Tuesday, January 30, 2007

Q4 06 Intel and AMD: Who Won?

Sometimes it is easy to pick a clear winner and sometimes it is more difficult. The fourth quarter of 2006 for Intel and AMD is hard to call because of conflicting numbers for revenue share and volume share.

I am always sceptical of changes in share in a rising or falling market. For example, if AMD gains in a falling market, the numbers can be inflated and make AMD appear to be doing better than it really is. Likewise Intel can gain temporarily in a rising market (if AMD hits its capacity ceiling) and appear to be doing better than it really is. I figured I would have to wait until Q1 07 to see if Intel held onto its revenue share. However, now I'm faced with conflicting numbers so it looks like the actual scenario is different.

In order for Intel to have a temporary gain in Q4 06 they would have had to gain in both volume and revenue share, but this didn't happen. Intel gained revenue share but lost volume share. So, it does not appear that Intel had a temporary gain. We have to assume that the revenue share increase was genuine and not temporary. However, AMD gained overall in volume share. Since we know that Intel was not at a capacity ceiling this gain too must be genuine. This is confusing. However, it gets even more confusing because Intel and AMD appear to have swapped strategies. In the past, Intel has had a priority strategy of taking volume share, usually by lowering the price of its chips. In contrast, AMD has targeted the server market and has taken server share for the past three years. Yet, in Q4 06 AMD cut its prices (reducing its margins) and gained volume share in both desktop and mobile. Meanwhile, Intel gained back a small amount of share in servers. Clearly, AMD's actions reduced its margins while Intel's gains were most likely due to the improvements with Woodcrest and Tulsa over older Xeons.

Since both gained there is no obvious winner in Q4. So, we have to consider where AMD and Intel can move from here. In other words, if one company gained a temporary advantage that would be lost in Q1 07 while the other had a gain that could be built on in the following quarter, the company with the temporary gain would be the loser. Basically, half a step forward followed by another half step is better than one step forward followed by one step back. So, let's look at where each company is now after Q4 06 and where they can go in Q1 and Q2 07.

AMD's profits were only about 1/3rd of what they were in Q3 06 so the obvious need is either to raise prices or reduce costs. Raising the prices after lowering them would be out of the question so the only option would be reducing costs. FAB 36 is still ramping and as it ramps a greater volume of the production is on 300mm which does reduce cost. Likewise FAB 36 is converting to 65nm and this too reduces cost. So, by continuing with the same strategy AMD could expect to have better profits in Q1 07 although they may not quite match the Q3 06 numbers. Presumably, AMD could introduce Barcelona in Q2 with a higher price tag and with continually dropping costs on FAB 36 this should make the Q2 07 numbers better than Q1 07. I can't see any problem with this strategy so it looks like AMD could hold onto its current volume.

Intel lost volume share in desktop and mobile. However, we want to re-examine the issue of capacity. Although Intel was not limited overall, it was limited in terms of Conroe and Merom. Some could speculate that Intel gained where the market segment was covered by C2D (servers) but lost where capacity was constrained (desktop and mobile). This could then lead someone to assume that Intel would simply take back mobile and desktop share as production capacity of Merom and Conroe increased. I suppose this is possible but I don't think it is likely. For example, Intel's server gain was only 1.4%. If there were a strong preference for C2D then it seems that this gain should have been larger. I think a more likely scenario is that desktop is a function of performance and price, mobile is battery life and price, and servers are more complex with performance, power consumption, density, price, and support all being factors. Essentially, at the bottom range of the desktop segment, price would be dominant with performance secondary, but performance would become more important as we go up until it becomes the dominant factor at the high range with price secondary. Likewise, with mobile, cost is important at the bottom with battery life dominating at the top. With servers price and support probably dominate at the bottom with power consumption and density becoming stronger factors at the top. Performance also becomes a stronger factor as average workload increases. For example, a 10% increase in performance matters much more for a server that spends most of its time at 70% capacity than one that only gets above 20% a few times a day.

It is possible that AMD only gained temporarily due to capacity contraints of C2D for desktop and mobile. If this is the case, then Intel wouldn't have to do anything since their C2D capacity is steadily increasing. If the market prefers C2D to K8, Intel would simply gain the lost desktop and mobile share back as it is able to supply more of Conroe and Merom. So, if this is the case then Intel's strategy is solid and it will gain back its lost share. However, it is more likely that price is the biggest factor on the desktop so Intel would have to lower prices in addition to increasing volume of Conroe. Likewise, it appears that AMD will released a lower power chipset for mobile and this will take away Intel's battery life advantage. This would mean that Intel would have to change strategy to gain back desktop or mobile share. And, Intel's gains in server share could be in jeopardy once Barcelona is released in Q2 07. Basically, AMD will release a new architecture in 2007 just as Intel did in 2006. However, in spite of its new architecture Intel shows very little in terms of market gains. With increasing competition from AMD in 2007 both from a new architecture and new chipset offerings it really looks like Intel will have to be the one to change strategy. In this regard I would have to put AMD slightly ahead in Q4 06 but not by much. AMD will still have to deal with increasing volumes of C2D during the year so AMD is not guaranteed to be able to continue taking share. However, overall I think it is most likely that AMD will continue to nibble at Intel's share each quarter while Intel steadily grows to regain its former 2005 levels of revenue.

Overall, in 2006, there is no doubt which company is the winner. AMD added a second FAB and the ability to produce its own chipsets while gaining revenue and share while Intel lost both revenue and share in spite of releasing a new architecture. This puts AMD in a much stronger position in 2007 than it was in 2006 and it gains a new architecture on top of that. This makes for a very strong offering by AMD in 2007. However, if we take Intel's bag of tricks for 2007, turn it upside down, and shake it hard, not a lot falls out. Intel will increase the ratio of quad cores on its server chips and it will introduce a robust albeit large and expensive quad FSB chipset. It will also add a server chipset that uses DDR2 registered DIMMs instead of FBDIMM to reduce system cost. Apparently these are to try to avoid losing more server share in 2007. There really doesn't seem to be anything new for either desktop or mobile. AMD however will have new offerings for server, desktop, and mobile. For servers specifically, Barcelona should remove Intel's advantages with power consumption, SSE performance, and Integer performance. Barcelona should match Intel in terms of cache so this advantage (particularly in benchmarks) disappears as well. Intel has also enjoyed advantages in benchmarking because reviewers have been using operating systems that have poor NUMA support. This gives Intel an obvious advantage on multi-threaded code. However, having four cores on a single socket removes this advantage since a native AMD quad does not require NUMA. This should mean that Intel will lose almost all of its performance advantage when Kentsfield is benchmarked again versus a desktop version of Barcelona.

The only fair edge that Intel could gain would be by having a higher clock which is certainly possible. However, I'm guessing that Toms Hardware Guide will put its thumb on the scale several times to make sure that Intel has additional advantages. These cheats will probably include using the Intel Compiler instead of the PGI compiler. Whereas the Intel Compiler is only optimized for Intel processors and even turns off some hardware features for AMD processors, the PGI Compiler can generate code optimized for either Intel or AMD, or code that runs well on both. Since it supports both 32 and 64 bits, C++ and Fortran, Linux and Windows, and generates parallel code automatically while costing the same as the Intel Compiler, PGI will almost certainly become the standard compiler for software development. Using PGI instead of the Intel Compiler would go a long way to establishing credibility for review sites that claim to be impartial. I don't expect Toms Hardware Guide to do this and probably not Anandtech either. And, THG has other cheats that it can use. They will probably use high clocking memory with slow timings since this favors Intel. I've never yet seen THG bother to get lower clocked memory with faster timings that would work better with AMD even though it is usually readily available for the same price. They probably will avoid loading the extra cores since this allows one benchmark to use all of the cache. This situation would be very rare for real code but can give quite an artificial boost to benchmarks. They will probably also use their favorite cheat of overclocking the Intel chip while leaving the AMD chip stock. Overclocking belongs in a separate article much like using aftermarket parts to boost engine horsepower does not belong in a regular car review.

