Tuesday, April 03, 2007

Intel -- The Monopoly Under Siege

It hasn't happened yet but there are now clear signs that Intel's monopoly is crumbling. It's a slow process but from what has been announced there is no reason why AMD cannot achieve this goal by the end of 2008.

Back in 1998 AMD scooped Intel by releasing the powerful 3DNow! instruction set which for the first time allowed SIMD floating point instructions on X86 processors. However, this was cold comfort to AMD since Intel simply ignored AMD's 3DNow! instruction set and released its own SSE instructions a year later. In some ways, like the ability to add numbers within the same register, Intel didn't catch up to 3DNow! until SSE3 was released nearly six years later. Intel's monopoly position was such that it could create standards even when it was trailing its competition.

In 2003 when AMD released the AMD64 64 bit extensions to the X86 instruction set Intel had planned to simply introduce its own incompatible instruction set much as it had with SSE. However, Microsoft refused to support another set of 64 bit extensions so Intel was forced to follow AMD's lead. It is significant that even four years later C2D's ability to handle AMD64 instructions is still lacking. This is a clear indication that Intel has not actually reinvented itself and is still following the old notion that it can do whatever it wants simply because it is Intel.

After all, being behind with Pentium M worked fine. Although Pentium M was actually less power efficient (even without AMD64 instructions) than AMD's Turion, the Centrino brand was labeled as being more advanced. The reason Intel was able to hide the fact that Pentium M was behind was that it produced its own mobile Centrino chipsets. These chipsets were so good that they more than made up for the higher power draw of Pentium M. AMD in contrast was left waiting and hoping for a proper mobile chipset from one of the 3rd party chipset suppliers like VIA, nVidia, or ATI. No such chipset was ever produced. However, now that AMD owns ATI this is no longer the case. AMD is fully capable of producing its own low power chipsets along with its mobile processors. In contrast, Intel's standardization of the C2D family prevented it from leaving out the AMD64 instructions from Merom and this increased power draw past Intel's original target. The mobile market is now much more competitive.

There are other signs that things are changing. Intel has tried to replace the ATX motherboard with BTX. However, ATX is still the standard and BTX is still not popular with OEM's. In contrast, AMD is introducing two new motherboard standards. For workstations, ATX is too small so AMD introduced an extended ATX standard with Quad FX. A new standard is badly needed since there is no current standard larger than ATX. And while ATX is too small for workstations it is too large for low cost desktop systems. So, AMD is also introducing the DTX standard which will allow smaller motherboards and cases. This standard seems particularly attractive for several reasons. The first is that the rise of USB as a common external port means that many ports like the parallel, serial, and even mouse and keyboard ports can be left off. This means much less competition for I/O port space at the back of the motherboard. Secondly, both Intel and AMD are talking about putting GPU's in the CPU package. With AMD's Integrated Memory Controller (IMC) this would completely eliminate the Northbridge chip which would similarly reduce competition for space on the motherboard itself. There is some indication that Intel is resisting this move for fear of losing chipset revenue. However, making the motherboard and case smaller while reducing the chip count and I/O ports will reduce costs for OEM's so DTX should become very popular.

Another big problem for Intel is server support from IBM. Intel currently enjoys good server support from IBM because IBM spent a lot of money developing its Hurricane chip for the Xeon FSB and is eager to sell as many systems with it as possible. The problem is that Intel's Nehalem processor has its own IMC and therefore no FSB which makes it incompatible with Hurricane. It would of course be possible for IBM to create another chipset to support Nehalem but this does not appear likely. The reason for this is that one of Intel's current directions is to use the same socket for both Nehalem and Itanium. This should reduce cost for Itanium development and particularly for IBM's sever competitor, HP. Apparently, in response to this, IBM is making its Power 7 family socket compatible with AMD's Opteron. So, it appears that the strong server support that Intel enjoyed because of Hurricane is now over and that IBM will strongly support Opteron instead. With Itanium on the Xeon socket and Power on the Opteron socket this leaves Sparc as the only competing server processor without leveraged support. It remains to be seen what Sun's response will be but without similar costs reductions Sparc could be pushed out of the market.

Other problem areas have included compilers and DIMMs. With its IMC AMD's K8 processors benefit from low latency. AMD would have therefore benefited from DDR speeds of 466 and 500MHz. However, Intel was able to push the development of DDR2 so DDR standard speeds stopped prematurely at only 400MHz. There is no doubt that this helped Intel while hurting AMD. However, there is currently some suggestion that AMD will be able to get DDR2 extended from 800MHz to 1066MHz. If this is true then it would be a real benefit for AMD. Coming with lackluster support for Intel's FBDIMM this would indicate genuine shift in influence. Likewise Intel enjoyed optimal code speed from its compiler while AMD without its own compiler often saw its full processor power unused. Now, however, the Portland Group has a compiler that produces good code for both Intel and AMD processors. This would appear to be the logical choice for software developers who want the maximum possible market for their products. However, increasing use of this compiler by developers is reducing another of Intel's artificial advantages. It isn't an overnight change but all of these things should steadily erode Intel's monopoly standards position and allow AMD to create its own standards independently of Intel.

If AMD does indeed succeed in breaking the Intel monopoly by late 2008 then IBM's increased server support will simply strengthen AMD's position. I've heard a lot of talk about Intel's lead in process technology. However, this too is changing. AMD will reduce Intel's lead at 45nm to just 6 months. But, Intel will continue with a 24 month cycle to 32nm while AMD is pursuing an 18 month schedule. This should mean that AMD will introduce 32nm at basically the same time as Intel in late 2009 or early 2010, a difference of perhaps a few weeks instead of months. It also appears that Intel's Simultaneous Multi-Threading strategy is going to be ineffective for some time. This should be a maximum benefit on dual cores but a decreasing benefit on quad cores or quad cores on dual sockets. This is true because both Vista and Linux would be hard pressed to produce the 8 or 16 threads necessary to fully occupy these chips. This really only competes with the limited market currently being pursued by Sun's T1 chip. All in all, SMT is a good thing for processors and it is commendable if Intel can release a good version. However it is ironic that this advance would put Intel in the same boat that AMD has been in since 2003 with its advanced 64 bit K8 hardware underutilized.

I think that Intel will eventually realize that it still needs a few philosophy changes. Essentially I think that Intel's current strategy of low prices is going to fall flat in 2008 when faced with Fusion, chipsets without Northbridges, and DTX since this is exactly the platform that can flourish with severe pricing pressure. In contrast, Intel seems to be holding out by continuing to use the old FSB on single socket Nehalems and these will not fare so well with low prices. I would say that Intel is heading for another round of reorganization in late 2008 or early 2009. I also think Intel will end up scaling back some 45nm and 32nm FAB upgrades. I'm not sure if AMD will actually pursue the NY FAB option. If AMD is still pressed for cash it may resort to some type of expansion in Dresden perhaps as it did with FAB 30 back in 2004. So, we should continue to see AMD and Intel moving closer together into 2010.

82 comments:

weird_mind said...

I dont really think that the non-usage of an Opteron-socket is an disadvantage for Sparc.

