Charlie showed some wit
He wrote: "Corporate Marketing is looking awfully redundant, how many people do you need to sell to a single account?"
I think Intel need to hire more marketing droids to sell to a lot of small vendors, 5 chips here and there. Patty will have a lot of fun messing around.
In other news, DELL's AMDT notebook is at the shipping dock. Nvidia is in a tough situation, sleeping with Intel would mean certain death, the only choice is to rely on an expanding AMD market.
How soon will Intel BK? That's an interesting question.
I did a capacity estimate long time ago -- note I was basing my calculations on dual core. My number was > 50%. However, depending on the ramp and yield, I later adjusted it down to 40%, which is very conservative.
posted by Sharikou, Ph. D. @ 10/12/2006 09:41:00 AM
45 Comments:
Since when has Intel ever marketed any of their IGP products as being for gamers? Isn't that like criticising the Athlon 64 FX-62 for being a cruddy mobile chip?
The Pretender said "I think Intel need to hire more marketing droids to sell to a lot of small vendors"
Funny I thought sales people did the selling and marketing people seeded the desire.
You Sharikou are clearly a marketing type as you can't close the deal. You try and try to seed the interent with desire but upon closer inspection we see it is all FUD.
Here again you got it wrong!
It's so bad. AMD will have 100% market share at 2008 and yet still give its 2000 hardworking employees pink slips very soon it is so heartless of not sharing the fruit of hardworkship. Why Hector is so cruel?
AMD will have 100% market share at 2008 and yet still give its 2000 hardworking employees pink slips very soon it is so heartless of not sharing the fruit of hardworkship. Why Hector is so cruel?
Once Intel is killed, Hector will hire those workers back with double salary. It's war, folks.
Sharikou:
I do not understand why in the original post you ask when will intel be bankrupt. By your own calculations, It is 8 quarters, and by now, it should be seven.... right? Or do you want to revise those calculations?
By the way. Where are the tests of Core2Duo Vs K8 on things like "Apache on Windows and Linux", "Oracle on Windows and Linux", "Exchange server" and "Sendmail". I mean, those are the ones which are relevant to people buying servers, right?
Cheers
AMD lay off is part lie. For the first time in a long time The IEEE Spectrum has a whole ONE Page ad by AMD.. since last year those quarter page wanted ads by Intel are missing :-)))
Once Intel is killed, Hector will hire those workers back with double salary. It's war, folks.
Is that what Hector has promised to those employees? He is so nice.
But why he needs to do that. Having the 100% means close to 10B a quarter, it is more than enough to feed 200k emplyee. Or he wanna take fruit for himself? Why so mean ...
I read you blog very often because I invest in AMD shares but regarding yelds and capacity you are wrong.
During last 6 months AMD made two major steps to increase the capacity with FAB36/Chartered but it is not enough to pace with market demand. Additional 5000-7000 wspm doesn't solve the problem as we can observe.
Hector did a mistake because they should overproduce X2s. Maybe they did - we don't know. But it is very strange that having three fabs (30,36 and Chartered) they are not able to satisfy Dell and others in the same time.
Concluding - AMD doesn't have the capacity for 50% of the market.
There is also second conculsion - this is very comfortable position to produce less then market requires and increase production in small steps.
During last 6 months AMD made two major steps to increase the capacity with FAB36/Chartered but it is not enough to pace with market demand.
No. actually, during 1Q and 2Q06 AMD had almost zero capacity increase in terms of wafer output. Look at the graph presented at tech analyst meeting. The capacity only starts to kick in in mid 3Q06.
Sharikou said...
"Look at the graph presented at tech analyst meeting. The capacity only starts to kick in in mid 3Q06."
Middle of the 3rd quarter would have been about the middle of August.
Some have stated that it takes 12 weeks to make a processor, so we will not see demand being accounted for until the middle of November.
Thats about a month away, and alot of time to not have enough capacity/stock, and alot of chips going to Intel.
If the stories are true and the are not meeting demand, the quarterly earnings for AMD in 4Q will be terrible.
If the stories are true and the are not meeting demand, the quarterly earnings for AMD in 4Q will be terrible.
No. You don't get it. AMD started shipping procs out of FAB36 in 2Q, but at low volume. The output -- note the word --output-- picked up only in mid 3Q06. Entering 4Q06, you will see 50% capacity increase as both FAB36 and Chartered FAB7 cranks.
"Once Intel is killed, Hector will hire those workers back with double salary. It's war, folks."