2007 shouldn't be bad for either company. I expect Intel to keep growing and gain the revenue that is lost from 2005. I expect AMD to keep slowly expanding and taking more share. I guess if both companies make more money then neither one really loses. The two companies' offerings should be very close in performance and features by Q3 or Q4 2007 unless Intel manages to maintain a substantial clock lead. I am really hoping for some genuine reviews because this should be the best horserace we've had since PIII versus K7. However, I'm not optimistic that either Toms Hardware Guide or Anandtech will see the handwriting on the wall by then. So, we may still have to rely on other sites to get any proper test results. Although I'm guessing that THG will still be chasing the Intel advertising dollars six months from now I'll mention again what proper testing would be: no overclocked chips, match the DIMM timings to the processor, do proper core loading for multi-core processors, publish the cpu activity graphs as Tech Report does, use the PGI compiler, and either compile each for its own best performance or compile for good performance for both. I'll admit that NUMA testing is more complicated because they would either have to publish both interleaved and non-interleaved scores to show whether the OS has good NUMA support or they would have to include some Linux comparison scores. Much like the Sea Biscuit story, this is a horserace that everyone has been waiting for. It's just a matter of finding a review site where Intel won't have a five length headstart.

69 comments:

Real said...

Overall, I think your analysis is rather foolish. It is fairly obvious what happened in Q4. Intel’s Core2 came in at the high end of the market that was formerly owned by AMD. Although AMD owns 25% of the unit share of the market, it had and outsized chunk of the server and performance DT markets. Intel came in took the sockets where performance mattered (these are VERY HIGH ASP sockets) and pushed AMD down the pricing stack. AMD faced 2 options:
1.) Simply give up on the sockets and push your chip from the $2K designs at HP, etc to the $1,500 price points and lose margin in the process
2.) Cut prices drastically and try to hold onto the $2K designs and lose margin in the process

The fact that Intel experienced and ASP increase means that AMD was pushed out of the top end designs and probably slashed prices to stay in other (servers). At the same time, Intel was able to throw the P4’s at the low end of the market where they competed with the low end and mainstream parts from AMD (the sub $50 CPU designs). Clearly AMD gained unit volume here, likely a direct result of the Dell win. However, AMD was cutting prices and losing margin at a much faster rate than they were picking up low end unit volume which carriers near $0 margin. The result was the disappearance of profitability of AMD’s chip business. Just image, AMD entered the quarter making $10M/week (Q3 NI/13 weeks) but then only average $2.5M/week for all of Q4. If you take that as a linear decline from Q3, you see that AMD existed Q4 losing $5M/week and that loss will continue into Q1 and likely worsen. Given Q1 should see an overall unit decrease from Q4 and C/W/M will make up a bigger % of the Intel shipments, I would expect a minimum loss for AMD of $5M*13weeks = -$65M. This will be in addition to the $125M loss on the IP writedowns and probably another $20M-$30M loss for ATI. Overall, AMD will lose about $210M in Q1.

Intel, on the other hand, increased ASP and gained unit share where it mattered – the high end and servers. Do you forget that AMD’s margins cratered in Q4 vs. Q3’06? Intel simply has to keep its foot on the accelerator and AMD will not be able to recover as K8L is too costly to produce and is too late to market. If Intel can maintain its mainstream performance advantage and continue to hold share or grow share in servers, AMD is completely done as CSI will alleviate the main drawback for Intel processors today. Intel has not entered a quarter looking better than they have entering Q1’07. It truly is the best of times for an Intel shareholder.

In conclusion, your premise is fundamentally flawed if you believe AMD is better positioned now than it was entering 2006. This is laughable as in 2006 AMD knew it has 9 months of clear sailing before Intel could launch a competitive product. Now, they are entering a quarter bungling a merger, hemorrhaging cash, and knowing that they will be at a huge performance disadvantage. Even if they manage somewhat more than a paper launch of the K8L in Q2’06, they will have missed the major design cycles and will be sitting on the sidelines watching Intel expand margins and grow unit share.

Scientia from AMDZone said...

real
Overall, I think your analysis is rather foolish. It is fairly obvious what happened in Q4. Intel’s Core2 came in at the high end of the market that was formerly owned by AMD.


That was my assumption, that Intel took share at the top where performance is more important than price. Since you agree with me does that make you foolish too?

However, AMD was cutting prices and losing margin at a much faster rate than they were picking up low end unit volume

Again, I think this is obvious. Less than 1/3rd the profit from Q3 in spite of higher revenue. You are agree with something foolish again, hmmm.

and that loss will continue into Q1 and likely worsen.

I'm not sure that is true; AMD has released a new chipset. I guess we'll see.

AMD will lose about $210M in Q1.

I'm not expecting this. I'm expecting about break even.

Intel simply has to keep its foot on the accelerator and AMD will not be able to recover as K8L is too costly to produce and is too late to market.

Too expensive? I guess we'll see.

If Intel can maintain its mainstream performance advantage and continue to hold share or grow share in servers,

Well, to do that it would have to release additional C2D's at both higher and lower clocks. I guess we'll see.

AMD is completely done as CSI will alleviate the main drawback for Intel processors today.

Wow, AMD is completely done when CSI is released 2 years from now? I think AMD will have a few things of its own by then like DC 2.0.

Intel has not entered a quarter looking better than they have entering Q1’07.

Right, not since Q1 06. Of course, Intel looked better all through 2005 but let's not mention that.

In conclusion, your premise is fundamentally flawed if you believe AMD is better positioned now than it was entering 2006.

Yes, I guess my big flaw is thinking that a second FAB and the ability to produce its own chipsets will actually help AMD. Afterall, how could having a 300mm ramping FAB possibly compare to the advantage of having a single 200mm FAB that is running at capacity? What was I thinking? And, I don't know how I could have mistakenly assumed that having a proper low power mobile chipset could help. AMD will have one soon but didn't have one in 2006. How could I have thought that would help?

This is laughable as in 2006 AMD knew it has 9 months of clear sailing before Intel could launch a competitive product.

9 months of clear sailing. I should have ignored the C2D previews, the launch before 9 months and the price drops.

Now, they are entering a quarter bungling a merger,

By doing everything according to plan. Wow, just think how bad it would have been if they had not followed the plan.

hemorrhaging cash,

From a one time merger charge.

and knowing that they will be at a huge performance disadvantage.

At the top end of the desktop only and then only until K8L is released on the desktop.

Even if they manage somewhat more than a paper launch of the K8L in Q2’06,

No, I'm sure it will be worse than that. I'm sure Intel will release 45nm with the new socket and CSI in Q2 while AMD probably won't have Barcelona out until 2008. Surely, that is how it will be.

they will have missed the major design cycles

Because there is a deadline on design cycles after Q1. Hmmm, didn't C2D launch later than that?

and will be sitting on the sidelines watching Intel expand margins and grow unit share.

Yes, I imagine AMD will be bankrupt by the end of 2007 as Intel takes back not only the share it lost but additional share.

Scientia from AMDZone said...

real

It's hard to take your post seriously. You've overlooked a lot. For example, K8L is not too expensive; it is the same proportionate cost for 65nm as X2 was on 90nm. You've also overlooked that AMD is growing faster in the commercial segment than the consumer segment. Intel's high end performance is not much of a factor there but AMD's ability to offer its own chipset is. You seemed to have overlooked that. You overlook that Intel will be facing competition in integrated graphics where it has never faced competition before and where it has no advantage. Intel's in-house costs are a bit lower but this is offset by having to make a northbridge as well.

Also, I have no idea where you got your "too late to market" idea. C2D came out 3 years after K8, if three years isn't too late then how could 1 year be?

Your accelerator analogy is flawed because AMD is also ramping production. Unless you ascribe to the argument that AMD only gained because C2D was at a capacity ceiling then you would have a tough time explaining why Intel didn't take more share in Q4.

Finally, it is still being suggested that Intel is straining its relationship with partners. This too could be a factor.

Greg said...

I'd like to point out that Intel is delaying the release of 45nm to 2008 (I'm pretty sure everyone was saying q3 2007) and that the current assumption is it will release with CSI. However, there will be no new socket, and they are holding off on immersion lithography till 32nm.

If AMD can keep or exceed parity, in terms of performance as a company, with Intel then it will be, as real called it "smooth sailing" with high-k and immersion lithography in their 45nm process that will be released at around the same time as Intel's (instead of substantially later).

AMD also has a modular core design system that will debut in 08, making them capable of far greater flexibility and profitability (due to the ability to reach very high margin segments with unique core-type designs while optimizing power consumption for specific load types).

AMD has lost money before, and though it was of great consequence in terms of their companies stability, it was also due to the fact that AMD significantly fell behind in terms of performance for far more quarters than they'll be behind at the release of barcelona, and the fact that AMD had no stable source of buyers (such as Dell, HP, and Gateway) at the time. AMD was also losing volume share back then, so obviously this situation is far different than then.

While I think 2007 will be a fairly interesting year for processor sales and performance, 2008 will be the year when things get really crazy, and the market will be extremely interesting.