For example look at Niagara II. Niagara goes in a different direction for cost and complexity reduction, as you have PCI-E on chip and 10GigEthernet on Chip and of course you have the four memory controllers. It doesn t really fit into standard opteron socket. The socket was designed for processors that need chipsets to build complete systems. But it doesnt really match to processors that only need external chips for housekeeping, (relativly) lowspeed connects) and of course memory.

There is more than one way to get rid of complexity.

Fujiyama said...

Scienta, do you really belive that AMD will be able to introduce 45nm chip in 12 months from now?
If it's true Shanghai should be ready in Q3 this year. AMD was not able to keep any of production promises 0.13, 0.9 and 0.65.
They all were late.
AFAIR 65nm was scheduled to start in the summer. It didn't.
I assume that 45nm is Q1 2009, with Fusion introduced even later.
So the production gap is going to remain.

Scientia from AMDZone said...

fujiyama
Scienta, do you really belive that AMD will be able to introduce 45nm chip in 12 months from now?

Well, 45nm basically a little before mid 2008; that could be 13 or 14 months.

If it's true Shanghai should be ready in Q3 this year.

That doesn't make any sense. Shanghai is the 45nm die shrink of Barcelona with a bit more cache. Shanghai would be the first 45nm chip so it wouldn't be out until 2008.

AFAIR 65nm was scheduled to start in the summer. It didn't.

It doesn't matter. 45nm is an 18 month schedule. 65nm was released Q4 06, 18 months later would be Q2 08.

I assume that 45nm is Q1 2009, with Fusion introduced even later.

Well, your assumptions are simply wrong. 45nm is Q2 08. The first version of Fusion is MCM and should ship in 2008 as well. The 2009 version is with the GPU on the same die.

So the production gap is going to remain.

No, the production gap should be nearly gone at 32nm.

gdp77 said...

Scientia, I don't have inside info regarding when AMD be able to reach 45nm. What I have is the past of AMD... Until now, AMD never succeeded in a so fast transition in die process. I think is safe to assume that AMD will reach 45nm at Q3 or Q4 08.

How can u be so sure that at 32nm the gap between Intel and AMD will no longer exist? What makes u believe that AMD will make faster transitions that Intel? When this happened before? What r your facts?

Fujiyama said...

Scientia
I really don't see this as positive as you write. Shanghai with DDR3 support (AM3) and 6MB of cache is not going to be only the die shrink.
I made assumption based on past events and it is difficult to recognize 45nm conversion in 18 months. Remember that even easy K8 DDR2 was late.
45nm design samples should be ready this year to belive that AMD is going towards 2008/45nm.

Scientia from AMDZone said...

gdp77

Until now, AMD never succeeded in a so fast transition in die process. I think is safe to assume that AMD will reach 45nm at Q3 or Q4 08.

I was thinking late Q2 so early Q3 wouldn't be much of a change.

How can u be so sure that at 32nm the gap between Intel and AMD will no longer exist? What makes u believe that AMD will make faster transitions that Intel?

18 months is AMD's stated plan for both 45nm and 32nm. Intel's stated plan for 32nm is 24 months. Now, I'm not assuming that AMD will hit the same day or week as Intel. But if AMD can shorten the release lag to 6 weeks this would be close enough.

When this happened before? What r your facts?

AMD has never had a project for an 18 month release before. What facts are you talking about?

Scientia from AMDZone said...

fujiyama

Because of your wording I misunderstood what you meant. If AMD is going to release Shanghai in Q2 then it needs to tape out soon. Typically it is 12 months from tapeout to production.

Actually, at this point we don't know if Shanghai will support DDR3. Originally it was supposed to but then AMD talked about pushing the DDR3 schedule back.

Erlindo said...

Once more, great writing Scientia.

I do believe that intel is really concerned about AMD capturing more market share at any cost. The more market share AMD captures, Intel will have a hard time regaining that back.

And about AMD introducing 45nm and 32nm in schedule, I believe one of the main reasons AMD acquired Douglas Grose from IBM was to achieve these goals in a timely manner.

Scientia:
I'm having a hard time to post in your blog. This is the third time I register with a different e-mail and I hope that my account will be functional this time.

Scientia from AMDZone said...

There aren't many things to go by. One item would be that AMD produced a 45nm SRAM test wafer 3 months after Intel. This would imply That AMD would follow Intel by only 3 months. Since we are currently estimating 6 months I think this is reasonable.

By the way, if you do have doubts about AMD's process ability you might want to know that AMD's 65nm process was rated #1 in the entire world in 2006 by SI.

Semiconductor Insights (SI), the world's leading technical advisor to the global microelectronics industry, today announced that it has awarded the 65nm energy efficient AMD Athlon™ 64 X2 dual-core desktop processors the 2006 INSIGHT Award for Most Innovative Process Technology.

SI has participated in virtually every major semiconductor licensing campaign since the company's inception in 1989 and clients include such hallmarks as Infineon, Intel, TSMC, and Texas Instruments.

gdp77 said...

Scientia:
I'm having a hard time to post in your blog. This is the third time I register with a different e-mail and I hope that my account will be functional this time.


Same thing happened for me... I had to sign up again, using different email, and the user interface wasn't so easy to make u understand what exactly u should do... I think many people will also have the same problem.

Scientia from AMDZone said...

I've only switched accounts once. I noticed that my old blogger name stopped working properly. So, then I started using my google account and that has worked fine.

enumae said...

Scientia

Do you think AMD will be able to maintain the market share gains it made in 2006, to the end of 2007?

Do you think AMD will be able to pay for 45nm and 32nm equipment if the ASP's stay low and if they are loosing market share?

Considering that Intel has plans to pay for there 45nm equipment in the first half of 2007, does this give them some room to apply pricing presure for Mobile, Desktop and Servers for the second half of 2007 against AMD?

How do the reports of AMD burning thru a billion dollars in the first half of the year effect the second half of the year when this is when they are projected to begin converting FAB 30?

If Intel is able to release Penryn with performance relative to Barcelona in general not just a few specific bench marks, how will this effect AMD's future plans?

I am really hoping you will take the time and answer these questions, and I know you don't have a crystal ball, but I feel some of these points are very relative to your post.

Thanks, enumae.

Roborat, Ph.D said...

Observations of the past and then quickly predicting the future without recognizing whats happening in the present isn't such a wise thing to do. I don't have to look too hard to find Intel's influence as strong as ever to prove how incorrect your post is (WiMax, VT, Robson, SUN, Apple, etc).

So, we should continue to see AMD and Intel moving closer together into 2010.

I can see you haven't learned from AMD's Q4 earnings report. It's hard to see an AMD past 45nm. Intel is hell bent in making sure AMD never becomes profitable ever again, well, until it joins the merry club of TMTA and VIA.

Roborat, Ph.D said...

Erlindo said: And about AMD introducing 45nm and 32nm in schedule, I believe one of the main reasons AMD acquired Douglas Grose from IBM was to achieve these goals in a timely manner.

LOL! I'm sorry, can you please explain to us how an IBM person can help AMD get to the next node quicker? so much for the alliance then huh?

Fujiyama said...

