Intel yard sale continues
The Dialogic unit, which Intel bought for $777 million was sold for $50 million. 600 workers were dumped. You wonder why Intel needs to sell this business for such a tiny amount and book a $0.7 billion loss. $50 million is just the price of 50,000 Conroe XE CPUs.
Intel has to sell 10 Con XE 6800 CPUs per month to make enough profit to feed one worker, unfortunately, Intel's FABs are unable to produce so many Conroe CPUs. It can make 2000 Netburst CPUs per worker per month, but the profit is not enough to feed one man and his family. Now, the 600 men division had to find a better owner.
The fire sale of these business divisions further proves that Intel is doomed. Intel is trying to improve efficiency. However, Intel's problem is not losing small money on these tiny side busiensses. Intel's problem is its main businesses are making less and less. These 600 workers may have incurred some loss, but it should be very small. The biggest loser in Intel is the flash business, and Intel isn't selling that to Spansion. In fact, Intel is pouring more money in the NAND ventures to compete against Spansion's NORAND. To solve Intel's problem, it must improve efficiency in its micro processor business. I can tell you that can't be done within 6 months. AMD spent three years on that and the process is continuing. Intel can't just lay off a bunch of workers and become more profitable. For instance, suppose there are 5000 Netburst design engineers at Intel, which of them can Intel lay off? That's a tough question. Each of these engineers may hold a tiny piece only he knows. And 65% of Intel's CPUs are Netburst. You can imagine that the Netburst camp still owns Intel today. You wanna fire a Netburst crew? You lose 65% of revenue and die. You bet on Woodcrest? Woodcrest is 2P only, and AMD will own 100% of 4P. So Netburst people are all in a life-death fight against CORE2 folks for their jobs and existence.
So, there is no way for Intel to improve profitability by cutting costs as its CPU revenue keeps declining at a fast rate. AMD will exit 2006 with 40% market share (run rate), Intel will be about 55%. Intel will see both units and unit price crashing all over the place. The result will be massive operating losses from 3Q06 onward. BK time is 5 to 7 quarters.
In other news, the AMD Athlon 64 x2 3600+ PIB is released. Expect massive flood of DELL and Lenovo AMD64 desktops and Turion X2 notebooks soon. AMD is heading to 40% market share and Intel heading to 50%.
posted by Sharikou, Ph. D. @ 8/09/2006 06:34:00 PM
87 Comments:
--Clueless as usual. Q3 forecasts show $8.6B revenue and ~49% gross margin, profits should be seasonally up. Hardly a losing proposition.
On the other hand, we have your genius prediction that a Q206 GAAP loss was pretty much "in the bag". What happened to that?
This means more cash with which to grind AMD to the bone with in a price war. If this is war of attrition, then I'd put my money on Intel, not the company who just sold their soul to the devil to stay in the platform game...
It’s sad to see Intel has failed in so many endeavors. Does anybody know how many businesses Intel has dumped? Maybe there business model doesn’t work for other types of manufacturing.
What a fanboy you are. AMD's ship has sunk with Conroe smashing it silly. Your ramblings and inane stories look like dementia to me.
The price point for the X2 3600+ has also been released.
http://www.hkepc.com/bbs/itnews.php?tid=647389&starttime=0&endtime=0
It's going to be $148. At that price, I can't see it selling with the X2 3800+ only like $10 more expensive. I also can't see it putting much pressure on $93 805Ds or $113 820Ds, although it might put a bit of pressure on $133 915Ds. Still if AMD wants the X2 3600+ to really sell they need a $10 discount into the $130s. At $148 the processor is really a nonissue to any current product.
INTEL FOREVER!!!!
Pathetic way to put it! 800 million dollars isn't enough to feel 90000 employees? Are you out of your nuts?
65 Watts? The other processors I see are 89 Watts or 62 watts. What is this?
cu65 Watts? The other processors I see are 89 Watts or 62 watts. What is this?
That's ok. a lot of really good talent is already looking to leave before they get a foot in the ass. That should help Intel feel better about profitibility.
Employers should keep their eyes peeled for some of the good talent leaving voluntarily.
New joke from Longan:
OMG! Sharikou got it wrong. Intel paid $.78 billion, not $.60 billion according to this link http://www.theinquirer.net/default.aspx?article=33575
Come on, all Inhel-fanbois unite. We can bring down Sharikou now.
-Longan-
P.S. I am wondering what kinda loss Intel will file for this quarter report after all the sale/loss were accounted for, plus the cost for severance paid for the laid off... Gonna be an ugly quarter report....
Check Intel's stock-holders' equity. That has been shrinking since 2005 to 2006. I don't understand how a company reporting billions of dollars in profits and yet the stock-holders' equity is shrinking???
Q3 forecasts show $8.6B revenue and ~49% gross margin, profits should be seasonally up. Hardly a losing proposition.
On the other hand, we have your genius prediction that a Q206 GAAP loss was pretty much "in the bag". What happened to that?
Intel Q3 will see revenue far below expectation as the channel is already saturated, no room to stuff. You will see massive operating losses for Intel.
Intel didn't report GAAP loss for 2Q06 because it delayed the booking of impairment of goodwill to 4Q.
Sharikou: Couple of clarifying questions for your predictions:
- Can you define/give a range of the massive operating losses Intel will see in Q3'06 (preferably before the actual earnings are released)?
- You keep saying you expected impairment of goodwill in Q2 (I think you were referencing the sale of some of the comm business to Marvel on this?). How much were you forecasting this for? You must believe this to be >800Mil, based on your GAAP loss prediction and the fact that Intels net profit in Q2 was >800Mil?
For the non-financial types among us could you define what exactly impairment of goodwill is? Is it what the Intel originally bought the companies for, or the delta between sale price and original purchase price?
Also based on your financial knowledge and keen sense of strategy - did you not have the forsight to see that Intel would delay booking the impairment until Q4?
I recommend Intel continue to flood the channels with Netburst, and make Conroe a vapor launch and continue the Netburst price war (OMFG $50 for PD930), forcing AMD to BK within 5 quarters. Intel fanboy, cheers!
Oh and continue to flood the market with expensive 32-bit Centrino Duo and rip profit from it.
Conroe/Merom should be used as "display unit only" for taking the crown on review sites as it is too costly (low yields) for Intel to mass produce.
Intel has to sell 10 Con XE 6800 CPUs per month to make enough profit to feed one worker
Do you have any numbers on how many ram chips, celerons or any midrange CPU's would it take to support one man? Another nice number would be the total number of CPU's sold by Intel per month.
Btw, assuming that high-end marketshare numbers are roughly the same as for AMD (~0.1%) then I see no prolbems with having enough CPU's to feed its workers and still bring home several hundred millions of profit.
