Sunday, March 11, 2007

AMD cutting back...and how Q1 will turn out for them

Couple of interesting reports:

First...AMD delays moving into their new facility to defer costs to 2007...and they state not to disrupt the Barcelona launch.
http://www.statesman.com/business/content/business/stories/technology/03/08/8amd.html

Second...AMD is reviewing their plans to build the NY fab:
http://timesunion.com/AspStories/story.asp?storyID=570662&category=BUSINESS&BCCode=HOME&newsdate=3/10/2007&TextPage=1

I'm pretty sure the report will come back in 12 weeks a little less upbeat than AMD has sounded in the past on building this fab but will defer any decision till late this year or next. Their game plan here is to give themselves a face saving exit strategy from this investment as they continue to lose money and will undoubtedly have to go back to the capital markets.

As I have been saying...and saying. AMD will have a tough time managing cash and debt as they invest in ATI and start losing money again. Defering non-essential expenses is the right thing to do. However, as they lose money, they will be forced to resort to lay offs. It is not feasible for a company of 10K people making money to become a company of 16K people losing money and sustain their work-force.

Q1 will see them hit hard as they lose server market share which reduces their ASPs, their margins and hence their ability to sell client CPUs at low prices. They will gain unit share in mobile but unfortunately not a proportionate amount of revenue share due to deep discounting. Specially to Dell. They will lose desktop market share primarily in the channel but the overall ASPs in this segment are declining rapidly so this might be a good thing. The one reason it may not be good is if they are not using up all their mfg capacity. The other thing that will drag them down is they now need to combine the lower margins in the graphics business. Though I'm pretty sure Bob Rivet will continue to show the CPU and graphics businesses together and seperately in an effort to show that their CPU business remains robust.

AMD started this as a street-fight and they were winning through a series of sharp paralyzing jabs that were weakening Intel. However, Intel has now turned this into a full scale battle. Unfortunately for AMD, they have not upgraded from the knife they had in the street-fight (read - the ATI acquisition should have been done earlier) and Intel has brought out the entire arsenal. Had they acquired ATI 12 months earlier, they would be in a position to buffer their losses in the CPU only biz right now with a platform approach. Alas...hindisght is 20/20, even for me.

You can read the frustration of AMD management who have made significant (and commendable) progress through they way they continue to spin Barcelona without a real demo, complain about Intel's benchmarking when their own dubious practises have allowed Intel to use their outdated benchmarks against them and the whining tone of voice over the lost Intel records. All these signals indicate they don't really know what to do next and should Barcelona not be the all out winnner they predict it is, they will be a 10$ stock for a long...long time.

41 comments:

Unknown said...

While I'm still looking into your article to see exactly what I think of it, you're still going the path of "these people don't know what they're doing, because I don't like what they're doing".

Obviously AMD will take time considering the new york fab whether they are tight on money or not. It's a huge investment, and not one to be taken lightly. So your point that AMD is having problems because they're going to make sure they think about their costs is pretty weak.

Also, if you knew you were going to have a lot more profitable product in a few months, wouldn't you put off any unnecessary spending until that product was already there? I'm pretty sure I'd do this, whether I could swallow the expense now or not.

Your characterization of his epic battle is, yet again, short sited, and in no way takes into account how far ahead these companies have planned their product launches and their expenses.

Care to elaborate on AMD's "dubious practices"? I'm fairly certain, if AMD has ever done any sort of "unfair" benchmarking, it's in no way to the magnitude that Intel has, both through their "sponsored" benchmarks through Anand and THG, and other such websites.

I'd be frustrated if I worked at AMD too. It would seem very unfair that you couldn't operate as effectively as your competitor because of their shady business tactics, an uncooperative market, and customers who dislike you purely because you weren't there to begin with. Trust me, if you've ever had to run any organization that no one liked purely because of stupid misconceptions or ignored because they'd never heard of it, you'd get really pissed off when you tried to make a difference for them (in this case, by offering a choice). (If you were wondering, I've run 2, and taken part in runnin 1 other).

Anonymous said...

I think one thing the Sharikou, Scientia, et al, keep forgetting is that manufacturing costs between Intel and AMD are not the same, despite Sharikou's ridiculous calculation of chip cost based on revenues...

The assumptions are that yields are similar (if not better for AMD) because of APM3.0, when they know absolutely nothing about this automation technology, nor the competing automation/feed forward controls other suppliers use. While it is an advanced technology (and probably better than any other IC manufacturer uses), feed forward/closed loop systems are used in some form at every manufacturing fab in the world and there are far more significant yield levers than "tweaking equipment". If you develop a stable process which has a fairly wide process window, you are less reliant on feed forward contol. Of course if you do the CTI approach and make substantial process changes every 3 months than you need this control. If you use Intel's approach which is develop the process while implementing vitrually all of the improvements prior to releasing to HVM, there is less of a need to run the process on the edge of a cliff.

Intel's yields are actually better, coupled with smaller die size this is just one advantage Intel has (unfortunately I have no link to prove the yield #'s).

