Monday, July 24, 2006

AMD+ATI analysis: mobile is the key

My previous analysis on July 21, 2005 on AMD+ATI merger had a lot of unknown parameters, now the details of the merger are out, let's take another look. None of the analysis will change the projection that Intel will be slaughtered, but we will have more accurate insight on the near term impact.

I. Financial impact on AMD

AMD is going to pay $4.2 billion cash and issue 57 million shares of AMD stock. AMD will take a loan of $2.5 billion. For AMD shareholders, this is a great deal. With about 12% dilution, AMD now also becomes the world's largest GPU maker. This is a major boost to AMD's share value. Since ATI will add more than 12% in profit, the EPS impact on AMD will be positive.

AMD's cash position will be tough. However, Hector Ruiz must know something we don't, which makes him extremely confident. The DELL-AMD contract must be huge. I won't count AMD's monetary recovery from its anti-trust lawsuit at this point, but that's at least a few billion dollars in the bag.

The deal won't close until end of 2006, or even 1Q07. So there won' t be an impact for the current quarter. We should expect AMD's revenue and profit to increase for 3Q06.

II. Impact on AMD vs. Intel death struggle

The deal is definitely a major win for AMD. We should expect some major innovations come out from the AMD+ATI merger. This is not just about putting CPU and GPU on the same die, but fully integrating them in computing. Now AMD can provide full set of solutions in the PC: CPU+chipset+graphics at the highest end of the spectrum for commercial client and mobile market. Again, Intel is left to eat AMD's dust.

AMD(ATI)'s mobile chipsets will be a major boost to the Bulldozer platform due next year. The timing is perfect. Bulldozer will frag Merom like there is no tommorrow. INQ reported those Merom chipsets consume 13 watts. I expect AMD to FAB its mobile ATI chipsets on the same lower power 65nm SOI process in FAB36. Since mobile is the fastest growing segment of the PC market, Bulldozer+AMD Chipset+AMD graphics will be a major major frag.

One thing Intel will also hate is, AMD Crossfire Chipset+GPUs will milk the Conroe market for profit.

ATI had about $100 million revenue from Intel chipset business, that's a small piece to be thrown away. The revenue made from expanding the market (especially mobile) will be billions.


III. Impact on Nvidia

AMD will not ruin Nvidia's business. Nvdia owns 90% of the AMD64 chipset business with its proven Nforce technology. But this business is mainly in desktop and server, as AMD's mobile market share is very small, and ATI already has a big footprint in mobile AMD chipset. Therefore, expect Nvidia to continue to dominate the AMD desktop chipset business.

However, AMD+ATI merger does put Nvidia in a very tough position. Nvdia has nowhere to go, but to fully commit to AMD64.

(1) Nvidia can't get Intel's highend desktop/mobile chipset business. Intel has 100,000 mouths to feed and a large number of those depend on Intel's own chipset business.

(2) Intel won't and can't acquire Nvidia. Intel already owns 35% of the graphics market, adding Nvidia will cause a lot of overlap and will not pass HSR Act review.

(3) In the discrete graphics market, Nvidia's competitiveness will be dependent on performance, which is unrelated to this merger. On PCI-E, both Nvdia and ATI are on equal footing on both AMD and Intel platforms. On Torrenza, I expect AMD to license Torrenza to both Nvidia and ATI, even if there wasn't an AMD+ATI merger. If ATI produces some major innovations on Torrenza, Nvidia will be impacted--but that's not a result of the merger.

(4) Therefore, Nvidia's only oppurtunity lies with an expanding AMD64 market.

IV. Impact on the PC industry

Now everyone can depend on AMD to provide the full set of solutions to break the Intel monopoly. Expect AMD to ODM motherboards from Taiwanese and the circle is complete.
AMD will make CPU+hestsink/fan+chipset+graphics+motherboard. The only things you need to make a PC are memory, hard drive, case and power supply.

The HP+Compaq merger saved HP from certain death, HP and Compaq was a weak+weak merger.

The AMD+ATI merger is a strong+strong merger. #1 CPU technology vendor merges with #1 GPU vendor, very little overlap. It's a killer enterprise.

74 Comments:

Anonymous Anonymous said...

I think your timelines for integration are a little off but its exciting times for sure.

As they stated that integrated products won't be out until 2008, Conroe will be on its way out with Nehelm(or whatever annoying name they are giving it). And while I do agree Conroe inventory will be much higher than any new cpu arch, it may not end up being the Nehelm killer you may expect. We shall see on that point.

Torrenza support, or even 4x4 if they open that up to 1CPU/1GPU would be interesting as well.

Do you think they will introduce the ability to utilize GDDR4 in a CPU with this move?

10:58 AM, July 24, 2006  
Anonymous Anonymous said...

Correct me if I'm wrong, but I thought Intel was the biggest GPU maker not ATI or AMD/ATI. I know this is probably because of IGPs, but it still counts.

11:32 AM, July 24, 2006  
Anonymous Anonymous said...

ATI won't likely be fabing on AMD for a while. GPUs need to be custom designed for Fabs which is why ATI needed to convert the RV515 (TSMC) to RV516 for UMC for example. This won't likely happen for R600 since it's unnecessary and contracts with TSMC and UMC are probably already done. We may see ATI fabbing on AMD for the R700 generation though, but that'll be 2008. ATI will probably be using 65nm (Fab 38?) instead of 45nm (Fab 36) since the newer fab would probably be too busy with CPUs. Just like at Intel how chipsets are 1 generation behind CPUs.

11:35 AM, July 24, 2006  
Blogger Bruno Dieter Chan said...

'Correct me if I'm wrong, but I thought Intel was the biggest GPU maker not ATI or AMD/ATI. I know this is probably because of IGPs, but it still counts.'

It feels that way but consider this; Microsoft IE must then be the most installed browser for the PC. That is by default it comes with Windows. Knowning Intel they may have sweeten the deal for most mobo makers to include their IGP with their chipsets just to capture market shares. After all, they'll those companies that if the customer doesn't want IG they can disable it in bios and slap on a GC. There is nothing to lose.

11:42 AM, July 24, 2006  
Anonymous enumae said...

What happens if AMD's future products are not strong enough to beat Intel?

If they are not strong enough will they BK? Big loans, lose market share, pricewar these are all things that can hurt AMD.

You failed to point out that the 35% share of the graphics market that Intel holds is only the desktop market. You failed to point out that they hold between 60 - 70% in the mobile market.

What kind of loss are they going to take to have someone else produce their chipsets, GPU and MB, I am not saying design, but someone has to build them and they don't have the FABS.

To the final point... I am guessing we will never see you withdrawing the statement about Intel being BK in 5 quarters, from what you have said today it would seem Intel will be around for quite some time.

11:45 AM, July 24, 2006  
Anonymous Wirmish said...

"ATI will probably be using 65nm (Fab 38?) instead of 45nm (Fab 36) since the newer fab would probably be too busy with CPUs."

Chartered (65nm in mid 2007) can produce chipset and GPU for AMD...

"GPUs need to be custom designed for Fabs which is why ATI needed to convert the RV515 (TSMC) to RV516 for UMC for example."