Hi. Everyone realizes this loser is an ex-intel employee who was fired for system abuse? He doesn't even have a Ph.D. Sharikou - move on and get a life. You're pathetic.
So, Sharikou, do u estimate > $1B for impairment losses in 2006?
Intel paid 6B for all the acquistions combined. Some of the acquistions are still going to pan out and will likely erase the impairment from others.
It's odd how these things work out. Intel had trouble competing in the 2003-2005 time frame, but during this time they managed to pretty much deny AMD any portion of the market.
Now the tables have turned, with Intel "leaping ahead" and simply dominating with their C2D. And with this newly released technological wonder, what do they do? They begin selling off non-core businesses and laying off employees. Perhaps it would have made more sense to do this when their product line was not as strong?
And now that AMD is in the far weaker position technologically, they decide to spend $5.4B on ATI. You'd think this move would have been smarter back when they still had the steam from their 2003-2005 technological advantage.
My prediction on market share in the coming few years: things will remain the same. Even if AMD pull a rabbit out of a hat and suprise Intel with some grand chip, it would take them at least two years (possibly more) with the current price war and their $4B+ debt to save enough money to invest in more capacity.
This link: http://www.canada.com/ottawacitizen/news/business/story.html?id=14f8d1be-cc82-4c6a-828f-b33ab5e908bd says the price was $50 million U.S?!
""It was probably a fire sale," said a Bay Street analyst who asked not to be named."
Well...
AMD is seeing shortages for their chips. Search google, the news is all over the place. This is a bad thing, but at the same time, it is a sign of how the market feels about AMD.
You can say that Conroe has "fragged" AMD all you want, but the numbers tell a different story. AMD's chips are selling so fast that AMD can't make enough of them. Conroe is having no effect on AMD's sales at all. With the new fab36 and the new deal with chartered, AMD can make more chips today than they have ever been able to in the past; and by a large margin. Because AMD is running into a couple of shortages, this means the AMD is selling everything they make, and then some. AMD is set for a record breaking quarter in units sold.
Intel really doesn't have a chance against AMD+ATI in the long run. They are simply too far behind. The market tells the true story.
Since the channels are stuffed, it's a great time to buy an Intel processor.
Look for a surge of new, lower cost higher performing Intel based machines to hit retail stores and drive off AMD's market share.
Intel's going to do just fine.
Looks like we need to check the yields of AMD a little more closely...
[LINK]
"Since the channels are stuffed, it's a great time to buy an Intel processor."
I'm not sure of such claim. You sound like a Intel distributor who's being stuffed to choke.
The reason of the channel being stuffed in the first place is that these processors are consumes too much power for the performance. How could you recommand anyone to buy such things and see them obsolete in a few months if not already?
"Look for a surge of new, lower cost higher performing Intel based machines to hit retail stores and drive off AMD's market share."
Lower cost, yes. But new? What about the 4 billion inventory and the stuffed channel? And they do not perform, simple as that. Anything below than P-D 940 is no comparison to X2 4200+ for performance/price.
Yeah, it's a great time to buy a Netburst processor, but why would an educated consumer buy one of those?
dba said: "Based on AMD's roadmap, 3M AMD [64] X2 does not seem satify the market. have aguess which OEM get all those chips? (Dell, lenovo) ?"
True. It's really interesting to see how AMD play things out. When Intel started to cut prices like tomorrow, and everyone expect to see processor surplus, AMD is making deals with these tier-1 pc manufacturers and selling more Athlon 64 more quickly.
People said 90nm vs. 65nm, but who cares? It seems that AMD was right, that they do still have a lot of mileage in the 90nm and K8. They must've given those pc manufacturers very good deal for them to sell Athlon 64 X2 and to deny Intel unloading it huge outdated inventory.
Frankly, I wouldn't recommand anyone buying a P4 or P-D box today. I actually pity them when seeing people buy those. Less than a year they'll regret, feeling the thing too hot and slow (bye-bye to good Vista performance). Less than a year later, they wouldn't even have a place to sell these things, not even salvation army would accept obsoleted PC as donation nowadays.
It's so obvious that Intel do not really want to sell any division now. Compare the 600 to 100k total, you see the point. Intel is only selling negligible part to pose as it's restructuring, and to make its shareholders happy.
The fact is a company has to spend more to lose weight. It's financially bad for short-term but potentially much better for long-term. Intel OTOH thinks itself having short-term problem (Pentium obsoleting while C2D not ramping fast enough).
A simple google shows that in 1998 Intel layoff 3000 employees; in 2001 it layoff another 5000 (source: FACE Intel). Compared to those this 600 is nothing, keeping in mind that the total Intel headcount then was a lot less than it is now.
Intel didn't report GAAP loss for 2Q06 because it delayed the booking of impairment of goodwill to 4Q.
Factual question 1(trying to stay data driven): How do you reconcile delayed booking of impairment with GAAP rules and standards?
Followup, opinion based question:If this violates standards, do you really think that Intel would risk breaking the law just to p
Factual question 2 and hint for #1: Given your existing or newly-formed understanding of GAAP rules, is it possible that much of the impairment has already been recognized since the GAAP rules regarding impairment recognition were last updated?
I eagerly await your response. And Edward and Nyx-spare me the "what's the big deal arguments". There isn't a lot of wiggle room here- accountants, the SEC, and the IRS are really funny that way. :)
It's odd how these things work out. Intel had trouble competing in the 2003-2005 time frame, but during this time they managed to pretty much deny AMD any portion of the market.
Then how come AMD's share in the commercial sector grows from almost zero to over 20%?
..and AMD's US desktop sales from below 50% to 60%+%?
Yeah, it's a great time to buy a Netburst processor, but why would an educated consumer buy one of those?
Most people won't know the difference. IE: Most consumers are not educated. It will handle their home videos, porn, and emailing Aunt Sally. That is what they care about.
You guys are all basing your predictions of the market based on what YOU know about processors, but you're all (especially Sharikou) forgetting who the consumer is.
Intel is a premium brand name in the eyes of the consumer and that's all that matters. The consumer will spend a little bit more for a brand name they trust. You all may be wise to processors, but you all forget the power of branding.
Sharikou's market share predictions might be plausable if even half the people thought like him, but most people don't know or don't care. They buy what's on the shelf and from a name they trust or recognize - and there definitely will be a lot of Pentium machines on the shelf.
Expect AMD's market share to drop and Intel's to increase. Power hungry users will buy Core 2D's. Midrange shoppers will buy clearout inventory of Pentiums. And budget consumers will pick up Dell crap, which is Celeron, and also Semprons from their local builders.
I'm not biased either way, but I do know the power of brand in the consumer's eyes.