In other words "You ain't seen nothing yet!"

Erlindo said...

Scientia Wrote:
real

It's hard to take your post seriously.


It's becasue he is unreal. ;D

Scientia from AMDZone said...

greg

There will be a new socket; CSI doesn't help much without one. The new socket has been planned for a long time. It will arrive first with Itanium in late 2008 but not until 2009 for C2D. If Intel delivered CSI without a new socket the best it could do would be to take I/O traffic off the FSB.

Greg said...

I just wanted to give scientias point about CSI in late 09 with this article.

http://theinquirer.net/default.aspx?article=37373

While it is the inquirer, and thus not the most reputable resource for processor information, it's at least something, and nothing in it sounds fishy.

sharikouisallwaysright said...

Could someone please explain how the ASP of Intel has changed since the C2D launch last year?

Isnt it a selfconstucted ASP-rising that bases on the massive pricecuts b4 and therefore only PR-stuff?

Maybe i will become enlightended...

enumae said...

Nice article, but why ramble about THG?

Where to start?

I am really glad to have found your blog. Since finding your's and the Dr's I have learned alot, not because of what you say, but because I look up everything you say. I have been watching the sector very close now for maybe a year, and I have to say that since I started alot has happened.

There are a few things I believe you failed to mention, or take note of.

You failed to specify why AMD claimed market share after the release of Conroe and Merom, DELL, clear and simple.

Again congrats to AMD for gaining market share. I know you want AMD to keep growing, but they are about to hit a wall, and that wall is the side of FAB 36, or is it FAB 30/38...just kidding.

AMD can not afford to go after market share, they need revenue share, they need to expand capacity as fast as they can.

They will not get above 35% market share with two FAB's, you can say all you want about the next manufacturing node being comparable to the previous node, but that will end.

They will have about a three year wait from the day they decide to build a new FAB to the day they can ship processors from it, thats from looking at FAB 36. There has been no news about a new FAB as of late.

Intel knows this, and is why they are attacking AMD's margins and revenue, in an attempt to halt AMD's abillity grow. AMD's margins are going to drop in the first two quarters of 2007 due to the aquisition of ATI.

After that they will need to have a very good launch of Barcelona. When I say a good launch, more specifically they need volume, a very fast transition to 65nm, and with the projected performance they will be able to charge a premium, but with out volume they will not be able to capitalize on it.

----------------------------

So, I have to disagree with your statement about AMD being in a better position than Intel entering 2007, and as far as Q4 2006, I believe that Intel won that round.

Its all about money and what you can do with that money in the future. Intel taking money directly away from AMD's Servers is a clear example that Intel is indeed attacking AMD's wallet, and that it is working.

----------------------------

In your articles you show a somewhat balanced point of view. I do believe however that you are optomistic about AMD and a little pesamistic about Intel.

PS, I do not mean to beat a dead horse about AMD's capacity, but I do believe that is Intels plan of attack.

enumae said...

Scientia
Finally, it is still being suggested that Intel is straining its relationship with partners. This too could be a factor.

Could you link this?

Thanks

enumae said...

Greg
...with high-k and immersion lithography in their 45nm process that will be released at around the same time as Intel's (instead of substantially later).

I have a feeling that AMD's 45nm will be delayed about a quarter and even then it will most likely be very low volume, but we should have a good idea come Q3 this year.

How is AMD going to pay for 45nm, while ramping 65nm and 300mm, while Intel is attacking there Server market share/revenue?

Scientia from AMDZone said...

The new socket for Intel is socket B or LGA 1366. This socket is necessary for the IMC. But, like I said, I don't expect this until 2009.

Scientia from AMDZone said...

enumae
Nice article, but why ramble about THG?


Because good benches would be nice for for a change.

You failed to specify why AMD claimed market share after the release of Conroe and Merom, DELL, clear and simple.

You could be right but that still wouldn't change the fact that AMD and Intel have swapped strategies.

I know you want AMD to keep growing, but they are about to hit a wall, and that wall is the side of FAB 36, or is it FAB 30/38...just kidding.

I'm not sure what you mean. FAB 38 won't max out until end of 2009. That's nearly 3 years of growth.

AMD can not afford to go after market share, they need revenue share, they need to expand capacity as fast as they can.

Well, I think I've already covered this. AMD did go after volume and did lose margin and their profits will increase as they continue to reduce costs. And, obviously AMD is expanding as fast as it can.

They will not get above 35% market share with two FAB's,

Yes, that's about what I was thinking. I suppose it could theoretically hit 40% if Intel is using up some FAB capacity for chipsets but I'm only figuring about a 1/3rd right now.

you can say all you want about the next manufacturing node being comparable to the previous node, but that will end.

I don't know what you mean.

They will have about a three year wait from the day they decide to build a new FAB to the day they can ship processors from it, thats from looking at FAB 36. There has been no news about a new FAB as of late.

True, I'm thinking that to avoid a period of no growth AMD would need to get started sometime in 2007. However, AMD has been clever before; I'm not ruling out an internal expansion if necessary.

Intel knows this, and is why they are attacking AMD's margins and revenue, in an attempt to halt AMD's abillity grow.

It won't work; AMD has its avenues of attack as well.

AMD's margins are going to drop in the first two quarters of 2007 due to the aquisition of ATI.

Really? Why?

After that they will need to have a very good launch of Barcelona. When I say a good launch, more specifically they need volume,

Volume for Barcelona is Q3 07.

a very fast transition to 65nm,

by mid 2007.

and with the projected performance they will be able to charge a premium, but with out volume they will not be able to capitalize on it.

You need to re-examine history. AMD was able to capitalize on K8 in 2003 and K8L is better off than that in 2007.

So, I have to disagree with your statement about AMD being in a better position than Intel entering 2007,

Based on what?

and as far as Q4 2006, I believe that Intel won that round.

Strangely though I get the impression that if Intel had gained volume share while AMD gained revenue share you would say the exact same thing. Under what scenario would you have said that Intel did worse? BTW, if you'll recall, I said I would only put AMD slightly ahead.

Its all about money and what you can do with that money in the future. Intel taking money directly away from AMD's Servers is a clear example that Intel is indeed attacking AMD's wallet, and that it is working.

It did work in Q4 although it also cost Intel some volume; we'll have to see if it continues to work.

In your articles you show a somewhat balanced point of view. I do believe however that you are optomistic about AMD and a little pesamistic about Intel.

No. I expect Intel to grow all during 2007. I'm not sure what is pessimistic about that. And, AMD will launch a new architecture and new mobile chipset; that should be reason for some optimism.

+++++

Perhaps you should review a little history.

Intel launched the greatly improved Northwood core in Q1 2002 on 130nm. AMD moved to 130nm later and then followed with K8 in Q2 2003.

Intel launched C2D mid 2006 on 65nm. AMD moved to 65nm later and then will follow with K8L Q2 2007.

Do you see any similarities? Notice that the time from Northwood to K8 was larger than it will be from C2D to K8L.

Also, unlike 2003 when AMD only had a basic server chipset and had to wait and wait for desktop support AMD is now providing its own chipsets.

Scientia from AMDZone said...

enumae
have a feeling that AMD's 45nm will be delayed about a quarter and even then it will most likely be very low volume,


I expect 45nm to be delivered mid 2008. I suppose you could call it low volume until beginning of Q4 08.

but we should have a good idea come Q3 this year.

I assume you mean first silicon for a 45nm chip with DC 2.0.

How is AMD going to pay for 45nm, while ramping 65nm and 300mm,

Is this a real question? AMD will make more money as it ramps. Also, AMD will get some money by selling the 200mm tooling. If necessary I suppose AMD would find additional financing.

while Intel is attacking there Server market share/revenue?

For one more quarter. However, competition shifts in AMD's favor in Q2 and Q3.

Wise lnvestor said...

Brilliant post!

Scientia your CSI in 2009 is right on the money.

Keep it up!

Oh btw I suspected the server revenue drop fro AMD is largely due to customers holding on for K8L release. Sorry don't have proof for that, but it makes sense.

Roborat, Ph. D. said...

I have to agree with what REAL posted. You’re definitely too optimistic about AMD’s future. But don’t take my word for it. Just count many investors have upgraded AMD going into 2007. Q4’06 was a preview of what 2007 will be like. And according to AMD, Barcelona will only have an impact to its earnings in Q4’07. Your positive prediction is contrary to what the investors and AMD itself said.

In order for Intel to have a temporary gain in Q4 06 they would have had to gain in both volume and revenue share, but this didn't happen.
This statement is inaccurate. Intel did ship record volume parts in Q4’06. You’re forgetting that Intel ships to Apple which isn’t part of the Intel-AMD market share numbers.