Chartered have APM so they are compatible with all AMD designs...


I want a Torrenza GPU so I can upgrade it like I upgrade my CPU.

12:00 PM, July 24, 2006  
Anonymous bigbadwolf said...

i agree with shari one on this one now

Hector ruiz will bury intel once and for all.

Intel+ Monopolistgic pig

DIE a painful death

same to you intel fannies

12:45 PM, July 24, 2006  
Blogger Pop Catalin Sever said...

AMD is on the biggest gamble of all time. This certainly looks like verry strong future planing from AMD, but if Intel bites of his leg and atacks AMD with full firepower things could certaingly turn for the worse.

I think, althoug this long fought war, all depends on the short term high availability of Conroe and Memron with wich Intel can severily damade AMD.

For AMD's sake let's hope there are at least 6 months before Intel can flood the market with Conroes and Memrons.

12:49 PM, July 24, 2006  
Anonymous Graham said...

I agree with the Inquirer posting on this topic. This was something AMD HAD to do just to stay in the game just like the lawsuit is something they had to do because they can't really compete. They also needed to borrow $2.5 Billion (with a capital B) to get the job done. Paying interest on that is going to be a big drag on any future earnings. Add in the cost of building additional fabs and I would expect an AMD bankruptcy in not too long. Also, the deal won't realize anything useful until 2008 which is a long time in this industry. I would not expect NVidia to cozy up to AMD since AMD is now their main rival. (those ATI, discreet graphics cards aren't going to disappear overnight thus AMD is now NVidias chief competitor). I would expect NVidia to cozy up a little with Intel. Intel has to be laughing at this desperation move by AMD. Here is the Inquirer article:

http://www.theinquirer.net/default.aspx?article=33219

12:50 PM, July 24, 2006  
Blogger Ednella said...

Hey! I've been checking out your blog. My brother is a computer nut. We have AMD chips in all our computers.

Please take a look at my blog, Sound Out. I am interested in telling people about the one true way to heaven, and I'd love to answer any questions you might have about the Bible, God, heaven, whatever. If I can't answer it, maybe I can recommend someone who can.

~Nella

12:54 PM, July 24, 2006  
Anonymous Anonymous said...

Hi Sharikou,

I was first negative on this merger. I have seen a lot of mergers fail, but the more I hear about this one, the more I see it may have a lot of potential. After reading Rahul's take and the INQ, it seems there is more to meet the eye to this deal.

This merger appears to be a long term investment based on the premise that AMD will leverage a lot of technology from ATI. Personally, I don't like mergers, because most of them are more of trying to buy revenue then to actually grow business(buyouts for IP and nitches are good in my opinion). It seems that AMD is acquiring ATI for their talent and IP. Ideally, it would have been nice if AMD could have acquired a smaller company for such purposes, but in reality there are only so many players in the Graphics market.

Acquiring ATI from a purely financial standpoint is not bad. It is risky that AMD is taking in more debt, but ATI is financially very healthy. ATI has virutally no debt and they make a profit. Albeit, their profit is not as good as Nvidia's, but in a strategic, acquisition point of view, You don't want to necessarily acquire the best company. Plus, as bold and dynamic as Nvidia's CEO is, I doubt that such a merger would be possible(I think Rahul stated this). AMD/ATI may meld better as companies where as Nvidia is a better company partner then merger partner.


I think that the past 5 years has shown that AMD's leadership knows what it is doing. It is a risk, but if you read the INQ's take on it, it maybe a risk they just have to take. Plus, I think that AMD wouldn't have taken this if they didn't think they could make the money to do it...cough...cough...Dell...

12:54 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

it seems there is more to meet the eye to this deal.

1) AMD is definitely going to create a CPU with GPU on die with very good performance. The high clockspeed alone is going to have a boost. Imagine a notebook with such a CPU, the power requirements will be reduced substabtially.

2) AMD is going to integrate CPU and GPU functionality. The CPU is trying to add more and more SSE stuff, but the GPU has far more power than SSE.

3) Commercial clients can now depend on AMD to provide one stop and stable platform solutiom. Major win for AMD's commercial client business.

4) For shareholders, the dilution was only 12%. The added revenue is 40.

1:31 PM, July 24, 2006  
Blogger SmartM0F0 said...

So... does this mean we going to be seeing an "AMD" graphics chip running alongside an "Intel" computer...? lol

From going around various websites/blogs today, I feel a sense of excitement and enthusiasm as to where the tech industry as a whole is going... I think this acquisition will truly re-invigorate the industry, and invite all players to join in on the open platforma and AMD ecosystem!

The next 2-3 years will be VERY interesting and exciting to see.

1:47 PM, July 24, 2006  
Anonymous Anonymous said...

Sharikou, you just continue to prove that you are ignorant when it comes to anything related to processors.

There is a reason why GPU frequencies aren't anywhere near as high as CPU speeds, and it is a deliberate decision. There is a sweet spot between frequency and parallelism in the processor.

It is much likeK8 versus Netburst, except on a much larger scale because of the nature of GPUs.

1:54 PM, July 24, 2006  
Anonymous Anonymous said...

Doctor Doctor..

We can spectulate and guess as to the wisdom of this marriage.

Butt...

1) Please point me to a case where you have seen two larger companies merge successfuly where the sum was greater then the parts.

2) Also tell me where you every saw two companies merge when they weren't in trouble..

Never for both.. that is history. I find it funny Hector thinks he is so smart to avoid the pitfalls of all those past so smart CEOs'

That says it all.

The Doctor

1:58 PM, July 24, 2006  
Anonymous Anonymous said...

Tell me one thing: how is AMD going to compete with Nehalem? It will have all the goodness of Conroe micro-architecture, and it will have hyper-transport-like interconnect--most likely an on-die memory controller. It will be on 45 nm, something AMD won't have until Nehalem becomes a commodity. Looking at how Intel is increasing the cache size, Nhalem will most likely have 8 MB of smart cache--the same type that gives AMD a run without a point-to-point interconnect.

If this merger slows down AMD's schedule re introducing 45 nm and expanding capacity (which it very well could considering the revised credit rating), that would be the final nail in their coffin.

2:04 PM, July 24, 2006  
Anonymous Anonymous said...

Grasp straws much?

"I agree with the Inquirer posting on this topic. This was something AMD HAD to do just to stay in the game just like the lawsuit is something they had to do because they can't really compete."

Haha okay, more like keeping ATI afloat sounds like to me. The AMD-ATI is a smart one in the long term for one very simple reason: OEM sales! (complete product solutions, GPU, CPU and chipset)

Intel's anti-trust business tactics are long known and it was far time AMD did something about it. Even abroad Intel is finally getting slapped around for it's long monoplistic actions.(KK Intel Japan raids, investigations in Germany to name a few) The effects are already well evident as this is why AMD now commands 22% market share and it's growing(albeit slowly) even in light of Core2's release. The massive amount of evidence in this suit against Intel is staggering, they will be coffing up several billion or more when the settlement is made in 2008 I suspect.

"They also needed to borrow $2.5 Billion (with a capital B) to get the job done. Paying interest on that is going to be a big drag on any future earnings. Add in the cost of building additional fabs and I would expect an AMD bankruptcy in not too long."

