SemiAccurate Teases 10nm Failure Fallout

Meeho said:
That would high on the list of worst possible outcomes.
Click to expand...

I'd argue the worst possible outcome would be Apple snapping up the IP.

Then we'd have their patent trolling destroying the mobile industry for the next decade.

Google is fairly harmless. They'd likely just continue licencing the architecture just like ARM holdings has.
 
Based on everything I read on the theme I am almost sure now it's ARM. IF (and this is a giant IF) Semi accurate are ...accurate? = ) Qualcomm does not make sense at all, since they are purchasing and relying heavily on Samsung for the latest Snapdragons. That being said it leaves ARM up in the air and for some absolutely strange reason if this is good info it should be them. And let's hope fApple won't come in and buyout ARM and it's IP, because then they will start to fuck with everybody.
 
  • Like
Reactions: N4CR
like this
DTabakov said:
Based on everything I read on the theme I am almost sure now it's ARM. IF (and this is a giant IF) Semi accurate are ...accurate? = ) Qualcomm does not make sense at all, since they are purchasing and relying heavily on Samsung for the latest Snapdragons. That being said it leaves ARM up in the air and for some absolutely strange reason if this is good info it should be them. And let's hope fApple won't come in and buyout ARM and it's IP, because then they will start to fuck with everybody.
Click to expand...

IF this is accurate I see it becoming a bidding war between Apple and Google, as both of them have too much riding on ARM to let the other control it.

But ARM is owned by SoftBank, which has very deep pockets, so I'm not even convinced this would be the end of ARM, if they are the ones we are talking about. Certainly it would hurt their finances, but SoftBank could bail them out, drop their chip production and just return to R&D and licensing deals.

Their stock (both SoftBank and ARM Holdings) would absolutely tank, and there would likely be restructuring and job losses, but it doesn't necessarily mean the end of either company.
 
"A massive $20+ billion market cap tech giant" Well this tells us that the market cap of the company is probably between 20-30billion because any higher and it would probably be written as "30+ billion market cap". Massive probably means well known and has several fabs and has some through put.

The first company that comes to mind in Toshiba. After that maybe STMicro and then Infineon. But I have not heard anything about any of these companies doing anything with 10nm.

The usual big names in the business (Samsung, Micron, Broadcom, etc.) either skipped 10nm or are too large/diversified to let something like this ruin them like the article makes it sound.
 
viscountalpha said:
My main point is, we are likely getting taken for a ride.
Click to expand...

Highly doubtful. Despite his abrasive style, Charlie is well informed and has a certain amount of respect in the industry.

djand said:
The company is most likely ARM.
Click to expand...

That makes zero sense. ARM designs chipsets, doesn’t manufacture them. Even then, they’d still make crazy money just by licensing their designs as they’ve been doing for years. Qualcomm also doesn’t make sense as they use other foundries. I’m not sure who Broadcom uses these days. Either way, they cannot be dead as Charlie says, given that they own the majority of the phone industry: even if 10nm failed for them they could ride the wave with their current 14nm chips for another 2 or 3 years: not everyone buys an iPhone or a Galaxy.

If this affected someone so much, it must be another, more vulnerable company. Probably a company manufacturing and selling ARM chips, but who is doing that and has such a limited portfolio that may be sunk by this? Mediatek? Maybe others making FPGAs like Altera Or Xilinx? Texas Instruments?
 
Last edited:
Hopefully it's IBM or Google or Facebook. I thought ARM only designed chips for clients... so the client would be screwed, not ARM, per se.
 
Zarathustra[H] said:
That's what I thought about ARM too, but these things change so fast that I can't keep up anymore.

Qualcomm would make sense based on the description that "a lot of people in the industry rely on them". I guess it depends on what he means by a $20B company? If by revenue it's pretty close. Qualcomm was $24B last year. If its by market cap, Qualcomm is way too large, at some $85B.

To be honest, I don't even know what IBM does anymore. Consulting I thought?

I mean, they sold their PC business to Lenovo in like 2005, and sold their chip manufacturing to GloFo in 2014. I guess they still have some chip R&D?
Click to expand...

No no they design and have 3rd parties make POWER and Z System cpus and chipsets. That's not PowerPC though it's modern and extremely fast. In short it's mainframes and software/software services they make their money in.
 
Zarathustra[H] said:
I'm no expert on the subject matter, but in a world where you can short stocks and profit when they do poorly I assumed that pump and dump regulations also covered the reverse situation when you intentionally harm a stock in order to profit on its demize.
Click to expand...
Viceroy does exactly this (did it with AMD) and gets away with it.