Roborat, Enumae
AMD has money for 45nm conversion and it won't loose market share.
I observe more and more that all market projections are based on "who is the fastest now" status.
Having CPU faster 5% than competition you are the leader, 15% - competition is doomed.
CPUs below 120$ which are 80% of the market mean are the AMD territory in terms of price/performance.
Even Sharikou is right saying that X2 3600+ is better offer than Intel.
Let me give you an example.

I'm going to rebuild my S393 machine next month.
I have two options:
- buy S393 Opteron 165 for 700PLN
- rebuild the whole system selling my CPU,motherbord etc.
I decided to go with 2nd option.

From start Intel has no chance to come to my PC. Why? Because I'm able to define my needs and price/ performance numbers.
The only problem I have is the chipset because they all lack support for HT3/AM2+ and in the future I would like to upgrade just CPU using HT3 (Kuma or else).

I believe that my choice is reasonable, logic and cost-effective. Intel has nothing to offer:

Prices:
C2D E4300 671PLN
C2D E6600 900PLN
Motherboard Gigabyte
GA-965P-S3 Intel P965 413 PLN
(this is the cheapest motherboard)


A64 X2 3800+ EE 363PLN
Motherboard Sapphire XPress 1150
AM2 269PLN
(this is also the cheapest motherboard)

Summary:
Intel platform (E4300) 1084 PLN
AMD platform (3800+) 632 PLN

No, compare these two processors benchmarks here


So, who is going to pay 452PLN for faster movie decoding?
Even Paris Hilton says "I'm not stupid!" :)

enumae said...

Fujiyama
AMD has money for 45nm conversion and it won't loose market share.

If the reports are true that they will burn through 1.0 of the 1.5 Billion they had leaving 2006, announcing that Q1 will miss projected earnings, looking like Q2 will as well, ramping 65nm and 300mm wafers at FAB 30 at the beginning of Q3, while possibly loosing market share and facing lower ASP's...

Can you explain where they will get, or where they already have the money?

There have been reports stating that AMD inventories were building up in the channel.

Can you explain why you don't think they will loose market share?

CPUs below 120$ which are 80% of the market mean are the AMD territory in terms of price/performance.

First do you have a link specifying that 80% of processors sold are below $120?

Assuming your right about the 80%, this is where marketing comes into play.

People maybe hearing about AMD, but they already know Intel.

If people who know nothing about CPU's, read a review of Core 2 Duo, they could easily come to the conclusion that Intel has a better processor right now than AMD, and could therefore conclude that there other products are as well.

We know this is not true, but this is marketing, and is why Intel still sold alot of Netburst, name recognition, and the fact that the general public has no idea what they are buying.

Fujiyama said...

If you remember AMD profits in the past - AMD had low profits or none and was able to build FAB30, convert into 90nm. The same with 45nm.

Burning 1.5bln in very hard to believe. Are AMD costs higher than Intel?
I remember estimate cost for Intel 90nm CPU was 40$ with larger die than K8. So it should be similar or less for AMD at 90nm or 65nm dual-core.

I don't know any rules how people buy a computer. But I know that more and more of them knows that there is a small difference between Intel and AMD products.
Like everyone I look for best price and will never spend money for AthlonFX because it was never worth 999$. If you know the 80/20 rule you can put the price barrier somewhere and you recognize 20% of buyers who can spend more and 80% spending less. AMD is well prepared for this 80% with X2.
The other aspect is the status of the brand which was shot twice during last months. But this is lack of AMD response for Conroe.
No MCM quad-core, Brisbane with L3 cache (why not?), faster 65nm. There is silence and analysts see only troubles. But I think that in next two months everything is going to change. Penryn won't spoil the party and K10 show the green muscle. :)

enumae said...

Fujiyama
If you remember AMD profits in the past - AMD had low profits or none and was able to build FAB30, convert into 90nm. The same with 45nm.

Well from what I understand AMD did not have the debt it has now after the purchase of ATI.

Burning 1.5bln in very hard to believe.

I said the report was 1.0 Billion, not 1.5 Billion.

Are AMD costs higher than Intel?

I have no idea, and I doubt most, if anyone, outside of Intel or AMD do. People can speculate and back up claims with reason, but it is not usually proven with fact.

I remember estimate cost for Intel 90nm CPU was 40$ with larger die than K8.

I believe I had read that too.

So it should be similar or less for AMD at 90nm or 65nm dual-core.

That would make sense...

I don't know any rules how people buy a computer. But I know that more and more of them knows that there is a small difference between Intel and AMD products.

I believe that, but the Intel name is still very much accepted as better than AMD.

Like everyone I look for best price and will never spend money for AthlonFX because it was never worth 999$.

I simply can not afford it...

If you know the 80/20 rule you can put the price barrier somewhere and you recognize 20% of buyers who can spend more and 80% spending less.

I am not familiar with the 80/20 rule.

AMD is well prepared for this 80% with X2.

I do not doubt that AMD has a great lineup for the price.

At the same time Intel will as well come April 22nd.

The other aspect is the status of the brand which was shot twice during last months. But this is lack of AMD response for Conroe.

I agree.

No MCM quad-core, Brisbane with L3 cache (why not?), faster 65nm.

All of those things, if capable of being produced by AMD, I beleiev we would have seen them.

There is silence and analysts see only troubles.

Do you blame them?

But I think that in next two months everything is going to change.

It very well could, but again the silence is leading to alot of speculation.

Penryn won't spoil the party and K10 show the green muscle. :)

Looking at AMD's 42% claims at 2.3GHz, and those being on Spec_fp_2000, it very well could spoil AMD's party.

All of this depends on clock speeds and FSB.

Do you think if Penryn was clocked at 3.2GHz with a FSB of 1600, that it would be comparable to a 2.3GHz or 2.5GHz Barcelona?

Unknown said...

"I can see you haven't learned from AMD's Q4 earnings report. It's hard to see an AMD past 45nm. Intel is hell bent in making sure AMD never becomes profitable ever again, well, until it joins the merry club of TMTA and VIA."

This is called violating anti-trust law. I'm pretty sure Intel doesn't want the same status as Microsoft.

All of you seem to constantly forget (do you have alzheimers?) how much money AMD was losing during the k7 generation? I mean, scientia keeps bringing it up, so do you just ignore it? If AMD is doing better than then, actually has sizable marketshare, is now a mainstream product, is finally getting its own instruction sets standardized, bringing its production forecasts forward (which is the first step to getting the actual production up there), and is finally a recognizable product. I don't see how this puts them in a worse situation than they've ever been in. I realize Intel is still more recognizable, but it's still built as a company that is the only recognizable brand, which is a problem for Intel as it exists today.

"LOL! I'm sorry, can you please explain to us how an IBM person can help AMD get to the next node quicker? so much for the alliance then huh?"

Knowing how to push people to produce a product faster is not something you're taught in school, or something that everyone knows how to do. Bringing in someone from IBM who is known for this would obviously bring the same results for AMD. For you not to see this, roborat, proves you're either normally or being intentionally dense.

"Observations of the past and then quickly predicting the future without recognizing whats happening in the present isn't such a wise thing to do."

So making extremely general statements that have no backing or validity to discredit someone's material is a good idea? You apparently think so because you do it a lot.