A simple google shows that in 1998 Intel layoff 3000 employees; in 2001 it layoff another 5000 (source: FACE Intel). Compared to those this 600 is nothing, keeping in mind that the total Intel headcount then was a lot less than it is now.
This is inaccurate. Intel has already lost 3000 employees: 1400 in the Marvell deal, 600 in the current deal, and 1000 managers. And Paul O said that the process is halfway through. If the other half of the process lays off another 3000 employees, that will be 6000 total, 6% of the work force. Additionally, Intel roughly loses 1000 employees per year through attrition.
And yes, majority of Intel's problems (at least from AMD) are short-term problems. Once C2D ramps up, AMD better run for shelter. And no, K8L will not save the day for them. Simply because, it is an inferior architecture (3-wide issue, not 4-wide; no re-ordering of memory read/writes; no micro-fusion; no macro fusion; only 1 MB of stupid L2 shared cache that is not inclusive), and will be on inferior process at launch. It adds the capbility to perform two 16-byte L1 accesses in once clock cycle--a feature useful in some server workloads, mostly useless in desktop/mobile workloads. What's more, by the time AMD launches the K8L vapourware, Intel launches its 45nm compaction of C2D (Penryn), with probably quad-core being the norm. Kentsfield on 65nm is being launched at 1333 MHz FSB, so my guess is, Penryn will be 1600+ MHz FSB--again may not be good enough for MP servers, but more than enough for DP servers, desktops, and mobile.
BTW, Penryn tapes-out in few weeks. Expect to see some announcements in Fall 06 IDF, some samples in Spring 07 IDF, and again, another round of block-buster reviews following that.
AMD is toast. But does that mean Intel wins? No! Intel already owns close to 80% of the market. To win, it has to expand beyond the current markets. And that is what the reorganization is (should be?) about.
Cray Wins Another Contract w/ Opteron
http://www.marketwire.com/mw/release_htm...
Cray Wins $52 Million Supercomputer Contract With National Energy Research Scientific Computing Center
System at Department of Energy's Berkeley Lab Will Be One of World's Fastest
SEATTLE, WA and BERKELEY, CA -- (MARKET WIRE) -- August 10, 2006 -- Cray Inc. (NASDAQ: CRAY) and the U.S. Department of Energy (DOE) Office of Science announced today that Cray has won the contract to install a next-generation supercomputer at the DOE's National Energy Research Scientific Computing Center (NERSC). The systems and multi-year services contract, valued at over $52 million, includes delivery of a Cray massively parallel processor supercomputer, code-named "Hood."
The Hood supercomputer at NERSC will consist of over 19,000 AMD Opteron 2.6-gigahertz processor cores, with two cores per socket making up one node. Each node has 4 gigabytes (4 billion bytes) of memory and a dedicated SeaStar connection to the internal network. The full system will consist of over 100 cabinets with 39 terabytes (39 trillion bytes) of aggregate memory capacity.
"AMD and Cray continue to collaborate on innovative ways to leverage Direct Connect Architecture and HyperTransport™ technology," said Marty Seyer, senior vice president, Commercial Segment, AMD. "This innovation, along with Cray's supercomputing expertise and focus on scalable system architectures, has yet again resulted in a significant win. This is confirmation that customers believe that the design and performance of the AMD Opteron processor combined with Cray's superior system architecture provides a winning combination."
"Then how come AMD's share in the commercial sector grows from almost zero to over 20%?
..and AMD's US desktop sales from below 50% to 60%+%? "
You must be confused, as the 2003-2005 timeframe pretty much was uneventful for AMD in terms of Marketshare growth (except perhaps a slight growth in the server market). It wasn't until 2h2005 and into 2006 that AMD's growth became more prominent.
Case in point: in the middle of 2005, AMD files a lawsuit against Intel stating that with better products AMD was still unable to compete for marketshare. Since then, AMD has managed to grab large chunks of the server market as well as getting more spotlight with the desktop market. Good for them in that it increases their MSS, bad for them in that it removes the foundation of their lawsuit.
"Guess what happens next - the market begins to push Intel Netburst dual core into the consumer segment agressively."
If you look at AMD's forecast from HKPC, X2 takes only 25% of total volume. Forecast on the mature 90nm fabrication is demand-driven, that is, AMD would produce that many single-core CPUs because it estimates that many people demanding them.
"AMD will end up sitting on a lot of single core CPU inventory."
I don't know how you could reach this conclusion simply from high X2 demand. But even if this was true, the single-core Sempron/Athlon 64 are still better for a low-profile, power-efficient, 64-bit PC.
"Meanwhile, Intel will move their low cost 70-80$ Pentium 4's in emerging marekts like India, South East Asia, etc where the Pentium brand has a lot of recognition."
If you assume people in the 3rd world are stupid, then you are wrong. The 'Pentium' brand will live as long as it is a good deal.
Besides, the Pentium-brand doesn't help relieving the channel, which is stuffed with lots of Celeron already. Intel's aggressive Pentium branding in the future will actually choke the channel more on those aging Celeron chips.
"Intel rushed ahead, but it needs to burn 3 wafers to make one good chip." I still think its an interesting subject and I have been doing a considerable amount of reading into it.
Sometimes, you can't take everything literally. Does Intel need to burn 3 wafers to get a good chip, or is one wafer enough? It's just figurative speaking...
“My prediction on market share in the coming few years: things will remain the same.”
Oh contrary anonymous, Intel’s channels are stuffed with crap and there new stuff ain’t sell’en. AMD is selling every thing they make, which is a lot more than ever before. Rumor has it Dell is loading up on AMD causing the shortage of parts.
You must be confused, as the 2003-2005 timeframe pretty much was uneventful for AMD in terms of Marketshare growth (except perhaps a slight growth in the server market). It wasn't until 2h2005 and into 2006 that AMD's growth became more prominent.
So 2h2005 was not in the 2003-2005 timeframe?
Good for them in that it increases their MSS, bad for them in that it removes the foundation of their lawsuit.
There are other factors in recent AMD gains like:
=AMD convinced governments not to be
"Intel Only".
=Corporate buyers are seeing thru Intel/Dell's FUDs.
=The lawsuit has eased some vendors' fear of Intel.
The glaring point of the lawsuit is when AMD offered HP free chips just to get acceptance. The deal was done till Intel stepped in. HP canceled.
So IF AMD CAN PROVE IT, now the judge will just say all is forgiven now to Intel?
Sharikou, Ph. D shaped the electrons to say
Sometimes, you can't take everything literally. Does Intel need to burn 3 wafers to get a good chip, or is one wafer enough? It's just figurative speaking...
