Intel may bankrupt in seven quarters
DELL is going to market AMD like crazy. It's a business decision, business decision is based on money. Simple?
Intel execs are morons. They caused a price crash on Intel chips, instead of a price hike I proposed to them last year...What's the consequence of that? Dell's units and unit price will go down significantly. DELL's unit price will go down because Intel CPUs are dirt cheap now - a Pentium D 930 is only $100. DELL's units will go down because every small vendor can sell Intel PCes dirt cheap, DELL loses much of the pricing advantage...
Suppose DELL's PC unit price drop by 10% and units drop by 5%. The result will be 15% revenue drop. Can DELL survive that kind of spiral down? No. What can Dell do? Go AMD and expand the AMD market like there is no tommorrow. Since Dell's Intel business will suffer 15% drop, DELL must sell AMD like crazy to make up the difference. I have analysed this many many times.
Intel can be written off the book already. The prices of its chips have crashed by 60%, and units will be down significantly, as more vendors such as DELL are adopting AMD. The result will be continuing operating losses for Intel in the next 7 quarters. The end of that is an Intel at 1/10 of today's size, if it doesn't go straight to chapter 7.
posted by Sharikou, Ph. D. @ 6/26/2006 10:44:00 AM
80 Comments:
sharikou your diagnosis is plain stupid. Dell is going AMD because initially shipments of Intel's Core 2 CPU's will be limited. Dell will not get enough supplies from Intel - so to stay competetive they need AMDs.... otherwise they will be forced to sell obsolete Pentium 4 stuff while competition (Apple, Lenovo) will receive supplies of Core 2 CPUs....
its funny, that the most reliable source for this statements is theinquirer, and we know, everything that is posted in theinquirer is true...
Dell is going AMD because initially shipments of Intel's Core 2 CPU's will be limited.
The writing is on the wall. Dell have seen AMD's 65nm parts, which will be released in just 4 months. I told you 3Q06 and 4Q06 will be the time AMD kills Intel. DELL is just the first step.
the most reliable source for this statements is theinquirer
The problem with nerds is they think SuperPI scores are everything. For businesses, it's a money decision. DELL has nowhere else to go, but AMD.
Sharikou you should do us all a favor and warn Intel about what there future holds. If you do not I and many others will hold you solely responsible for the collapse and demise of the CPU industry, without Intel there is no need for processor research and development at AMD...LOL, are you kidding. There is no way for AMD to supply all the processors in the USA, let alon the universe... That is what you were hoping isn't it...j/k.
There is no way for AMD to supply all the processors in the USA
That's outdated information. Right now, AMD's capacity can supply 30% of the world's demand. By the end of 2006, 55%. By 2008, 100%.
I really dont hope Intel go bust
There will be no competition then:/
Thats means extremely expensive cpues:P
The Best would be a 50 50 situation which would make sure both company is strong and able to compete:)
@Teroedni
I really dont hope Intel go bust
Once a company like Intel goes down, it won't be able to come back. It's like the crumbling of Roman empire. Intel folks don't have capacity to fight in a fair and ethical way.
You better check how much AMD's stock has fallen today, then tell us who's going bankrupt.
In next 12 months AMD will not be able to supply 65nm CPUs in volume....
How many 65nm factories do they have now?
Once a company like Intel goes down, it won't be able to come back. It's like the crumbling of Roman empire. Intel folks don't have capacity to fight in a fair and ethical way.
it takes a bit more than several bad quarters to bring intel down. you forecast the death of intel in 3q06 and 4q06:
I told you 3Q06 and 4Q06 will be the time AMD kills Intel.
we will remember you at the end of the year about this statement, if intel is still around.
by the way. you already noticed, that amd has only few new products in the pipeline for this year? the next interesting ones are k8l and 65nm process somewhere at in 1q07...
wow... a company that has 8.66 billion net income last year will go bankrupt, and a company that has 96 million net income will suddenly flourish . nice logic you got there...
"sharikou isn't paid by AMD, even AMD is shaking their heads and saying "WTF is this guy smoking?"" - shinigamiX
"The prices of its chips have crashed by 60%"
ahem, outdated chips, which soon to be replaced by Core 2.
you DO know that intel has businesses other than processors?
"Right now, AMD's capacity can supply 30% of the world's demand. By the end of 2006, 55%. By 2008, 100%."
oh right. proof? don't quote one of your own blogs, because those are not considered substantial evidence.
"Once a company like Intel goes down, it won't be able to come back. It's like the crumbling of Roman empire. Intel folks don't have capacity to fight in a fair and ethical way. "
humm...a big company has everything to lose, and nothing to gain, while a small company has everything to gain, and nothing to lose.
are you sure this logic is correct?
i wonder where you get that PhD, Palacky University of Olomouc for mentally challenged?