Also the wafer production costs are just not the same. AMD/IBM uses an extra metal layer than Intel for a given technology node, the SOI/SSOI added cost is substantial (~15% to finished wafer cost), they also use more strain technologies which require extra processing steps on 65nm. While folks like to say this is more sophisticated, Intel gets the same/better PROCESS performance using a simpler and cheaper process flow.

The other factor is when you consider the WAFER starts on 200mm compared to 300mm, 200mm continues to be at least 1/3 of the production. Industry estimates are as ~30% cost reduction using 300mm. As most of the equipment is probably depreciated by now the cost delta may not be as substantial, but for a significant volume of wafers it is actually cheaper to BUY all new 300mm equipment than to re-use 200mm equipment. Also as Intel buys substantially more equipment they can get better volume pricing (this is true for Si wafers and process materials as well).

Until AMD gets their 2nd fab fully converted, they will not be in the ballpark for costs. And at the time they do this Intel once again will be 1/2 to a full technology node ahead (which will maintian a manufacturing cost advantage).

Finally MCM is a much more cost effective approach to quad core and gives more manufacturing flexibility as the decision between dual and quad core can be made after the wafer is finished production as opposed to before 0- this is a 13 week advantage. As AMD has been very public about there issues being able to preict market mixes the last few quarter, people underestimate the value of this.

I'll let folks debate about the technical merits, but from a cost perspective there is no sane argument regarding MCM vs monolithic approach. Sheer die size alone drives yield benefits. Bin splits for both speed and power will also be better for MCM as you can match up 2 dual cores. For monolithic your bin speed for the chip is driven by the slowest of the 4 cores.

In my opinion, AMD's only chance is to have a significantly superior product which will allow them to set the pricing (this was obviously the case in the Opteron/Athlon days pre-Core2). If the performance is similar enough the only place where AMD will be able to set the price is in the 4P+ space - while people (specifically AMD fans) like to talk about this a lot it is a very small piece of the revenue pie.

The NY fab is probably irrelevant, by the time it would be contributing any capacity it will be late 2009/2010. AMD is better off expanding Dresden with another fab as they have expertise and economies of scale in place there. The downside is from a risk mitigation perspective - if anything happens there (natural disaster, etc), AMD is sunk.

At this point, I would think AMD would/should work on expanding it's relationship with Chartered and or start working with UMC/TSMC. While this eats into there margin some, it will give them expandable capacity without the cash outlays. It's also not clear where the graphics production (both IPG and discrete) is happening - does anyone know if they are still outsourcing this? I remember reading somewhere where AMD said they will start doing this inhouse in 2007, but given capacity I would think they would be better off continuing to outsource this.

Anonymous said...

Call me crazy, but I think at this point it is probably best not to release Barcelona benchmarks.

If they are only marginally better than Core2, AMD's market share in server may crash as people who are at the upgrade cycle with "older" Opterons are probably holding off to Barcelona. If it is not much better and is still 6 months away, they would be more likely to switch to Core 2. If they find out say a month or so prior to product availability, they may be willing to wait that time even if it's only marginally better.

And in the case Barcelona is significantly better - that will crater most new Opteron sales. While most fanboys say it is a drop in and customers can buy an Opteron now and drop in a Barcelona later, you are still throwing your money away on those temporary Opteron chips. If there is a vast improvement I don't see people buying Opteron systems now and upgrading to Barcelona in 6 months - I see them just waiting those 6 months and limping along with their current systems. This would create an even greater cash crunch at AMD.

I would suspect the major customers who buy significant server chip volumes have access to either engineering samples and/or real benchmarks - AMD has little to gain and potentially a lot to lose releasing these in a widespread fashion this far in advance of actual product availability.

One last ramble - consider desktop customers which are at looking at least a quarter or 2 behind the server chips. If Barcelona turns out to be 10-15% better than Core2 would the high end consumer be willing to wait to late Q4/Q1 to get this advantage or just get a Core 2 today and deal with the 10-15 delta (hoping by then Intel might also have a better 45nm product as well)?

Anonymous said...

I think your predictions for Q1 is dead on - server share will continue to erode, I think notebook will be flat (Intel is starting to ramp Core2 now in this space which may mean the spread of the price war to mobile), and desktop I think will be flat sharewise. Of course Q1 is generally lower than Q4, so even if AMD is simply able to maintain share and ASP's there #'s will be down, if you couple in expected share loss and likely a continued erosion in ASP's I think Q1 (and Q2) will be miserable.

I don't see significant layoffs - layoffs actually cost money short term (3-6 months) due to severance packages and savings only kick in after that. Also it is not like AMD is running fat like Intel was (and still is?), a workforce reduction will likely have a real impact to AMD (whether it be R&D, manufacturing or marketing/sales).

Keep in mind Intel reduced workforce by 10K for a projected savings of (I think) ~1Bil/year. As AMD is starting with 16K, what realistically could they cut without having a substantial impact to operations? 20% would seem to be absolute max, no? 20% would be ~3000 people which would be on the order of $200Mil/year if you use Intel's estimate.

$200Mil is a drop in the bucket...to put this in perspective 200Mil would be ~5 new 45nm litho immersion tools.