So then you readily admit to having pulled the numbers out of your posterior? As to the baclpedaling, "is one wafer enough?"... apply the same analysis I held your hand and walked you through in the other topic: if it is 1 die/wafer yield, Intel is currently burning 115k-143k wafers a week to hit their stated shipments. Again, all while shipping Preslers and Yonahs out of the same 2 65nm fabs? Good Lord, do they have huge fabs or what?!!!
Sharikou, if that's the best you can come up with, that's pretty freaking weak. Where did you get your PhD again? That sort of logic shouldn't have made it past qualifiers.
BTW, still waiting for your dissertation on GAAP treatment of goodwill impairment. Here's some reading material for you to help answer the questions: http://www.hlhz.com/download.asp?fid=400
Sharikou said "Sometimes, you can't take everything literally. Does Intel need to burn 3 wafers to get a good chip, or is one wafer enough? It's just figurative speaking... "
Your entire blog is figuratively speaking.
"If the other half of the process lays off another 3000 employees, that will be 6000 total, 6% of the work force."
That's really a big and thorough restructuring. See my point now?
"And yes, majority of Intel's problems (at least from AMD) are short-term problems. Once C2D ramps up, AMD better run for shelter."
If Intel could make C2D as efficiently as AMD making X2, and AMD still stay at 90nm (or poor 65nm yield), then yes, AMD will be in trouble.
Somehow I don't see that happening before 2007. After 2007, that'd be a much less threat to AMD whose 65nm production should ramp up already.
"And no, K8L will not save the day for them. Simply because, it is an inferior architecture (3-wide issue, not 4-wide; no re-ordering of memory read/writes; no micro-fusion; no macro fusion; only 1 MB of stupid L2 shared cache that is not inclusive), and will be on inferior process at launch."
These are laughably wrong:
1) K8 already has micro-op fusion, where macro-op fusion only works with limited x86 branch+compare, depending how well branch and value predictions of these instructions are performed.
2) K8L will have shared L3 cache among all 4 cores. In comparison, Kentsfield will have two L2 cache shared by two dual cores separately. Please tell me which one makes more senses? (smarter?)
3) Direct comparison of pipeline width between Conroe and K8 is invalid, because IIRC C2D has asymmetric decoders while K8 has symmetric ones.
4) Also, in reality few if any x86 program sustains 3 issues per clock cycle. The 4-wide decode/issue probably makes more sense for macro-op fusion, where 15% time an extra branch instruction could be decoded/issued with the preceding compare. Yet the actual benefit there would be less than 10%, probably even 5%, and definitely not 33% like ignorant people suggested.
"It adds the capbility to perform two 16-byte L1 accesses in once clock cycle--a feature useful in some server workloads, mostly useless in desktop/mobile workloads."
But that wrong! K8L increases L1D cache to dual 128-bit load/cycle mainly because it can execute TWO 128-bit SSE2 instruction per cycle. Isn't SSE2 for desktop and mobile, too?
"What's more, by the time AMD launches the K8L vapourware, Intel launches its 45nm compaction of C2D (Penryn), with probably quad-core being the norm."
Good, let see how quad-core becomes the norm when C2D occupies only 1/3 of Intel market share (early 2007 I mean).
I've heard people claiming 75% dual-core by end of 2006. After seeing AMD's AND Intel's production forecast, we all know how rediculous that is. Pure enthusiasts thinking, with no grasp of market reality.
"Kentsfield on 65nm is being launched at 1333 MHz FSB, so my guess is, Penryn will be 1600+ MHz FSB--again may not be good enough for MP servers, but more than enough for DP servers, desktops, and mobile."
Great! Dream on.
"Intel Fans unite"
Yes, you "Intel Fans" youre right. Intel needs your support this time. Probable more than ever. And also don't forget to wait for reward. Because, this years Intel was playing with your money, new cpu-new motherboard lots of marketing bullsh*t and bad business practicess.......
"Sometimes, you can't take everything literally. Does Intel need to burn 3 wafers to get a good chip, or is one wafer enough? It's just figurative speaking..."
No, it's just bullshit. Like the rest of your blog.
It would too kind to call it a lie. But that also is what your blog is: a bunch of self-serving, self-aggrandizing lies.
Someone with authentic analytical skills and honesty would be able to see that AMD has NOT executed well as of late.
AMD may be ramping up in market share due to massive fire sale pricing, but from a technological standpoint, the company is dead in the water.
There has been nothing innovative from AMD since the original Opteron design, now 5 years old. The team that did that good design is long gone and AMD is in trouble.
Intel has delivered to the market Conroe and Woodcrest systems. You can go and purchase a Mac Pro which has two Woodcrest chips. And these Mac Pro machines are selling like hotcakes.
Meanwhile, AMD announces more and more vaporware: 4x4, KL8, and Torrenza.
It all adds up to AMD having massive execution problems and Intel doing a good job. Maybe Intel is not ramping up as fast as they'd like, but Conroe and Woodcrest are selling today in good volumes.
I have three AMD systems and am not buying anymore. AMD decided to screw their customers by making all their old systems obsolete. Paying top dollar to upgrade my Socket 940 Opteron systems is not going to happen. I can buy a complete system with TWO dual-core Woodcrest Xeons at 2.66Ghz for less money than just the chips cost from AMD. And if I want virtualization support? I have to throw away my system anyway and get something new based on Socket F. For even more money! And it's not even shipping until late this year, possibly next year! To hell with that.
Take away AMD's dumping and fire sales, and AMD has no value proposition. Their chips are overpriced and the technology is 5 years old with no sign of improvement. Even "4x4" is just a poor man's 2P system. ZERO INNOVATION.
There are companies that pretend to be more than they are. And AMD is one of those companies. Businessmen know this. Fanbois don't.
"People said 90nm vs. 65nm, but who cares?"
It's not only about performance - if AMD had made this transition when Fab36 started up as they had originally planned, they would be getting roughly double the output of F36 and be able to supply a lot more chips throughout this year.
Edward: "Less than a year they'll regret, feeling the thing too hot and slow (bye-bye to good Vista performance"
Do you honestly believe Joe consumer will be upgrading to Vista next year if they buy a computer this year? Most people upgrade their OS when they buy a brand new computer with one installed on it...Sure the enthusiast space will do this, but the average consumer will not care (or possibly even know) what their operating system is other than it is "Microsoft".
Also do you think the average ocnsumer will think their computer is hot and slow if they have no other reference point to compare it to?
"It can make 2000 Netburst CPUs per worker per month, but the profit is not enough to feed one man and his family. Now, the 600 men division had to find a better owner."
Sharikou, you must be smarter than this, no? You are saying Intels DEG group shold be subsidizing the communication groups which are losing money? In business one does this for a little while to see if that business unit will turn around or if it makes sense in terms of grwoing the overall market; however if the business unit continues to lose money and has little prospects to start being profitable, why continue to keep it?