Sharikou
I do have a vested interest in AMD, but your analysis of what Intel should have done is flawed. Intel can't go around raising prices on an inferior product just because they can at the time. It will create so much bad-will that they will never recover from it. Not to mention legal issues because Intel is still more or less a monopoly...
Mike
"oh right. proof? don't quote one of your own blogs, because those are not considered substantial evidence."
Well, it's good to see the idiot intel fanboys are out in full force today. Nice to see you demand some proof (which, I'm sure Sharikou will be more than happy to supply you with) without actually providing any of your own in defense of Intel. Intel is a dying breed, biting the hands that feed it. AMD will be there to take those very same hands -cough Dell cough- and run Intel into the ground.
Sharikou, do you honestly think that a near-total AMD monopoly is going to be a good thing? I mean, which of the two following scenarios do you think is more likely?
A) AMD carries on making super-innovative processors at low, low prices because, gosh, they're just such great people!
B) AMD realises that they have no competition in the desktop market, pump prices sky high and only release new processors every few years - driven by IBM's competition in the server market more than anything else - because there's really no point putting much effort in now that the only alternative is Via's niche offerings.
Not to mention that a total AMD monopoly would last all of five minutes before the DoJ split it up.
It is funnny to read comments from some pathetic 70 year olds (who can barely walk ..let alone use a computer ) who brought Intel stock in 1985 ..
"you DO know that intel has businesses other than processors? "
Yea... the majority of them lose $$!! They have other businesses to uphold the monopoly they have established.
"wow... a company that has 8.66 billion net income last year will go bankrupt, and a company that has 96 million net income will suddenly flourish . nice logic you got there..."
Free cash flow, money owed in taxes, and expenses needed to keep all fabs running will dwindle down that cash flow quick.
btw... last I checked, AMD has $165million in profit just Q4'05.
Why are Intel fanboys so keen on AMD dying and disappearing off the face of the earth?? The only reason we all have cheap CPU's today is due to the competition of AMD...
"I really dont hope Intel go bust
There will be no competition then:/
Thats means extremely expensive cpues:P"
I'm sorry, but anybody who believes that 1 company means higher prices, is an idiot.
Look at Wal-Mart for example. Wal-Mart works by either A) Buying the manufacturing companies themselves or B) Forcing them to sell at a lower price or they won't do business with them, thus forcing them out of business. Wal-Mart is lowering prices constantly (albeit 1 or 2 items a week) and they have no competition (K-Mart, Target...don't kid yourself).
Wal-Mart even runs the food supplies to some foreign 3rd world countries, basically OWNING the counties. Wal-Mart is almost THE sole food supplier for groceries...are they outragously priced? No. If AMD were the only company, you'd have CPU's that were the same price now...why? Because raising the CPU's all of a sudden would piss people off and the BBB would likely step in (how much they do is unlikely).
The EU and JU would also likely stop AMD or any corporation from pricing too high in a sizable market. Do not think because a company is alone they charge alot for their product all of a sudden, though Intel most definately would ($999 Pentium 4 = perfect example).
"humm...a big company has everything to lose, and nothing to gain, while a small company has everything to gain, and nothing to lose.
are you sure this logic is correct?"
Read your history; the logic is indeed not only correct, it is well proven.
Look up:
Dinosaurs vs. Mammals
Empires vs. Upstarts
It is only in the modern age of total corruption that we see so-called "imcumbents" winning vs. their political foes. This is due to rigging the system via gerrymandering and other corruptions.
Hence in the political world, if Intel keeps bribing governments all around the world, they will be harder to win against.
In the technical world, besides a refreshed Pentium Pro design, there is little at Intel to be proud of. The talent is gone. The culture of innovation is gone. The company is dead on the inside even if its momentum and mass carries it forth in the market. Intel is soul sick.
AMD is the new beginning with all the virtue and energy of new life. Barring further illegal deception and corruption from Intel, AMD will be the winner.
"I told you 3Q06 and 4Q06 will be the time AMD kills Intel."
we will remember you at the end of the year about this statement, if intel is still around.
Intel won't be dead at the end of 2006, but it will be mortally wounded and never able to recover. The end time is early 2008.
"Right now, AMD's capacity can supply 30% of the world's demand. By the end of 2006, 55%. By 2008, 100%."
oh right. proof?
Try google "AMD capacity" and hit the News button
Triple whammy loss for Intel.
One Dell-AMD pc sold in lieu of an Intel equates to Intel's lost sales in cpu, chipset and MB.
I don't have a problem with this site, but I'm sorry I have a problem with a total bag site on Intel. Intel is good for competition (and yes, I use an AMD chip) and I wish you'de start writing some interesting stuff rather than garbage dissing intel constantly. It's become boring - not to mention anyone who looks on both sides (like myself) finds it insulting.