Well considering AMD just made a huge deal with Dell, I think that's a drop in the bucket when doing business with arguably the top OEM dealer. Aquiring ATI will allow AMD to have a complete solution of CPU, GPU and chipset from top to bottom to satisfy Dell.(much less other OEM's they do business with) Once additional FAB capacity is in place(FAB in NY, FAB 38) along with it's APM software, you are going to see AMD with 4 of world's most cutting edge FABs on the planet.(FABs 38, 36, one in NY and FAB 7) The costs these FABs will save and potential market share K8L/K10 and possibly the ATI R600/R700 series will earn by next two years with these FAB will quickly settle those debts due to total cost of manufacturing and I expect AMD to continue eroding Intel's marketshare at thier expense.

The bottom line is this though:

The server market is the most lucrative and most profitable. AMD will continue to command it at the most important point at 4P and up indefinately due to Intel's reliance of it's pathetically aging FSB techology. I won't even bother going into detail about Hypertransport 3.0, Torrenza co-processors and AMD's 45nm at that point,(even with Intel at 32nm)
it will be too much for Intel to throw anything at in the server market IMHO. Whether Intel wants to admit it or not, the server market is changing thanks to AMD. AMD is clearly showing the future with HT 3.0, Torrenza, Pacifica virtualization and if Intel can't or won't adapt them, it will become near or outright extinct in the server market.

"Also, the deal won't realize anything useful until 2008 which is a long time in this industry. I would not expect NVidia to cozy up to AMD since AMD is now their main rival. (those ATI, discreet graphics cards aren't going to disappear overnight thus AMD is now NVidias chief competitor). I would expect NVidia to cozy up a little with Intel."

Actually I think that conclusion is fulla crap as the Inquirer seems to think contrary as I do to your conclusion for several important and understanding reasons:

1. Intel is fixing to get into the high end GPU race. That means Nvidia = unimportant to Intel when they have thier own chipset business (and soon GPU business) to prop up.

2. AMD doesn't burn bridges with thier partners in the industry. There is still plenty of business for Nvidia to be had in AMD's segment of the market, with chipsets and especially high end GPU's.

"Intel has to be laughing at this desperation move by AMD. Here is the Inquirer article:"

Yep, I read the article, again your conclusions aren't even agreeing with what they came to a conclusion. Nice try though, feel free to play again Intel fanboy!

2:12 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

Also tell me where you every saw two companies merge when they weren't in trouble..

The immediate example you have is AMD and ATI. Both companies are doing great. AMD is killing Intel and ATI is the #1 GPU vendor.

A success story of large merger is HP and Compaq. Today, DELL is being slaughtered. DELL will follow Enron unless it act real quick. HPQ is thriving.

2:14 PM, July 24, 2006  
Blogger netrama said...

"I think, althoug this long fought war, all depends on the short term high availability of Conroe and Memron with wich Intel can severily damade AMD."

This view point expressed by some folks is totally wrong..if this were true, AMD should have had the whole CPU market when Athlon FX came out. At the most AMD might loose a little sales in the high end desktop/mobile before 4X4 and K8L comes out.
Some folks (especially those crappy Intel fanboy Analysts ) still view AMD, like it is the same AMD back in the 1990's under Jerry..they need to take a fresh look at AMD

On the merger - A combined solution for the mobile space is what is really needed. what the heck...I want all the devices on a single chip and it should save Power and run longer and be better and weigh less, instead of Intels multiple chips monopoly/crappy mobile solution(graphics etc).

I wouldnt be surprised if Intel announces a combo mobile chip tomorrow using their famous copy and paste solution (more like their dual and quad core solutions :)))

2:25 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

This view point expressed by some folks is totally wrong..if this were true, AMD should have had the whole CPU market when Athlon FX came out. At the most AMD might loose a little sales in the high end desktop/mobile before 4X4 and K8L comes out.

What matter is the main stream market: the $150 price range. Even though AMD was 60% faster than Intel for the past three years. Intel maintained its market share because at the $150 range, both were comparable. The Conroe is actually a good thing for AMD, as it acts as a marketing message to Intel customers that Pentium 4 is indeed junk. Such a message will help strengthen AMD's market position. Now people know AMD was telling the truth and Intel was lying about clockspeed/hyperthreading/power.

2:30 PM, July 24, 2006  
Anonymous Anonymous said...

"Tell me one thing: how is AMD going to compete with Nehalem? It will have all the goodness of Conroe micro-architecture, and it will have hyper-transport-like interconnect--most likely an on-die memory controller."

It will take Intel this long to finally reverse engineer AMD's technology, wow. Bout damn time I say. Still, K10 will blow Nehalem out of the water. How do I know this? Just like you know Nehalem will blow K10 out of the water.

In short: get a clue. Nobody knows how either will perform till they get here.


"It will be on 45 nm, something AMD won't have until Nehalem becomes a commodity."

AMD doesn't need to be a same process as Intel does to remain competitive, AMD at 130nm while Intel was at 90nm with Precott has proved this. A shrink in process smaller than your competitor does not secure a performance edge.

"Looking at how Intel is increasing the cache size, Nhalem will most likely have 8 MB of smart cache--"

As far as cache is concerned, it is nothing more than a kludge IMHO for it's pathetic aging FSB.(the only thing that keeps Conroe from choking on it's own memory bandwidth, then there's the issue of cache thrashing) The more cache you cram on a die, the higher chance of lower yields. This is why AMD sold Athlons with only 512k L2 for much cheaper instead of all of them being 1Meg L2 like the higher end parts like the FX's.

"the same type that gives AMD a run without a point-to-point interconnect."

IF CSI ever shows up that is, it's already been pushed back to 2009 last I read. It still has nothing over what HT 3.0 will have though by that time or especially HT 4.0 I wager.

"If this merger slows down AMD's schedule re introducing 45 nm and expanding capacity (which it very well could considering the revised credit rating), that would be the final nail in their coffin."

Gloom and doom, gloom and doom. Please. The truth is always somewhere in the middle I find. As much I enjoy the rhetoric as I read from Sharikou about Intel's demise, it's a bit overly exaggerated, much like your "expert" analysis here. As I said, AMD's 65nm will surpass Intel's own 65nm and will compete well with Intel's own 45 intitially as well I suspect if history proves itself again.(AMD 130nm vs Intel 90nm)

2:30 PM, July 24, 2006  
Blogger symbiansn said...

"There is a reason why GPU frequencies aren't anywhere near as high as CPU speeds, and it is a deliberate decision. There is a sweet spot between frequency and parallelism in the processor."

You are the ignorant.

Both Intel and AMD are researching new technologies where they could make distinct parts of the cpu run at different clockspeeds and different voltages one to another.

2:49 PM, July 24, 2006  
Anonymous Anonymous said...

"A success story of large merger is HP and Compaq. Today, DELL is being slaughtered. DELL will follow Enron unless it act real quick. HPQ is thriving. "

Talk about getting the rope and putting the noose around your own neck, tying it around a tree and jumping... LOL

This is laughable. The CEO that engineered this was fired! The idea was horribled concieved and the execution was disaster.