And to pay that double salary the CPU prices will increase several times. Yay for monopolies
Sharikou said...
"Entering 4Q06, you will see 50% capacity increase as both FAB36 and Chartered FAB7 cranks."
Maybe I am missing something, but we are in the 4th quarter now, right?
AMD is releaseing the 3rd quarter results on the 18th of this month.
The supply issues have been claimed this quarter.
Also when you say "50% capacity increase", do your mean 50% more production, or just the ability to have 50% more production?
Is your 50% based off of the 15,000 wspm for Fab 36, which, as you know is not until the end of the 4th quarter, about three months from now?
And how much of that 50% coming from the die shrink to 65nm?
Here’s a quote from Jon Markman who is a financial reporter for MSN money.
“Let's start with Intel. Ugly does not begin to describe the shape of the world's largest maker of semiconductors. It's got too many people, too many factories, too little innovation and too little energy. Its top managers, promoted from within, are mostly a bunch of pansies who have tried to nip and tuck the company into shape with cosmetic changes when the company really needs open-heart surgery.”
Jon also gave good reasons this may be a good time to buy stock in this loser company with Vista around the corner, but the main point to me was there are annalist that know the truth about this company.
Also when you say "50% capacity increase", do your mean 50% more production, or just the ability to have 50% more production?
Look at the AMD wafer out graph presented on June 1, AMD sees 50% more wafer outs starting near the end of Q3, from finished wafers to CPUs may needs 4 weeks, so once we are in November, we should see CPU supply increase by 50%, by the end of the year, the supply will increase by about 75%, by 2Q07, AMD will be able to supply 55% of the chips.
With regards to chipsets, looks like Woodcrest may not be such a good way to go:
http://www.theinquirer.net/default.aspx?article=35047
Not to be the purveyor of good things Intel and bad things AMD, but I'm sure you're aware that Intel has 4 (FOUR) fabs currently up on 65nm: D1D, D1C in Oregon, Fab22 in Arizona, and Fab24-2 in Ireland. Sadly, they won't be able to keep all four up on 65nm because D1D is transitioning to 45nm with production out starting in 8-9 months. And by then Fab32 in Arizon will be up in parallel with them with Fab28 in Ireland due to bring make the 45nm trifecta in early 2008. Interesting fact: 'treble' 65nm and 'treble' 45nm = 'treble trouble' for AMD.
You should go back and re-calculate your figures from your March 2006 post that you linked back to where you only listed 2 Fabs @ 65nm.
“Let's start with Intel. Ugly does not begin to describe the shape of the world's largest maker of semiconductors. It's got too many people, too many factories, too little innovation and too little energy. Its top managers, promoted from within, are mostly a bunch of pansies who have tried to nip and tuck the company into shape with cosmetic changes when the company really needs open-heart surgery.”
And lets look at how this horrible company whith pansie managers did this past year...
Operating margin of 25.17% where they have to invest billions every year in capital.. amazing!
7.64 Billion cash on hand!.. Wow what a bunch of pansies.
5.4 billion leveraged FREE cash flow.. WOW a company near BK.. LOL
Net income of > 6 billion.. yup a company in BIG trouble.
A more accurate assessement is that with more diciplined and focused management All those numbers would make MSFT blush.. but unfortunantly Craig Barret really screwed up
Yo Sharikou you want to get a INTEL insider fired? Post the IP and the source of the silly insider from INTEL that posted on
5:03 PM, October 12, 2006
He must either have gotten his pink slip or was already walked...
http://www.eetimes.com/news/latest/showArticle.jhtml?articleID=193004733
"The firm estimates that dual add-in graphics boards (AIBs) generated $86.2 million US dollars in the last twelve months, and $28.4 million the second quarter— 0.5 percent of the overall AIB market."
Usual speculation by Inquirer. If SLI/Crossfire generated 0.5% of graphics revenue, considering the exponential costs, the unit share should also be exponentially less also.
"And to pay that double salary the CPU prices will increase several times. Yay for monopolies"
No...
AMD doesn't need a monstrous number of employees as Intel apparently does; they can more than afford to pay a damn nice salary to a small work force and keep chip prices low rather than bloating themselves with employees. Look at them climbing closer and closer to 50% market share with a mere 10k strong workforce.
