Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
Justin,
Unlike some of the other guys here, I am NOT a process guy, just an old, black hearted sales and marketing guy.....who can read the minds of these guys (I swear I can:)) But take what I say with a grain of salt. Couple of things come to mind "Startup" and "momentum". Both other approaches have problems and not small problems. SOI is expensive. It used to be called SOS (Silicon on Sapphire) and because no one could ship anything but promises on SOS, we began calling it "Same old Shxt." Obviously those days are gone. bottom line is that Intel has something that works and is relatively cheap. I have to think that a graduated doping is not an easy process.
I hate to sound like a negative old fard, but I have seen a lot of "new" things that never worked. If Suvolta's process turns out to be THE answer, Intel will be involved.
Moore's observation will end someday because it is, after all, an observation. These aren't simple shrinks anymore. It took strained silicon to get to 90 or 65nm (my memory is getting bad) and HKMG to get to 45 and 32nm. Now it is all of those and Trigate to get to 22 and 14nm. Intel claims to have line of sight to 10nm...but maybe that is just to mind-doodle the competition (they do that, ya know). Taking a 160sq. mm Ivy Bridge to 10nm would make an 60-80sqmm chip.....with 1100 pins...gotta be pin limited.
There is still 3D to go....and who do you suppose will be the leader there?
WRT the power thing on IB vs. Haswell. Haswell will come with an in-package power control chip that has 320 control points. this chip will turn off most of the Haswell most of the time. There are Haswells out in the field, on mother boards that can really be tested for power consumption (and performance.) I haven't seen a thing on power.....could Intel have these guys so well handcuffed that they don't dare leak the actual power specs? If so, could the Haswell power consumption be even much lower than we have been lead to believe? could that be why ARMH is down 10% in three days?
Intel has multiples of the capacity required for sustaining a flat PC CPU business. And the two largest facilities aren't on line yet! That fact alone tells me that there is some kind of tsunami coming.
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
Justin,
Well, I'll be dipped.....basically what the article says or rather what Morris Chang says in the article, is that if TSMC's technology can't keep up with Intel, the consequences are "too horrible to imagine".
When I first read the article I nearly wet myself! check with all the old guys on this board, I'll bet NO ONE has ever heard that kind of frankness and that kind of language from any leader in the semiconductor business.
The building could be burning down and these prima donnas would say they have a "minor process blip that will be fixed in two weeks."
What is happening with the foundries and Intel is the end of Moore's observation for the foundries while Intel, through billions in research can stretch it a couple more nodes.
Intel will use its manufacturing lead to snarf up on the mobile business. I think we can now see that they will own the tablet business. The smartphone business will be next. Figure all phones will be smartphones. Figure 6.5 billion people, most of whom are not trading their phones every two years like we do with the subsidized business model in the US. So, figure the phone biz peaks at 1.5-2 billion units per year. At 14nm (10nm is not a gimme yet) an application processor with an Intel digital radio baseband (yup, goodbye QCOM) will be about 50 sq. mm. That will be 2500 good chips on an $8000 450mm wafer. That's a $3.20 cost for the guts of a phone. Anyone stuck at 28nm and 300mm wafers will have a manufacturing cost of $13 for a part that is lower performance and higher power. Intel would sell those $3.20 parts in a heartbeat for $13.....75% gross margin. I'm sorry, I just don't see how anyone else wins or even plays. Getting to 22 or 14nm or finfets four years late is the same as not getting there at all.
Maybe it takes another couple of years for Intel to sweep up the business, but sweep it up they will.
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
The major foundries (TSMC and GloFo) are in a terrible place. Neither of them have anything but words about nodes below 28nm and finfet.