Toms Hardware Guide will put its thumb on the scale several times…
THG’s benchmarks didn’t matter at all when AMD was in the lead years ago so I don’t know why you’d think it’s worth noting. The people who read these websites are smart enough to check the consistency of results from other review sites.

2007 shouldn't be bad for either company. I expect Intel to keep growing and gain the revenue that is lost from 2005… The two companies' offerings should be very close in performance and features by Q3 or Q4 2007 unless Intel manages to maintain a substantial clock lead.
You seem to conveniently skip the hellish 2 quarters that AMD needs to go through before Barcelona. I’d pay more attention to what REAL said about the rapid pace of AMDs ASP and revenue share decline. The worst scenario in this market share strategy is that at one point AMD would be shipping at its peak capacity and yet never make a profit. And guess what, this is where AMD was before 1995 - where they had more market share than today. And we all know what happened next after that – AMD couldn’t expand.

S said...

"It's just a matter of finding a review site where Intel won't have a five length headstart."

Your head thinks with a bias against Intel. And it shows in you r article.

S said...

I think both AMD & Intel are nearly equal in the architecture front as both seem to be well positioned to be trading the top spot between them for the next couple of years (Barcelona-Penryn-Fusion-CSI etc..)

It is really down to execution - how well they will run their business. Intel has long track record of doing well. It is also getting more focussed by getting rid of non-performing distractions. AMD's track record is debatable and so is ATI's. They need to get through the merger unscathed.

Does anyone know the details behind the $500 million merger related cha rge. Sounds very high to me, especially since ATI is a fabless company. Is it inventory write down ?

Real said...


Does anyone know the details behind the $500 million merger related cha rge.


The vast majority of it was for 'in process R&D' - this is required to be written off within 1Q of the acquisition under US GAAP. Basically, when AMD agreed to the purchase price, they needed to itemize how they were assigning their purchase price. They assigned >$400M to in process R&D (alternative would have been to assign it to goodwill). As a result, they were forced to write it off vs. carrying it at cost until it was impaired if they would have classified it as goodwill. I guess from AMD's perspective, they figured it would be better to take the hit in Q4 to obscure the drastic tanking in their processor core business rather than carrying it as goodwill and being forced to take a write down in Q3/Q4 of this year when they hope to be making a profit.

enumae said...

I have removed a previous comment because I felt that I had been rude, and quick to respond, to anyone who had read it, I appologize, and will respond in a more civil manner later today.

Scientia from AMDZone said...

roborat

Well, Motley Fool doesn't seem to agree with your investors:

AMD Best Tech Stock 2007

AMD is in the running for the best tech stock for 2007.

this scrappy underdog is better equipped to fight the good fight against chip giant Intel (Nasdaq: INTC) than most investors seem to realize.

Don't forget that AMD is used to making a living off slim margins, in the same class as Wal-Mart (NYSE: WMT) or Dell (Nasdaq: DELL). Its substantial investments in manufacturing infrastructure leads to Texas Instruments (NYSE: TXN)-sized depreciation and amortization -- and more than $1 billion in positive operation cash flow through the first three quarters of 2006. But management tends to spend all of that cash on expanding its business. As a result, AMD is keeping pace with Intel in processor performance, making for a give-and-take race between two worthy competitors.

in AMD's case, together with IBM (NYSE: IBM). There are worse partners than the perennial world leader in new patents.

Over the past three years, Intel has grown revenues at a plodding 5.5% annually. Texas Instruments did better with 13.2%, but AMD smokes them both with 17.1% annual sales growth.

Intel shareholders aren't likely to want the price war extended beyond 2007, which is management's current plan. When the pricing lid comes off later this year, AMD's earnings will take off again, as its highly efficient operations creates earnings leverage when revenues and gross margins expand. And that depressed valuation will follow suit.


++++++

Roborat, please talk about what I actually wrote instead making up arguments. I specifically said that Intel's revenues were geniune and not temporary.

THG’s benchmarks didn’t matter at all when AMD was in the lead years ago so I don’t know why you’d think it’s worth noting

Who are you talking to, robo? My opinion about the testing at THG and Anandtech is documented in my two articles: Toms Hardware Guide Sells It's Soul and Anandtech Melts down. I state that their attitude about AMD started in 2002 and continues to today. You are again making up arguments.

You seem to conveniently skip the hellish 2 quarters that AMD needs to go through before Barcelona.

No, but you do seem to have skipped my mention of history. BTW, AMD will be less affected in Q1 and Q2 in 2007 than Intel was in the same two quarters in 2006.

And guess what, this is where AMD was before 1995 - where they had more market share than today.

I have no idea what you are talking about. AMD did not have that much share before 1995. AMD's previous record volume share was in 2001 but the volume share today is higher than that. AMD has never in its entire history had more share than today.

Scientia from AMDZone said...

real
I guess from AMD's perspective, they figured it would be better to take the hit in Q4 to obscure the drastic tanking in their processor core business rather than carrying it as goodwill and being forced to take a write down in Q3/Q4 of this year when they hope to be making a profit.


The only one who posts here who has zero credibility is Pointer. He claims to have all kinds of industry inside technical information and yet he spends his time collecting AMD jokes on his blog instead of writing technical articles explaining why he prefers Intel. Zero credibility.

Real, your above statement is nonsense and I'm sure even you realize how biased it is. I said in my article that it is possible that Intel will take back share as C2D production capacity increases. I don't think this will happen but it is possible. History does not agree with your assessment.

Scientia from AMDZone said...

enumae
have removed a previous comment because I felt that I had been rude, and quick to respond, to anyone who had read it, I appologize, and will respond in a more civil manner later today.


You probably aren't aware of this but when I remove the deleted comments I can still read them. I don't want you to get the impression that I think that your opinion is not valued. My jab about giving Intel a win no matter what could also be considered rude and it wasn't intended to be. My apologies.

I tolerated Red for a long time but he is now gone and I have no intention of ever letting him back. He demonstrated that he has no respect for my opinion and no integrity of his own.

However, I also gave theKhalif a warning in the last thread when he started flaming the other posters. I guarantee this blog will not turn into ForumZ.

Roborat, instead of giving rhetoric about THG, how about addressing the points that I brought up? Do you disagree that THG typically puts OC'ed Intel chips up against stock AMD chips, do you disagree about their using slow timings, do you disagree about their not publishing cpu activity meters? What I'm saying is that I gave specific reasons why I don't like the testing at THG; it isn't just a blanket dismissal. You need to address it on that basis.

+++++

Enumae, Real, and Roborat. The three of you are not Pointer or Red. I do not want you to think that I want a blog that is only composed of AMD supporters or people who agree with me. On AMDZone I disagreed with Kaa all the time and yet I think he is one of the sharper contributers to the forum.

Read the Motely Fool piece and take into consideration the history I've already mentioned. Consider these points:

Recall that when Northwood was introduced in 2002 AMD got slammed. AMD lost about 25% of its share and didn't turn a profit from then until Q3 2003. Today, AMD is in much better financial shape than it was at the beginning of 2003 and is at a record volume share.

AMD only had a 200mm FAB while Intel was enjoying the cost benefits of 300mm. Today, FAB 36 is running at 50% capacity and growing.

AMD had a lot of trouble transitioning to 130nm SOI in 2003 and had the worst starting yields they have ever had. Today, 65nm is working smoothly.

AMD started with a server chip even though it had only a very tiny presence in servers from Athlon MP. Today, AMD's server presence is much greater and well established. Also, AMD had only a very tiny presence in supercomputers whereas today it is among the highest ranked.

AMD had no real competitor to Intel's Centrino platform. Today, Turion is a genuine competitor and will get a boost from a new chipset soon. You must be aware that the lack of a good mobile chipset has been Turion's problem from the start.

AMD is growing faster in the commercial segment than in the consumer segment. Being able to offer an in-house chipset as well as having solid reference systems should help AMD gain in this area in 2007. Intel has less advantage here since this segment is not driven primarily by top level performance.

So, in 2003, AMD had lost money for 4 quarters in a row, had lost some credibility because K8 was delayed, was scrambling to deal with Intel's unexpected doubling of the FSB speed (hence the hasty introduction of socket 939 later in the year), had no support for K8 beyond its one server oriented chipset, had no competitive mobile product except mobile Athlon in the value segment, had been completely shut out of Dell and Gateway because of their Intel only stance, and did not have a single tier one manufacturer supporting their servers.