I do not think Charlie is doing this though.
 
viscountalpha said:
Pump and dump is the opposite of what i'm saying. They are speculating intel is doing poorly and then suddenly it's doing great. it's much harder to prove and prosecute.

"A pump and dump scam is the illegal act of an investor or group of investors promoting a stock they hold and selling once the stock price has risen following the surge in interest as a result of the endorsement. ... Pump and dump scams tend to only work on small and micro-cap stocks that are traded over the counter."

My main point is, we are likely getting taken for a ride.
Click to expand...

Poop and scoop is the term you're looking for.
 
Mediatek has a market cap of $17B. Calling them a 20B company would be fairly correct. Rounding down to $10B won't be right. All smaller cellphone companies use them. There are reports that they are moving from TSMC to GF, but they did use 10nm. There were rumours of going to 7nm, but what if they actually signed a contract with Intel last year, or early this year? I know that TSMC and Intel's process differ a lot and 7nm would make a lot more sense, but maybe Intel offered Mediatek a deal they could not resis. Anything is possible.

But would this kill Mediatek? Probably not, they could survive on older process since they anyway feed lower end customers, unless they burnt their bridges. TSMC could just say no after losing the contract.
 
Charlie told you all enough with his hints who this "foundational tech giant" is. It's none other than Hewlett Packard. Their market cap right now is around 22 billion. Charlie said the company in question is 20+ billion in market cap. Dell and HP are the two biggest customers of Intel chips but Dell is a newcomer compared to HP who practically created Silicon Valley. It IS a "foundational tech giant". But unlike Dell, HP has been in financial trouble for years. They spun off the Printers division and the Consumer PC division and they are focusing on Enterprise level computing. They can not afford to miss a product cycle due to Intel's fumbling 10nm.
 
What would HP have being manufactured there at 10nm? Cisco has long been an Intel foundry user and with the shift from hardware tied network equipment to more a software license a stumble could be bad there too. They're $200b market cap and $48b revenue though.
 
Jumbotron said:
Charlie told you all enough with his hints who this "foundational tech giant" is. It's none other than Hewlett Packard... They can not afford to miss a product cycle due to Intel's fumbling 10nm.
Click to expand...

Wait, you think HP is in trouble if they miss 10nm? LOL. The great majority of their customers buy computers with 768p screens and i3 processors... but yeah, that 10nm will definitely do it, we all know how well educated their general consumer is... 10nm or bust for that Pavilion, I say!

:D
 
What used to be HP is two companies now. HP which sells PCs and printers, and HPE which sells servers and network equipment.

HPE does have some custom ARM stuff going on for their servers (not sure if they were going to use Intel 10nm), but I guess they could survive on selling Intel/AMD/Qualcomm CPUs in servers.
The more I think of it, the more the speculation about Nokia betting 5G on Intel 10nm exclusively makes sense. Surely we will see in Nokia's next quarterly report.
 
Kurakka said:
Here is a conspiracy theory straight from an RV parked across the street form the parking lot right outside of left field. Intel purposely faults 10nm production to tank said company for their chip design patents (assuming they are mobile chips) so Intel can then buy said company for billions less to make sure they have something to help their struggling mobile chips.... Any takers?
Click to expand...
This but to kill ARM which is the biggest future threat to them in server and general marketspace. Perhaps even more disruptive than AMD EPYC.

Now only Samsung and Apple will have a strong ARM (pardon the pun) which will fragment the user base successfully.
I'm sort of relieved they are not targeting AMD like they used to, bigger fish to fry.
 
N4CR said:
This but to kill ARM which is the biggest future threat to them in server and general marketspace. Perhaps even more disruptive than AMD EPYC.

Now only Samsung and Apple will have a strong ARM (pardon the pun) which will fragment the user base successfully.
I'm sort of relieved they are not targeting AMD like they used to, bigger fish to fry.
Click to expand...

Sorry, this... just doesn't make any sense. How is ARM going to die from Intel's 10nm production issues? ARM doesn't manufacture anything, they create architecture designs! Also, if ARM died, explain to me how on Earth Samsung or Apple would have any way to keep designing SoCs since they are virtually ALL designed following the ARM architecture? You're confusing architecture with process and manufacturing in market positions that don't have any logic.
 
djand said:
The company is most likely ARM. ARM bet heavily on quick custom production @ 10nm with pre-designed modules. They were selling this capability to existing customers two years ago and it still hasn't delivered. ARM was relying on Intel for a quick transition to 10nm and were using Intel's reputation for solid engineering to draw customers for low power/high performance custom chips. Think of the smaller companies that have been waiting; the fallout is not going to be limited to ARM...
Click to expand...