While there may be layoffs - if this is being done to impact profitability and/or cash flow it is a waste of time and a poor strategic decision. The only reason to do layoffs in AMD's case is if they feel there is some excess fat post ATI acquisition (which there probably isn't much). If anything the continued F36 ramp and F30/38 conversion will likely put increased pressure on existing headcount.

180 Sharikou said...

Greg:

1. I am not implying AMD has problems...I'm stating it. If you cannot see that FACT then I can't help you.

2. I think AMD management knows exactly what they are doing. They are taking whatever short term measures they can to ensure their quarterly numbers don't look like Intel is beating the stuffing out of them. This means cutting short term costs.

3. Barcelona will not help AMD's financials this year. I trust you read the interview with Hector where he said that. Secondly - there is no co-relation between the release of Barcelona and the decision to go ahead with the NY fab which will be commissioned in 2009. Stepping away from the NY fab is a long term decision, much more long term than Barcelona. It will be driven by AMD's results for the rest of this year. If they don't regain profitability they will not have the money to fund the NY fab.

Put the pieces together - if they dilute equity through a share placement to protect themselves against a buy out then they cannot also go to the capital markets to borrow money to build a fab. The strain on their balance sheet will be immense.

3. Let me tell you that Hector and team get to keep their jobs not because they stand in front of the board and say - "Yup...we have a plan where the stock is 10$ for the next 2 years and then we will kick ass." It's not short sighted - it's called reality. For that matter, neither does Otellini keep his job for not delivering consistent results. CEO's job is to deliver both long and short term results. Specially when you've already been in the driver's seat for a few years.

4. Have you not read Sharikou's blather on the benchmarks and his link to George Woo's blog. If not, go read it.

5. I thought you said a few weeks ago you're still in college when someone was poking at you. Now you've run two companies. Whatever dude...I'm the King of Norway and run a country. Argue the logic buddy, don't get emotional.

Unknown said...

I said organizations, not companies, so get the banter right. Also, I was pointing out that it's easy to get emotional when it seems like the world is against you (which you were saying the management of AMD was doing, but attributed it to them being misguided).

I'm sorry, I don't read sharikou's blog. Unlike some people, when I find a news source to be worthless in terms of intellectual value, I quit wasting my time with it.

Why would AMD be at risk for buyout? Have you not read the OC'ers.com article? Buyouts occur when a company can make a lot of money, but has big management issues. AMD has problems making money right now, and has pretty good management compared to most companies that get bought out.

I don't get what your third point is about. Are you talking about contingency plans or are you talking about planning on the fly (which last I checked, was an oxymoron)?

Quad core barcelona will not be AMD's only new part. True, all the new parts wont immediately turn their financials around, but it will renew investor confidence, and get AMD supply wins in the channel (which you said they really needed, last I checked). Also, as most people here know, there is far less demand in q1-3 then there is in late q3 and q4. AMD doesn't have to have large amounts of Barcelona to take a respectable portion of market share with it (note I said respectable, not large).

Sorry I was not more specific. I was talking about management and planning problems. I realize they are really tight in terms of what they can spend and what they can lose, but I don't think this was something they didn't realize was coming. Which, referring to your previous point about how Hector would keep his job (I still don't see what exact point you were making) would mean they wouldn't blame Ruiz for their current position, because they knew part of it, if not all of it was inevitable (though if it wasn't, or even if it was, you should write an article about it, since there seems to be a lot of division on this issue).

Anonymous said...

"Trust me, if you've ever had to run any organization that no one liked purely because of stupid misconceptions or ignored because they'd never heard of it"

This is true of many new companies - and you know how they generally overcome it? Superior products which MEET market needs (AMD HAD this with Opteron, but this lead has evaporated) and sales/marketing which allows you to get your products visibility and address teh "stupid misconceptions". Let's face it AMD's best marketing scheme to date is to whine about how unfair Intel is - if they focus on why they have a better product more people may listen.

Handing out a [fill in the blank] book for dummies is catchy and may get some press, but is that really helping AMD reach new customers? Did those books help generate a single sale? Might the money on crap like this be better spent elsewhere? Might there be a better way to educate your customers about these "stupid misconceptions".

"Also, if you knew you were going to have a lot more profitable product in a few months, wouldn't you put off any unnecessary spending until that product was already there?"

No...if I truly KNEW I was going to have a profitable product I would continue to spend. If the spending is "unnecessary" than why do it at all? If it made sense to consolidate the Austin facilities in one location 6 months ago, have the technical merits of this decision changed? (other than cash position).

This we don't want to distract Barcelona launch is complete crap - did they not know Barcelona was launching at this time when they made this decision previously? As you point out, "how far ahead these companies have planned their product launches and their expenses." This is a pure cash saving move during a cash crunch, no matter how much they want to try to dress it up as a "we're focusing on Barcelona" thing.

For you to state otherwise just shows you lack of objectivity.

Anonymous said...

"AMD has problems making money right now, and has pretty good management compared to most companies that get bought out."

How do you know this (good management)? What attributes are you evaluating the "good" on? Are you an analyst who evaluates management teams? Do you do VC? Perhaps your past roles in running organizations gives you some insight you can share with us? (not meant to be sarcastic)

AMD's price to book is ~1.34, while I'm not an analyst I would tend to think THAT may be more of a reason why VC's are looking at them.