These guys are desperate. If either of them could show the world a test chip using 20nm planar or 16nm finfet or whatever, the smoke would be billowing off their tennys, running to get to a demo conference. Nothing is happening. Scenario: What if TSMC really knew that they might never (in an economically rational timeframe) have finfets? What if they knew that the best they will ever do is 28nm because 20nm planar is too leaky to provide any benefit? Does Morris Chang (one of the straightest shooters in the business) do? Does he go public with this? No, not even Morris. He will promise whatever his customers need him to promise that will allow them to sleep at night. If he would suggest difficulties, a busload of his customers would be riding over to GloFo, who has no compunctions about outright lying. What do the have to lose? Nothing. But, if lightning strikes, and they are able to solve a problem or two, maybe they pilfer some TSMC customers. The foundries are in the biggest PLAUSIBLE lie mode. It is kind of like we know the world will end...and we can't stop it, but we can delay it a little or we can keep the information from the l'il people so they don't worry. Do we delay and withhold information? Of course we do. When TSMC begins to tell their customers that they have tons of 28nm capacity but nothing smaller. Those customers will have to do the best they can and some will cease the bleeding edge technology. That's what creative destruction is all about.
The facts are that Intel has been shipping and improving Trigate for a year and a half and will be moving to 14nm Trigate this year. The best that we have seen from any other fab is 28nm planar, no 20nm test chips, no finfet ARM mock-up SoCs for show and tell...nothing.
Again, I post the this link: http://bit.ly/166M64T Morris understands the problem, he also understands that TSMC is now in the "horrible consequences" stage of the problem.
Well this is interesting. Eight days into this experiment and the five stocks have gained almost 12 times the S&P gain. eight days - big deal:) The interesting thing is this morning, at the market high, The stock gain was only 3.5 times the S&P. Apparently the afternoon bloodbath affected the S&P much more than the five stocks. Maybe the formula stocks protect against losses??
WEIGHTING BEGIN PRICE Value Current Price current value CHRW 0.974 56.07 54.61218 59.11 57.57314 NOV 0.807 67.65 54.59355 69.17 55.82019 CVI 0.94 58.2 54.708 66.9 62.886 WCG 0.96 56.96 54.6816 51.72 49.6512 DK 1.59 34.41 54.7119 35.87 57.0333 Total 273.29 273.30723 282.77 282.96383
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
"the better tech mousetrap doesn't always win"
I've heard that comment by a bunch of guys who really have no idea what they are talking about. The world can survive some sloppy design, but every time an engineer selects an inferior solution they weaken their company. Enough of that and the company and its products cease to exist.
We really don't know whether ARM is the best solution or even a "good enough" solution for mobile because ARM has been the ONLY solution available when the prime criteria for the SoC is low power consumption. MIPS wasn't a solution. ARMH was the only solution for fabless semiconductor guys who needed a processor core and didn't have the time or skill to design their own. The guys at ARMH are as surprised by their success in smart mobile as anyone.
The mobile processor and SoC world is now changing at a high rate...and not necessarily to ARMH, QCOM, NVDA, and TSMC's liking or advantage. Full mobile SoC products with superior performance on speed as well as power consumption from Intel will be the first real test of how "sticky" ARMH's technology really is.
In the face of an Intel capable design and a 1.5-2 node process advantage, I think that ARM will prove to be much less sticky than many think.
Nomura downgrades SanDisk (SNDK -2.3%) to hold on valuation grounds, leading shares to take a breather following a 38% YTD gain. Though granting SanDisk is benefiting from "disciplined" NAND flash supply growth and improving product mix, Nomura still sees potential risks, such as capacity additions from rivals, high capex due to technology transitions, and the potential for Samsung (SSNLF.PK) to gain an edge down the line from its lead in developing 3D NAND chips. [View news story]
The stock has a fair value of about $50, so it is overpriced. the problem is that they are painted into a corner on growth. No more capacity coming on short term. The 3D issue isn't worth worrying about. Toshiba is thought to have a good line on 3D. Micron, backed by Intel might be more of problem for Sandisk and Samsung on 3D NAND.
ARM Holdings: Learning From My Mistakes [View article]
ARMH is worth $17/share under perfect conditions. Their growth rate or net margins stumble just a tad and it is worth $10.
The efficient market is anything but efficient.
All of my stock playing is inside a Roth IRA, so I can't short, but I can buy puts, and that is exactly what I will do next week.
When the wheels come off thing stock, the decline will be breathtaking.
The human critter is three times more sensitive about losing something they have than about not gaining something they desire. It is a corollary to the Scarcity Principle in the Six Principles of Persuasion.
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
Did you know that Intel added 500 million transistors from Sandy Bridge to Ivy Bridge. IB was like a tick+.
