

 

[moinmoin]

You're playing the doom and gloom card entirely WAY too early. Every one of AMD's processors released after Intel came out with Core 2 Duo were sub par. The bled money, people/talent were coming and going left and right and they still managed to finally release a decent product with Ryzen, which only puts them at best, at equilibrium with Intel in terms of mainstream performance.

This has happened before. AMD hit equilibrium with Intel with the release of the Athlon64, then they took the performance crown with the X2's. They held that performance crown for just over 1 year before Intel took it back for the next 12

Intel's opponent isn't really AMD per se here but GloFo, TSCM, Samsung etc. It's why Intel was pulling out of mobile chips race (foundry competition was too strong at a low margin), and why AMD's revitalization now actually hurts them (combining a competitive design with said strong foundry competition). AMD's outlook wouldn't look half as good without the 14nm yield and the prospect of being able to move to 7nm pretty soon.

 

[2is]

By 1 you mean 3 years and 5 in the datacentre. Even then products like the x6 where quite competitive.

Even with bulldozer if you look at the components of the uarch, they where just as advanced, they just completely miss judged the target/ had development hell.

But Zen 2 is looking really good simply because IBM's 7 nm looks the goods. 14nm lpp is 3rd best by quite some distance.

What a Zepplin refresh looks like in 4-6 months time we will have to wait and see, bd to pd delivered around 400-500mhz, hopefully this refresh can to.

By 1 I mean 1. X2 was released in 2005 based on my quick googling and core 2 was released 14 months later

 

[maddie]

that had nothing to do with manufacturing. it was intel restructuring from "being a PC company" to focusing more on IoT, AI and Autonomous Vehicles. entire divisions closed, products axed, and many low-scoring employees were fired along with early retirement of senior employees. that number also does NOT include the McAfee spin-off. so add 7-8k more people to that.

since then Intel has acquired MobilEye and other small companies (as it usually does) and hired new people.

BK is leading the shift from a CPU centric company to a "almost everything else" company. Forget about Intel as you knew it, Intel of the future might be a slightly smaller but leaner company, which intents to put their chips in all electrical appliances, cars, buildings, etc'. ChipZilla is no more.

I learn new things every day. Thus is magic management at work.

A company repositioning itself to be an ""almost everything else" company is going to be "slightly smaller but leaner".

 

[epsilon84]

By 1 I mean 1. X2 was released in 2005 based on my quick googling and core 2 was released 14 months later

Actually AMD was ahead of Intel for 4 out of 5 years between 2000 to 2005. I clearly remember the original AMD Athlon beat the Pentium 3. AMD held the lead until P4 Northwood came along and hit 3GHz, which was enough to give Intel the crown back for approx 1 year. P4 Prescott was a disaster and AMD again overtook Intel with the Athlon 64, which later became the X2 as you said. But once Core 2 was released it's been pretty much all Intel in the high end... until Ryzen came along.

 

[2is]

Actually AMD was ahead of Intel for 4 out of 5 years between 2000 to 2005. I clearly remember the original AMD Athlon beat the Pentium 3. AMD held the lead until P4 Northwood came along and hit 3GHz, which was enough to give Intel the crown back for approx 1 year. P4 Prescott was a disaster and AMD again overtook Intel with the Athlon 64, which later became the X2 as you said. But once Core 2 was released it's been pretty much all Intel in the high end... until Ryzen came along.

Athlons and the P3 traded blows with a slight edge to the Athlons. A64 and P4 Northwood traded blows. AMD winning the ST crown Intel edging out in MT (thanks to hyper threading) It wasn't until the X2's where AMD had a clear advantage across the board. That lasted 14 months.

Were they competitive for 4/5 years? Sure, but that isn't the same thing as saying they were ahead of Intel.

 

[epsilon84]

I think you are painting a rather rosy picture for Intel during the early 2000s. Athlon was definitely a win for AMD as it had the clockspeed advantage over P3 (first to 1GHz), as well as slightly higher IPC IIRC. P4 Northwood mainly competed with Athlon XP, not Athlon 64. Athlon 64 competed with P4 Prescott and, whilst not winning every single benchmark, definitely had the edge overall. Athlon X2 as you said was a step above Pentium D.