Fujiyama said...

Enumae
There is silence and analysts see only troubles.

Do you blame them?


Yes, I blame them because they see only financial and current market situation. AMD evaluation doesn't take serious ATI takeover seeing only debts.

BTW if you look at AMD history there were debts smaller and bigger and they were still able to buy new toolsets and design new chips.

Regarding performance gain I see 40% improvement was compared to 2.3GHz Barcelona - cause AMD is not able to produce anything faster right now.

Peryn 3.2GHz/3.4 will probably match this chip.
So what? You can always rise the bar clocking another 100MHz and another 100MHz where Intel has no chance to go...

Azmount Aryl said...

Fujiyama said...
Peryn 3.2GHz/3.4 will probably match this chip.
So what? You can always rise the bar clocking another 100MHz and another 100MHz where Intel has no chance to go...


I found interesting article describing Intel Qaud's wattage:
HP, and other OEMs, should have Clovertown gear ready on the 14th. Our sources inside HP say the chip is eating between 140 watts and 150 watts, so please wear your protective gloves when making component changes.LINK

Also, some unofficial sources been saying that 3.0GHz quad-xeon found in This Apple machine are 150W parts.
Now i know as a fact that while yields do improve for intel, wattage does not, as so, i do not see how they will be bumping those 100MHz past certain point, even for penryn. What do you have to say to this?

enumae said...

Azmount Aryl
What do you have to say to this?

Who is this directed towards?

enumae said...

Scientia

Any response from my questions?

Unknown said...

I find all of these pro-AMD comments really, really funny in light of the most recent revenue warning of both units shortfall and ASP's in the gutter. On top of that, $500M will be cut from capital expenditures - how do you think that will impact the 45nm ramp at AMD? I think removing $500M more or less pushes 45nm by 6 months to a year at a minimum as AMD is still ramping 65nm now. I think you should all go back and read my previous posts about AMD struggling and the optimistic predictions/arguments most of you made then have been proven to be totally wishful thinking BS in light of the most current information. Granted, this blog is a little more realistic than Sharikou's, but that is not saying too much....

Roborat, Ph.D said...

it's funny that whenever Scientia posts something rosy for AMD, AMD never fails to immediately announce a revenue miss or an awful news.

Bradley Coleman said...

sci, its brad/whorush from amdzone. i love your blog.

i really like this trend towards minimalism: DTX, FUSION, IMC, USB, etc. i think it will put the PC into new markets, by bringing the price/wattage down a lot. i think it will be one of the big steps towards the maturation of the PC.

i'm not a performance user, i just want something simple and good enough.

btw, i'm using FIREFOX on linux and this website really really creeps! there's something buggy going on somewhere in the website ... btw, i dont get this bug on ANY OTHER website, but i always do here.

Crainial said...

gosh, an Intel fanboi...

Unknown said...

Well, you can laugh at the AMD fanboys, and we can laugh at you for hating AMD. As for that 45nm being pushed back 6 months because of 500mill, I don't really see how you can immediately connect a specific production time table and that cost, though if you can cite some sort of source or specific reasoning for your argument I'm sure the "fanboys" would appreciate it.

roborat, I don't see how your post is in any way relevant, seeing as scientia is talking about problems at Intel. It is actually possible for both AMD and Intel to face hard times at the same time (though you seem to find this very hard to understand). Also, scientia has predicted short falls at AMD for this and next quarter, though I'll leave it to scientia to back this up if he feels the need to.

Anonymous said...

Well written article & thorough. I have followed Intel & AMD since the days of 133 mhz cpu's which by the way was my last Intel. I believe AMD will prevail in this pricing war as Intel has appx 99,000 employees compared to Amd's 16,500. Lot of checks to write when revenue is 80% lower then expected. Intel will suffer the most under these conditions. Period. Just my 2 cents worth

Anonymous said...

Intel has had a problem with heat almost ever time they ramp up the mhz. No different now in their cpu's.

enumae said...

only amd
Lot of checks to write when revenue is 80% lower then expected. Intel will suffer the most under these conditions. Period.

Where do you get 80% lower than expected?

Last few quarters Intel has been able to bring in profits of...

Q3 2006 = $1.3B
Q4 2006 = $1.5B

I am unable to see your connection here, please explain.

Periander said...

greg:

All of you seem to constantly forget (do you have alzheimers?) how much money AMD was losing during the k7 generation? I mean, scientia keeps bringing it up, so do you just ignore it? If AMD is doing better than then, actually has sizable marketshare, is now a mainstream product, is finally getting its own instruction sets standardized, bringing its production forecasts forward (which is the first step to getting the actual production up there), and is finally a recognizable product. I don't see how this puts them in a worse situation than they've ever been in.

Two points. One, AMD's balance sheet and debt load is much worse now than it was then, even during AMD's darkest days just before the K8 introduction, AMD had 739 million in cash vs. 1.6 billion in debt.

http://tinyurl.com/39smvn

As of the end of Q4 AMD had 1.5B cash vs. 3.7B in debt, and it's quite possible that half of that cash will be wiped out by losses by the end of this quarter.

Worse, much much worse, are the idiotic terms imposed by the 2.5B loan they took on to buy ATI.

Read this:

http://tinyurl.com/2poc5g


"Under the terms of that credit agreement, if AMD raises money through a new debt issue, it is required to use all those proceeds to help pay down the Morgan Stanley loan. If AMD issues new shares, 50% of those proceeds would go toward the loan, according to Securities and Exchange Commission filings."

This horrible horrible loan drastically impacts their ability to patch up their balance sheet. They can't borrow money to fix the problem, they have to issue equity, and instead of needing to issue 1B worth of shares they need to issue 2B. We're talking massive dilution, and very tough sell when your whole market cap is 7B and falling.

Worse is that the spectre of major dilution helps drive down the stock price, which means and even larger dilution, which puts even more downward pressure ect ect. It's a death spiral effect.

AMD is in terrible trouble. They need to renegotiate the MS loan (good luck with that) or they need someone to buy them out. Otherwise, Chapter 11 is a very real possibility.

lex said...

Under siege?

The only thing under siege right now is AMD's future.

Re-org and 500M expenditure reduction will be in 65nm capital and 45nm R&D.

There is NO way AMD is going to be pulling in 45nm. If anything they will be pushing it out.

INTEL already has multiple CPU designs out. Its been a quarter or so since the first Penrym so I figure that they have final production designs already running and thus 2nd half 2007 is a done deal.

AMD/IBM are still struggling, note INTEL annouced fully functional and yielding 45nm more then a year ago. AMD / IBM annoucment was very very vague and said something about functional SRAM blocks. That means their 45nm design has yet to yield a fully fuctional SRAM at 45nm designs. Without that there is nothing. AMD/IBM is 2 years out at best.

Lastly they will have little performance pop at 45nm. Without advance strain and HighK metal Gate the abilit to scale without penatly for gate oxde leakage or draing leakage will be minimum. Their 45nm technology will simply be a lame shrink with little peformance advantage.

AMD is cooked

Azmount Aryl said...

lex said...
There is NO way AMD is going to be pulling in 45nm. If anything they will be pushing it out.