"Also, I was pointing out that it's easy to get emotional when it seems like the world is against you (which you were saying the management of AMD was doing, but attributed it to them being misguided)."

Not to defend 180... but is being emotional and being misguided mutually exclusive? I tend to think they were misguided not emotional. CEO's/management teams of this large of a company don't get there by not having the ability to remain objective and unemotional when it comes to making business decisions (except maybe Sanders...). If they are getting emotional than the board should step in and replace them.

"I realize they are really tight in terms of what they can spend and what they can lose, but I don't think this was something they didn't realize was coming"

This is the biggest joke of what you said - if this is in fact true, then it means they realized what was coming and BLATANTLY LIED to the financial community in advance of Q4'06 earnings and in their Q1 update where they gave guidance on sales only to downgrade it later - I find this hard to believe.

So which is it - they either saw this coming and lied to investors/analysts or do you think perhaps, just maybe, they didn't really see the magnitude of what was coming? (I tend to think the latter)

Unknown said...

Anonymous, have you been asleep for the last 3 years? Did you not notice how AMD's products were pretty much completely unused outside the channel? Did you not notice how the k7 wasn't a bad product, but pretty much everyone refused to use it because it wasn't an Intel processor?

AMD doesn't get to release a new, superior product whenever they feel like it. Getting a new architecture out the door takes time, lots of researchers, and often, lots of money. AMD has been short of much of this due to their previous "circumstances". As such, AMD has dropped prices to keep demand up and fab usage up as much as possible.

Yes, AMD is whining, but it's wrong to allow someone to just walk all over you. What AMD is doing is far from the incessent and pathetic whining most people do. They'd be talking about how much more money they'd have or how much more market share they'd have if they were really whining like that. All they're complaining about is the fact that the market is now unhealthy and not consumer-centric. Yes, this is all marketing and rhetoric, but it is all true.

An office transition will inevitably lead to more of the missed opportunities and product misplacement AMD had with Dell. If they have any of that with a new product, it will be disastrous with their consumer relations. This is the "distracting from Barcelona launch crap" that you are referring to. Ya, they've known for months that this will be launched. But Dell, HP, and everyone else will wait to make orders, because it's inconvenient for them to make too early of an analysis on demand and don't want to make space for a product they're not perfectly sure will be there and can't be perfectly sure of the numbers. It's all business. Also, you quote me saying it is about money, but then seem to believe I don't think it's about money.

Roborat, Ph.D said...

Greg said: “All they're complaining about is the fact that the market is now unhealthy and not consumer-centric. Yes, this is all marketing and rhetoric, but it is all true.”

“Consumer-centric” another consultant invented pseudo word that doesn’t really mean anything. “The consumer doesn’t have a choice and we’re here to provide choice!” – yelled AMD. Thank you AMD for the noble cause. I can see from your earnings report that you’re not really in it for the money.

Let’s talk about the consumer then. I suggest you go to a PC store and watch clueless individuals figure out what’s the better choice, the PC with a shiny mouse or that laptop with that nice movie playing in it. If anything there is just TOO much choices when it comes buying a PC for the average consumer. Workstation class desktop, gaming, home-office, budget, performance laptops or budget laptops from 10 different companies and not to mention there’s that stand alone Mac. PC’s ranging from $400-$3000 with thousands of confusing peripherals with a thousand more confusing standards (PS/2, USB 1.0, USB 2.0, Bluetooth, Firewire, SD, SO-DIMM, PCI, PCI-E, AGP, AGP4x, WTF, etc, etc, etc.)
Seriously, is there really an unhealthy PC market or is AMD simply crying “pro-choice” just to make it look like they’re fighting the battle for us? Hard to believe the latter when AMD hardly complained about the market in the better part of 2006 when they had accelerated gains. It looks to me like Intel’s dominance is simply a convenient excuse for AMD’s poor performance.

Anonymous said...

"Anonymous, have you been asleep for the last 3 years? Did you not notice how AMD's products were pretty much completely unused outside the channel?"

No I haven't been asleep - read my statement you need a superior product AND sales/marketing to sell the product! To date AMD's main marketing initiative is buy from us we're the underdog...Intel sucks...let's chaffeur (sp?) some folks around in limos during Intel's IDF's and handout non value added pamphlets

Unless you think AMD's marketing of Athlon/Opteron has been good?

"An office transition will inevitably lead to more of the missed opportunities and product misplacement AMD had with Dell."

This is complete, and unsubstantiated crap. Please tell me the "inevitable missed opportunities" by moving offices within the same state? Is the moving truck going to crash? Are the employees going to get lost going to the new site? SPECFICALLY what do you think some of the lost opportunities may be? This is not their production facility, how exacly are they going to have product misplacement from this geopgraphy?

So you are going to sit here with a straight face and say the PRIMARY reason they postponed the office ove is because they don't want the distraction. I noticed you completely ignore the point that AMD should have had a pretty good idea of when Barcelona was launching when they made this decision... I don't mind folks being fans of a company but it is completely illogical (or ignorant) not to understand that the primary reason AMD postponed the move was for cash.