YOu had to bring in Hurd to fix the bitches screwup. If you look at how and why HP is humming now it has NOTHING to do with the COMPAQ assets. If anythign it continues to cause brand confusion with the multiple lines.

The Doctor

3:41 PM, July 24, 2006  
Anonymous Anonymous said...

STILL NO ATHLON X2 3800 EE SFF CPU AT ANY SITE DEALERS? ANYONE CARE TO POST A SITE THAT HAS IT? IF YOU CAN, THANKS ALOT.

HINGSUN

3:47 PM, July 24, 2006  
Anonymous Anonymous said...

sharikou,

You should start using the analogy "clean kill" now, along with "fragged" - remember Herb's statement about the original hammer?

"We are right now focused on the server space…and Hammer absolutely gives us a 'clean kill'," Herb also said. "It will be the best server processor."

3:49 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

If you look at how and why HP is humming now it has NOTHING to do with the COMPAQ assets.

As I
wrote here last year
, The HP+Compaq saved HP from certain death. With the exception of printers, most of HPQ's enterprise business came from Compaq. Without Compaq, HP would be selling a bunch of printers and PA-RISC running HP-UX.

4:11 PM, July 24, 2006  
Anonymous Anonymous said...

"As I wrote here last year, The HP+Compaq saved HP from certain death. "

Why are you referencing your own article as proof??

4:36 PM, July 24, 2006  
Anonymous Graham said...

Some anonymous coward wrote: There is still plenty of business for Nvidia to be had in AMD's segment of the market, with chipsets and especially high end GPU's

Hmmm, what does NVidia make? Chipsets and high-end GPUs. So I guess AMD is just going to roll-over and let NVidia have a share of their market just out of good old altruism? I don't think so. Your conclusions are about as thinly grounded as Dr. Sharifraud's....

5:01 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

Why are you referencing your own article as proof??

It just shows that I have far more foresight than most analysts. I can see through things. Most dudes can only be led by the media like a bunch of sheeps...

5:04 PM, July 24, 2006  
Anonymous Anonymous said...

I’ve bought and sold a few companies myself and the financing is the funniest part, because the debt is always serviced by the increase revenue of the purchased company. In other words it’s free or money you wouldn’t have had. The hard part is always the cash out lay, which AMD has and can get more of in the next six months. The shares traded on the deal will be money AMD prints. It’s easy, done everyday.

Let’s have no more crying in the Intel fanboy peanut gallery, I feel your pain.

5:06 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

Hmmm, what does NVidia make? Chipsets and high-end GPUs. So I guess AMD is just going to roll-over and let NVidia have a share of their market just out of good old altruism?

As I have analyse, Nvidia's only future lies with an expanding AMD64 market. But that's a given, as we know Intel will BK in 5 to 7 quarters. By then, AMD will take 80% of the CPU market, and 60% of the GPU market. Nvidia will remain a viable solution.

5:07 PM, July 24, 2006  
Anonymous Anonymous said...

Not to change the subject, but I see a lot of replies in Tom's Hardware Guide (THG) forum defending Intel and suggesting that this move will not affect Intel and may actually hurt ATI and AMD.
Almost 99 % of posters there have negative attitude toward AMD. Do you think THG is a paid Intel internet site? and could it be that THG have created an Intel PR front that encourages Intel fanboysm?
Their "game is over for AMD" sure sounds like an Intel slogan.

5:26 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

Almost 99 % of posters there have negative attitude toward AMD. Do you think THG is a paid Intel internet site?

It doesn't matter whether a few Intel fanbois say--they don't have the $$ to buy a good CPU any way. AMD+ATI merger is about breaking into the commercial client and mobile space.

5:31 PM, July 24, 2006  
Anonymous enumae said...

"Intel will BK in 5 to 7 quarters. By then, AMD will take 80% of the CPU market, and 60% of the GPU market."

You never stop do you?

Even if what you said was true AMD can not supply 80% of the CPU market, they wont even be able to do it when the New York fab opens, which is well over 7 quarters away.

5:32 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

Even if what you said was true AMD can not supply 80% of the CPU market, they wont even be able to do it when the New York fab opens

No. AMD will be able to suppy 80% of the market with three 300mm FABS cranking at 65nm: FAB36, FAB38, Chartered FAB7. Total 50,000wspm.

5:34 PM, July 24, 2006  
Anonymous Anonymous said...

"It just shows that I have far more foresight than most analysts. I can see through things."

Is that what you got fired for? Why is Warren Buffett second richest guys on this planet, not you Dr. Sharikou? lol...

5:34 PM, July 24, 2006  
Blogger Bruno Dieter Chan said...

Has anyone notice this but when AMD buy patents or technology they are within more or less a year of completing their new technology? That the big buys they make is the final piece to their puzzles?

Take for instance IMB's SOI process but now they are more ambitious so they need both teams of engineers (gpu and motherboard design) for intergration of ATI gpu into Torrenza platform and added manufacturing abilities and know-how to produce this platform.

Otherwise they will have the same issues they had when they just partnered up with IBM on finally being able get SOI to work.

5:46 PM, July 24, 2006  
Anonymous Anonymous said...

Friday June 09, 2006 :

http://www.hardocp.com/news.html?news=MTk0MzIsLCxobmV3cywsLDE=

5:46 PM, July 24, 2006  
Anonymous Anonymous said...

"I expect AMD to FAB its mobile ATI chipsets on the same lower power 65nm SOI process in FAB36."

When? Isn't this fab needed to generate the 50% market share you claim by the end of this year or was it next year?

Amd is going to use it's capacity to make low end/low maargin mobile chipsets over high margin CPU's? Wouldn't it be better to continue to outsource the chipsets to UMC/TSMC/etc...

5:57 PM, July 24, 2006  
Anonymous Anonymous said...

"AMD doesn't need to be a same process as Intel does to remain competitive, AMD at 130nm while Intel was at 90nm with Precott has proved this."

There was a wide disparity in microarchitecture - as Intel clearly has a more competitive with the Core miocroarchitecture now I don't think the example above is still valid.

"The more cache you cram on a die, the higher chance of lower yields. This is why AMD sold Athlons with only 512k L2 for much cheaper instead of all of them being 1Meg L2 like the higher end parts like the FX's."

AMD reduced cache size to increase capacity (more die/wafer due to smaller dies size). As AMD has "perfect yield" (reference: Sharikou) due to APM, yield is not an issue.

6:10 PM, July 24, 2006  
Anonymous Anonymous said...

"It just shows that I have far more foresight than most analysts. I can see through things. Most dudes can only be led by the media like a bunch of sheeps..."

You always post your so called "analysis" after it has been discussed elsewhere and your track record is horrible. You always post to a site that will have a few words that back up your theories. I seriously doubt you own any of the equipment you discuss which explains why you are always linking to some specific chart that agrees with your less than accurate statements. You also censor any statements on the website that would counter your views in a factual manner.

8:46 PM, July 24, 2006  
Anonymous Anonymous said...

As I wrote here last year...


And in two years we will see Sharkiou PhD write a case study for the Harvard Business Review of the bankruptcy of INTEL... LOL...

Its funny to watch a lunatic believe all that is imagined in his mind become reality.