Yet, AMD pulled things together in late 2003. What I keep trying to figure out is why you feel that AMD's situation today is worse than it was in 2003 and why you believe that AMD won't be able to do substantially what it did then.

If you are still pessimistic about AMD then I need to know why you discount the above points.

Greg said...

enumae:
Well, considering the fact that most of their research is done on 45nm and that they continue to research into it, I'd have to say that they basically already have. If they need to continue paying for it, it'll be through their R&D budget, which is already factored into their current financial analysis. As such, yours is a moot point.

Whether AMD is delayed a quarter with 45nm or not, this will still be a huge amount of ground for them to catch up to, in terms of achieving parity with Intel's stability.

I'd like to point out, to everyone who feels that going after volume at the expense of profitability is a bad idea, that if AMD has a much larger portion of the market and then decreases costs, the effect will be huge, compared to decreasing costs with a small portion of the market. If AMD decreased its portions of the market (which would be necessary to maintain profitability in this scenario) and increased ASP, there would be no gain at all in terms of profitability and there might even be a loss of profitability due to fab and ATI aquisition expenses (as both of these are absolutely essential to AMD's continued existence as a company). If they then increased volume, it would take longer, and the effect would be smaller, as it would immediately come at the cost of ASP's (you can't go after high volume sellers like Dell without losing ASP). So AMD's course is the only correct one, regardless of what Intel choses from here on out.

Real said...

Real, your above statement is nonsense and I'm sure even you realize how biased it is.

Nonsense? It is a pretty normal proceedure call "The Big Bath". For your education, here is a link:
http://financial-dictionary.thefreedictionary.com/Big+Bath

As to whether they expect to make a profit in Q3/Q4 next year, I would point you in the direct of Goldman Sachs, Lehman Bro's, etc which foreast losses for AMD in Q1 and Q2 with a return to very moderate profitability in Q3/Q4'07. If AMD follows this trend, it will have shown red for 3 straight quarter until Q3'07 where it would definitely NOT WANT to take the impairment charge.

Now, as to the original premise of my post about taking the bad news in Q4 so people would not be able to distinguish how bad the core business had become, the analysts had to work backwards to calculated a 'normalized' loss of $0.04/share, which was pretty much ignored as AMD management only wanted to talk about 'write-offs' and gaining unit share. In that sense, they were very successful which is exactly what the 'big bath' hopes to acheive. This is especially true when looking at severance and accuring expenses for future losses. This equated to about $80M of the charge on Sales and Marketing. Basically, what AMD did was recognize $80M in charges in Q4 which will not be seen until Q1 and Q2 - however, when they are realized in Q1 and Q2, they will not be seen on the income statement. Rather, the only impact will be on the balance sheet where the accural will reverse and cash paid will reduce the cash balance.

After reading through your blog entries, I can see that you are not very finance savvy so I hope that I did not use terms you did not understand. If I did, I would be happy to better define the terms for you so it can be a productive discussion.

abinstein said...

Hi Scientia, you have good points in your analysis. However, IMHO, I believe your article would be easier for people to read if they are more structured - like using some numbering or bullets. Thanks.

Real said...

Today, AMD is in much better financial shape than it was at the beginning of 2003 and is at a record volume share.

AMD has a negative cash/share of about -$4.50. AMD has more debt than it ever has had in its history. AMD will likely need a cash infusion of between $1.0 - $1.5B for the year due to negative cash flow from operations. While I agree that Motley Fool is 'cute', you really need to read the analyst reports from the more respected authorities. I would recommed the piece by Goldman Sach's about AMD's business model being flawed. In short, AMD is entering a period of sustained losses with a higher debt load (in absolute as well as in $'s/share) than it has ever had in the past. Expect an additional convertible debt offering in the Q3 timeframe which will further punish their stock price. As to unit share, believe what you want but I know the numbers provided by Mercury are not accurate.

Today, 65nm is working smoothly.

'Smoothly' in what sense? AMD management and fans have more or less admitted the 3Q slip in shipments on 65nm. Not only that, but their 65nm chips are at lower clockspeeds and underperform the 90nm versions in terms of IPC. AMD may acheive a clockspeed increase or lower power over time, but currently 65nm is still a mess. Further, I would take any promises of 'smoothness' from AMD management with a giant grain of salt.

Yet, AMD pulled things together in late 2003.

AMD pulled things together by leveraging the design knowledge of the Alpha folks while Intel was muddling around with the P4. AMD launched their best design to date against what was unquestionably Intel's worst (Itanium excluded). This time, K8 is way behind in performance and K8L will at most bring back parity until Intel releases Penryn or chooses to ramp clock speed. Lastly, in 2003, the industry came out of a growth slump and volumes surpassed Intel's production capacity - "a rising tide lifts all boats". In 2007, the exact opposite will happen with AMD and Intel producing at about 1.5x industry demand. Intel has a lower cost product, cash reserves, and the knowledge that failure is not an option now that it divested all other potential growth businesses. Call it whatever you want, but AMD's future has never looked more bleak than right now.

Erlindo said...

real wrote:
'Smoothly' in what sense? AMD management and fans have more or less admitted the 3Q slip in shipments on 65nm. Not only that, but their 65nm chips are at lower clockspeeds and underperform the 90nm versions in terms of IPC. AMD may acheive a clockspeed increase or lower power over time, but currently 65nm is still a mess. Further, I would take any promises of 'smoothness' from AMD management with a giant grain of salt.

AMD pulled things together by leveraging the design knowledge of the Alpha folks while Intel was muddling around with the P4. AMD launched their best design to date against what was unquestionably Intel's worst (Itanium excluded). This time, K8 is way behind in performance and K8L will at most bring back parity until Intel releases Penryn or chooses to ramp clock speed. Lastly, in 2003, the industry came out of a growth slump and volumes surpassed Intel's production capacity - "a rising tide lifts all boats". In 2007, the exact opposite will happen with AMD and Intel producing at about 1.5x industry demand. Intel has a lower cost product, cash reserves, and the knowledge that failure is not an option now that it divested all other potential growth businesses. Call it whatever you want, but AMD's future has never looked more bleak than right now.


This doom and gloom is typical from Jumpingjack. :D

Real said...

This doom and gloom is typical from Jumpingjack. :D

Is this a joke in a language outside English or a country outside the US?

Scientia from AMDZone said...

erlindo

Smaller quotes please for one line comments. Thank you.

Scientia from AMDZone said...

Real

Sorry, that was my fault. The only part I meant was biased was the "tanking" comment. Dropping price to gain share is a valid business strategy. Tanking is when you lose both volume and revenue. Now, I've already addressed the part about AMD's margins. If AMD holds onto its share then it will increase margin as its costs continue to drop. You haven't said anything to counter this.

Jumping Jack is a regular poster on ForumZ who is legendary for his anti-AMD bias. He has a personal grudge against AMD because he said that AMD screwed him somehow.

Real said...

Dropping price to gain share is a valid business strategy. Tanking is when you lose both volume and revenue.

I think you need to look at the trend here. AMD dropped prices and I can guarantee you that this was not done in the first week of the quarter but got progressively lower as the quarter went on. My forecast is that AMD exited the quarter losing money in their core CPU business. Further, I believe Intel will continue the price war so ASP for AMD will drop further. Basically, AMD can not cut costs fast enough to outrun their falling ASP's which caused margins to fall to 36%. In week 1 of Q4, the margins were somewhere around 51% based on Q3 average. If the average for all of Q4 is 36%, I would think margin for the last week of the quarter is closer to 25-30%. Given Q1 should be less units overall (AMD guided down for -3% to -10% for revenue), fab underload comes into pay to further offset any other cost savings. I agree with the Goldman analyst that AMD's strategy is fundamentally flawed. We shall see this in 2007 ending in a large net loss.

Scientia from AMDZone said...

I read these two. Perhaps they cover the things you were talking about, Real.

Weak Pricing, Strong Competition

Price War

If that covers the Goldman Sachs position then I understand what they are saying. They forecast reduced demand and overcapacity.

Their rough financial picture is accurate but what is missing from their assessment is any real technical understanding about microprocessor production. I gather that GS would have been happier if AMD had not bought ATI and were not expanding FAB 36/38. If this is their view then their assessment cannot be taken seriously.

Now, some corrections. Intel does not have a lower cost product. On 65nm AMD's and Intel's costs are about the same.

Secondly, your characterization of AMD's 65nm as a "mess" is incorrect. It is proceeding much more smoothly than either the 130nm or 130nm SOI transition did.