Unless of course you think the move was planned FURTHER in advance than the launch of AMD's first new architecture in years and AMD "just" realized, hey, this moving thing might be a distraction - do you really think their planning is that bad?

I'm done arguing on these points - I guess people will see what they want to see. (ignorance must indeed be bliss)

"Yes, AMD is whining, but it's wrong to allow someone to just walk all over you."

You're right whining is the best solution. My bad. If they whine to enough people and do it frequently, I'm sure will start to listen - I think I read that in 7 habits of highly effective people. As someone who has "run" several organizations, do you subscribe to this theory as well (When getting walked over, whine)? Whine maybe once, but then fix/confront the problem or move on....

Unknown said...

The most recent anonymous, do you think management would tell their investors a quarter ahead of time that their financials would be weaker than the previous quarter? When most of those investors pretty much want them to fail?

What I believe is that AMD saw a possibility that they could do what they projected, but did not assume that would happen. Ya, this means they basically lied, but what company doesn't basically lie about what they project for themselves and their competitors? Being excessively optimistic in this type of report is not a crime if it is possible for what you project to happen.

I don't get what point you're trying to make with your third paragraph. I'm fairly certain that 180's emotional argument was in reference to AMD's complaints about the current market situation and Intel's forcing the market to be more exclusive. This then lead to the whining argument. Point being, I don't blam them, and I don't think it makes sense not to complain about it.

As to management, considering what the current management has accomplished (getting AMD processors in all of Dell's lineups, in Gateways servers, in Founder's and Lenovo's computers, and getting AMD's market share to the highest it has ever been) I'd have to say they haven't been doing a bad job. Thus, the management isn't bad, the company is just facing some very hard times because it's trying to expand in a market that is hostile to them and trying to expand their business in a way that will eventually finally make them stable.

180 Sharikou said...

Greg - what you describe in your first two paras just above is either fraud or incompetence and it was less than 30 days before they announced their Q4 miss...the exact date was 14th Dec. Either ways, AMD management is asking to be sued if they do this again.

Second - I was not referring to AMD's complaints about market situation but to their complaint on how Intel used their own outdated benchmark and then showed how Intel beat them on the benchmark they are publicly using.

pointer said...

Greg said ... Thus, the management isn't bad, the company is just facing some very hard times because it's trying to expand in a market that is hostile to them and trying to expand their business in a way that will eventually finally make them stable.

perhaps you used the wrong tense. the market was sort of hostile, not is hostile to AMD. The hostility had a reason, its brand, its supply capability, and its CPU perceived capability. All these have sorta gone with K8 and AMD's OEM penetration but AMD still has supply issue.

Another thing that you forgot to mention why AMD has to fight so hard is it currently lose the performance leads almost all ends to Intel, and not to mention the manufacturing prowess that is useful in price war and process development.

Anonymous said...

"The most recent anonymous, do you think management would tell their investors a quarter ahead of time that their financials would be weaker than the previous quarter?"

No - but that is not what I said. AMD in DECEMBER in an anlayst's meeting, made not mention of earnings/gross margins issues. This was OVER 75% of the way through the quarter at the time and they had no indications they're Q4 GPM was falling off a cliff?

Then at end of JANUARY when they announced Q4 earnings, they forecasted Q1 revenues, ~ 1 month later they announced a likely shortfall. Somehow in Q1 they could forecast the results 2 to 2.5 months into the quarter, but in Q4 they couldn't. Here's a hint - in Dec timeframe AMD was trying to secure additional loans/financing! (announcing a major shortfall then would have significantly impacted the rates). Now this is just speculation on my part, but others have speculated this as well.

Please get your facts straight.

"Ya, this means they basically lied, but what company doesn't basically lie about what they project for themselves and their competitors?"

Very sound reasoning - this of course is bordering on lawsuit worthy for investors (if in fact they did lie - I think this was more a convenient omission, rather than a lie).

As to management goodness - I'd argue that AMD's gains were in spite of their management. They had a tremendously SUPERIOR product for the better part of 3 years and in my mind management failed to fully capitalize on it. While some will argue that this is due to Intel - I'd argue that AMD was selling all of their available capacity so if there were more potential customers in this time period, it wouldn't have mattered as AMD couldn't produce anymore chips.

My point on the mgmt is they failed to expand capacity early enough - people again will argue that was because Intel held them down and they didn't have cash. But look at how much debt they took on for the ATI acquisition. With such a superior product (especially in server space) - they should have taken on more debt and expanded. If you say they couldn't have done this then they should have had initiated the foundries a lot earlier which would allow them to effectively expand without the capex spending. Sure this would have hurt margins but at the time they were >50% and this would have allowed them to capture even more market share.

In my view any ADEQUATE management team could have gotten the results AMD has gotten given that they had such a superior product. A good management team would have had better strategic planning (capacity) and AMD would not have lost profitability like they have recently. Don't judge "good management" by market share alone. And while everyone holds up the Dell deal as a major coup - this has seriously hurt AMD and was done a year too early. It has limited all manufacturing flexibility (see Ruiz comments the last 2 quarters on iability to get the right product mix), crippled the margins/profitability (again look at last 2 quarters) and has allowed Intel to change out low margin Dell business with better margin channel business (Intel's Q4 margins were flat despite being in a price war while AMD's dropped >10)%.