Sharkiou tell me what do you do in the real world to make a living besides getting handouts from AMD and the AMD fanbase. Any credible accomplishments on the resume that you care to justify your qualifications to make forecasts with data of credible and logical bases.


The Doctor

9:24 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

You always post your so called "analysis" after it has been discussed elsewhere and your track record is horrible.

so far, my track record is almost flawless. If you dude find something wrong, point it out.

9:28 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

And in two years we will see Sharkiou PhD write a case study for the Harvard Business Review of the bankruptcy of INTEL... LOL...

I accurately projected the following
1) Intel's 1Q06 revenue would fall 15-20% from 4Q05
2)Intel's 2Q06 would fall more
3) DELL would have to go AMD before the end of 2Q06
4) Intel would have to establish uniform pricing
5) Intel would start mass layoffs

I expect the following to come true
0) DELL to go AMD around Conroe launch time

1) Intel will start posting losses from 3Q06 and continue for 7 quarters, until it files for chapter 11 BK

9:37 PM, July 24, 2006  
Anonymous Anonymous said...

Pretty amusing stories written here by Sharikou. It does seems that AMD is sweating it out here and doing everything it can to survive the onslaught from INTEL. AMD ATI merger is only going to worsened it's condition. AMD would be debt ridden.

10:11 PM, July 24, 2006  
Anonymous Edward said...

"so far, my track record is almost flawless. If you dude find something wrong, point it out."

Actually... I recall right before AMD's Q1 press conference you expected in one of your comments a huge surge in AMD's revenue and profit which didn't come true. Some other guy said in a following comment that you are too optimistic, and it turns out he was more correct.

It's some time in Fabruary, but I'm too lazy to find out the post, though.

10:17 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

Actually... I recall right before AMD's Q1 press conference you expected in one of your comments a huge surge in AMD's revenue and profit which didn't come true,

Yes. It was about the only mistake I made. I incorrectly estimated 20% ASP growth and 20% unit growth for AMD's 1Q06.

10:29 PM, July 24, 2006  
Anonymous Anonymous said...

"Has anyone notice this but when AMD buy patents or technology they are within more or less a year of completing their new technology?...Take for instance IMB's SOI process..."

This would be a good example/analysis save for the fact that AMD was shipping SOI product on 130nm BEFORE they entered into the deal with IBM on 90nm and on. So in this case they bought technology AFTER thery were alreay in the market. Only you neglect the fact that the IBM purchase was for node scaling not a pure SOI play, so it was a good investment as AMD could not do the scaling on their own.

10:36 PM, July 24, 2006  
Anonymous Jeach! said...

"1) Please point me to a case where you have seen two larger companies merge successfuly where the sum was greater then the parts."

a) Definatly Compaq/HP
b) Five top Canadian banks (RBC, BMO, etc buying U.S. banks.
c) Canadian Manulife buying U.S. John Hancock
d) GE buying over 50 companies per year
etc, etc, etc...

Carly Fiorina was the architect of the HP merger but the shareholders got her fired before the flowers blossomed leaving the current idiot to take all the success. Just like Otellini is currently taking the heat for Intel.

"2) Also tell me where you every saw two companies merge when they weren't in trouble.."

a) Most financial institutions
b) Most insurance institutions
c) Most large Pharma/biotech co.
d) Gold/steel/coal/oil/aluminum giants merging

"I find it funny Hector thinks he is so smart to avoid the pitfalls of all those past so smart CEOs'"

Thats because you don't know what Hector knows. The DELL deal must be a huge long term contract. Heck, I wouldn't be surprised if DELL is not behind this deal in some way. Watch and see if it won't be them that buys the stock AMD will be selling.

10:44 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

Carly Fiorina was the architect of the HP merger but the shareholders got her fired before the flowers blossomed leaving the current idiot to take all the success. Just like Otellini is currently taking the heat for Intel.

It's a real pain to listen to Mark Hurd talking. Paul O is taking all the blames for Craig B.

10:48 PM, July 24, 2006  
Anonymous Anonymous said...

"It just shows that I have far more foresight than most analysts. I can see through things... ..."Yes. It was about the only mistake I made."

I put this in another blog but perhaps I will refresh your memory:

1) "My whipser number for Intel's 1Q06 revenue is $7.67 to $8.13."

That's from your blog " 1Q06 notebook sale drops by 35-40%". Actual Q1'06 - $8.9.

2) "The result will be continuing operating losses for Intel in the next 7 quarters."

From blog: "Intel may bankrupt in seven quarters". You must have meant after Q2? (Blog was before Q2 earnings and in this particular one there was no mention of impairment on goodwill)

3) "My advice to Ben Lynch: dump INTC at a loss and buy AMD, you can still triple your money."

This was written 4/7/2006, Blog: "Deutsche Bank AG to lose $1 billion on INTC". Had he followed your advice his investment would now be down ~40%, Intel is not a whole lot better at down 5% over that same time period, but down a net of -35% is a bit different than +300%.

4) "I now project 2Q06 GAAP loss for INTEL due to charges associated with impairment on goodwill. I project operating losses for Intel's 3Q06, 4Q06, 1Q07 and beyond."

From Blog: "Intel to slash Pentium D prices by another 60%"

Now you will go on to make excuses for why these didn't come true, but whatver the reason, the simple fact is: none of these predictions were correct.

10:51 PM, July 24, 2006  
Blogger Sharikou, Ph. D said...

1) "My whipser number for Intel's 1Q06 revenue is $7.67 to $8.13."


Intel stuffed the channel by quite a few hundred millions. I posted stories on that. Some companies were stuffed to death.


2) "The result will be continuing operating losses for Intel in the next 7 quarters."

wait and see.



3) "My advice to Ben Lynch: dump INTC at a loss and buy AMD, you can still triple your money."

wait and see.


4) "I now project 2Q06 GAAP loss for INTEL due to charges associated with impairment on goodwill. I project operating losses for Intel's 3Q06, 4Q06, 1Q07 and beyond."


Well, Andy Bryant decided not to book the impairment on goodwill in 2Q06 -- I estimate at least $1.6 billion impairment. I never said Intel will suffer operating losses in 2Q06.

10:57 PM, July 24, 2006  
Blogger Bruno Dieter Chan said...

This would be a good example/analysis save for the fact that 'AMD was shipping SOI product on 130nm BEFORE they entered into the deal with IBM on 90nm and on. So in this case they bought technology AFTER thery were alreay in the market. Only you neglect the fact that the IBM purchase was for node scaling not a pure SOI play, so it was a good investment as AMD could not do the scaling on their own.'

That cause they didn't perfect the process enough that they had to obtain additional expertise (IBM). In this case with ATI I believe AMD decided they will not wait and come to that stage but get implimented fast and right on the first go.

12:10 AM, July 25, 2006  
Anonymous Anonymous said...

"There was a wide disparity in microarchitecture - as Intel clearly has a more competitive with the Core miocroarchitecture now I don't think the example above is still valid."

This just further proves my point when it comes to comparing CPU manufacturing processes to the poster I originally responded to, thanks. ;^) As I said, a manufacturing process is no guarantee of a platform's overall performance, the design of the architecture has more to play into that. (see AMD64 vs. P4)

"AMD reduced cache size to increase capacity (more die/wafer due to smaller dies size). As AMD has "perfect yield" (reference: Sharikou) due to APM, yield is not an issue."