I understand your confusion about the 65nm clock levels but I can assure you that 65nm today is capable of clocking to the same speeds as 90nm without any additional changes.

The transistors that will be used in Barcelona are different from current Brisbane and should reduce power draw quite a bit. These new transistors may however take some tweaking to get the clock speeds up.

Your rising tide scenario is cute but you seem to forget that AMD gained in 2001 with a falling market.

Your Alpha knowledge statement appears to be a lame attempt to avoid giving AMD any credit by giving all of the credit to DEC. The biggest problem with this notion is that the Alpha tech including the Alpha bus was incorporated into K7, not K8. There was no further infusion of Alpha technology for K8. If you want to give DEC credit for K7 that is fine but trying to give DEC credit for K8 shows that you are either technologically inept or ignorant of history.

Scientia from AMDZone said...

Real

Yes, I agree if you make some assumptions and then make projections based on those assumptions that you can arrive at a pretty gloomy forecast. However, I have no reason to think that the management of AMD is as stupid and inept as they would have to be to chase your scenario. There are other factors that you have ignored.

Basically, I would expect AMD's margins to recover somewhat in Q1. Apparently, you expect their margins to be worse. I guess we'll see what happens in a couple of months. If AMD shows further drops in margin and loses volume share then I would also be more pessimistic about their future. But, right now, I don't believe that 2007 will be anywhere near as bad as you suggest.

Real said...

If you want to give DEC credit for K7 that is fine but trying to give DEC credit for K8 shows that you are either technologically inept or ignorant of history.

LOL, I suggest you look into the EV7. The Opteron basically added IMC along with hypertransport. Where do you think that idea came from?

I understand your confusion about the 65nm clock levels but I can assure you that 65nm today is capable of clocking to the same speeds as 90nm without any additional changes.

LOL^2...

here is a link to the lower IPC http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2893

As to the lower clockspeed, you only need to refer to AMD's roadmap and the top binning parts that remain 90nm.

It is proceeding much more smoothly than either the 130nm or 130nm SOI transition did.

I can agree with that statement to an extent. .13u almost killed AMD. The 65nm misstep is much more entertaining as fanboi's were expecting 65nm shipments back in Q1 after all the press releases and technology days from AMD hyping their transition. There is a reason why AMD management has lost credibility when it talks about ramping 65nm in 2007 and launching 45nm in 'mid'08'. As to Intel, I believe their goal remains to launch 45nm processors in Q4 - if they do, I would not be surprised is K8L and 45nm Intel chips ship in the same quanity for the quarter.

Scientia from AMDZone said...

Real
The Opteron basically added IMC along with hypertransport. Where do you think that idea came from?


Are you serious? The IMC predates Operon by more than 20 years. It predates not only the EV-7 but Alpha itself. And HyperTransport dates back to the Athlon MP when it was originally called "Lightening Data Transport". The LDT consortium predates K8; it was later renamed the HyperTransport Consortium.

here is a link to the lower IPC http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2893

Yes, there was a 17% increase in cache latency with Brisbane. C2D's cache latency increased 40%.

The top binning 90nm parts are not because of 65nm's limitations. 65nm can be clocked to 3.0Ghz today although you can probably hit 3.2Ghz with 90nm.

as fanboi's were expecting 65nm shipments back in Q1

Q1 of 06 or 07? And, why are you talking about fanboi's?

Intel should deliver a small quantity of Barcelona's in Q2 and should have FAB 36 nearly converted to 65nm by mid 2007. It was rumored that Intel would deliver some 45nn in Q3 but lately the talk has been Q1 08. If Intel launches 45nm in Q4 it will not match the volume of K8L.

Scientia from AMDZone said...

Real

I have no idea where you get all your misinformation from. Here is a link to an EETimes Article. This is when AMD changed the name from Lightning Data Transport to HyperTransport. The article is dated 2/20/2001 which predates the launch of K8 by more than 2 years.

The EV7 did indeed include interprocessor communications links when it was released in 2003.

However, the Athlon MP included the APIC bus which allowed communications between the two processors and 766 southbridge in 2001.

Greg said...

Lol^3, isn't respected analysts an oxymoron?

Like I posted and linked, 45nm will be mid 08 from Intel (no, I'm not switching the companies names on accident like scientia just did). IMC and CSI will not be implemented until 09 (well, the article I posted only says CSI, but you can't have one without the other).

AMD isn't releasing as high of clockspeed 65nm parts because they're trying to milk 65nm as much as they can in terms of savings. 90nm has so much transistor technology improvement invested in it that it's fairly easy for them to get higher clockspeed parts to turn out than 65nm when they first turn out 65nm. When 65nm is matured more, most of those improvements should have transfered to it, and will probably allow them to clock it even higher than the 3.1 ghz people are getting out of pretty much every 65nm processor out there (at least, when they know what they're doing, which most don't).

In a monopoly turning duopoly situation, the smaller company should conceptually benefit from market oversupply due to the need for the market to regulate and even out product distribution. Obviously, this isn't the type of closed and ideal situation I just described, but if there is oversupply due to market shrinkage, AMD will inherently benefit due to their limited capacity and exceedingly inaccessible demand, and due to pc makers wanting leverage against their own suppliers to ensure better products.

Also, real, is this "largest debt they've ever had in terms of percentage of income and capital or other such numbers, or simply in terms of gross monetary volume. If it's the latter, I suggest you seriously re-evaluate the value of such a statement.

Why would their be a price war in q1? What growing market would their be to fight for? I mean, seriously, everyone just got their new budgets (that they're going to stretch as much as possible) and everyone just spent way too much money on christmas presents, and while Vista did just launch, it's obviously not making anyone salivate enough to buy an entirely new computer for it. As such, ASP's would surely increase, as their is little incentive for either company to lower their own ASPs by engaging in a price war.

Also, while AMD did lose money in q4, that does not mean they couldn't use saved capital to pay for such losses, and thus you can't know if they're in debt, or how far they're in debt. Now, maybe I'm just ignorant of the situation, but this does at least sound like a somewhat valid point.

Woof Woof said...

I know Scientia likes the discussion to stay on track, but there seems to be a lot of confusion abt AMD's transition to 65nm, which I think bears further discussion.

I guess AMD has had to change the way they design/tweak and improve their process and their designs.

There's a pretty verbose article in AMD's website
http://www.amd.com/us-en/Corporate/AboutAMD/0,,51_52_9999_10000,00.html

But the gist (when I talked to some people outside of AMD who have worked with APM) is that it is very different from the way that Intel uses.

The first area of difference that I can see is the concept of a gradual/modular evolution. This is not a process yield improvement per se.

Rev E to Rev F provided an evolutionary process, with just tweaks to the core. Both on 90nm SOI process. With improvements made on 90nm SOI, the transition is fast.

Rev F to Rev G transited from 90nm to 65nm. As Scientia has suggested, it appears the transistor design of the Rev F does not match directly to the new 65nm SOI process.

Over time, this process will be tweaked. Just like how the first 130nm SOI process improved over time, and later then 90nm SOI. Which kinda explains why 90nm parts are still higher clocked than the current 65nm variants though 65nm parts already yield better TDPs especially on idle. But expect the performance lead to be overtaken by 65nm parts once there is more experience with the new process.

So when Barcelona launches with a pretty much new design/core, the risks associated with the transition to 65nm AND a new core is mitigated since half the work is already done.

In part, this difference in idealogy is because AMD started off a lot smaller than Intel. They don't have different lines/fabs to do completely new lines or products. Using APM allows them to improve their designs over time and mitigate transition risks.

The other area is the concept of modular designs, which I think is pretty cool. Memory controllers/Cores/Caches etc are designed in modular blocks and then can be laid out, almost mix/match variations depending on the task at hand. So, the new Barcelona designs can be with dual or quad cores. The cache sizes and levels of caches can be adjusted (I think this is why there is an increase in latency today even if the products continue to have the same cache levels as before). We don't have full visibility of AMD's roadmaps beyond this year or so, so why this change was made, is anybody's guess for now. My guess is that it is for a product further along in the roadmap.

So what does this have to do with how AMD's position going forward?

I think it does coincide with Scientia's analysis earlier.

AMD took a hit in margins in Q4, reducing their ASPs to gain market share.

I think that it is a calculated gamble, because as their process improves on
a. 65nm (smaller dies)
b. 300mm wafers (bigger capacities)
So will their cost structure.

So the increasing share strategy should pay off, and their margins will recover.

And with the transition to Barcelona cores, it should make the fight much more closer than it was in the months ending 2006.