180 Sharikou said...

Good post Anonymous...well said.

Anonymous said...

"Ya, this means they basically lied, but what company doesn't basically lie about what they project for themselves and their competitors?"

lol *cough cough* Intel is evil, AMD 4EVA!

Anonymous said...

$16.1Mil for Ruiz... thats about 1 Million for every $600Mil in market cap that AMD has dropped.

Seems reasonable, no? Could you imagine what Ruiz would have gotten paid if he actually was able to make money for the owners (stockholders) of AMD? Well at least the stockholders can cash those quarterly dividend...er, maybe not...well at least Hector got his money,...

Apparently AMD is not cutting back in executive pay - Ruiz' pay alone could mean the difference between a red or black quarter! And yet folks are blaming earnings on some silly price war (sarcasm)

Roborat, Ph.D said...

Due to the poor response of pro-AMD posters paralleling AMD’s poor response to Intel’s offering, allow me say something in defence of AMD. Firstly, when it comes to capacity expansion, I thought AMD has done a good job getting outsource Fabs qualified. I think this is a clever move considering AMD’s poor position and should shield them from underutilized Fabs in case of a downturn. Also, bear in mind that building Fabs to increase your capacity involves looking forward several years and making high risk decisions. 5 years ago you wouldn’t have imagined the gain’s AMD had in early 2006.

On the Dell business, I believe it was also a good move. The idea here is to gain a foothold on the commercial business to make AMD’s sales more stable and predictable. Having a predictable revenue stream allows you to be able to look further ahead, and maybe plan new Fabs with less risk. The volume committed by AMD to Dell is what’s questionable. Either way, one can easily understand AMD’s joy in getting Dell on board.

In general though, I have to agree mostly with what Mr Anon Y. Mous said.

Anonymous said...

"Also, bear in mind that building Fabs to increase your capacity involves looking forward several years and making high risk decisions. 5 years ago you wouldn’t have imagined the gain’s AMD had in early 2006."

This is a good point - however could you have started to see this 2 years ago? which is the time from breaking ground to initial wafers out. I think at end of 2004/beginning of 2005 AMD may have been able to see things a little better and while risky, perhaps not as high a risk as you indicate. And I will reiterate - they could clearly have gone with a more aggressive foundry approach which would have required SIGNIFICANTLY less time and much less capital risk.

"Having a predictable revenue stream allows you to be able to look further ahead, and maybe plan new Fabs with less risk."

AMD had a pretty good revenue stream prior to Dell (in fact some may argue it was better and more profitable). While the move will eventually be a good one, the timing was horrendous as AMD did not need the extra demand and simply swapped out channel sales with lower margin Dell sales.

Your last point about volume to Dell is a very good point! All AMD really needed was a small volume ar Dell across the various products - this would enable them to take the PR win and establish more credibility as well as establish a relationship with Dell that they could later add to if/when they do have the capacity. My suspicion though is that Dell would not bother to do a deal for a small volume as they work on economies of scale and it is not worth complicating their supply chain, validation, sales, etc.. if AMD were only to be a small part of the puzzle.

Though for AMD this probably would have been the best of both worlds as they could have remained loyal to the channel which made them what they are, instead of jumping ship to the prettier girlfriend when they started making more money. I think now they maybe starting to realize that the old girlfriend may not have been so bad, though it's difficult to say whether the old girlfriend will answer the doorbell when AMD comes knocking again!

Anonymous said...

An interesting article:

http://www.eetimes.com/news/semi/showArticle.jhtml;jsessionid=ZN01V0CU5EBIUQSNDLRCKHSCJUNN2JVN?articleID=198001579

Note - this is a fairly non-biased site that covers a lot of different aspects of the industry...as an example the other day they published an article/analysis asking if Intel was evil based pn their participation in the goring internet cafe market in China.

A few short excerpts for those who don;t want to cut/paste the link:

'AMD is taking prices down to levels the industry has never seen before in the low-end,'' said analyst Doug Freedman of American Technology Research Inc. ''We believe poor product mix and excessive channel inventory are limiting the impact of price cuts. We also believe that recent price cuts have been pushed out as they have proven unsuccessful in clearing the inventory or boosting top-line results.''

'...we believe ASPs remain pressured by AMD's hell-bent strategy on getting to 30 percent MPU market share (24 percent at present) at any cost and as quickly as possible,'

''We believe AMD double stuffed the channel in 4Q06 and January '07, resulting in excess inventory that even aggressive prices are unable to clear.''

Funny wasn't this being said by all of those AMD fans about Intel and P4? I wonder if any of these reports/articles will show up on Sharikou's site?

Didn't AMD report substantial unit market share gains in Q4 and then project market share losses in Q1? Sounds eerily consistent with the quote above about the stuffing of the channel, no?

Unknown said...

Anonymous, capacity doesn't just pop up because you decide your management is going to focus on it. And buying ATI is expanding. As commentors on this board have mentioned before, it takes 3 years to build a fab (possibly a little conservative?) and you still haven't ramped it.