Sharikou, knows he is right just by the number of Intel fanbois screaming 'it aint so'. If he was wrong they wouldn't even pass water on this blog.
I wonder how many employees of Intel's HQ are writing those comments. Should check their IP.
It is not surprising that once a company even a large one is on a downward spirel they mostly likely never be able to break out of it. You either have to make some sort of crash landing and hope everyone lives or change the pilot and hopes he can bring the plane out crash and makes it to the airport.
Intelis probably hoping to survive the crash. Then face their shareholders with the words, 'There, that wasn't so bad? Right?' And hope they can get by like AMD did during AMD's black years when they owed so much cash.
http://www.theinquirer.net/?article=32617
I guess Google dumping AMD is just a last gasp from Intel? We'll wait a few years, but I seriously doubt we'll see Intel bankrupt. It's not like it's a good thing for the economy anyways. Intel does a lot more than microprocessors and the ripples would be felt in every industry throughout the world.
On capacity, you may say AMD will have the ability to supply 55% of the market by the end of this year, but I don't see it. Certainly the fact that AMD needs to drop all 1MB desktop processors and cut the S939 line a year earlier than originally planned doesn't speak to a company with spare capacity. Neither does it build consumer confidence especially in the case of S939 support, which I believe AMD originally promised well into 2007. (I guess it's still true althoug the selection is limited to say the least.
Dell is partly going with AMD because with Vista coming out Intel's IGP is not making it for heavy use desktops. As a developer I hate Dell because of their onboard video and cheap prices (companies think it's OK to lessen productivity if it saves money).
I don't see them going out of business but they're crazy if they think this isn't now a two horse race.
I would say it depends on how much belt tightening they're willing to do.
I'm sure that PLENTY of OEMs would like to stick it to Intel.
Someone please read the last paragraph of this link...
ITWire
Does this show any of you what your own company wants? AMD is perfectly content to stay number 2, or maybe its because they know they cant match capacity. 2 FABS vs 11 or 12 for Intel. 1 fab for 65nm by the end of 06 vs. 3 fabs already up and running.
Alot of you are saying Intel will be bankrupt in 2008, well it takes years to build factories, so scratch that thought from your heads.
Oh yeah, I would not like to be AMD (or a fanboy) in late 2007, 65nm, CSI, IMC on the server class. You all love to talk about AMD and there superiority in servers, what happens if they loose that market share, and still have all those fabs with no one wanting there chips?
Not that this would be a good thing, but how about AMD bankrupt in 2008?
"Oh yeah, I would not like to be AMD (or a fanboy) in late 2007, 65nm, CSI, IMC on the server class. You all love to talk about AMD and there superiority in servers, what happens if they loose that market share, and still have all those fabs with no one wanting there chips?"
I'm sorry that you are a complete Intel fanboy moron and don't know wtf you're talking about. Intel will not have IMC or CSI in any processor until late 2008 in the earliest, and by that time, Intel fanboy, DCA2.0 and HT4.0 + K10 will lay the smack down on any Intel fanboys candy ass.
Does this show any of you what your own company wants? AMD is perfectly content to stay number 2, or maybe its because they know they cant match capacity. 2 FABS vs 11 or 12 for Intel. 1 fab for 65nm by the end of 06 vs. 3 fabs already up and running.
Intel makes NICs, mobos, chipsets, IGPs, Flash and CPUs(x86, mobile, etc.) They have to have all that Fab sapce.
Madmodmike, wow you are an angry man...j/k.
I may be an Intel fanboy, but you always want your choice (team) to show improvement. Hence why so many people are excited about Conroe.
Visiting a site like this thats so "PRO AMD", its hard to read any post and try to think that they are posted without any biast.
Then, you make a post like yours and trying to debate any subject (not PRO AMD) is obviously useless.
I had already posted the previous comment prior to correcting the post I made on your site and the article I read said early 2008, so if the article is wrong, my bad, but a moron NO.
Let me ask you and all the other people her one question...
Does Intel have the resources and development ability to make a chip with an IMC and CSI that would compete with "DCA2.0 and HT4.0 + K10"?
Not trying to argue, but debate.
I may be an Intel fanboy, but you always want your choice (team) to show improvement. Hence why so many people are excited about Conroe.
That's OK. But Intel was obviously cheating, and you Intel fanboys shouldn't endorse that kind of unethical behaviour. Intel is worse than Tonya Harding. Imagine what if AMD does the same: picking up a set of benchmarks which show AMD having 40% lead and then pay reviewers to write about it.
"Does Intel have the resources and development ability to make a chip with an IMC and CSI that would compete with "DCA2.0 and HT4.0 + K10"?
Not trying to argue, but debate."
Intel doesn't have the people to do any advanced work because good people don't want to work at Intel.