Yes this is true pertaining to more dies per wafer but what I said was also accurate, at least in particular to Intel's case. Intel doesn't have APM to rely on to keep thier FABs yields high as AMD can though. The more cache they cram into the Core2 the lower thier intial yields will be, this will improve over time I imagine, but still it's going to limit thier yields vs. what AMD is currently hitting with their mere 2-3 FABs.(near perfect) This is why Intel needs so many FABs while AMD is still getting by with only a few and still capable of getting near Intel's output.

7:22 AM, July 25, 2006  
Anonymous Anonymous said...

"There was a wide disparity in microarchitecture - as Intel clearly has a more competitive with the Core miocroarchitecture now I don't think the example above is still valid."

This just further proves my point when it comes to comparing CPU manufacturing processes to the poster I originally responded to, thanks. ;^) As I said, a manufacturing process is no guarantee of a platform's overall performance, the design of the architecture has more to play into that. (see AMD64 vs. P4)

"AMD reduced cache size to increase capacity (more die/wafer due to smaller dies size). As AMD has "perfect yield" (reference: Sharikou) due to APM, yield is not an issue."

Yes this is true pertaining to more dies per wafer but what I said was also accurate, at least in particular to Intel's case. Intel doesn't have APM to rely on to keep thier FABs yields high as AMD can though. The more cache they cram into the Core2 the lower thier intial yields will be, this will improve over time I imagine, but still it's going to limit thier yields vs. what AMD is currently hitting with their mere 2-3 FABs.(near perfect) This is why Intel needs so many FABs while AMD is still getting by with only a few and still capable of getting near Intel's output.

7:24 AM, July 25, 2006  
Anonymous Anonymous said...

"Sharikou, Ph. D said...
... Without Compaq, HP would be selling a bunch of printers and PA-RISC running HP-UX."

4:11 PM, July 24, 2006

1stly, I thought that the 'bunch of printers' that HP is selling is in fact its most profitable division. In fact, HP uses it to finance its losses in other divisons, PC divison is one I think...

2ndly, mergers are rarely as smooth as people hoped. The 1980s had seen lots of corporate buyouts all for GOOD reasons (synergy and other crap I am sure... but few actually survived. The mergers were later termed 'di-WORSE-ification'. Hence, it is too early to say that any AMD/ATI merger would turn out good. You are indeed too optimistic and we can only HOPE for the best at the most.

3rdly, stop predicting intel's death, it wouldnt just die. With 80% of the laptop market is good enough reason it wouldnt just die out already and AMD mobile chips are lagging behind even Core1 Yonah. The server/desktop market will probably be a DUOpoly provided that AMD doesnt screw up the restructing that follows every merger (Otherwise, its a monpoly Intel). Time will also be needed to implement integrated graphics onto processors and ATI doesnt actually have that good a track record as Nvidia on making AMD mobos. Hence, Intel wouldnt just die, no fanboi-ism involved, just being realistic.

9:33 AM, July 25, 2006  
Anonymous Edward said...

Many people are falling to a serious logic pitfall here. They say AMD+ATI will lose 80% of graphics/chipset market buy not optimizing for Intel FSB. They also say this will be very bad to nVidia because it will have hard time competing with AMD+ATI. But if the first is true, nVidia should easily grab a big chunk of that 80%, and the second must be false.

In fact, it is obvious from this interview that AMD+ATI do not intend to have adverse effect on current ATI products. If anything 'bad' to ATI's Intel support happens, it must come from the Intel side. But if so, then nVidia is really the beneficiary, because now it is the only 3rd party on both Intel & AMD platforms.

So in the short term, this merger has little negative effect on nVidia or AMD+ATI. In the long term, nVidia will be hard to survive anyway, not more due to AMD+ATI than Intel's own high-end graphics/chipset development. Unless, of course, nVidia keeps innovate and beat AMD/Intel on graphics & chipset performance.

12:29 PM, July 25, 2006  
Anonymous Gandalf said...

So in the short term, this merger has little negative effect on nVidia or AMD+ATI. In the long term, nVidia will be hard to survive anyway, not more due to AMD+ATI than Intel's own high-end graphics/chipset development. Unless, of course, nVidia keeps innovate and beat AMD/Intel on graphics & chipset performance.

intel is developing high end graphics? please point me to the link, this is news for me. up to now they just produce entry level graphics...

1:47 PM, July 25, 2006  
Anonymous Edward said...

Also look at what Henry Richard in the DigiTimes interview:

Q: So can we continue to anticipate AMD based motherboards that carry Nvidia graphics and Nvidia core logic?

A: I surely hope so, absolutely. I come as you know from a background at IBM, and one of the principles I learned there, and I think this is very important, is that if internal divisions are to remain competitive, you should not shield them from external competition. ... I know for a fact that artificially shielding them from external competition would probably be bad business practice.

From time to time ... it may happen that Nvidia has the best chipset. And if that’s the case, there’s absolutely no reason why they shouldn’t reap the benefit of their innovation. ... Also, unlike the competition, I don’t foresee putting in place artificial marketing barriers to force our customers to make certain choices that are not necessarily good for their business.


I guess many Intel fans, or people thinking from Intel's perspective, are too narrow-minded to grasp this before. Look at Intel's response to this merger; look how it pulled ATI's license. We are looking at two very different companies, AMD and Intel. Even if AMD+ATI substitutes Intel to become the dominant force in the computing industry, AMD with its mindset will not behave as badly as Intel.

2:36 PM, July 25, 2006  
Anonymous Edward said...

"intel is developing high end graphics? please point me to the link, this is news for me. up to now they just produce entry level graphics..."

Is Intel produces only entry-level in the future, there is really nothing nVidia should fear about AMD+ATI. It will be the ONLY player for the 80% Intel graphics market.

Unless you assume Intel will happily allow AMD+ATI graphic cards to work well on Core 2 Duo systems?

Either way, AMD+ATI and nVidia win. Intel here is forced to change it monopolistic way of doing business. I'd say it's a great move for AMD & ATi.

5:25 PM, July 25, 2006  
Anonymous Anonymous said...

"Look at Intel's response to this merger; look how it pulled ATI's license."

So Intel should continue to license it's technology to what is now a direct competitor? Perhaps they should donate a couple of fabs too, so things will be "fair" in terms of manufacturing capacity?

5:30 PM, July 25, 2006  
Anonymous Anonymous said...

"This is why Intel needs so many FABs while AMD is still getting by with only a few and still capable of getting near Intel's output."

Intel is supplying ~80% of the world CPU's; AMD ~20% (this for the math challenged is a ~4:1 ratio).

Now keeping in mind that Intel produces chipsets, wireless products, embedded products, flash - only ~4 fabs are producing CPU's vs AMD's ~1.5.

Ratio seems about right given the ratio of market share segment...if yields were lower as you suggest Intel would need >5 fabs producing CPU's only. So much for your yield argument...

5:35 PM, July 25, 2006  
Anonymous Edward said...