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
Unlike some of the other guys here, I am NOT a process guy, just an old, black hearted sales and marketing guy.....who can read the minds of these guys (I swear I can:)) But take what I say with a grain of salt.
Couple of things come to mind "Startup" and "momentum". Both other approaches have problems and not small problems.
SOI is expensive. It used to be called SOS (Silicon on Sapphire) and because no one could ship anything but promises on SOS, we began calling it "Same old Shxt." Obviously those days are gone.
bottom line is that Intel has something that works and is relatively cheap. I have to think that a graduated doping is not an easy process.
I hate to sound like a negative old fard, but I have seen a lot of "new" things that never worked. If Suvolta's process turns out to be THE answer, Intel will be involved.
Moore's observation will end someday because it is, after all, an observation. These aren't simple shrinks anymore. It took strained silicon to get to 90 or 65nm (my memory is getting bad) and HKMG to get to 45 and 32nm. Now it is all of those and Trigate to get to 22 and 14nm. Intel claims to have line of sight to 10nm...but maybe that is just to mind-doodle the competition (they do that, ya know).
Taking a 160sq. mm Ivy Bridge to 10nm would make an 60-80sqmm chip.....with 1100 pins...gotta be pin limited.
There is still 3D to go....and who do you suppose will be the leader there?
WRT the power thing on IB vs. Haswell. Haswell will come with an in-package power control chip that has 320 control points. this chip will turn off most of the Haswell most of the time. There are Haswells out in the field, on mother boards that can really be tested for power consumption (and performance.) I haven't seen a thing on power.....could Intel have these guys so well handcuffed that they don't dare leak the actual power specs? If so, could the Haswell power consumption be even much lower than we have been lead to believe? could that be why ARMH is down 10% in three days?
Intel has multiples of the capacity required for sustaining a flat PC CPU business. And the two largest facilities aren't on line yet! That fact alone tells me that there is some kind of tsunami coming.
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
Well, I'll be dipped.....basically what the article says or rather what Morris Chang says in the article, is that if TSMC's technology can't keep up with Intel, the consequences are "too horrible to imagine".
When I first read the article I nearly wet myself! check with all the old guys on this board, I'll bet NO ONE has ever heard that kind of frankness and that kind of language from any leader in the semiconductor business.
The building could be burning down and these prima donnas would say they have a "minor process blip that will be fixed in two weeks."
What is happening with the foundries and Intel is the end of Moore's observation for the foundries while Intel, through billions in research can stretch it a couple more nodes.
Intel will use its manufacturing lead to snarf up on the mobile business. I think we can now see that they will own the tablet business.
The smartphone business will be next. Figure all phones will be smartphones. Figure 6.5 billion people, most of whom are not trading their phones every two years like we do with the subsidized business model in the US. So, figure the phone biz peaks at 1.5-2 billion units per year. At 14nm (10nm is not a gimme yet) an application processor with an Intel digital radio baseband (yup, goodbye QCOM) will be about 50 sq. mm. That will be 2500 good chips on an $8000 450mm wafer. That's a $3.20 cost for the guts of a phone. Anyone stuck at 28nm and 300mm wafers will have a manufacturing cost of $13 for a part that is lower performance and higher power. Intel would sell those $3.20 parts in a heartbeat for $13.....75% gross margin.
I'm sorry, I just don't see how anyone else wins or even plays. Getting to 22 or 14nm or finfets four years late is the same as not getting there at all.
Maybe it takes another couple of years for Intel to sweep up the business, but sweep it up they will.
Micron: The Electronics Industry Just Dodged A Bullet [View article]
For the real dickheads it should be septic tanks and straws:)
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
These guys are desperate. If either of them could show the world a test chip using 20nm planar or 16nm finfet or whatever, the smoke would be billowing off their tennys, running to get to a demo conference. Nothing is happening.
Scenario: What if TSMC really knew that they might never (in an economically rational timeframe) have finfets? What if they knew that the best they will ever do is 28nm because 20nm planar is too leaky to provide any benefit? Does Morris Chang (one of the straightest shooters in the business) do? Does he go public with this? No, not even Morris. He will promise whatever his customers need him to promise that will allow them to sleep at night. If he would suggest difficulties, a busload of his customers would be riding over to GloFo, who has no compunctions about outright lying. What do the have to lose? Nothing. But, if lightning strikes, and they are able to solve a problem or two, maybe they pilfer some TSMC customers.