Why did AMD spin off Spansion? Couldn't they have just made more CPU's to feed the Spansion families?
Other thing to note about your astute analysis; I believe profits are generally after workers have been paid (although I'm no accountant). A company doesn't rely on profit to pay employees.
Otherwise, great analysis as always...
I'm sure the errors, as always, were just "figurative" statements :)
Sharikou - please define the massive operating losses you are predicting for Q3.
Do you mean >500Mil net operating loss for Q3?
This means more cash with which to grind AMD to the bone with in a price war. If this is war of attrition, then I'd put my money on Intel, not the company who just sold their soul to the devil to stay in the platform game...
Don't you mean they just upgraded the one they have?
"AMD is seeing shortages for their chips. Search google, the news is all over the place. This is a bad thing, but at the same time, it is a sign of how the market feels about AMD."
It could also me Capacity and yields problems.. When is the next fab coming? .. oh
"You can say that Conroe has "fragged" AMD all you want, but the numbers tell a different story. AMD's chips are selling so fast that AMD can't make enough of them."
I agree.. AMD can not make enough of good ones.. But them make a lot of scrap.. Just ask Sharikou's sources
What a fanboy you are. AMD's ship has sunk with Conroe smashing it silly. Your ramblings and inane stories look like dementia to me.
But 95% of AMDs chips are fatsre than 95% of Intel's chips. You do the math.
Core 2 OVERALL is around 10% of shipments - especilly considering - that the 50% Xeon Otellini promised will barely dent a 3 month + surplus with more HeatBurst models having been announced.
if it is 1 die/wafer yield, Intel is currently burning 115k-143k wafers a week to hit their stated shipments.
You guys are so serious. My estimate is Intel can make 40 good Conroes off one wafer. That matches the expected 1 million/7 week shipment figure. In comparsion, AMD makes 330 good X2s off a 300mm wafer.
In five days (TUESDAY) socket F is released. It should make a difference for Q3. We’ll see.
"Now the tables have turned, with Intel "leaping ahead" and simply dominating with their C2D."
Dominating with that puny quantity? LOL. Wake up.
"Yeah, it's a great time to buy a Netburst processor, but why would an educated consumer buy one of those?"
Looks like Intel is depending and betting on their branding and brainwashing. OMFG Intel Conroe frag AMD, PD must be quite good too LOL. And when some average plain Joe/Jane asking their famous handy nerds in their neighbourhood, guess what he says on Intel PD.
"and Intel has built up huge Netburst single/dual core inventory because they think they can sell it all so at the end of the day these are just estimates."
Intel built up huge Netburst inventory because it had no alternative. Beofre Q2 this year Netburst is all Intel could mass produce.
AMD OTOH can adjust percentage of its dual-core however it likes. It produces 25% of them in volume not due to capacity limit, but due to customer demand estimate.
The two above are vastly different cases.
http://www.tgdaily.com/2006/08/10/amd_opteron_socket_f/
socket F out in august 15.
"socket F out in august 15."
My monkey told me AMD is burning 100 wafers to get one F-chip.
When my horse heard this, he stomped his hoof 5 times which means AMD will BK in 5 quarters.
So there you have it. The definitive shanalysis.
And Conroe frags AMD again:
"AMD’s best processor, even when overclocked, still struggles to compete with Intel’s mid-range Core 2 Duo processor. The Core 2 Duo E6600 can deliver roughly the same performance as the Athlon64 FX-62 processor while costing roughly half as much, requiring 1/3rd less power, and remaining far more overclockable.
This certainly leaves AMD in a tough spot. Even their newly launched Socket-AM2 processor lineup is getting beat in terms of performance and price, and even after their massive price drops, enthusiasts are still flocking towards Intel. These are the same enthusiasts who were so adamant about using AMD products in their prior systems, and now we’re seeing a similar move towards Intel. AMD could drop prices even further, but this is unlikely to help much. AMD needs to provide a faster product to the market as soon as possible. With no such product in sight until the end of the year (or perhaps later), its Intel’s market to take at this point. As fast as they can pump out these Core 2 Duo products, it’s likely that they will be sold. "
AMD has lost the enthusiast market. No one will want 2 crappy chips instead of 1 good chip. 4x4 is dead on the vine.
And after the AMD fire sale is over, AMD's market share will plummet.
My horse has revised his estimate and now he says AMD will BK in 4 quarters.
The horse knows!
Hmm... before Q2 Intel mass produced Netburst is understandable. But now Intel still mass produce Netburst and stuff more Netburst. Reviews, benchmarks already show the 32-bit prowness of Conroe, the new x86 performance crown. Now with a better chip from Intel, Intel still crunch out Netburst. WTF with Intel?
The things that I can think of are:
- low yields for Conroe, hence costly to manufacture, thus low profit margin. So Intel is better off (profit wise) with Netburst for the moment.
- if the flooding of Conroe is realized, then everybody would know Netburst is utter crap, and nobody would buy Netburst. Ok, that leaves to Intel fanboy to buy. Support Intel right?
Do you honestly believe Joe consumer will be upgrading to Vista next year if they buy a computer this year? Most people upgrade their OS when they buy a brand new computer with one installed on it...Sure the enthusiast space will do this, but the average consumer will not care (or possibly even know) what their operating system is other than it is "Microsoft".
You are incredibly right.
Joe Consumer won't care about Windows Vista. 64 bit means nothing to most normal people. I've seen beta versions of Vista and from what I see, I wouldn't buy it as it was unless I had a 64 bit machine myself.
The consensus is that most people who planned on buying a computer in the last few years have already done so.
For these people, there should be no legitimate reason to upgrade their CPU's (which would mean pretty much a completely new system), nor would more people be interested in what Vista has to offer.
Consumers haven't seen a dramatic NEED in more horsepower for a while.
Vista and 64 bit CPU's will really only matter to the enthusiast and fan market, which if I remember correctly was around 12 percent of the desktop market.
Even if Vista comes out in the first half of next year, it won't really make any impact for an additional year or two.
I can understand why AMD might have a few fans (faster, cheaper, smaller company). I can not understand how Intel has any supporters. For the last few years they offered slower, more expensive chips. They tried to push Rambus memory, have crappy integrated graphics, and try to crush their competition by locking distributors into Intel only deals.
What is there to like about Intel? What is there to dislike about AMD?
I was talking to my friend who runs a branch PC sales and repair shop for a medium size company about the sales volume of Intel vs. AMD.
Note that in Malaysia especially KL and Penang there are lot of PC shops around. In one mall you'll have a minimum of 10 (small mall) to 25 (large mall).
A few days back we had one of those PC fairs in Penang that we have about 3-4 times a year.