Besides the personal opinion of a hopeless intel fan boy is there any source that you can use to back that statement? I see, FUD it is then.


lex said...
INTEL already has multiple CPU designs out.


They don't have multiple designs out. What they do have is one version of duel core design working with different FSB frequency and voltage in different packages as well as upcoming (but far from being out) shrink of that same design.


lex said...
AMD/IBM are still struggling, note INTEL annouced fully functional and yielding 45nm more then a year ago.


Wrong. AMD/IBM SOI process already gives their chips more advantage than high-K metal will give intel on 45nm, while double patterning on intel's 45nm translates into disadvantage as it worsens yield and requires more runs.

Unknown said...

Just a quick post before I have to run to class, but I find this interesting.

http://www.theinquirer.net/default.aspx?article=38781

Turns out THG has been sold. No idea who the buyer is, but it would be hilarious if it were Intel (j/k).

Unknown said...

A close friend of mine, senior manager in Intel sales has perspective re: this latest round of battle between Intel/AMD.

Prior to 4Q05 Intel miss, the environment at Intel was tangent strategy such as Viiv, beat Sun (accomplished), end user marketing, mobile innovation, etc. Headcount increases(very easy), job redundancy was evident in S&M.

The cracks appeared in server share and ASP as Intel was caught off guard by poor netburst strategy among other things. Since the reorg (not handled very well/decreased moral) and price war, the renewed focus is paying dividends. Customer wins and improved roadmap are strategic focus as wimax/other strategies become secondary(not forgotten). The giant is awoken as the US after Pearl Harbor. However, AMD needs Intel and converse also. Even employees realize this. Both companies have strengths requiring the other to innovate.

Intel marketing machine is a monster (#5 brand in the world/ worth $32billion, AMD not even near top 100, however INTEL dropped from $35 billion in 05). x486/Pentium 1-3 dominance continuing to pay dividends 10 years later.

On consumer side, case studies show brand dominance puts Intel in leading position even with competitive AMD chip. Market followers need superior offerings for years to erode market leadership. Sustained 2-5 year leadership (depending on brand), not occasional 1-1.5 years leadership. Microsoft(#2 brand) vs. Linux, even more lopsided, for example. Research reports indicate consumers walk into Best Buy/Circuit City/CompUSA/ preferring Intel by a massive margin (10 to 1), but purchase AMD system many times due to price/performance/education (ratio decreases significantly).

Example: My doctor brother (complete non tech) didn't want to buy HP/Sony/Dell system, he wanted Intel system; that is the power of brand dominance from Pentium days. Intel brand more dominant than even OEMs brands.

As in the past, Intel seems to have responded in time from marketing perspective to prevent further brand erosion. AMD system on Best Buy shelf must vastly outperform Intel to change the 10 to 1 to more favorable ratio en masse.

Long term 75/25 mix is realistic as Intel continues to exploit its process and brand advantage. Price war continues from Intel as long as stockholders stay on the boat, which seems to be happening. Innovation continues as industry matures and both become significant dividend paying stocks.

You can debate the technical merits of product roadmaps, tech specs, and both sides will always have merit. Those who buy from Newegg, read Tom's hardware/Anadtech remain in the minority and marginalized due to Intel $200 million ad campaigns.

Link to brand info:
http://tinyurl.com/yt9v95

Scientia from AMDZone said...

ho ho

You implied that I wanted to use the Portland Group compiler because it would favor AMD. There is nothing to indicate that this compiler favors AMD.

Saying that you are sorry that I don't like free speech or that I can't take criticism is not an apology. To be honest your later comments suggesting that I then want to cheat by using different DIMMs makes me doubt that you have any intention of admitting you were wrong.

You do not have posting priviledges here until you apologize.

Unknown said...

Tom, you might want to bother to read over scientia's reply to what is literally the exact same accusation (which apparently hoho completely ignored). I don't blame you since you just joined us, but ho ho should know not to do something so stupid.

Your points on Intel's brand recognition are great, and are part of why it's important for AMD to use this time to increase market penetration. I have the feeling that AMD might back off of their insane race to 30% market share for a while, and cut back production at 90nm a lot. They might also use Intel's strategy and start ramping parts of production, then moving onto the next part and stopping with the others.

While tom's is read by a very small portion of the market, it affects a much larger portion than that by word of mouth. Generally the people who sell these products are in the minority that reads review sites like Tom's, and thus they'll recommend products based on what they've read. Also, enthusiasts that consumers trust will pass on what they've read via word of mouth. Eventually, this caused erosion of Intel's market share to a certain extent, which is why HP took a chance and started selling AMD 64 before anyone else.

Scientia from AMDZone said...

enumae

Do you think AMD will be able to maintain the market share gains it made in 2006, to the end of 2007?

Honestly? I could see a loss of a few percentage points in Q1 and Q2 and maybe gaining back some in Q3 and Q4. AMD could be even or down slightly at the end of the year. But, I still do not understand the extremely negative views. In Q4 2001 AMD was at 23% by volume and two quarters later in Q2 2002 was at 8%. I don't see that happening.

Do you think AMD will be able to pay for 45nm and 32nm equipment if the ASP's stay low and if they are loosing market share?

Well, 45nm, absolutely. There is no question that 45nm tooling has been ordered and will be installed. This equipment should begin testing mid year. There is no possibility that this equipment will not be installed. I don't know about 32nm. If AMD did have low revenues through 2007 and 2008 then it could push 32nm back.

Considering that Intel has plans to pay for there 45nm equipment in the first half of 2007, does this give them some room to apply pricing presure for Mobile, Desktop and Servers for the second half of 2007 against AMD?

No. The second half of the year will be better for AMD than the first half. The second quarter will be maximum pressure.

How do the reports of AMD burning thru a billion dollars in the first half of the year effect the second half of the year when this is when they are projected to begin converting FAB 30?

AMD cuts capital spending

AMD has announced that it is reducing its capital expenditure plans for 2007 by $500 million. Initially, AMD planned to spend $2.0 billion this year on adding capacity at Fab 36 and initial production ramp of its converted facility Fab 30 to 300mm wafer production.

The company also stated that it would be restructuring the business in an effort to reduce operating costs and a virtual hold on discretionary expenses and hiring.

AMD said that the reduction in capital spending would not ‘materially impact capacity plans for the year'


It's hard to tell. This could mean that FAB 38 will ramp more slowly. But I figure that AMD will give more details at the Analysts Meeting in June.

If Intel is able to release Penryn with performance relative to Barcelona in general not just a few specific bench marks, how will this effect AMD's future plans?

I don't think it would effect AMD's plans beyond perhaps slowing AMD's rate of expansion. AMD is already committed to DC 2.0, a new mobile core, a modular core, MCM and on-die GPU, HT 3.0, 45 and 32nm.

I am really hoping you will take the time and answer these questions, and I know you don't have a crystal ball, but I feel some of these points are very relative to your post.

You need to understand the time span, which seems to be something that other posters here have had trouble with.

Currently AMD has no quad core and is losing ground on the desktop and probably mobile too. However, the new chipset will help with mobile and low end desktop. This should help with volume but isn't going to make up for revenue loss.