The foundries are in the biggest PLAUSIBLE lie mode.
It is kind of like we know the world will end...and we can't stop it, but we can delay it a little or we can keep the information from the l'il people so they don't worry. Do we delay and withhold information? Of course we do.
When TSMC begins to tell their customers that they have tons of 28nm capacity but nothing smaller. Those customers will have to do the best they can and some will cease the bleeding edge technology. That's what creative destruction is all about.
The facts are that Intel has been shipping and improving Trigate for a year and a half and will be moving to 14nm Trigate this year. The best that we have seen from any other fab is 28nm planar, no 20nm test chips, no finfet ARM mock-up SoCs for show and tell...nothing.
Again, I post the this link:
http://bit.ly/166M64T
Morris understands the problem, he also understands that TSMC is now in the "horrible consequences" stage of the problem.
Test Of Formula Investing [View article]
The interesting thing is this morning, at the market high, The stock gain was only 3.5 times the S&P. Apparently the afternoon bloodbath affected the S&P much more than the five stocks. Maybe the formula stocks protect against losses??
WEIGHTING BEGIN PRICE Value Current Price current value
CHRW 0.974 56.07 54.61218 59.11 57.57314
NOV 0.807 67.65 54.59355 69.17 55.82019
CVI 0.94 58.2 54.708 66.9 62.886
WCG 0.96 56.96 54.6816 51.72 49.6512
DK 1.59 34.41 54.7119 35.87 57.0333
Total 273.29 273.30723 282.77 282.96383
5/14/2013
Gain 3.54%
S&P 1650.34 1655.35 0.30%
5/22/2013
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
I've heard that comment by a bunch of guys who really have no idea what they are talking about. The world can survive some sloppy design, but every time an engineer selects an inferior solution they weaken their company. Enough of that and the company and its products cease to exist.
We really don't know whether ARM is the best solution or even a "good enough" solution for mobile because ARM has been the ONLY solution available when the prime criteria for the SoC is low power consumption. MIPS wasn't a solution. ARMH was the only solution for fabless semiconductor guys who needed a processor core and didn't have the time or skill to design their own.
The guys at ARMH are as surprised by their success in smart mobile as anyone.
The mobile processor and SoC world is now changing at a high rate...and not necessarily to ARMH, QCOM, NVDA, and TSMC's liking or advantage. Full mobile SoC products with superior performance on speed as well as power consumption from Intel will be the first real test of how "sticky" ARMH's technology really is.
In the face of an Intel capable design and a 1.5-2 node process advantage, I think that ARM will prove to be much less sticky than many think.
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
http://bit.ly/166M64T
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
ARMH down 5% today in an up market.
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
7160
Structural Change In The Mobile Processor Marketplace: Intel Wins; ARM, AMD Lose [View article]
Great article!
I'm betting there are a lot of non electronics guys wondering what a "FET" is:)
Nomura downgrades SanDisk (SNDK -2.3%) to hold on valuation grounds, leading shares to take a breather following a 38% YTD gain. Though granting SanDisk is benefiting from "disciplined" NAND flash supply growth and improving product mix, Nomura still sees potential risks, such as capacity additions from rivals, high capex due to technology transitions, and the potential for Samsung (SSNLF.PK) to gain an edge down the line from its lead in developing 3D NAND chips. [View news story]
Micron, backed by Intel might be more of problem for Sandisk and Samsung on 3D NAND.
Test Of Formula Investing [View article]
Equally weighted stock list up 3.57%, S&P up 1.03%
We'll see what next week brings.
ARM Holdings: Learning From My Mistakes [View article]
The efficient market is anything but efficient.
All of my stock playing is inside a Roth IRA, so I can't short, but I can buy puts, and that is exactly what I will do next week.
When the wheels come off thing stock, the decline will be breathtaking.
The human critter is three times more sensitive about losing something they have than about not gaining something they desire. It is a corollary to the Scarcity Principle in the Six Principles of Persuasion.
Elpida Acquisition Will Be Accretive To Micron, Part II [View article]