At this fair my friend was at lasted for 3 days and have 50 over stalls. Doing that time his stall sold 12 P4 D, 15 AMD AM2 and 10 AMD X2 AM2 and 1 X2 3800 939.
KL has a much larger crowd and possible has about X4 the sales as Penang but as you can see the buying public usually go for the mid-range for their CPU choice and AMD seems to fit that bill quite nicely.
As for Conroes? No one has seen one yet. Here or there.
So if you think Intel really can off load all those old P4s you better think again. Cause they are only selling half of what AMD sells.
To suppliment the earlier post I made I also like to mention that almost all the PC shops do a bit of viral marketing here for AMD. This is unintentionally. They advise customers (and I ask a lot of shops about this) against getting the any Conroes for several reasons:
1. You're not gonna get it when you want it in any reasonable time. (That's their way of saying don't when we're get it.)
2. There is almost no motherboard support for it now.
The second point is important as most shops really don't want the hassle of tech supporting customers getting the Conroes to work on the current chipsets.
Basically, what am saying is over here almost all PC shops advise you to get an AMD cpu. Whether you want a dual core or a cheap single core. Its easy to install, runs cool and trouble free.
And you would say why not Intel? Its way cheaper than AMD. Well there is a mentallity that among group of cheatest stuff you can get you would never go for the one on the end of the list.
nyx said...
"They asked me why all these review sites benchmarking the differences between the AMD 64FX-62 and the E6700 and X6800 in a multitude of games are testing at resolutions of 1028 x 768?"
I know you are spinning this to make your point, but at the same time you and I both know why this is done... or maybe you don't and its scary that people are asking you these types of questions while your ill informed.
I think its called... oh yeah, NOT HAVING THE TEST GPU LIMITED.
If you want to trully know how these chips perform you have to take the GPU out of the equation.
I can not remember where I read this but the quote was something like this...
"todays top of the line processors are already performing at a higher rate than any single GPU can deal with , and almost as much as Crossfire or SLI"
For a real test, and at the resolutions you want, you would have to go quad SLI, maxed out, and on say your 30" widescreen, then and only then can the performance crown be attained.
So until you or possibly someone here can point me to a test setup that is equal or surpasses that, I will believe that C2D is beating the FX62, maybe not by as much as Intel would like but 10-20% is better than loosing by 40% with P4 vs FX62.
PS:
"AMD 64FX-62 : $ 831.99
Intel Core 2 Extreme X6800: $ 1,255"
Why not look at it correctly and compare equaly performing chips when comparing prices...
AMD FX62 = $ 831.99
Intel E6700 = $699.00
Looks a little different now doesn't it, let alone the mark up price vs MSRP on Intel.
"What is there to like about Intel? What is there to dislike about AMD? "
Because there are a lot of people still holding onto Intel stock or stock options that's down a bunch from its purchase price. They are hoping that Conroe will boost Intel back up so they can at least break even.
To understand their plight: if you were an Intel employee hired in the last 8-10 years, today you will make exactly ZERO on your thousands and thousands of stock options.
-Jack
"For the last few years they offered slower, more expensive chips. They tried to push Rambus memory, have crappy integrated graphics, and try to crush their competition by locking distributors into Intel only deals."
The biggest joke got to be the Centrino wireless. That low-performing thing lived its life totally out of Intel's monopolistic marketing. I know engineer working at Intel's networking, and engineer working at Broadcom. They agree on one thing: Centrino doesn't perform well.
"What is there to like about Intel? What is there to dislike about AMD?"
Intel shareholders blame AMD for Intel's loss. They so desire the growth of Intel they saw in the late 80's and 90's. What they don't understand is that AMD has little to do with Intel's slow-down - it is the blue company itself and its monopolistic thinking that destroyed the prospect of this industry.
Remember Intel's Itanium and 10GHz P4 efforts? The resources that Intel pulled into those could probably have bought/established two more AMDs. Every cent of those is investor money.
Yet, to give credit where it is due, Intel does have the best fabrication in the world. It is almost one generation ahead of all competitors, while it is not needing to use SOI. Intel's Core 2 Duo has a nice core, though its architectural scalability still need some work.
"As the AMD rep says, they're seeing unprecedented demand. Translate that into unforseen demand. I rest my case...they called the dual core forecast wrong. And the reason this happened is the agressive Intel pricing."
Interesting point. So if what you're saying is correct, then Intel is trying to make AMD's demand forecast wrong by badly hurting its own ASP. I say they learn such tactics from the suicide bumbers. I mean which normal person would think of crashing into a building to make his point? Brilliant! ;-)
Besides, if that indeed was the cause of high X2 demand, then as early as now AMD must have realized what Intel's doing. How slow do you think AMD & channel can react on that? 2 months? 3 months? That's still before the Thanksgiving, isn't it?
Edward wrote:
"Remember Intel's Itanium and 10GHz P4 efforts? The resources that Intel pulled into those could probably have bought/established two more AMDs. Every cent of those is investor money."
This is probably the most insightful paragraph on this blog for a long time. AMD fanbois should to read this again. And Intel fanbois must read this TEN times.
-Longan-
P.S. My home town is in Portland, Oregon. Let me add an insightful piece of information from my friends who were ex-Intel or current (and in fear) Intel.
There are three tiers of engineers in Intel. The best/top-tier were pursuing communication and Itanium. The second tier engineers were on the Pentium. And the third tier were the chipset engineers.
So, this is what you got when you pit the second tier Intel engineers vs the best of AMD engineer. That is for Pentium vs Opteron.
I am not talking about the Core 2 Duo and I have utmost respect for the Israel engineers.
Edward wrote:
"Remember Intel's Itanium and 10GHz P4 efforts? The resources that Intel pulled into those could probably have bought/established two more AMDs. Every cent of those is investor money."
This is probably the most insightful paragraph on this blog for a long time. AMD fanbois should to read this again. And Intel fanbois must read this TEN times.
-Longan-
P.S. My home town is in Portland, Oregon. Let me add an insightful piece of information from my friends who were ex-Intel or current (and in fear) Intel.
There are three tiers of engineers in Intel. The best/top-tier were pursuing communication and Itanium. The second tier engineers were on the Pentium. And the third tier were the chipset engineers.
So, this is what you got when you pit the second tier Intel engineers vs the best of AMD engineer. That is for Pentium vs Opteron.
I am not talking about the Core 2 Duo and I have utmost respect for the Israel engineers.
"AMD OTOH can adjust percentage of its dual-core however it likes. It produces 25% of them in volume not due to capacity limit, but due to customer demand estimate."
This is simply not true - AMD is capacity constrained (at least in the near term), even Sharikou acknowledges this! Otherwise why make a lower margin ditributing ~10% of your chips outsourced to Charter (rather than making in house) and cut cache from 1Meg to 512.