By Q3 AMD should have a solid quad offering but will still be lagging on the upper desktop in dual core. So, mobile will probably be okay with AMD only holding the lower desktop.

By Q4 AMD should have a dual desktop offering however Intel will have Penryn. AMD may begin slipping in mobile before the new chipset is released.

In 2008, AMD releases the new mobile core and the 700 series chipset which includes Intel's Robson. AMD releases DC 2.0 and 45nm.

Okay, this should mean in 2008 that with DTX and MCM GPU AMD will be solid in the low end desktop. Mobile should be solid. AMD could be slipping in the upper desktop quad core but would probably be fine in dual core.

Servers are tough to tell. I would guess that AMD should be fine in servers once it gets K10 out. It is also possible that Intel's 1600Mhz FSB could put a serious damper on its enthusiast market but this wouldn't be much of a loss of revenue. I figure Intel will hold the upper desktop quad area but some of this also depends on AMD's Stream computing plans. Again, these things should be clearer after the Analyst Meeting in June.

Scientia from AMDZone said...

periander

You've manage to link to a three year old article from 2004 which is the same one ho ho found. This article does not refer to any particular preference from Portland towards AMD. What it actually refers to is that AMD64 was brand new in 2003 and the Portland Group started working with AMD in 2002 so that it could support AMD64 right away when K8 was released. Recall that Intel did not officially support 64 bit extensions at first so it would have been impossible to have been working with Intel.

If you do any checking at all on Portland Group you will find that they have no preference for AMD and support Intel processors equally with code optimized for each.

Scientia from AMDZone said...

periander

I understand why you and others would be confused about the debt situation. It's an easy mistake to make. This is the actual situation:

AMD

Q1 2003
Assets - $7.2 Billion
Debts - $3.0 Billion
Balance + $4.2 Billion

Q4 2006
Assets - $17.1 Billion
Debts - $6.5 Billion
Balance + $10.6 Billion

Periander said...

Portland Group is still quoting AMD, and not Intel, spokespeople in their current press releases and bundling AMD software with their own.

http://www.sys-con.com/read/347571.htm

... and a bundled version of AMD's ACML, a library of highly optimized numeric functions for mathematical, engineering, scientific and financial applications.

"AMD multi-core processors with Direct Connect Architecture, along with PGI's compilers, deliver incredible performance for many widely-used 64-bit engineering and scientific applications such as geophysical modeling, mechanical engineering, non-linear dynamics, computational chemistry and high-energy physics," said Margaret Lewis, director, Commercial Solutions, AMD. "Working closely with leading technical innovators like The Portland Group to ensure ongoing support for the latest processor innovations is a key component of our strategy and our commitment to HPC."


Of course they are going to claim that they are fully optimized for both processors, but given their history of working closely with AMD, their past and current statements, I find a claim of their complete impartiality dubious at best. Find a Portland Group press release where an Intel spokesperson is quoted about "working closely" with Portland Group, and I'll be happy to retract my statement.

Periander said...

Scientia:

Q4 2006
Assets - $17.1 Billion
Debts - $6.5 Billion
Balance + $10.6 Billion


These figures are rather dramatically different from ones given in:

http://www.amd.com/us-en/assets/content_type/DownloadableAssets/FinancialsQ406.pdf

Care to explain why the figures provided by AMD are wrong?

Scientia from AMDZone said...

lex

Re-org and 500M expenditure reduction will be in 65nm capital and 45nm R&D.

It has no effect on 45nm R&D although it could effect 32nm.

There is NO way AMD is going to be pulling in 45nm.

AMD won't have to pull in 45nm. Mid 2008 is the normal schedule.

If anything they will be pushing it out.

This won't happen either. However, if you wanted to argue that it could push out 32nm then you would have a point.

INTEL already has multiple CPU designs out. Its been a quarter or so since the first Penrym so I figure that they have final production designs already running

Then you don't understand chip design. The chips that were shown were Alphas. You can go through 2 or three Alphas before you get a good Beta. At that point they can only produce from D1D.

and thus 2nd half 2007 is a done deal.

Yes, Q4 2007, not Q3. 2 months after production at D1D they can begin production on other FABs.

AMD/IBM are still struggling,

Struggling? Actually AMD produced a 45nm SRAM test chip just 3 months after Intel. To stay on schedule AMD's 45nm chip doesn't have to tapeout until June. Are you seriously trying to suggest that you have information about AMD's pre-tapeout status?

note INTEL annouced fully functional

Now you are making a fool of yourself. When Alpha chips are tested they use BIOS patch code to keep them from crashing. These chips are neither fully functional nor stable.

and yielding 45nm more then a year ago.

Are you referring to the SRAM test which AMD did 3 months later?

AMD / IBM annoucment was very very vague and said something about functional SRAM blocks.

This indicates that your knowledge of the development process is even worse than I thought. Process testing always begins with SRAMs.

That means their 45nm design has yet to yield a fully fuctional SRAM at 45nm designs.

From April 2006
AMD Update

AMD also gave us a brief update on 45nm, stating that they have successfully produced a SRAM test wafer at 45nm. It's very popular to produce a wafer full of SRAM chips as you're bringing up and validating any new process, since the circuits are simple enough to actually make the chips producable but complex enough to get useful feedback on your process from the test wafer.

AMD's 45nm SRAM test wafer follows a little over 3 months after Intel demonstrated a similar 45nm test wafer.


Without that there is nothing. AMD/IBM is 2 years out at best.

You are out of touch with reality.

Lastly they will have little performance pop at 45nm. Without advance strain and HighK metal Gate the abilit to scale without penatly for gate oxde leakage or draing leakage will be minimum.

This shows that you are not familiar at all with AMD's historical development patterns. 45nm will have a low clock at release however AMD will push 65nm clocks higher. Intel doesn't do this.

Their 45nm technology will simply be a lame shrink with little peformance advantage.

No. The initial release will not have faster clocks. Look for the 45nm clock increase to be in 2009.

Scientia from AMDZone said...

tom

Thank you for the marketing information. This is a good example of how marketing statistics can be misinterpreted.

For example if you look at the value numbers for brand recognition for Coke and Pepsi you will see a rather surprising ratio of 5:1 in Coke's favor. This is a polling statistic and has no actual connection to reality. The real ratio in terms of sales of product between Coke and Pepsi is 4:3 in Coke's favor. It is interesting but bears no relation to either current or future sales.

Scientia from AMDZone said...

periander
Care to explain why the figures provided by AMD are wrong?

Sorry, I had a mistake in my spreadsheet. This should be correct.

AMD

Q1 2003
Assets - $5.4 Billion
Debts - $3.0 Billion
Balance + $2.4 Billion

Q4 2006
Assets - $13.1 Billion
Debts - $6.5 Billion
Balance + $6.6 Billion

enumae said...

Scientia

Thank you for taking the time and answering my questions.

Periander said...