ARM already demoed Cortex chips on 7nm time ago and they must be close to doing a tapeout on 5nm, so not sure what are you saying.
 
777 said:
What a load of absolute rubbish! Francois Piednoel confirmed that Intel's 10nm problems were all but fixed.
Click to expand...

Both Charlie and Francois are extremists. The truth about 10nm is somewhat in the middle between their public statements. Neither everything has been fixed, nor 10nm is that unfixable distaster that Charlie pretends.
 
juanrga said:
Both Charlie and Francois are extremists. The truth about 10nm is somewhat in the middle between their public statements. Neither everything has been fixed, nor 10nm is that unfixable distaster that Charlie pretends.
Click to expand...
So what is your position on the theory that there is a harmonic disturbance at that dimension in silicon?
 
gigaxtreme1 said:
So what is your position on the theory that there is a harmonic disturbance at that dimension in silicon?
Click to expand...

tenor.gif
 
euskalzabe said:
Sorry, this... just doesn't make any sense. How is ARM going to die from Intel's 10nm production issues? ARM doesn't manufacture anything, they create architecture designs! Also, if ARM died, explain to me how on Earth Samsung or Apple would have any way to keep designing SoCs since they are virtually ALL designed following the ARM architecture? You're confusing architecture with process and manufacturing in market positions that don't have any logic.
Click to expand...
Not that ARM (corp) per-se is going to die but because now the only two big, established players are Samsung and Apple, if the 10nm BS affected another large (server-aiming) ARM-based customer which was supposed to be launching soon/now, year or so before 7nm ARM products are out, then it will have taken the wind out of the sales and cause a slow build-up. Basically now Intel (or AMD) is a better choice due to lack of availability from established providers due to 10nm option (if rumours are true) failing.
 
mdswish said:
I should have been more specific. I was meaning Intel "mobile" chipsets. ASUS, for example, uses Intel mobile chips in their ZenFone line.
Click to expand...

yea but this would be someone who exclusively depended on intel 10nm. Like a chip maker who is using intel. Asus doesn't make their own chips. Even if their mobile division failed they wouldn't be in much trouble as I bet most of their revenue comes from other stuff like motherboards, GPUs and ROG branded products.
 
juanrga said:
Both Charlie and Francois are extremists. The truth about 10nm is somewhat in the middle between their public statements. Neither everything has been fixed, nor 10nm is that unfixable distaster that Charlie pretends.
Click to expand...
The truth is that you should not hold your breath for a fixed version of 10nm any time soon.
 
juanrga said:
Where did I say anything about soon? Nowhere.
Click to expand...
Not everything people write in reply to you is something they wish to pin you down on :) .

But on that subject 2020 or 2021? Certainly not 2019 ...
 
Pieter3dnow said:
Not everything people write in reply to you is something they wish to pin you down on :) .

But on that subject 2020 or 2021? Certainly not 2019 ...
Click to expand...

Maybe, only maybe, you would have some familiarity with what other people writes before making assumptions about what they think. FYI I wrote this

TSMC 7nm is ~0.35x scaling of 16nm. Cavium could duplicate cores and still reduce power consumption a bit. At 7nm, TSMC has dedicated HPC version which targets ~4.5GHz.

On the other hand Intel will not have 10nm server until 2021 or so.
Click to expand...

Does that enter in your definition of "soon"?
 
euskalzabe said:
Highly doubtful. Despite his abrasive style, Charlie is well informed and has a certain amount of respect in the industry.



That makes zero sense. ARM designs chipsets, doesn’t manufacture them. Even then, they’d still make crazy money just by licensing their designs as they’ve been doing for years. Qualcomm also doesn’t make sense as they use other foundries. I’m not sure who Broadcom uses these days. Either way, they cannot be dead as Charlie says, given that they own the majority of the phone industry: even if 10nm failed for them they could ride the wave with their current 14nm chips for another 2 or 3 years: not everyone buys an iPhone or a Galaxy.

If this affected someone so much, it must be another, more vulnerable company. Probably a company manufacturing and selling ARM chips, but who is doing that and has such a limited portfolio that may be sunk by this? Mediatek? Maybe others making FPGAs like Altera Or Xilinx? Texas Instruments?
Click to expand...

It does make sense. ARM has been selling modular designs that exclusively use INTEL 10nm fab services. ARM and their design customers have a huge investment in INTEL's custom fab service. since 2016.
 
You must log in or register to reply here.
Back
Top