"So Intel should continue to license it's technology to what is now a direct competitor?"

I don't understand what makes that not possible? If Intel had an open approach toward platform architecutre like AMD, it can certainly license its "technology" to its competitors.

And what "technology" are we talking about here? Hardly any IP or real implementation, but just the bus protocol! People like you have the narrow mindset of Intel where monopoly is gold and competition is bad. This is exactly how Intel ruined the PC industry over the past few years (except kept being pushed forward by AMD).

11:41 PM, July 25, 2006  
Anonymous Anonymous said...

"DELL will follow Enron unless it act real quick."

Sharikou, I think you need to do a bit of research into the Enron scandal as you give the impression of being an idiot when you say things like that.

12:48 AM, July 26, 2006  
Anonymous Anonymous said...

"Intel is supplying ~80% of the world CPU's; AMD ~20% (this for the math challenged is a ~4:1 ratio).

Now keeping in mind that Intel produces chipsets, wireless products, embedded products, flash - only ~4 fabs are producing CPU's vs AMD's ~1.5."

I know that Intel has 4 65nm FABs right now, that's why they NEED 4 OF THEM to get near AMD's 2 FABs that they have pumping 65nm.(FAB 36, FAB 7) In short, Intel's FABs are low yielding right now due to massive use of on-die cache! Yields will eventually improve, but at the startup, they suck compared to what AMD is capable of with APM at it's startup.(near perfect) There is a reason why Intel said Core 2 will only be at 30% at year's end, because they are still pumping P4's off those lines!

"Ratio seems about right given the ratio of market share segment...if yields were lower as you suggest Intel would need >5 fabs producing CPU's only. So much for your yield argument..."

Yea, so much for your paper tiger there genius, care to try again? As Sharikou already has projected, AMD needs only 2 FABs at max output with high yields(this does not include FAB 7) to supply the majority of the market. It takes less FABs that you think to supply the world's market demand when those FABs are very efficent and Intel's are not a good yardstick by any means.

9:43 AM, July 26, 2006  
Anonymous Anonymous said...

"I know that Intel has 4 65nm FABs right now, that's why they NEED 4 OF THEM to get near AMD's 2 FABs that they have pumping 65nm."

You're kidding right? Intel is supplying 3-4X more of the world's supply of chips than AMD, so at similar output/fab they would need 3-4X the # of fabs. (This of course is not true due to the fact that one of AMD's fab is 200mm and 90nm and their other fab is still mostly 90nm right now, but let's put that aside for the time being...)

If Intel's yields were lower they would need even more fabs to continue to produce 3-4X the # of chips that AMD supplies. How can you not see this basic capacity analysis?

"As Sharikou already has projected, AMD needs only 2 FABs at max output with high yields(this does not include FAB 7) to supply the majority of the market."

If this were true than AMD would eithre have 50% of the market right now (which when last I checked they don't) or their inventory would be building substantially (which it is not).

If AMD's is at ~20% market share and their 2 fabs can produce the majorirty of the market (>50%)why would they go to Chartered for additional capacity? (and earn lower gross margin?). If this masterful capacity assessment by Sharikou were true, AMD would have very large inventory levels right now which according to their latest financial reports, they don't or they would own 50% of the market. Where is this extra 30% capacity going if they are not selling them or buildng inventory?

Not all of Intel's 65nm fabs are operating at high volumes right now, at least one of them (D1d in Oregon) also supports the 45nm development line. You need to look at overall wafer starts per week, (or per month) and die sizes to do the yield comparison you are (crudely) trying to do.

1:25 PM, July 26, 2006  
Anonymous Edward said...

"If AMD's is at ~20% market share and their 2 fabs can produce the majorirty of the market (>50%)why would they go to Chartered for additional capacity? (and earn lower gross margin?)'

A few reasons:

1) AMD's marketshare is increasing. It used to be 20% desktop, 0% server, and 5% notebook; it is more than 20% desktop, 25% server, and 15% notebook now. AMD's overall processor market share, in terms of silicon area, probably doubled during the last year.

2) AMD is lowering price for Conroe's release. That means a percentage increase of previously high-end (and thus larger) products. In other words, AMD wants to sell more high-end products at about the same price; Intel can only sell the same craps at lower price.

3) AMD is making 200mm to 300mm and 90nm to 65nm transitions on its fabs now. This might impact its production volume.

4) AMD might be preparing for 4x4 or K8L later this year. We all know that up till end of Q2 every single AMD chip was sold like a hot cake. Once it introduces quad-core, however, it won't be able to produce as many Opteron DC and Athlon64 X2. It might be stockpiling now for later.

Conclusion: AMD does have a tight capacity. However, once fab 36 and 38 are turned to 300mm/65nm, which produces 100% more than 200mm/90nm, plus Chartered, it is not impossible for AMD to supply more than 50% of world demand.

That, my friend, is before the end of 2008.

5:19 PM, July 26, 2006  
Anonymous Anonymous said...

Edward - are you just making this up as you go along...you are so far off on all of your points.

"1) AMD's marketshare is increasing. It used to be 20% desktop, 0% server, and 5% notebook; it is more than 20% desktop, 25% server, and 15% notebook now. AMD's overall processor market share, in terms of silicon area, probably doubled during the last year."

Uh, wrong... Market share segment reports all CPU sales latst I checked not "silicon area" (whatever that means)... so double market share is a bad analysis. Are all 3 of those markets equivalent in size? I think there was a 1% MSS increase for AMD in Q1? If AMD is now about 22-23% that would mean they were ~11% last year, which they weren't if you look at the MSS reports.

"2) AMD is lowering price for Conroe's release. That means a percentage increase of previously high-end (and thus larger) products. In other words, AMD wants to sell more high-end products at about the same price"

If yields are NEAR PERFECT as Sharikou contiunes to claim (with no actual data to support it), only the FX62 part is bigger die size due to the larger cache. All other X2's I believe are the same size so actual capacity will not change if the mix between X2 3800's and X2 5000's changes (higher speed doesn't necessarily mean bigger die!). Of course yields are not perfect...and these parts are speed binned which all IC manufacturers do, but Sharikou doesn't believe this.

"3) AMD is making 200mm to 300mm and 90nm to 65nm transitions on its fabs now. This might impact its production volume."

The 200mm fab (F30) is a completely separate building from 300mm fab (F36,) how again is startup of a 300mm fab impacting produuction in a completely separate building right now?

AMD is adding 65nm capacity to the existing 90nm startup capacity they already have in F36. The conversion from 90nm to 65nm in that building starts next year. So that capacity, right now, also shouldn't be impacted. With all of the claims of APM and perefect yields there must be 0 impact converting to new technologies?

"It might be stockpiling now for later."

This is actually the best one of your comments! This would show up as inventory on the financial reports (which it doesn't according to latest Q2 report). And do you really think if every single chip is "being sold like a hotcake" they would be stockpiling them? Instead of selling them for more revenue where they could potentially reduce the 2.5Bil loan for the ATI purchase?

"AMD might be preparing for 4x4 or K8L later this year."

4x4 is same (or similar?) to Althon and is completely independent of 200mm, 300mm, 90nm or 65nm technologies.