Intel is also poised to do well (again, this is not something that Scientia is disputing).

Intel's move to increase their mix of Core 2 products will help them maintain their margins over the next few quarters (acc to Watch Impress it will be 100% by Q4 07), but their performance/cost structures are likely to stay the same because the trump cards:
a. 65nm process
b. 300mm wafers
c. new improved cores
were already played in 2006.

Intel's 2 remaining cards are: a 45nm process and the bigger 6MB shared cache (Wolfdale Dual Core and Yorkfield MCM Quadcore), which according to Watch Impress' roadmap (usually uncannily accurate), won't be in till 2008

http://pc.watch.impress.co.jp/docs/2007/0202/kaigai333

Scientia from AMDZone said...

boon
But the gist (when I talked to some people outside of AMD who have worked with APM) is that it is very different from the way that Intel uses.


Intel uses Copy Exact.

The first area of difference that I can see is the concept of a gradual/modular evolution. This is not a process yield improvement per se.

Okay, neither of these are related to APM. And, AMD does not have a modular design yet.

Rev F to Rev G transited from 90nm to 65nm. As Scientia has suggested, it appears the transistor design of the Rev F does not match directly to the new 65nm SOI process.

That's not quite what I am saying. Rev G (Brisbane) is just a die shrink of Rev F. What I was referring to was Rev H (Barcelona) which has substantial transistor changes from Rev G even though both are 65nm.

The other area is the concept of modular designs, which I think is pretty cool.

This doesn't show up until 2008. Barcelona is not a modular improvement.

because the trump cards:

a. 65nm process
b. 300mm wafers
c. new improved cores
were already played in 2006.


Right which is why AMD's costs should drop more quickly than Intel's. Intel will reduce costs though but not as quickly because they did not start out on 200mm.

Intel's 2 remaining cards are: a 45nm process and the bigger 6MB shared cache (Wolfdale Dual Core and Yorkfield MCM Quadcore), which according to Watch Impress' roadmap (usually uncannily accurate), won't be in till 2008

In 2008, AMD will have a modular design, DC 2.0, and something from Fusion as well as its own 45nm.

The real advantage of APM is two things: First, it includes expert level production checking and correction software. The software is capable of making corrections as a tool wears and moves outside of the original spec. This is what Intel does not have. However, they apparently have plenty of engineers to do this manually.

The second improvement is not fully realized yet. APM is capable immediately starting production on another similar APM system rather than taking two months like Intel. Unfortunately, this only helps with Chartered at the moment. However, once AMD has two 300mm FABs this will allow production to move from one to the other in one day.

Fujiyama said...

I read this discussion and see that there are two important factors - the time and capacity. The design doesn't matter because the customer is going to get faster and faster chips - this is not an issue.
P4 is slower than Athlon64 which is slower than Conroe which will be slower than Barcelona which.....

What really ponders me is the capacity and market share gains with better design. AMD is showing better and better numbers every quarter in terms of revenue but the old problems are still unsolved. Dell casus is not going to happen again and the capacity choke is the last one - hopefully. But if Barcelona demand is higher than output I would say AMD should finally make some changes in production planning

TheKhalif said...

However, I also gave theKhalif a warning in the last thread when he started flaming the other posters. I guarantee this blog will not turn into ForumZ.


I don't flame people. I also don't appreciate 3rd party references. Yes, this is your blog but please refrain from mentioning me when I'm not involved.

Scientia from AMDZone said...

theKhalif
I don't flame people.


I can copy and repost what you said. It was a flame.

I also don't appreciate 3rd party references. Yes, this is your blog but please refrain from mentioning me when I'm not involved.

Some might suggest that I banned Red because he was pro-Intel. I was making the point that you got a warning for flaming even though you tend to be more pro-AMD. The point I was trying to make is that I am not letting AMD fans say anything they want and only attacking pro-Intel comments.

TheKhalif said...

Some might suggest that I banned Red because he was pro-Intel. I was making the point that you got a warning for flaming even though you tend to be more pro-AMD. The point I was trying to make is that I am not letting AMD fans say anything they want and only attacking pro-Intel comments.


But they were being a-holes.

enumae said...

Interesting article.

Scientia, could you speculate on this processor design?

Thanks.

Scientia from AMDZone said...

enumae
Scientia, could you speculate on this processor design?


I'm not sure what there is to speculate about; this processor is not very interesting. This is a System On Chip (SoC) design which is common with embedded processors. It's also a lighweight design meaning that it is designed for low cost rather than performance. It is 32 bit only and has very little cache. Being only a 32 bit design makes it similar to AMD's x86 Geode LX series and IBM's embedded version of Power.

enumae said...

Thanks Scientia, I wasn't quite sure what I was looking at.

Just to be clear though, if this is based on Pentium M, and it is competing against AMD's Geode, shouldn't this be pretty impresive performance wise in this segment?

Azary Omega said...

With VIA out of picture - yeah sure it will be....

By the way, if VIA didn't had to go then by the time this chip arrived they would have something better. You just keep buying intel chips - antiprogress needs your money to compete with progress.

Greg said...

Via's not out of the picture yet, though it's sad that Intel will make them that way soon. Seems to me like Intel's embedded product is fairly inferior to AMD's, seeing how it runs at about 1.5x the TDP, which is a much larger concern for those that would be interested in this product (13watts compared to 9). You also have to consider Intel's rigged TDP scheme (which may or may not have actually been used).

It'll be interesting to see how this product's IPC compares to Geodes.

enumae said...

Azary Omega
You just keep buying intel chips - antiprogress needs your money to compete with progress.

Please explain how Intel is anti progress?

Azary Omega said...

In this case intel puts out effort to take VIA out of picture to make way for their own product when in reality intel's product can only compete with VIAs C3 chip which is pretty old and would definitely could not stand a chance to what VIA could have came up with by the time intel's new SoC chip would be released. Because what intel did industry will stand on one place for a substantial amount of time. If intel wouldn't have pursued VIA to exit x86 industry then the we would have better CPU design in this specific field.

Thus intel is antiprogress.

enumae said...

Thank you for an explanation Azary Omega.

Greg
Seems to me like Intel's embedded product is fairly inferior to AMD's, seeing how it runs at about 1.5x the TDP, which is a much larger concern for those that would be interested in this product (13watts compared to 9).

Well I was looking at AMD's Goede NX, and the 9W parts (667MHz and 1.0Ghz) chips do not include everything that Intel Tolopia does.

AMD's power that you are comparing this to is just the processor, and not the chipset.

As I understand this, Intels TDP includes both processor and chipset on a single chip (excluding the GPU), while using a processor based on the Pentium M, which used to compete with Athlon, so please explain how this is inferior.

Thanks

Aguia said...

I think I have read that VIA is going to release a new core on a new socket or bus because of Intel.

But i think i would prefer the Transmeta

Anandtech pre-preview

Greg said...

Ya, I have a friend who does gps systems solutions for trucking companies, and they use transmeta as well due to wicked low power requirements. If you want the ultimate in low power low cost, you still use transmeta.

Ah, good point enumae, I forgot to consider all the onboard parts. Still, it's technically just as good then, which means it's slightly worse than via's products, but probably cheaper.

Scientia from AMDZone said...

Intel's product is faster than Geode LX although it uses about 3X as much power. It is not quite a fast as Geode NX although it should use less power.

Embedded processors are common. There are embedded versions of ARM, MIPS, Power, and X86. The new core that you are talking about from VIA is a niche product.

Actually, VIA's toughest competitor was AMD, not Intel. When AMD launched Geode it only drew half as much power as the equivalent VIA chip.

Intel missed an opportunity when they didn't buy Transmeta. If they had, Itanium would be in much better shape today. As it is, I don't expect Itanium to grow. Miss guess is that they will become marginalized within a few years. Certainly, it is difficult to see how Itanium will handle Opteron's native 32-way design in 2008.

Real said...

Real said Even if they manage somewhat more than a paper launch of the K8L in Q2’06,

Scientias responded:
No, I'm sure it will be worse than that. I'm sure Intel will release 45nm with the new socket and CSI in Q2 while AMD probably won't have Barcelona out until 2008. Surely, that is how it will be.


You should not have used sarcasm here....

http://images.dailytech.com/nimage/3796_large_Opty-list.png

As I said, looks like a paper launch of even the high end parts and no real volume until 2008. That is, unless they figured how to go from production to release of products in under a month....

enumae said...

Just something I think will reinforce my position about Intel's ability to govern AMD's margins.

DigiTimes

Aguia said...

Real I also expect AMD to paper lunch the new processor in all its forms single/dual and quad; mobile, desktop and server.