Intel's culture is morally bankrupt and their business practices are based on deception, bribes, and illegal activities.
Maybe Intel gets lucky and builds a decent chip sometime. Good people who know the facts will not buy from Intel because Intel poisons the computer industry with their evil ways.
One must remember that it is Intel's sickness of the soul that got them to this bad place. Intel fired all the smart people who would not say "Yes, my Neo-Fuhrer!" to Intel's militant CEOs. So all that is left is a bunch of dummies and some Isrealis who want to play corrupt chip monopoly.
Intel is a good example, right in front of us, of what happens when you sell your soul. Rot, disease, and malaise set in quickly. Hey, that's the New World Order.
"In the technical world, besides a refreshed Pentium Pro design, there is little at Intel to be proud of. The talent is gone. The culture of innovation is gone."
How do you know? Do you work at Intel, or do you used to work there?
I think P-III and P-M are nice chips, and Conroe's not bad, either.
AMD is going down and down.. and will be possibly bankrupt before end 2007. Didn't you see the signs?
And there's nothing you can do about it but spread slander and blanket statements about Intel. That's pathethic!
I think P-III and P-M are nice chips, and Conroe's not bad, either.
Conroe hasn't even been released and you're backing it already.
Fair enough if it's proven, but it's not.
At least Sharikou provides evidence and proof (especially when asked). So if you guys wanna debate, debate with a little bit of evidence at the slightest bit? Even if it is the Inquirer.
I will LMAO when the Intel's Nehalem will simply obliterate AMD.
I'd also like to add that this is speculation. Intel could go bankrupt today, tomorrow, next week blah blah. No company is safe from bankruptcy these days and I'm sure you can find plenty of info on it.
The author has obviously written it in a way to predict given the CURRENT information.
Tables could turn, who knows. Conroe is the "next" biggest thing, it's only a month away so let's see what happens.
"How do you know? Do you work at Intel, or do you used to work there?"
I have friends and acquaintances that used to work at Intel who give me all the dirt. Some of these people were senior Intel employees. They know what is going on.
There is not one ex-Intel person who I know who likes how the company is today. Most of these ex-Intel staffers are rooting for AMD as they view today's AMD in the same light as Intel when they used to work there. AMD is innovating and making things better for people and mostly Intel is not.
If you want some more information how Intel is a rotten company, check out this site. The information is a bit dated, but it is quite chilling nonetheless.
I guess Google dumping AMD is just a last gasp from Intel
did you read this update?
This news might delay Intel's bankropcy in few months....
*DJ Marvell To Pay $600M To Intel
*DJ Marvell To Also Assume Certain Liabilities Under Purchase
Intel's communications and application processor business includes approximately 1,400 employees involved in a variety of functions, including engineering, product testing/validation, operations and marketing. It is expected that the vast majority of these employees will become employees of Marvell.
I guess Google dumping AMD is just a last gasp from Intel
did you read this update?
I hear Intel put a ton of work to custom design systems for Google.
so Intel is now ready to do custom designs for ONE cutomer's requirement, to roll back some of the PR whiplash. hmmm, we can only guess how Intel must have lost on this deal.
you DO know that intel has businesses other than processors?
err, did you mean this or this or this
Visiting a site like this thats so "PRO AMD", its hard to read any post and try to think that they are posted without any biast.
bias, did you mean a post like, like this. dont forget to the check the date of postings
or you meant these posts. the dates matter here too.
Intel should sue Sharikou one of these days for being there before the press
"Intel has lost the ability to be innovative, and that is why AMD has been the performance leader for 3 years, and will be for some time."
Thats why AMD is still using 90nm, its because their so innovative...lol
I know they have a new design coming on 65nm, but how long will they use that one, Intel is supposed to have a new design every two years (so they say). It will be great when they beat AMD and have no claims of faking/falsifying (though I sense a site like this one would still find a way) the results.
It would be great for Intel lovers and all of you AMD lovers if this happens, then AMD would have to respond, thus driving development.
Sepculation - AMD's going bankrupt before end of 2007 if the price wars keep up and Intel continously pump the market with cheaper 65nm processors in big volumes.
Remember the transputer? Remember the ARM processor? Remmeber MC68K processors? Remember the PowerPC? Some prime examples...
AMD expects to have the capacity to supply a third of the market for the dominant x86 microprocessor by 2008
obviously Preston Snuggs, Manufacturing Systems vice-president at AMD is a liar bought by intel.
AMD expects to have the capacity to supply a third of the market for the dominant x86 microprocessor by 2008
That was old news. AMD changed its plans.
You dudes are not good at google. See this
http://www.itwire.com.au/content/view/4749/53/
"Conroe hasn't even been released and you're backing it already.
Fair enough if it's proven, but it's not."
Sure, it's not "proven" because it's not released, but come on! Do you really think Conroe will be slower than AM2 when it's released?