I'm sure you're aware that Intel has 4 (FOUR) fabs currently up on 65nm
http://www.bloomberg.com/apps/news?pid=conewsstory&refer=conews&tkr=AMD:US&sid=aJX88TKAnbiQ
Three of the largest notebook manufacturers (OEM) have confirmed that they are making more laptops with AMD chips, a area that was supposedly owned by Intel
Calculating from the Compal figures thats quoted
2005 = 20% x 9.8 mil = 1.96
2006 = 30% x 14.5 mil = 4.35
If the percentage is for the entire year, thats a jump of more than 200%, however it could also be percentage of present run which will bring down the figure to more modest nos,
extrapolating those nos to the other two manufactures, its no wonder AMD has so much problem meeting demand,
If this trend continues Intel may have to layoff one or two fabs next year
I do not understand why in the original post you ask when will intel be bankrupt. By your own calculations, It is 8 quarters, and by now, it should be seven.... right? Or do you want to revise those calculations?
No. Sharikou said 8 quarters back in Q2. Since we are now in Q4. It should be 6 quarters. I am sure as that gets closer and closer and apparent that Sharikou was DEAD WRONG he will backpedal.
Sharikou, I really liked the first line of the post you linked to...
"...furthermore, 65nm products will be shipping as early as June 2006. Clear indication that AMD is gunning to take over 50% of the x86 market by the end of 2006."
You were very optommistic, as you are now...
1. AMD has already said "they want 40% of the server market by 2009."
2. Have a look, "Sayer admitted that the outfit sold "below 20" percent share of the global computer microprocessors market by revenue. It hopes to increase this to a third of the market by 2008.", it would seem they would be happy with 33% by 2008.
3. At the same time... Nevertheless, AMD was surpassed by its challenger – Intel on the notebook specialized market, the latter’s shares increasing from 86.3% in Q1 to 86.7% in Q2, as AMD’s share decreased to 13.3% from 13.6%.
I think that AMD is in trouble, but we won't know until the 18th.
My E6400 just failed today speaking of witch. Am I unlucky or what. Part of the cpu shorted out when I looked at the problem, parts of the caps on the chip blew, there are 7 fat ones and 3 of them blew. Was pretty tho seeing green and orange sparks flying from under the heatsink then abit of smoke. Didn't damage the mobo as far as i can tell, but some of the pins are singed as well as the bottom of the inside of the socket.
Could have been worse I guess, but wow didn't go with a bang more like a wissle. DX Pretty week fireworks tho. May as well go back to my AMD and sell the mobo. ;__;
Hey Sharikou... I have a question for you: How can a company like Intel go bankrupt when they are still profitable? SGI has been bleeding cash for years now and is still around!
If this trend continues Intel may have to layoff one or two fabs next year
Notebook market is growing at a much faster rate at the expense of deskops. So if AMD is gaining market, it is not at the expense of Intel's numbers.
with the AMD stock moving in positive direction, some poor guys are getting the pink letters. So sad, more profit is going into Hector's pocket instead of theirs.
And to cut 'low performer' at the AMD side? it's time to point finger to the peers and create the office war. It was time to AMD, externally and internally. So sad. Everyone won;t trust each other after that.
with the AMD stock moving in positive direction, some poor guys are getting the pink letters. So sad, more profit is going into Hector's pocket instead of theirs.
And to cut 'low performer' at the AMD side? it's time to point finger to the peers and create the office war. It was time to AMD, externally and internally. So sad. Everyone won;t trust each other after that.
Not only Hector, the ATI CEO get hired with 1.7mil annual wage, and yet his old folks has to reduced by 1200. What a good CEO.
And also, to get pink letter from AMD sides, it is going to hurt twice. First, you lose job. Second, you are labeled low performer. WTF.
Heard that the seperation package is bad too as compared to Intel's. Again more money channel to Hector's pocket.
Intel's worst nightmare
http://money.cnn.com/magazines/fortune/fortune_archive/2006/08/21/8383598/index.htm
" Anonymous said...
Hey Sharikou... I have a question for you: How can a company like Intel go bankrupt when they are still profitable? SGI has been bleeding cash for years now and is still around!
11:47 PM, October 12, 2006 "
how many times SGI went into restructures?
the name of SGI isnt that valued anymore, since itsjust "another expensive intel workstation manufacturer"
no longer has its power as before.
in short words, SGI is not valued and its not the same.
“Hey Sharikou... I have a question for you: How can a company like Intel go bankrupt when they are still profitable?”
When AMD reaches 62% cpu market share, do the math.