Dual core means fewer die/wafer than single core and as AMD shifts mix toward more dual core it's effective capcity decreases (of course they are ramping F36 and switching to 65nm to counteract this).
I don't think people realize the real reason Intel kept pushing dual core early on - it was not really for performance it clearly hurt AMD's capacity further. Every dual core chip AMD produced is 1.5-2 single core chips they could have produced and for a company that was capacity constrained on 200mm ,90nm at the time this was a big deal (hence why AMD's X2 percentages are still relatively low compared to single core).
"You guys are so serious. My estimate is Intel can make 40 good Conroes off one wafer. That matches the expected 1 million/7 week shipment figure"
There is NO WAY you could possibly calculate this without knowing how many wafer starts are being used for Conroe vs P4 vs mobile vs server.
Are you assuming ALL of Intel's manufacturing capacity is making Conroe? If not what % are you assuming such that you can say that your 40 Conroes/wafer matches the 1mil/7 week figure?
Also I assume you are aware of the cycle time of a lot in a 300mm fab, correct? I believe it is now >100 days for both Intel and AMD; fator in sort, test and packaging and the chips that are being shipped out 7 weeks from now had to be started over 1-2 months ago!
To put this in perpective for everyone even if Intel changed every wafer start to Conroe today (or similarly for AMD when their new chip comes out) you wouldn't see that output from that until at least 3-4 months from now.
nyx I am sorry if I was a little aggresive about your statements, its not meant to come across as a slam.
"The simple truth is, for someone who only games and does so at a res of 1600x1200 or higher, it makes very little to no difference at all as to what CPU you use."
I agree, sort of, we are talking about Intels next generation processor, and in doing so we should also look ahead at next generation video cards.
Anyone building a high end system should be aware or made aware of future prducts.
There is an article, HERE, which would be the ideal test setup, and it is showing the performance of next generation GPU performance, coupled with the X6800. It would be nice to see the same sytem with an FX62.
Currently FX62, while using todays hardware is about 10-20% behind X6800, next generation should widen the gap... to prove the statement look at the reviews showing CPU limited performance between FX62 and X6800.
Do you agree?
"When did I refute that in my last post?"
Your previous post states...
"Most sites though seem to insist on running games at 1024 X 768 which amazes the hell out of me.
Of course Intel won't be able to sell their new processor lines to gamers if this information is made public."
You make a statement like this, and I took it at face value.
"It was a simple statement of truth. That's all, nothing more."
I have one other question did you explain to your friends why they test at 1024x768?
I do not mean for these comments to come across in a negative way, but I just do not like to assume or leave holes in staements, so direct works for me.
Thanks.
"Every dual core chip AMD produced is 1.5-2 single core chips they could have produced and for a company that was capacity constrained on 200mm, 90nm at the time this was a big deal (hence why AMD's X2 percentages are still relatively low compared to single core)."
I do not dispute that AMD's capacity is constrained. But it's not a bad thing to have constrained capacity if they can have someone else (Chartered) to fulfill the demands. At least AMD's winning lots of tier-1 PC makers as customers, which is unprecendented.
If you read the thread that I wrote and you commented on, I was explaining why AMD won't become like Intel sitting there with loads of aging single-core processors. If single-core CPUs don't sell, AMD can simply produce more X2.
So AMD's problem isn't constrained capacity - Chartered is covering for prehaps another 5-10% market share. The problem isn't aging single-cores, either, where AMD seems to be able to adjust production volume very flexibly. The problem IMO is price. If Intel could lower its Core 2 Duo street price so much that would seriously hurt Athlon 64's profit margin, then AMD's in trouble.
In order for this to work, not only Core 2 Duo's tag price must be low, it's volume must also be high (we knew that Netburst couldn't hurt K8's margin). So my conclusion is that Intel won't be able to hurt AMD much (more than hurting itself) with aggressive Netburst pricing; Intel could hurt AMD only by producing enough Core 2 Duo quickly (e.g., 30% or so before Christmas).
I wonder if any of the 600 workers got laid off later, would they got any severance pay?
Long time ago, I heard a company sold a group for $25 Millions but then also paid $10 Million so that the buy would take another junky/unwanted group. The net sale was $15 Millions. Well, it could have been a urban-legend...
-Longan-
Damn, everytime I read this blog I read Intel will be BK in 5 - 7 quarters.
It was like that last quarter as well, shouldn't it be 4 - 6 quarters now?
Or will Mr Sharky keep saying 5 - 7 quarters for eternity? :/
You guys should all lighten up and read this blog for pure entertainment value. It's just people's opinions - what do you want for free?
P.S. 8-ball, are you sure you are reading your horse right? Mine pawed 5 times, meaning Sharikou will BK in 5 quarters.
Like flies to shit...
Another BS post.
It is truely amazing that the Board hasn't fired Craig Barrett and his side kick Paul Otellwennie for wasting billions of INTEL's money on silly purchases in the 90's.
Can you think what a monster INTEL could have been if it had another 6 billion in the bank. And for the past 6 years was a 75K employee company. Barrett and Otellwennie have been trying their best to ruin the company.
But here are the facts
3 300mm 65nm factories in full production making 200Million+ Merom/Conroe/Woodcreast
3 300mm 90nm factories making chipsets for them Merom/Conroe/Woodcreast
4 200mm factories making chipsets
2 300mm 45nm factories in construction.
75% marketshare and they will be back to growing server by Q1'07
With the price war INTEL profits will be down from 8 billion last year to probably about a billion this year. Yes so SAD only ONE BILLION dollars.
AMD on the other hand will exit the year losing money.
Yes.. AMD is going to take over the world.
What is scary is that INTEL is getting back to basics. Once they fire Otellwennie they will be a real monster again, but even with incompetent managers they still win over AMD everytome.
Maybe in the Sharikou's wet dream.
Keep jacking off there jack.
What and where did you get your PhD.
The PhD Prentender..
let me show you guys where this why this guy is the ultimate joker.
"The result will be massive operating losses from 3Q06 onward."
I'll be hear to remind the PhD how he missed the boat. he'll have an execuse as good as his fictious degree.
"In fact, Intel is pouring more money in the NAND ventures to compete against Spansion's NORAND'
INTEL isn't competing with Spansion you PhD prentender. INTEL is making sure it has a presecnce against Samsung. They want to play in the fastest growing semiconductor growth market directly without being saddled with the capacity. They want to make sure Samsung doesn't become in NAND what INTEL is in CPUs. Can you comprehned that you pretender Doc?
"10 Con XE 6800 CPUs per month to make enough profit to feed one worker" At least they can feed an employee. Look at what AMD is selling its CPUs for. They need to sale what 1,000...