So AMD was more than competitive during those 4 years, they actually had a performance lead, and more often than not, a price/performance lead as well. Being behind the curve didn't seem to impact Intels prices that much from what I remember.

 

[tommo123]

I think you are painting a rather rosy picture for Intel during the early 2000s. Athlon was definitely a win for AMD as it had the clockspeed advantage over P3 (first to 1GHz), as well as slightly higher IPC IIRC. P4 Northwood mainly competed with Athlon XP, not Athlon 64. Athlon 64 competed with P4 Prescott and, whilst not winning every single benchmark, definitely had the edge overall. Athlon X2 as you said was a step above Pentium D.

So AMD was more than competitive during those 4 years, they actually had a performance lead, and more often than not, a price/performance lead as well. Being behind the curve didn't seem to impact Intels prices that much from what I remember.

this is what i recall too. iirc intel held video encodings crown but for everything else (more or less) it was AMD. until conroe around july 2006 but we got an inkling on it's power when apple swapped over to intel and everyone went "whaaaa?".

there were signs of conroe being a player when iirc, people bought merom laptop chips and used them via adapters of some kind in their PCs. or am i misremembering there.

 

[WhoBeDaPlaya]

there were signs of conroe being a player when iirc, people bought merom laptop chips and used them via adapters of some kind in their PCs. or am i misremembering there.

Not at all. A bunch of folks were using Yonah chips in desktops back in the day.

 

[2is]

I think you are painting a rather rosy picture for Intel during the early 2000s. Athlon was definitely a win for AMD as it had the clockspeed advantage over P3 (first to 1GHz), as well as slightly higher IPC IIRC. P4 Northwood mainly competed with Athlon XP, not Athlon 64. Athlon 64 competed with P4 Prescott and, whilst not winning every single benchmark, definitely had the edge overall. Athlon X2 as you said was a step above Pentium D.

So AMD was more than competitive during those 4 years, they actually had a performance lead, and more often than not, a price/performance lead as well. Being behind the curve didn't seem to impact Intels prices that much from what I remember.

What rosy picture? What part of what I said is inaccurate? P4 Northwood was better than Athlon XP, at everything. Athlon 64 was largely better than P4 Northwood and Prescott but P4 still held the lead in several benchmarks/applications. Photoshop, video encoding/transcoding, and general multi tasking. There were still very legitimate reasons why one would pick a P4 over A64 up to and including this point. Again, it wasn't until the X2 where AMD had an advantage across the board, including the Pentium D's and that across the board advantage lasted 14 months, far less than the 4 years you're giving AMD credit for. That 14 month span is the only time AMD held the type of advantage over Intel that Intel has held over AMD for the last 12 years.

I have zero reason to be bias here. I owned AMD during that entire time span including the time when P4 Northwood was kicking AMD's ass. I had a Pentium 200 MMX and everything after that up until my Q6600 was powered by AMD. Starting from a Thunderbird 1GHz to a XP 1900+, to a mobile Barton 2600+ to Athlon 64 3800+ (IIRC) and finishing off with an X2 4400+ before finally moving to a Q6600, and even that wasn't until it dropped from $1000 to $300.

I don't think either of us is disagreeing with history here, what I see happening is you think AMD's advantage over Intel was far greater than it really was.

http://www.tomshardware.com/reviews/amd,817-17.html

Pretty comprehensive list of benchmarks there. I counted 16 victories for P4 and 22 for A64. Advantage AMD to be sure, but if you don't consider that trading blows, we'll just have to agree to disagree. 3 of the 4 years of what you consider to be AMD dominance looked just like that

Moving on to the X2, which is where I feel was the only time AMD actually dominated Intel

http://www.tomshardware.com/reviews/amd,817-16.html

Athlon X2: 24
Pentium D: 6

AMD only had 14 months where the performance delta looked like that. This is the type of advantage Intel has been enjoying for 12 years.