The Q4 Balance Sheet from AMD's own website shows

Assets: 13.1B
Liabilities: 7.4B
Stockholder Equity: 5.8B

The problem with this balance sheet is that no less than 4.4B of the assets are Goodwill and "Acquired Intangible Assets". These are accounting fudge factors to account for the difference between the price AMD paid for ATI and the tangible assets they acquired in the transaction. They have no meaning unless AMD intends to turn around and sell ATI off. Not a likely event at this point.

AMD's net tangible assets dropped as a result of the acquisition from 5.0B all the way down to an alarming 1.4B. Even in Q1 2003 it was the same as Stockholder Equity, i.e. 2.4B. Combine this with the insane covenants included in the Morgan Stanley loan they used to finance the acquisition and AMD's room for financial manuvering is drastically restricted.

Scientia from AMDZone said...

periander

Well, I'll give you credit for effort but no credit for results. Intel has the same thing called Intel Math Kernal Library.

ACML is optimized for AMD but supports Intel.

IMKL is optimized for Intel but supports AMD.

Both include LAPACK and BLAS.
Both include Fast Fourier Transforms.
Both include vectorized Transcendental Functions.

You can use the PG compiler with either library. The reason ACML is included but IMKL is not is that Intel charges $400 for IMKL.

Unknown said...

Scientia

Brand value is a relative, not necessarily absolute, metric. Sales ratios should not be construed from brand ratios. However, please understand that brand analysis is a tool used by many in the financial community. By maintaining a strong brand, Intel has marketing momentum inherent to their chips; a multiplier effect for their marketing campaigns. It can help Intel recover from mis-steps and take the brunt of superior AMD products.

I guess my point is AMD will have extreme difficulty overcoming a superior product AND brand. Brand awareness can lead sales. I'd like to see how AMD's brand awareness has increased over the last few years, but it's probably not available.

Great blog.

Scientia from AMDZone said...

periander

The problem with this balance sheet is that no less than 4.4B of the assets are Goodwill and "Acquired Intangible Assets".

Not quite. I'll agree with you that we can leave out goodwill which is $3.2 Billion however the intangible assets include the research and intellectual property from ATI. AMD lists this as $1.2 Billion. Since my rough estimate of this value would be $1 Billion I would say that AMD's figure is more than reasonable.

They have no meaning unless AMD intends to turn around and sell ATI off.

This statement is rather foolish. You could argue that a FAB has no value unless it is sold. Intellectual property has value in the same way that FABs do in that they allow you to make a product that has tangible value. Try making any IC chip with no intellectual property. We'll leave out goodwill which roughly includes assembled design teams and management plus relationships with foundries and established relationships with customers. I'm willing to discount this since this is difficult to quantity. For example, how much was lost with the loss of Intel as a customer? So, we'll treat the value as zero.

Once again:

AMD

Q1 2003
Assets - $5.4 Billion
Debts - $3.0 Billion
Balance + $2.4 Billion

Q4 2006
Assets - $13.1 Billion
Goodwill - $3.2 Billion

Adjusted Assets - $9.9 Billion
Debts - $6.5 Billion
Balance + $3.4 Billion

sharikouisallwaysright said...

From the Inq with love:
http://uk.theinquirer.net/?article=38836

R6xx ordered in tons from Dell and other OEMs...

Scientia from AMDZone said...

fujiyama

CPUs below 120$ which are 80% of the market mean are the AMD territory in terms of price/performance.

Intel's ASP is $130
AMD's ASP is $79

This puts the market average at $117. Now $120 is pretty close to $117 but 50% of the chips below $117 is a lot less than 80%.

Scientia from AMDZone said...

enumae

How do the reports of AMD burning thru a billion dollars in the first half of the year effect the second half of the year

Since I've been asked to provide links (and I have) I would love to see a link where AMD has said anything like this. Or was your source merrill lynch or forbes? Both of these have demonstrated serious gaps in understanding about processor and gpu development. In other words, taking advice from these two would bankrupt most companies.

Scientia from AMDZone said...

real

"in light of the most recent revenue warning of both units shortfall and ASP's in the gutter. On top of that, $500M will be cut from capital expenditures"

Yes, I've mentioned these specifically.

" - how do you think that will impact the 45nm ramp at AMD?"

It won't effect the launch. 45nm will begin mid 2008. I suppose it is possible that this could cause a slower ramp. I don't think this is possible on FAB 36. However, I suppose it is possible that it could effect the ramp on FAB 38.

" I think removing $500M more or less pushes 45nm by 6 months to a year at a minimum"

Not even close. However, I guess I can't fault you too much for making this mistake because I've seen other "professional" analysts do it too.

"as AMD is still ramping 65nm now."

Then I assume you don't understand that by the time Barcelona is launched AMD will no longer be making 90nm on FAB 36.

enumae said...

Scientia
Since I've been asked to provide links (and I have) I would love to see a link where AMD has said anything like this.

I think the only link you will get from AMD will come after April 19th, if the reports turn out to be true it will be in there balance sheet, right?

Like I said, "How do the reports of AMD burning thru a billion dollars in the first half of the year effect the second half of the year..."

Christian Jean said...

sharikouisallwaysright said...
http://uk.theinquirer.net/?article=38836

R6xx ordered in tons from Dell and other OEMs...


100 million? How much do each of these cost? Or more importantly, does anyone know how much money AMD will make per chip? Are we talking about profits of $0.01 each, $0.10 each, $1.00 each or $10.00 each?

Azmount Aryl said...

Found something here:

The guys from Boot Daily claims they've got their hands on the pricing details of AMD's upcoming graphics cards:

# R600: $499.00
# R630 XT: $199.00
# R630 Pro: $179.00
# R610: $99.00

(link)

Scientia from AMDZone said...

jeach

There is no way that AMD is going to sell 100 Million chipsets by the end of the year. 50 Million, possibly. Probably between $8 and $15 profit each.

Wise lnvestor said...

Scientia from AMDZone said...
Both of these have demonstrated serious gaps in understanding about processor and gpu development. In other words, taking advice from these two would bankrupt most companies.

Wise words. Sometimes I find it smart to do the opposite of what they say.

Jeach, Scientia I heard from somewhere GPU is a close to 9 Bil $ industry. And it's going strong with double digit growth while the rest of the tech sector show signs of slowing. I believe when nVidia posted their earnings, their margin was 48% +-.

From what I heard the R6xx GPU's are very impressive. The process effectively been 65nm with a few add in goodies(sound card)(HDMI).

On the other hand, nVidia wont be getting 65nm of their own till end of 07. The R6xx could run fast and furious till the end of year. And remember nVidia has no fab nor APM , AMD have them both.

APM could help TSMC quite a bit, my instincts tell me AMD is looking for some sort of alliance/partnership with Taiwan.

On the side note. Essential for Vista, Vista Ready failed for what they claim to be. So Dell's move along with other OEM's didn't surprise me.

Scientia from AMDZone said...

wise inv

On the other hand, nVidia wont be getting 65nm of their own till end of 07. The R6xx could run fast and furious till the end of year.

At the end of the year, AMD will be looking at R700.

And remember nVidia has no fab nor APM , AMD have them both.

AMD's FAB won't be a factor until Fusion.

APM could help TSMC quite a bit, my instincts tell me AMD is looking for some sort of alliance/partnership with Taiwan.