The only reason AMD isn't bankrupt is because INTEL doesn't decide to put them there.
Keep up the shitting Sharkikou!
The only reason AMD isn't bankrupt is because INTEL doesn't decide to put them there.
This is the biggest bunch of bullshit ever, and the true sign of a hard core Intel jerkoff.
Intel has tried everything in their power to put AMD under, but has failed.
Sharikou,
Intel is not competing against Spansion. Its against Samsung and Toshiba.Sapansion is 2 years behind Intel in NOR. That NorAnd product has become the joke of the industry. It is costly to manufacture and sells for 30$% loss.
"And 65% of Intel's CPUs are Netburst. You can imagine that the Netburst camp still owns Intel today. You wanna fire a Netburst crew? You lose 65% of revenue and die."
So just so I, and the rest of your readers understand, if Intel lays off the Netburst design team they can't continue to manufacture or sell them? Especially when they are converting their product line to Core 2? Thanks for that pearl of wisdom!
Oh and the Intel flash business that is losing money is NOR; NAND flash is the fastest growing memory segment and the competition in that space is Samsung, not Spansion! Thanks for that pearl of wisdom too!
What percent of the overall x86 market is 4P+ servers again?
"But it's not a bad thing to have constrained capacity if they can have someone else (Chartered) to fulfill the demands."
Actually this is a bad thing because the margin made on a chip produced by Chartered will be less than one produced in house, unless:
A) Chartered is selling them at cost and has become a non-profit company.
B) Chartered's productions cost are lower than AMD's.
Neither of these are likely, thus any chip production outsourced is lost margin opportunity (i.e. AMD will still make $ on these, but they will make less of it). Moving to a greater mix of dual-core when you are capacity constrained therefore means less volume produced in house and lower margin due to increased ratio of outsourcing to Chartered.
Edward you also stated: "AMD OTOH can adjust percentage of its dual-core however it likes"
While AMD can do this theoretically, they really can't do this practically - my point is this further reduces AMD's effective chip capacity (from producing more dual core chips). Yes they can go to Chartered but as you say the real issue is pricing and producing chips at Chartered puts further pricing pressure on AMD as they have to be paying something above cost from Chartered.
Until AMD's capacity increases significantly (F36 converted over to 65nm and at high yield); they can continue to try to capture higher market share more quickly (via low end single core) or make fewer chips but focus on mid-range dual core which will give them more margin and money to invest in the future. I think the higher dual core approach is better, much like when Ruiz said they will focus on server and mobile not too long ago - in the short term they may sacrifice a little additional market share (they may still take market share, but producing more dual core will slow this rate somewhat since they are capacity constrained), but this will help grow their financial health and allow them to pay back debt incurred from ATI acquisition, F36 depreciation, F30->F38 conversion, and NY fab construction.
Doctor Doctor Calling the Doctor
Tell me again how many CPUs AMD will sell?
Tell me what AMD's MS will be of the total CPu market?
Tell me how much money AMD will make.
And pleaseeeeeee tell me how much INTEL will lose this year, as they are just 7 quarters away from banrkuptcy.
While you are telling us all of this where did you learn your data deep data analysis skills?
http://www.xbitlabs.com/articles/cpu/display/athlon64-x2-3600_11.html
Overclocked to 2.6ghz makes it a nice cheap solution....
"Actually this is a bad thing because the margin made on a chip produced by Chartered will be less than one produced in house, unless:
A) Chartered is selling them at cost and has become a non-profit company.
B) Chartered's productions cost are lower than AMD's."
Chartered is a foundry. If you had some idea on how semiconductor foundry works, you would have known that sending production off to foundry could be more profitable than producing yourself.
The deal between AMD and Chartered, as I perceive, is a multitude of things. Chartered wants 90nm SOI, AMD wants extra production. People don't understand how that could be beneficial to both parties - because people mostly don't understand the semiconductor industry. There is a reason for AMD not going to TSMC and UMC...
Edward you really should stay away fromyour manufacturing analysis.
"Chartered is a foundry. If you had some idea on how semiconductor foundry works, you would have known that sending production off to foundry could be more profitable than producing yourself."
This is true for folks who are not selling enough products to fill up a full factory (not applicable in AMD's case) or you don't have the epxertise to do it yourself (since Chartered/AMD/IBM are co-developing technology this is also not applicable) or the foundries yield is so much better than our own that it is chaeper (I assume this is not the case).
I did not say that foundries don't make sense at all! Are you honestly saying it is better for AMD not to build their own capacity? It might be better short term or useful when factories are being converted, but counting on this continuous cuts into your profit margins. If this helps AMD margins then they should be outsourcing 100% of their production!
You originally said it is not a bad thing AMD is outsourcing production to Chartered, I'm saying it is not a good thing as it cuts into their potential profit margin (assuming they could produce it in house). As Sharikou keeps saying AMD has 50-100% capacity for the world's chips why outsource anything? (obviously because his analysis is wrong and they are capcity constrained)
"If you had some idea on how semiconductor foundry works, you would have known that sending production off to foundry could be more profitable than producing yourself."
So you are saying AMD would be more profitable outsourcing to Chartered. I think it is time to fire AMD's manufacturing planning as they clearly should be outsourcing 100% of their chips.
You are talking about companies whoe either can't fill a fab up or do not have the research capabilities. AS AMD is already doing there own 65nm this is not the case. I'm talking about AMD and you're tallking abstractly about other ASIC's.
Do you think AMD gets more profit off chips the have produced at Chartered?!?
"This is true for folks who are not selling enough products to fill up a full factory (not applicable in AMD's case) or you don't have the epxertise to do it yourself (since Chartered/AMD/IBM are co-developing technology this is also not applicable) or the foundries yield is so much better than our own that it is chaeper (I assume this is not the case)."
1) The extra demand volume AMD needed does not fill up a full factory... not until 2008, according to AMD. That's why AMD wants flexible capacity instead of partly filled factory capacity.
2) The Chartered alliance could bring more funding to the process development, which could be important to IBM/AMD.
3) You simply don't know, and can't assume, that AMD produces chips cheaper or not than Chartered. Note that not just fabrication, but testing, packaging, transportation, all these cost $$$.
"I did not say that foundries don't make sense at all!"
Good. So in AMD's case, sending off production to Chartered does make sense, and probably won't lose them money, either. That's the original point of the thread of my comments.
The funny part is that your claims actually supported my analysis, except where you made some off-chart assumptions... ;-)
"You are talking about companies whoe either can't fill a fab up or do not have the research capabilities."
No. Foundry usually has better yield and lower production cost - that's what foundry is specialized for. At some price point, factories such as Intel's would simply exit the market not to compete with foundries, for example, the DRAM.
Post a Comment
<< Home