K8L will be using 65nm technology which also presumably will be used for Athlon/Opteron. It is only a litho mask change to switch from one product to another in the fab. You also might want to check AMD's own statements - K8L is only going to be demo'd at end of 2006; launch is "mid-2006" according to AMD. But you may be right they are probably just keeping a bunch of tools/fab capacity idle "preparing for K8l" rather than producing more revenue producing chips today!


"However, once fab 36 and 38 are turned to 300mm/65nm, which produces 100% more than 200mm/90nm, plus Chartered, it is not impossible for AMD to supply more than 50% of world demand.

That, my friend, is before the end of 2008."

...and so how will this lead to 50% market share the great Sharikou prior to the end of 2007? On a side note I do think your 100% estimate is about right (Sharikou, I think, claims either 200 or 400% but ignores larger die size as higher mix goes to dual/quad core and addition of L3 cache to K8l)

8:32 PM, July 26, 2006  
Anonymous Anonymous said...

Edward - my apologies. I thought you were the idiot who said that AMD (TODAY) can produce 50% of the markets chips with F30 and F36. I realize now that you were just answering the question of why they would use Chartered.

(But reason #2 and #4 on dies size and stockpiling is still wrong for reasons given in last post)

8:39 PM, July 26, 2006  
Anonymous Anonymous said...

Sharikou, you need to stop stating the obvious saying that intel is dead, we know!

11:16 PM, July 26, 2006  
Anonymous Anonymous said...

I've been reading the site for a week or two now, and it really gives me good laughter time and again.

The fun starts with the referred blogs on the left side, where we have authorities such as Fomitchev, who proved his wisdom when writing about Vista and GPU usage a while back. Needless to say, that gave good laughter too and I wasn't the only one.

Then we go on into Pervasive 64-Bit computing. Based on this blog, I have no frigging idea what that means. I guess it should be something on-the-edge or fancy. Then the author speak about 50$ CPUs and X2s. Yes there are some more articles, but seems the whole question is about consumer base. Most of the comments deal with game performance. Which happens in 32-bits.

Then we have the bashing that is pulled straight from the authors ass, with a reality distortion field even greater than that of Steve Jobs.

Still we have almost a religious Ph. D. who thinks a little too much of himself and seems to be on a crusade. There are no academic merits of any sorts in his arguments and the references are to his own writings, writings that don't support his arguments or that are only rumours, second hand information or things that might not even exist. Oh, and let's not forget speculations about specs, even after we've seen many times how those are proven wrong by practice. Live and learn, neh?

And finally we have the clowns in the audience that actually believe this guy. If somebody tries to make some sense out of this crap, he is automatically an Intel fanboy. Whoa, again a nice proofe of academic skills.

Sharikou, I'd like to know where I could buy titles such as Ph. D. like yours. I know I've two to come still, but I could cash out a few more in, say, East-Asian languages or such.

But keep on the good work, I love the laughter you give me. Only concern is that somebody might take this for real. Then again, smoking kills too.

12:45 PM, July 27, 2006  
Anonymous Edward said...

"Uh, wrong... Market share segment reports all CPU sales latst I checked not "silicon area" (whatever that means)... so double market share is a bad analysis."

First, you can just treat silicon area as roughly the product of number of chips and ASP. When we talk about fab capacity, we cannot ignore the size of each chip (which in 1st order approximates the price).

Second, AMD's sales grow almost 60% year-on-year for both Q1 and Q2. Given the price war in Q2, AMD's marketshare growth in terms of size*unit could be as high as 1.8x. 2x is an exaggeration, I admit, but it's a lot more closer than your MSS estimation (which suggest, what, 1.1x increase?)

"... only the FX62 part is bigger die size due to the larger cache. All other X2's I believe are the same size so actual capacity will not change if the mix between X2 3800's and X2 5000's changes (higher speed doesn't necessarily mean bigger die!)."

Let me be patient to tell you why your analysis on X2 is plain wrong.

First, the majority of AMD processors sold, if you didn't know, is Sempron. We observe a much diverse mix there, and the trend is toward the high-end architecture, with 64-bit, IMC, SSE, Cool&Quiet, etc.

Second, even Athlon64 out-sell X2 at this point, but over the past year X2's has been increasing.

Third, no to mention that Opterons generally are larger than Athlon64, and Turion64 X2, which just recently debut, takes twice as much area as Turions.

Still think that size isn't a factor here?

"The 200mm fab (F30) is a completely separate building from 300mm fab (F36,) how again is startup of a 300mm fab impacting produuction in a completely separate building right now?"

Read this news article, first sentense of the second paragraph, and think again.

"AMD is adding 65nm capacity to the existing 90nm startup capacity they already have in F36. The conversion from 90nm to 65nm in that building starts next year. So that capacity, right now, also shouldn't be impacted."

They are ramping 65nm right now. Go to AMD's Q2'06 CC transcript and search for "ramp 65-nanometer." Read before you make more erroneous claims.

"With all of the claims of APM and perefect yields there must be 0 impact converting to new technologies?"

APM drastically reduces impact of transitions. AMD never said it's zero. Even 25% impact is very low for the industry. Intel's copy-exact, OTOH, probably requires a few days if not weeks of down time on the fab.

"This is actually the best one of your comments! This would show up as inventory on the financial reports (which it doesn't according to latest Q2 report). And do you really think if every single chip is "being sold like a hotcake" they would be stockpiling them?"

Again, you have no clue, and I pity you, seriously. Go again to AMD's Q2'06 CC transcript and search for "expect inventory" and read the Q&A, then search for "inventory up" and read the Q&A, before you make more of your amateur claims.

"K8L will be using 65nm technology which also presumably will be used for Athlon/Opteron. It is only a litho mask change to switch from one product to another in the fab."

I start to feel wasting time to explain to someone like you who has no idea how the fabs work in practice. You probably just read it off a textbook and assume you knew. Wrong again! To prepare for K8L is to prepare for its production, which means a flow of wafers has to be run to fine-tune the many parameters on the tools, for 65nm, for K8L. This will inevitably impact the current production.

You were just spreading false claims without proof, much like what you critisized Sharikou for. Please read the references I showed you and tell me which I said was wrong? Tell me that you're not ashamed of yourself?

4:25 PM, July 27, 2006  
Anonymous Anonymous said...

To the anonymous long post who laughs a lot.

Jealousy rings loud in every line of your post. You put down everyone and everything about the site. I assume you must know it all and haven’t learned anything after reading the site for two weeks as you claim. You must be a world class tech wizard. You for got to leave us some of your wisdom. You neglected to leave facts. You made no helpful contribution to any of your degrading claims. Where is this other wonderful site that impresses you so much? Inquiring minds want to know.

Your post turned out to be a LOSE – LOSE. You lost because the site wasted your time for two weeks with out a gain and I lost because I took the time to read your post.

What we have here is a failure to communicate.

5:35 PM, July 27, 2006  
Anonymous Anonymous said...

"Intel's copy-exact, OTOH, probably requires a few days if not weeks of down time on the fab."

Can you back this up with a link, because "probably" is not going to support your claims?

12:19 PM, July 28, 2006  

Post a Comment

Links to this post:

Create a Link

<< Home