No. APM wouldn't help. The process at TSMC is completely different. It works with Chartered because they use the same process as IBM/AMD.

Scientia from AMDZone said...

ho ho

When you found that announcement about Portland Group from three years ago I'm sure you felt that you had found a smoking gun that showed that I wanted testing to favor AMD.

Obviously you didn't understand the context of the announcement nor do you or periander seem to understand PG's business. Unlike MS which sells OS's and applications and unlike Intel which sells processors and chipsets PG depends on its compiler. The notion that it would build a general purpose compiler to favor AMD is nonsense and goes against even the most basic understanding of business. It should be common sense that PG would not design its compiler for 25% of the market. Logically, they would make the compiler as robust as possible for both and have access to the 99% of the market that doesn't include VIA. PG gives no indication that they optimize for VIA.

So, if you want to post here stop talking about freedom of speech and how I can't take criticism; stop trying to compare your posts with everyone else's (red was the most annoying and he no longer posts here). Instead, simply explain how you came to such a wrong conclusion about my motives.

LG said...

Hello scientia. Great blog. I was wondering if you had time to read this article posted at B3D and give your analysis on it?

LG said...

Oops, I guess a link would help! :(

http://www.beyond3d.com/content/articles/32/1

abinstein said...

"This puts the market average at $117. Now $120 is pretty close to $117 but 50% of the chips below $117 is a lot less than 80%."

It's 50% wrt dollars, but could be 80% wrt units.

abinstein said...

"If you remember AMD profits in the past - AMD had low profits or none and was able to build FAB30, convert into 90nm. The same with 45nm."

AMD had no cash and made no profit because it had to spend all the money it earned on the fabs it built.

The development of 45nm is basically over. AMD should be and will be able to afford 45nm. 32nm or 22nm will be the real challenges for AMD both technical and financial wise.

Unknown said...

There is no way that AMD is going to sell 100 Million chipsets by the end of the year. 50 Million, possibly. Probably between $8 and $15 profit each.

You have absolutely no idea about chipset pricing if you honestly believe this. Chipsets will run around $10-15 in ASP for the type of chipsets ATI provides. Unless you think there is zero cost, your 'profit' numbers are laughable.

Scientia from AMDZone said...

real

If you have the wholesale prices for each chipset in the range then please post them.

Scientia from AMDZone said...

abinstein

Yes, it is possible that 80% of the units could be below 117 but it doesn't seem likely given a normal statistical spread. The real number is probably at least 2/3rds but I'm not sure that it would be as high as 3/4.

Scientia from AMDZone said...

lg

I looked at that article and others at beyond 3D. There really isn't anything in the revenue article. The author simply gives AMD's estimate and then spends the next several pages wondering if the drop is due to ASP, volume, or both and then wondering if the drop is in processors, chipsets, or consumer products.

Since we don't know what the cause of the drop is, we don't know how bad it will be in Q2 or when it will get better. We'll find out.

The article about the GPU direction at Intel is somewhat interesting but no one seemed to notice that Intel's presentation compares GPU with VLIW. And, VLIW is of course, Itanium. So, this does not appear to be a presentation about desktop or consumer graphics. Perhaps this was intended to suggest a different direction in HPC.

Scientia from AMDZone said...

I might mention that people might judge the popularity of one of my articles by the number of comments. This article has fewer comments. However, looking at the traffic logs I can see that there has been a flood of viewers from Intel. So, apparently, this has been interesting to the people there.

Unknown said...

Scientias - I am not at liberty to post price lists but my range is correct for mainstream chipsets (Viiv, Live, etc) and are not low end product price points. I am sure you have some contacts within the industry where you could try to confirm my 'estimates'. As such, you need to seriously revise your profitability expectations from chipsets.

Scientia from AMDZone said...

Okay, let me try it this way. The current highest priced AMD chipset motherboard on NewEgg runs $160. Assuming a wholesale price of $80 doesn't leave room for $15 profit from the chipset, $8 would be closer. However, Intel's current highest priced motherboard runs $250. Assuming $125 wholesale this would allow for $15 profit for the chipset.

So, I would say you are correct if AMD's motherboard prices stay where they are now.

LG said...

Scientia.

Yeah, those were sort of my thoughts on that article as well. Another question I have concerns the alleged buy up of R6xx by Dell. If that were true, would AMD be allowed to publish numbers for an unreleased product? Is it possible that the sales are there, but revenue can't be reported until an official release?

Scientia from AMDZone said...

R600 chipset sales would be Q2 so this wouldn't help the Q1 numbers. AMD only has R500 chipset sales for Q1 and obviously R500 isn't quite as competitive.

Polonium210 said...

"One item would be that AMD produced a 45nm SRAM test wafer 3 months after Intel."
This is a common misconception-AMD produced a test SRAM wafer in January 2006.

Polonium210 said...

Scientia:"I think you need to take a much harder look at reality."
Whoever needs a reality check, it is not me. Check slide13 of Daryl Ostrander's presentation in June 2006, which is titled "Great Progress on 45nm!" and which states:"Working SRAMs Achieved in January 2006"

All of the interviews and reports you link to were conducted in April 2006, so when AMD confirmed that they had produced test SRAMs, the reporter assumed it was in April when in fact AMD was simply confirming the January test.

I am surprised that you, of all people, have fallen into the trap of believing what is written on the sites you quoted without checking. It also goes to prove that no matter how often an error is reported on the Net, and no matter how many links someone may be able to produce to the said errror, it proves nothing except that people don't check the facts before writing.

Most of these sites simply feed off each other and so any FUD quickly spreads.

Scientia from AMDZone said...

No, I misread your original post as January 2007 which seemed ridiculous.

January 2006 seems reasonable and as you say it was in the slide. So, apparently AMD must be having a bit more trouble with 45nm than Intel.

Lately, I've even seen Intel enthusisasts suggest that AMD wouldn't have 45nm until 2009 or that Nehalem would be out before AMD's 45nm.

Polonium210 said...

Scientia:"So, apparently AMD must be having a bit more trouble with 45nm than Intel."

IIRC Intel too produced its first test SRAM in January 2006. This being the case, how could AMD's 45nm process be in trouble?

Scientia from AMDZone said...

Because it looks like AMD will produce 45nm chips about six months after Intel.

My guess would be that AMD was delayed because of the need to get FAB 36 up to speed. In other words, if FAB 30 had been able to use 65nm wafers I would bet that AMD would be up with Intel now.

Polonium210 said...

Scientia:"Because it looks like AMD will produce 45nm chips about six months after Intel."
Fair point, but cost may have more to do with it than technical factors.

Why have you deleted your post with all the links. It was an honest mistake on your part and you only damage your reputation by deleting it.

Scientia from AMDZone said...

polonium

No, it wouldn't be a matter of cost. AMD was producing chips with FAB 36 earlier but these were 90nm chips. According to AMD they had to extrapolate from IBM's 300mm FAB and FAB 30's 200mm data to get the initial settings for FAB 36. As far as I can tell, the equipment was capable of 65nm but it took a bit of time to get things set properly.

The links only supported the April announcement of the test by AMD. My original statement is in your quote so there was no reason to leave a long post which has already been corrected.