From the beginning I've been saying that Conroe would be 10-15% faster than current Athlon64 in average. I also think those "Conroe's 20-40% faster" and "Intel's beating AMD" sites are BS, but I stand by my prediction that Conroe will be (proved) faster, in average, than today's Athlon64, when it is released.
You may say "it's no big deal after 3 years," or "its availability will be limited," etc. Fair enough, and maybe true. But still, again, Conroe will be 10-15% than current Athlon64.
Woodcrest OTOH might be different, since server apps are different from desktop ones. K8 is originally a server chip, where Core2 was from desktop/mobile. Woodcrest might have only small if any advantage on high-duty web, DB, or data-center performance. Sure it looks hell better on specint & specfloat. I still remember SUN or IBM claiming world records on those benchmarks by twicking one compiler option.
"obviously Preston Snuggs, Manufacturing Systems vice-president at AMD is a liar bought by intel."
And you have bad logic.
The ability to supply 1/3 market share BY 2008 doesn't imply the inability to supply 30% market share in 2006, it also does't rule out the ability to supply 100% market share in 2008.
"it also does't rule out the ability to supply 100% market share in 2008."
WOW, 2 fabs to produce 100% market share. FAB 30 must have put magical dust on their chips because only here in Sharikou land can you hear such fairytales.
2 Fabs = 100% fairytale future.
11 Fabs = 75% - 80% currently.
I know I'm not good at math, and obviously you are, so I'll believe it cause I heard it here...lol
"I know I'm not good at math, and obviously you are, so I'll believe it cause I heard it here...lol"
Not only you're not good at math, but also logic, and probably also (middle school?) reading.
Go back and read my comment again. I said it didn't rule out the ability, I didn't say it will have that ability.
I don't know whether AMD will have that capacity, and I don't care; no matter how much capacity it ends up having, as long as it's higher than 1/3 world demand, it has no conflict with Preston Snuggs's claims. The guy who said Preston Snuggs is a liar bought by intel apparently has bad logic and reading capacity, just like you.
As for my own estimation, AMD could support up to 50% world demand with fab36 + fab38 + Chartered. Intel has so many fabs because 1) it manufactures a lot more than processors, 2) it constantly rotates fabs to upgrade them, 4) some fabs are used for R&D and not for production, 3) it probably has lower yields than AMD.
Both sides of the fanboy arguments are wrong:
"By 2008, 100%" (referring to AMD supplying the world chip market)
How? F36 will be fully ramped, F30/38 will be in conversion and ramping 300mm, NY fab will still be being built. Factor in an increasing mix to dual and quadcore further reducing # die/wafer. How are 1.5-2 300mm fabs going to handle the entire world supply?
"http://www.itwire.com.au/content/view/4749/53/"
This article makes no mention of 100% of world supply!
On the other side of the equation:
"11 Fabs = 75% - 80% currently."
The 11 Intel fabs are not producing only chips - there are basically anywhere from 3-6 fabs doing this (others are doing development, chipsets, flash and other misc products).
edward you said "Go back and read my comment again. I said it didn't rule out the ability, I didn't say it will have that ability."
It does rule out the ability, 2 Fabs even as of right now can only supply 30% as was stated, the New York fab will take a while to be completed, at a minimum 3 years. We are nearing the middle of 2006 + 3 = middle of 2009 at a minimum.
Hence my numbers are fine, and your still in diapers, and wearing a bib. Do you want me to read you a bed time story...something with small words?
"edward you said "Go back and read my comment again. I said it didn't rule out the ability, I didn't say it will have that ability."
It does rule out the ability,"
What's the problem with you? First you have poor logic, then you don't read well, now you simply can't read.
Do you know what the "it" was in my original comment? "It" was Preston Snuggs's claims about AMD supplying 1/3rd of world demand. "That" doesn't rule out the ability for AMD to supply 100% market. Not the number of fabs, not whatever craps you're saying, but that (again, that's Preston Snuggs's claims).
I personally made no claim about whether AMD can meet 100% world demand or not. I said I don't care, I even said my estimate is up to 50%. Don't you have eye balls on your head? Or do you have a brain inside it?
It's hard for me to understand how the expensive American education system can come up with morons like you. It's really a shame. You should indeed go back to your 9th grade and learn how to read and do abstract thinking.
"It's hard for me to understand how the expensive American education system can come up with morons like you."
The American "Educational System" is a joke.
This is what I thought you had posted.
"Edward said...
And you have bad logic.
The ability to supply 1/3 market share BY 2008 doesn't imply the inability to supply 30% market share in 2006, it also does't rule out the ability to supply 100% market share in 2008."
This was your comment yes? I will not say anything more until you answer.
If this was not your comment I will appologize for everything I said.