Regarding how you seem to be hinting at what a good management team would have done, let's say that they decided to stick with the channel. They're still screwed because they have to drop prices to remain competitive. They'll lose market share as Intel floods the channel, and they'll lose respect for their product because it's not faster AND no one except channel customers use it.

By securing Dell, Founder, and IBM, AMD now has constant sources of income that will eat up all of their chips. Yes, they didn't adjust their inventory to guard against mis-orders, but they now have far more market penetration, and are now a valid brand to most customers. Yes, it's expensive and it is hurting them financially, but I challenge you to specifically describe a plausible situation where they actually come out better than they have now.

As for pointer's comment, the end consumer is AMD's market. To this day, a computer without an Intel processor is, to most consumers, a questionable purchase. They haven't heard of AMD, haven't seen someone who uses a computer with AMD in it, and don't know if they're "stable" or "reliable".

In reference to the lying comment, you are right, it's more of an ommission, and I was mainly trying to use my counterpart's argument against them (cause I was being to defensive). You're also right that that's not what you said, thank you for laying out the facts very clearly.

As to roborat, I have to say this is the most reasonable argument you have ever made, and have to wonder if someone hasn't used the little trick 180 outlined before to post in your namesake. Very good points though, and thank you for bringing them up.

Unknown said...

Also, not to try to look stupid for posting something linked off of sharikou, but this is actually a good article.

http://blogs.zdnet.com/Berlind/?p=366

Seriously... Find somewhere where AMD has done anything like that.

Anonymous said...

http://jonnyguru.com/news_details.php?id=206
Ring the alarm, price cut for the 64574th time!

Anonymous said...

By the way for all of those AMD fans who have repeatably stated 65nm was and F36 was on schedule here was AMD's original plan:

"AMD’s manufacturing leadership is further demonstrated by its technology collaboration with IBM, and AMD Fab 36, which is expected to be ready with 65nm manufacturing capability in mid-2005."

Source - AMD website, Jun2004 press release. This means they were over 1 year optimistic? And this was only a 1 year time projection.. Again this is typical AMD parsing of words - perhaps 65nm manufacturing capability meant 65nm DEVELOPMENT capcbility (but then again most of the development was done in Fishkill..)

http://www.amd.com/us-en/Weblets/0,,7832_8366_7595~86455,00.html

Anonymous said...

"As commentors on this board have mentioned before, it takes 3 years to build a fab (possibly a little conservative?) and you still haven't ramped it."

Intel F32 (AZ) - ground breaking in Aug/Sept'05 (meaning start digging into the ground). PRODUCTION wafers out by Q4 2007. I don;t have full ramp date but I would speculate full ramp by ~Q4'08 which would trash your 3 years just to build a fab timeline...

This is about 2 years (a little more) for initial volume wafers out and this also includes a fair amount of ramp in the factory - by end Q42'07 F32 will be capable of >1500 WSPW or ~7K WSPM. Keep in mind this is while Intel is also building and ramping other fabs simultaneously...

My point remains AMD's management has been very conservative when it comes to adding and ramping capacity, whether it be more aggressive foundry work, building another fab, extension to an existing fab or building F36 to a bigger overall capacity. It is because "capacity doesn't pop up over night" that I think AMD's management team has done a poor job planning in this area. I just looked through an AMD foil which said NY fab would be could be fully operational between Dec 31, 2012 - Dec 31,2014! WTF!

If you look at their potential construction start times, they are estimating ~5.5 years to build and ramp the fab!

Also keep in mind while total fab cost may be ~4Bil, year 1 costs are ~1Bil for shell construction (minus whatever subsidies they get which are typically front loaded) Total capital equipment spending is ~2.5 Bil which is spread out over the course of the ramp (in AMD's timeline this would be spread over 4 years).

So please don't come back with they didn't have the money arguments they just took out loans for what 2.5 Bil (or more?), they could have spread that same loan out over 2 years and had a fab that was pushing wafers out (allowing it to finance the rest of the capex spending) shortly? Or perhaps a bit earlier then say, I don't know, Dec 2012?

Source for AMD NY fab timeline(page 18): http://www.amd.com/us-en/assets/content_type/DownloadableAssets/Dec-06A-Day20BobRivet.pdf

Anonymous said...

"By securing Dell, Founder, and IBM, AMD now has constant sources of income that will eat up all of their chips."

Eat up I think is a good term. But I'm having a hard time remembering, please help me out here....before the Dell business (which was Q3'06 for DT/Mobile or maybe Q2'06 for servers), was AMD not selling out all of their chips, getting better income AND gaining market share? While your statement is technically accurate, it doesn't make the inverse statement inaccurate - without Dell, AMD still would have had a constant source of income and a market for their chips.

In my view AMD went from selling out all of their chips and earning a decent income without Dell, to selling out all of their chips and earning less margin with Dell. Of course it is impossible to tell how much of the margin hit was due to Dell and how much to the price war - but if you think the margin hit was only due to the price war you might want to look at Intel's Q3 to Q4 margin trend which was flat, while AMD's fell off the table.

So once again, I think AMD will need Dell EVENTUALLY if they truly want to expand by the numbers they say they want to, but they had absolutely no financial need for Dell in 2006 for income or sales of chips, in fact they would have been better off from a margin perspective not to be working with Dell in the short term.