HAHAHA... I'm completely flattered about this:
Anonymous said...
Yo Sharikou you want to get a INTEL insider fired? Post the IP and the source of the silly insider from INTEL that posted on
5:03 PM, October 12, 2006
He must either have gotten his pink slip or was already walked...
7:37 PM, October 12, 2006
I'm no more an Intel insider than Sharikou is on AMD's payroll. That being said... my information came straight from link back in this post to the his previous post right here where he life the following post script:
PS: There is a great secret on Intel's capacity, if you can see it, please post your findings in the comments area.
I wonder if I would look as sexy as Charlie did in a bunnysuit?
"When AMD reaches 62% cpu market share, do the math."
Would you please actually do the math and tell us roughly when will AMD reach >60% market share. Given AMD's current market share growth I'm not seeing that happening any time soon.
Just looking over your previous linked blog in this blog..
Is K11 still on schedule as you mentioned? If so could you let us know what this schedule is?
Is AMD's performance lead still widening as 2006 progresses? And this is measured by what metrics?
Ant the 25Watt Turion X2 - you must have been referring to non-mainstream parts?
Also noticed you were assuming a 55% yield back then, now it seems to be up to 100% on a new technology node that is just starting to ramp and doesn't even have all of the transistor improvemtents implemented yet...but of course doing things like introducing new strain techniques on the transistor will obviously have no yield impacts.
Sharikou:
Your capacity estimate is fundamentally flawed. You proclaimed a die size of 85mm2 for Socket AM2 Athlon 64 X2. Let me connect you with something called reality. Scroll down here and don't hesitate to call Hot Hardware a bunch of incompetent liars for pointing out that the die sizes are a little larger - well, if you can call 2X "little":
http://www.hothardware.com/viewarticle.aspx?articleid=822&cid=1
But I say old boy, let's also call AMD incompetent and a liar because gosh, they also think their smallest X2 die size is 147mm2 going all the way up to 218mm2:
http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_9485_13041%5E13076,00.html
Flaw #2 (I don't have time to look for more) is your Conroe die size of 155mm2. Unfortunately, not even the Core 2 Extreme hits that die size and the volume sellers (E 6300/6400) are just 111mm2:
http://www.tomshardware.com/2006/07/14/core2_duo_knocks_out_athlon_64/page4.html
Dude - get a grip and don't make up so much stuff. Yet again your ability to call the future is compromised. Stop drinking your own Kool-aid and try and provide a more realistic assessment. It will go a long way in boosting your credibility.
"Anonymous said...
Since when has Intel ever marketed any of their IGP products as being for gamers?"
Having a spanking new notebook that doesn't even play 2-3 yr old games is kinda sad.
But I think the real kicker is when Vista launches and all those spanking new Intel notebooks suddenly need to disable all those next-gen 3Deffects on the GUI.
"But I say old boy, let's also call AMD incompetent and a liar because gosh, they also think their smallest X2 die size is 147mm2 going all the way up to 218mm2"
You are comparing the size of AMD CPU @90nm with that of Intel CPU @65nm.
Anonymous said...
"Having a spanking new notebook that doesn't even play 2-3 yr old games is kinda sad."
Of the games listed which ones are 2-3 years old?
"But I think the real kicker is when Vista launches and all those spanking new Intel notebooks suddenly need to disable all those next-gen 3Deffects on the GUI."
You need to look up the Intel integrated graphics a little more, they have already been given acceptance for being able to run Aero.
Charlie showed some wit...How about Longan shows some wit???
Don't mean to scare Intel fanboi, but you will be scared after I said this:
1. Intel forges ahead with expensive 45nm process, thus they need the dominant and near monopoly position to rip the benefit. But now day, they need 45nm to remain competative. What if they fail/delay the 45nm?
2. AMD's process go with the mainstream. Charter's 65nm is AMD's back up plan. Does Intel have a partner with 45nm back up plan??? Of course not.
3. The day of going solo is over. Wait till you hit 32nm, you need an alliance to do it. Intel for so long has been in a track deep on their own. It's hard for them to join an alliance. The day Intel loose the process advantage, their inefficient design will be exposed.
Try to imagine Core 2, Conroe/Woodcrest designed in 90nm instead 65nm. OMG, what a thought.
4. Loosing the profit margin, loosing near monopoly will no longer support the raging ahead with advanced processing. Loosing advanced processing will come back and haunt Intel. It's a spriral collapse of their current business model (monopoly - one process ahead).