Pls explain to me why AMD's VP would say 30% when they can actually do 100%.
Search "capacity" on this blog, you will find articles analysing AMD's capacity. It's just a matter of calaculating how many good dies AMD can produce given the information on die sizes, wafer outs and estimated yields.
Why would AMD say 30% if they can do 100%?
My guess is the discrepency was because of the anti-trust lawsuit. If Hector Ruiz said AMD could get over 50%, Intel lawyers would tell the judge: see they can do 50%... case should be dismissed.
I really doubt AMD can collect any money from Intel at the end of the trial -- Intel may have BKed well before that.
What happens to AMD capacity wise short term post Intel's July 27 price drop on dual core.
You should read AMD's slides in their June 1 tech analyst meeting and pay attention to AMD's expansion plans. Basically, by now, AMD's capacity has increased by 50% from 1Q06 level, by 4Q06, AMD's capacity will more than double from the 1Q06 level. As for price war, Intel is commiting suicide. AMD's ASP was $95, Intel was $150. AMD's cost per CPU was about $40. I projected that AMD will exit 2006 with 40% market share with 50% an attainable goal.
"AMD's ASP was $95, Intel was $150. AMD's cost per CPU was about $40."
What is Intels cost per CPU? I have heard you say this many times, yet you never state Intels price, to back up your statements please inform us on Intels price.
Thanks
What is Intels cost per CPU?
AMD's cost was easier to estimate, because it only makes CPUs.
I estimate an Intel CPU+Chipset combination costs $140.
"I estimate an Intel CPU+Chipset combination costs $140."
But arn't we talking about CPU's?
Wouldn't it be fair to assume Intel would be equal if not less, due to the fact of 65nm vs 90nm?
More chips per wafer, right, also 300 vs 200 on wafer size?
When you calculate ASP for AMD why not include the chipset there as well, the end user would have to purchase one anyways, or would that raise the ASP for AMD, and then it would not look so good?
Thanks.
Wouldn't it be fair to assume Intel would be equal if not less, due to the fact of 65nm vs 90nm?
More chips per wafer, right, also 300 vs 200 on wafer size?
Dude, you don't have a clue at all. Intel is already at 65nm. It's chipsets are made on 90nm now. Intel gets no cost advantage from 65nm, because its chip die sizes aren't shrinking. From 130nm to 90nm to 65nm, Intel chips are at pretty much the same die size. This was because Intel's outdated architecture requires more and more cache to compensate its architectural inefficiencies. A Conroe has 4MB cache, its die size is over 160mm^2. An AMD 90nm X2 5000+ has a die size of 180mm^2, because the X2 5000+ only needs 2x512KB cache. There is also the question of yields. Apparrently, Intel's yield is only a fraction of AMD's yield. You have to rememebr AMD took 22% of the PC market with one 200mm FAB.
AMD doesn't make PC chipsets. It makes some server chipsets, but not many. Intel makes both CPUs and chipsets. With the estimated $140 for the CPU+chipset pair, it's safe to assume the CPU costs $120.
"Dude, you don't have a clue at all."
No I dont, which is why I read blogs and ask questions, to understand something I know very little about. But if you would like talk about tennis or golf I'm sure I could make you feel pretty fuc##ng stupid, so don't be an A$$hole?
I do not know about die sizes and the intricacies of the processors, I have only picked up little bits here and there. I have also noticed how biased you are. You really hate Intel, so anything you post has to be looked at very closely, and taken with a grin of salt.
Thanks for answering the question though.
"I do not know about die sizes and the intricacies of the processors, I have only picked up little bits here and there. I have also noticed how biased you are. You really hate Intel, so anything you post has to be looked at very closely, and taken with a grin of salt."
No need for name calling here kids... :)
To get a rough idea of what Intel is going to struggle with. 90nm, 65nm, or 45nm... With a 1MB L2 cache of the past, the L2 cache took up close to 50% of the die area.
Them going to a 4MB L2 means that is EATING UP prescious real estate (300mm silicon) per CORE!! You can make an educated guess here that the L2's x2 (8MB Total) is close to ~80% of their die, with 20% being logic functions of the microprocessor.
This all translates into an enormous die ($$COST$$)... which is the reason they MUST transition to 65nm, and 45nm ASAP!!
AMD had one 200mm fab (Fab30), running 90nm parts and took 22% market share alone!! They don't need to increase their L2 to insane amounts due to better design(less $$).
When they go to 300mm, they get 33% more chips just by switching from 200mm to 300mm (12inch wafers). When they go to 65nm, they get even MORE Dies Per Wafer.
Now factor in, Fab36 ramping up, Chartered, and in the future... Fab30-Fab38 conversion, plus the NY Fab in 5 years, you get an idea of what AMD is capable of.