They will use that same trick as Intel did with Core 2 Duo, to force Intel lower their prices (profits).

I don’t see a problem because wasn’t the Core 2 Duo paper lunch too?
6 months after their production (or commercially availably) was lower than 23%. Isn’t that a paper lunch that forced AMD to cut prices in half?

AMD only needs the new processors to be at least 20% of their production by the year end.


And enumae Intel isn’t govern AMD margins. AMD only priced their products according to the new Intel performance offerings. That’s why you see the X2 6000+ price about the same cost of Core 2 6600/6700.

If the new processor is as fast or faster than the Core 2, Intel will also have to lower their prices and lower their margins, leaving Intel waiting for 45nm.

enumae said...

aguia
And enumae Intel isn’t govern AMD margins. AMD only priced their products according to the new Intel performance offerings.

I agree, and how is that not controlling margins.

AMD could have done the same thing when Intel only had Netburst, but they needed money for expansion and the ATI aquisition, they could not provoke a pricewar.

Intel has lost market share, but has regained in the Server area due to good pricing and performance.

And with the push of Quad cores, competing against AMD's dual cores, the choice would seem pretty simple, you will go with quad cores.

If the new processor is as fast or faster than the Core 2, Intel will also have to lower their prices and lower their margins, leaving Intel waiting for 45nm.

But I am talking about April, and looking at the recent news about Barcelona roadmap (expected Q3), Intel will be able to offer quad core's at excellent prices while AMD will be forced to lower it's Opteron prices, thus lowering AMD's margins.

This is some recent news.

1. Intel quad core price cuts.

2. Opteron price cuts.

There was another article (asking for registration when I use a link to it) on DigiTimes about Intel increasing 2P Cloverton to 70% by years end.

AMD has to follow suit right now and what would be the next couple of quarters, and this is why I believe Intel is controlling AMD's margins.

Yes, things will change on the launch of Barcelona, but thats going to be a coulpe of long quarters for AMD.

Greg said...

But the price of clovertown is being lowered to make up for its assumed performance deficit against barcelona. So AMD doesn't have to lower its prices, because the demand for its product will be even higher. AMD could even price quite a premium because of how much server makers want to leverage them against Intel, making their ASP's much higher than was even normal in the past.

Also, even at the worst, AMD has had paper launch slips that lasted less than a quarter. Saying they would slip an entire two quarters is ludicrous and a show of either ignorance or contempt! Even if they did slip 2 quarters, they'd be slightly earlier than mass market perforation than penryn. As long as penryn doesn't smoke them, which isn't necessarily that likely, AMD will follow very close behind Intel in 45nm and will have immersion lithography. Also, barcelona is expected q2, so look again, and get your facts straight. Also, real, you might want to look at that chart again, as it shows q2, and also is in no way referring to anything that has to deal with availability and paper launching.

Aguia, core2 a paper launch, but only a short one. It was not as bad as AMD 65nm (as long as we're calling lack of supply due to massive pc manufacturer consumption a "paper launch").

I think we could see penryn and barcelona go head to head in terms of performance, and that barcelona could see a good 40% performance advanatage clock to clock. A recent video on youtube showed that AMD was increasing branch guessing efficiency and accuracy, meaning better IPC.

enumae said...

Greg
But the price of clovertown is being lowered to make up for its assumed performance deficit against barcelona.

Well looking at the way the prices worked last year, Server is the one that was not touched, not until a while after Woodcrest's release, so I can only assume that Intel is trying to apply as much pressur on AMD as it can.

Intel is going after AMD's wallet, the Server segment, and at all cost. If after Q1 Intel has not regained a substantial amount of the segment (3-5%, including the previous 1.4%) then AMD has weatherd the storm, if Intel has, they will report a bad quarter due to the pressure and it will continue into Q2.

...because the demand for its product will be even higher...

We need to see numbers first... don't we?

...leverage them against Intel, making their ASP's much higher than was even normal in the past.

I think you have that backwards, Server makers can leverage Intel against AMD to keep AMD from making their ASP's to high.

...either ignorance or contempt!

Who is this comment intended for?

...AMD will follow very close behind Intel in 45nm and will have immersion lithography.

I read on VRZone that it is about a year to a year and a half between test wafers and processors, could you or maybe Scientia elaborate on this?

Intel had done there 45nm test wafer in January of 2006, while AMD did there's in April.

If this is the case they should have working samples somewhat soon, correct?

Thanks.

Scientia from AMDZone said...

enumae
I read on VRZone that it is about a year to a year and a half between test wafers and processors, could you or maybe Scientia elaborate on this?


I assume you mean from the first SRAM wafer test. This is more like 2 years. Intel produced its first 65nm wafter in Jan, 2004.

Intel had done there 45nm test wafer in January of 2006, while AMD did there's in April.

So, Intel should be able to make 45nm chips in January 2008.

Scientia from AMDZone said...

real
As I said, looks like a paper launch of even the high end parts


I have no idea where you get this notion from. What about the roadmap suggests a paper launch?

and no real volume until 2008.

No. It would be real volume in Q3 07.

That is, unless they figured how to go from production to release of products in under a month....

Where do you get the one month figure from? Oh, I see, you interpret the Production column as the start of the chips on the assembly line. However, the Production column actually means production out. Tthen it takes about a month to get chips into the channel before launch. When AMD says Mid 2007 this could be as late as end of July.

So, anyway, some chips should arrive in Q2, volume in Q3 and desktop chips in Q4 with the 1xxx series Opterons. Actually, I'm surprised they will release 8xxx that early. In the past 8xxx was always a quarter behind 2xxx.

enumae said...

Thanks Scientia.

Greg said...

The comment was intended for real, enumae. I'm sorry if I worried you.

While server makers could use Intel against AMD, that would go against the will of their consumers, and thus hurt them if a server maker then decided to not use Intel against AMD. You can only really use a competitor to force a product's cost down if that product has inflated demand due to availability and market penetration. So this really probably doesn't work on AMD. Obviously, any new drop in product for a socket 1207 board that doubles the number of cores that vastly outperform the current processors in that socket on a core/core basis would definitely have a high amount of demand, and that's excluding demand from those who are currently using Intel systems, which would be the only indeterminable, but would still exist to a noticeable if not large extent.

enumae said...

Greg
...Obviously, any new drop in product for a socket 1207 board that doubles the number of cores that vastly outperform the current processors in that socket on a core/core basis would definitely have a high amount of demand...

Good point, I had a narrow view of the overall picture.

Real said...

Latest comments from ThinkEquity Partners (Eric Ross):

We continue to hear from our sources that Advanced Micro Devices (AMD) is going through an extremely tough quarter, with inventories piling up at the channel as Intel takes more shares in the server segment. We expect AMD to continue to experience share loss and ASPs erosion in the server space. Desktops and notebooks are also at risk.

I think this is exactly what I have been saying....

Hearing more chatter of unscheduled pricing cuts. We continue to hear from several of our channel sources that distributors are rapidly cutting prices in order to unload AMD parts. Some of these are in response to Intel's rebates and marketing dollars, but it is obvious price wars are coming, in our opinion.

Price war not ending but accelerating - yep, exactly like I predicted. Further, AMD ASP's are falling drastically as people shift to buying Intel parts that perform better at lower power.

Intel is taking more shares away from AMD in servers. In our opinion, Intel's product line up should remain strong as to enable the company to continue to put pressure on AMD's margins. We expect price competition in the server space to remain intense. We do not expect AMD to see much relief until it introduces its quad-core chip "Barcelona" expected in 2Q.

Only thing wrong here is assuming Barcelona will ship in any quantity in Q2. AMD is falling about as fast as any company I have ever seen. Didn't Henri say "the gains we made are irreversible" only like 6 months ago? Funny how AMD exec's always look foolish in hindsight....

Greg said...

I'm pretty sure Henri was talking about volume share, and he was right, as they obviously haven't lost any.

Real, how many times has anyone on this board said you need to look at this in relation to AMD's previous times of losses and ASP pains.

Also, if this is the fastest you've seen any company fall, you obviously haven't been around very long. Remember the end of the 90's? I kinda do, the whole no one having a job thing was a bit of a problem and a bit memorable. I'm assuming you just slipped there, but there were companies that fell much faster back then and they weren't even necessarily set up on flawed business models.

Again, respected analyst? And how?

Scientia from AMDZone said...

real
We continue to hear from our sources that Advanced Micro Devices (AMD) is going through an extremely tough quarter,


What quarter are they talking about? Q4 06?