-Longan-
P.S. I doubt Intel will go BK. They will have to adjust for life of duo-monopoly. (This is where I differ from Sharikou.)
Longan said...
"1. Intel forges ahead with expensive 45nm process, thus they need the dominant and near monopoly position to rip the benefit. But now day, they need 45nm to remain competative. What if they fail/delay the 45nm?"
Why do they need a monopoly to reap the benefits of 45nm?
Why do they need the 45nm to remain competitive?
You are assuming that a Rev H Dual core is going to save AMD from Conroe/Woodcrest.
Well what if it only delivers a moderate improvement and only equals Conroe, or if Quad core only equals Kentsfield/Cloverton?
What will you say then?
"2. AMD's process go with the mainstream. Charter's 65nm is AMD's back up plan. Does Intel have a partner with 45nm back up plan??? Of course not."
Are you talking about R&D or manufacturing?
AMD's backup for R&D is IBM, chartered is there manufacturing backup.
Does Intel need a bckup? They seem to be doing quite well.
“3. ...The day Intel loose the process advantage, their inefficient design will be exposed.”
Unless your talking 4P and up, whats wrong with the FSB?
By the way, Intel is going to incorperate CSI, and an IMC around the middle of 2008, meaning by the time they get to 32nm (2010), where is the inefficient design?
What you fail to account for is Intels ability to respond. When Intel incorperates CSI and an IMC, what advantage will AMD have in 4P and up?
“Try to imagine Core 2, Conroe/Woodcrest designed in 90nm instead 65nm. OMG, what a thought.”
Whats the point, do you want to point out how big the chip would be?
How about K8L on 90nm OMG, what a thought... lol?
“4. Loosing the profit margin, loosing near monopoly will no longer support the raging ahead with advanced processing. Loosing advanced processing will come back and haunt Intel. It's a spriral collapse of their current business model (monopoly - one process ahead).”
How can you assume that Intel will loose that much market share, keep in mind mobile is the fastest growing segment and Intel contols 87%, server they control 75% (80% is 2P or lower), and desktop is probably around 75% (I have links for mobile and servers if you want them).
Do you think they will just roll over?
Don’t forget about WIMAX...
"duo-monopoly"
I'm going go out on a limb and guess you don't have a economics background...
"AMD's process go with the mainstream. Charter's 65nm is AMD's back up plan."
Chartered process = AMD process = IBM process; they kind of jointly develop the technology :) Thus if there is a problem with AMD's process (not that there could be because don't forget AMD has APM3.0!), Chartered will likely have the same issue unless it is a factory specific yield issue.
Here's a scary thought for you - what would happen to AMD if there was an earthquake in Dresden? A worker strike? Political issue? Well I guess they have Chartered's 5000 wafer starts as a backup.
"Chartered process = AMD process = IBM process; they kind of jointly develop the technology :) Thus if there is a problem with AMD's process (not that there could be because don't forget AMD has APM3.0!), Chartered will likely have the same issue unless it is a factory specific yield issue."
If there's a problem with Intel's process, ... remember Intel has that copy-exact!
The good thing about join-effort is that you have more people working at the same problem from possibly different perspective.
Edward said...
"Chartered process = AMD process = IBM process; they kind of jointly develop the technology :) Thus if there is a problem with AMD's process (not that there could be because don't forget AMD has APM3.0!), Chartered will likely have the same issue unless it is a factory specific yield issue."
If there's a problem with Intel's process, ... remember Intel has that copy-exact!
The good thing about join-effort is that you have more people working at the same problem from possibly different perspective.
The good thing about the copy exactly is that you can get faster and equal improvement across the factory.
I strongly believe that for AMD's CPU, the performance such as overclockability for a similar product at the same time frame but different factory will exhibit some significant different statistical result. I really wish that someone can test my hypothesis out :)
of course, my evil intention is that once the user find out which factory is better, they insist on product from a particular factory and put pressue on AMD :)
Instead of waiting for ppl to test my hypothesis of diffrent AMD CPU quality from different factory. I'd like to create some opinion based analysis (with the same evil intention :))
Since the Chartered is new in making the AMD CPU, so I foresee that the German made one is better than the Singpore made one. If you really are die hard AMD fans and wanna buy slower CPU, go buy the German one and push AMD to let you know how to identify the CPU die origin :). If AMD refused, go buy intel's one (you might need to clock it down to meet your desired speed) :) hahaha
Post a Comment
<< Home