When they go to 300mm, they get 33% more chips just by switching from 200mm to 300mm (12inch wafers). When they go to 65nm, they get even MORE Dies Per Wafer.
Actually, a 300mm wafer has more than double the area, and 65nm reduces die size to 50% of 90nm. For the same chip, a 300mm wafer at 65nm produces more than 4x the chips you get from a 200mm wafer at 90nm.
"No need for name calling here kids... :) "
Thank you for the explanation.
"With the Pentium brand available at 50$ and dual core Pentium's at 90$, if the large consumer base in emerging markets like Asia go dual core crazy...assume by year end 70% of the desktop market is dual core, ..."
Not even $90 is low enough for the "emerging markets!" Do you expect 70% of those families starve for a whole months just to buy one CPU?
$50 or below is a more reasonable price for the emerging market. Tell you the truth, if they could, they'll even save a $15 fan and keep $30 instead of buying a high-grade power supply.
Pentium-4 doesn't have a big share in the emerging market... unless all those people get rich both economically and energy-wise and the world becomes a wonderland.
"Actually, a 300mm wafer has more than double the area, and 65nm reduces die size to 50% of 90nm. For the same chip, a 300mm wafer at 65nm produces more than 4x the chips you get from a 200mm wafer at 90nm."
Actually you keep forgetting that their 300mm fab will run ~50% fewer wafer starts per month than their 200mm fab (something you continuously ignore); ao they may get 4X die/wafer by your math but they will not produce 4X the die.
You also assume yield is the same between 200mm and 300mm.
You also assume the die size will keep scaling 2X each generation and yet ignore higher quantities of dual and quad core vs single core will mean die size will not go down 2X each generation.
You also assume next gen architecture (K8L or K10)will not increase die size - almost every new architecture ahs more transistors/die.
Other than that your calculations are spot on, as always....
"Edward - do you live in an emerging market or been to one recently? Perhaps the largest in the world like China or India? Starving people aren't buying PCs."
Sorry, but you have no clue. Starving people won't buy PC, but people will starve for a few days (NOT a month!) to put their children into college. That means they will buy a PC. Yes, I live in the U.S., but I've seen those examples with my eyes (thankfully I must travel and know much more around the world than you do).
The "middle class" who are getting financially rich in those areas (China & India etc) wouldn't care about the $100/$150 difference, if they can easily afford a $500 or $550 PC and the latter brings better value. The majority below that class, OTOH, will rather buy $40 instead of $80, will save $15 on fan, will save $30 on power supply. Fortunately, Dr. Hector Ruiz has a much better idea on the emerging market than you.
"My question was if in 2H 70% of the market toggles to dual core, with data pls tell me whether AMD have the capacity to supply their WW demand."
First, there's no way 70% of the market switch to dual core in 2H of 2006. People making such assumptions must be living in the enthusiasts' world. ;D
Second, almost 60% of the Athlon64 die area is the 1MB L2 cache. That means AMD saves die space 30% by cutting L2 cache to 512KB, and 45% to 256KB such as in dual-core Semprons. Thus AMD can produce as many dual-core Sempron64 as (single-core) Athlon64. That saves power, too.
"Actually you keep forgetting that their 300mm fab will run ~50% fewer wafer starts per month than their 200mm fab"
The bottleneck is not the number of wafers per hour that go through the pipeline (which is about the same for 300mm and 200mm wafers), but the rate they can test & verify the fabricated wafers. (Why would they have 4x testing capability for 200mm plant? They wouldn't.) As demand increases, AMD can remove the bottleneck by simply increasing that capacity.
"You also assume yield is the same between 200mm and 300mm."
Why not? Won't be initially, but why not within a month or two? I've seen experimental data that show wafer size pretty orthogonal to yield.
"You also assume the die size will keep scaling 2X each generation and yet ignore higher quantities of dual and quad core vs single core will mean die size will not go down 2X each generation."
eh.... don't know what you're talking about... :p
"You also assume next gen architecture (K8L or K10)will not increase die size - almost every new architecture ahs more transistors/die."
True, K8L will need more die size, just like Kentsfield.
"I also did not ask you if you thought the market would reach 70% dual core or not. I asked what if it does..."
My point was, since your assumption was false IMO, your question was bogus.
"and clearly AMD has announced a price drop on their Athlon dual core to combat the Intel price drop. I don't re-call them making any announcement of a Sempron dual core"
That's because AMD thinks that the moment the emerging market don't need dual cores. They need a nice, power-efficient computer. That could change in 2008, or maybe later 2007 though.
Besides, from your description (that you're living in and go visiting a PC market regularly), we seem have different definitions on the emerging market. I wonder what's the percentage of families already having a PC in where you live? If it's above 30%, I wouldn't consider it an emerging market.
"The more I see your responses the less interested I am in your perspective...or lack of it. I am interested in facts and intelligent analysis."
Just because you wouldn't listen to and accept them, doesn't mean that they are not facts and analyses.
Mojo - please see this news about Athlon64 X2 with 256KB L2 cache. It'd be interesting to benchmark this $140 CPU against the $90 P-D, both performance and power-wise.
You still don't think my analysis on cache size was valid? I just did it one step ahead of this news, man.
Mojo - FYI (information or interest, whatever), here's some more analyses on AMD pricing & dual-core market share percentage.
We know that AMD's ASP was $95; let's make it simpler to $100. Lets classify AMD processors to 3 classes:
1)The dual-cores and high-end single-cores, now priced from $300 to $1000; lets arbitrarily & conservatively assume the ASP of those is $400.
2)The better single-core CPUs ranging from $100 to $300; lets conservativley assume all were sold at $100.
3)The low-end CPUs from $50 to $100; again conservatively assume all were sold at $50.
How many more "class-3" processors above must AMD sell, compared to the "class-1" ones, in order to achieve $100 ASP? A simple calculation will give you about 8 times. That is, those high-end single cores and dula cores together occupy less than 15% market share today.
Plus, that's an extremely conservative estimation. A more likely reality is less than 10% at this point in time. Even with the coming 40% price cut, dual-core market share will still be less than 25% (assuming 2.5x as many dual core were sold). Do you still expect the market share of those jump to 70% in 2H this year?
Please just promise me that if after 7 quarters Intel is NOT bankrupt then you will go away.....
Intel won't be dead at the end of 2006, but it will be mortally wounded and never able to recover. The end time is early 2008.
Will you stop blogging your useless information if this speculation of yours is proven wrong..... This means we only have roughly a year and a half more of your garbage to deal with.
"The bottleneck is not the number of wafers per hour that go through the pipeline (which is about the same for 300mm and 200mm wafers), but the rate they can test & verify the fabricated wafers."
So if they are bottlenecked by test capacity why ramp more Si factories? Why go to chartered? Folks keep saying an added 300mm fab will 2X capacity but it will only do that IF the wafer starts are the same and the yield is the same. AMD has stated that the F36 WSPM capacity will not be the same as F30 - how can you not understand this extremely simple math?
""You also assume the die size will keep scaling 2X each generation and yet ignore higher quantities of dual and quad core vs single core will mean die size will not go down 2X each generation."
eh.... don't know what you're talking about... :p"
Just so I understand a migration to a higher mix of dual and quad core will not increase the average die size? And an increase in die size will not reduce overall capacity?
...you're right I don't know what I'm talking about...can anyone else help me out here?
"you contradict yourself by saying AMD will churn out Sempron dual cores if needed then say AMD doesn't think emerging markets need dual cores."
So there's no dual core Sempron, but just one Athlon64 X2 with 256KB L2 cache. Where's the contradiction?
"Assuming Intel has 75% MSS, this puts close to 60% of the market on dual core. With that kind of momentum, having the market at 70% dual core is not a stretch."
Ah! So your dual core occupying 70% world PC market share is based on a stretch on Intel's analyst day intent... why didn't you just say that in the beginning, so people could've known where the silly claims came from, and you could've stopped those bogus hypotheses and name calling on fanboyism?
"Now, I know you know exactly what Hector will do but I don't think your AMD fanboyism is going to have Otellini calling you to consult on how he intends to do this."
How do you know that, and what the h*ck are you talking about?
"If the best product were to win AMD would have 100% server/desktop and Intel 100% notebook market ownership."
Actually if you're talking about price-performance, it should be 100% Opteron on servers and 100% Turion on mobile.
"All this is background - my questions remains what will Hector do?"
You really should try to organize your arguments more logically - you don't argue correctly by just saying a bunch of backgrounds.
1) Intel won't release dual-core Celeron anytime soon, and AMD won't do that for Sempron, either.
2) Intel's going to sell Core2Duo under $160, and AMD going Athlon64 X2 under $140.
3) Intel's going to sell P-D under $100, but those are... craps. The only thing left in those are the 'Pentium' brand.
So all that you said was that Intel's going to lower price on P-D and attack by basically selling that brand name. All your arguments in the end only want to show that P-D will be the AMD-killer because it will carry the burden of 70% world PC market share.
However, you didn't question whether Intel could make that many P-Ds; you didin't question whether consumers (even those who are technical-challenged) would be willing to buy old generation PCs. If I follow your logic correctly, it's like --
a) "Intel's analyst day says 75% dual-core mix" -> "there must be lots of people buying P-D"
b) "P-D doesn't perform well" -> "it's okay because it's cheap and for emerging market"
c) "Intel has so many fabs" -> "it must be able to produce enough P-Ds"
Maybe it's me being picky, but I'm not convinced by such statements, though.
Post a Comment
<< Home