My opinion remains they would have been better off doing this around Q2'07 when the bulk of their F36 65nm conversion is done and their capacity truly starts picking up (this would also would have allowed them to tease Dell with early/preferential access to Barcelona capacity to perhaps get a bit better price bargaining position with Dell). This would also have given them plenty of time to build the Dell relationship in advance of the late 2008/2009 potential fusion products which I assume will initially be targeted for commercial space and will need OEM's like Dell.

Anonymous said...

"To this day, a computer without an Intel processor is, to most consumers, a questionable purchase. They haven't heard of AMD, haven't seen someone who uses a computer with AMD in it, and don't know if they're "stable" or "reliable"."

Greg your comments are proving all of my points - why do you think the end user doesn't know AMD? More importantly how will they get to know AMD better? Like it or not marketing is almost as important as the product performance to get market penetration. Product performance becomes more important when trying to maintain the market or if from a brand perspective you are similar to your competitors (which is clearly not the case between AMD and Intel right now).
AMD has to focus more of their marketing on why their products are a BETTER choice than Intel (whether it be power, performance, cost, upgradebility...) as opposed to the Intel is cheating on benchmarks, they are bullying us, etc. While you need to address this, if this is AMD's sole focus they will remain in a reactive marketing scheme to Intel. I think the better by design is an attempt at this (though I question how successful it will be) - at least it's a step in a positive direction.

Think about how successful Intel's marketing is (we can have a long debate over whether or not this is a good thing, but like it or not is is HIGHLY successful). How many other products on the market do you buy where the manufacturer of a subcomponent is as important/more important than the manufacturer of the entire product.

How many people ask what the decoder chip is in a DVD player, engine manufacturer is in a car, LCD screen on a TV. While some folks may factor this into their purchasing decision I assume it it very, very small percentage. This is probably not a great analogy as the CPU is probably a greater percentage of the overall product performance/perception than the other examples but it does give you an idea of what Intel has done from a marketing perspective.

As a final example of a company that is perhaps the best (in my mind) at marketing look at Apple - I highly doubt most of their market knew that IBM/freescale used to make their chips or that people who buy Apple now even ask who makes the CPU inside.

Scientia from AMDZone said...

Anonymous

I think one thing the Sharikou, Scientia, et al, keep forgetting


Sometimes the ideas attributed to me are rather curious (and incorrect).

The assumptions are that yields are similar (if not better for AMD) because of APM3.0,

I would assume that yields for Intel on 65nm are better since they've been using the process longer. I've also said that AMD won't get the full benefit of APM until they have two similar FABs and this won't happen until mid 2008 when FAB 38 is up and running.

Intel's yields are actually better,

As I've said.

coupled with smaller die size

What smaller die size? Are you comparing Intel's 65nm die size with AMD's 90nm size?

Also the wafer production costs are just not the same.

Again, I agree. AMD has higher manufacturing costs. K10 increases this to 11 metal layers.

WAFER starts on 200mm compared to 300mm, 200mm continues to be at least 1/3 of the production.

As of the end of 2006, half of AMD's production was on 200mm. It won't be down to 1/3 until mid year.

but for a significant volume of wafers it is actually cheaper to BUY all new 300mm equipment than to re-use 200mm equipment.

AMD is selling the old 200mm equipment. Some is being sold to a Russian chip manufacturer.

Until AMD gets their 2nd fab fully converted, they will not be in the ballpark for costs.

As I've already said, about mid 2008.

And at the time they do this Intel once again will be 1/2 to a full technology node ahead

Really? In what way? AMD will begin production of 45nm chips at midyear as well. This would put Intel at most 1/4 of a technology node ahead.

Finally MCM is a much more cost effective approach to quad core and gives more manufacturing flexibility as the decision between dual and quad core can be made after the wafer is finished production

Very true. This is why AMD is currently looking at MCM for fusion. AMD cannot do MCM with the processors because they don't use a shared bus. This also means that Intel will lose its ability to do MCM when it switches to an IMC/CSI architecture in early 2009.

from a cost perspective there is no sane argument regarding MCM vs monolithic approach.

Again true, but Intel will likewise have to give up MCM when it drops the shared FSB.

In my opinion, AMD's only chance is to have a significantly superior product which will allow them to set the pricing

This is not the case. Your view as 180's is a bit too pessimistic. AMD (as 180 is conveniently forgetting) survived being behind all during 2002 and into 2003 when K8 was released at low volume (with low yield). The span this time is less. To anyone with common sense, the situation today is better than it was in late 2002.

the only place where AMD will be able to set the price is in the 4P+ space

Not true. AMD should be able to set its K10 chip prices slightly higher than comparable Woodcrests and Clovertowns for 1P, 2P, and 4P.

The NY fab is probably irrelevant, by the time it would be contributing any capacity it will be late 2009/2010.

You are way too optimstic. It would be impossible for the NY FAB to be ready in 2009. Think early 2011 if construction starts in 2008.

AMD is better off expanding Dresden with another fab as they have expertise and economies of scale in place there.

You are failing to consider that AMD might do another interim expansion at Dresden even if they pursue the NY option. FAB 38 will be maxed out by end of 2009 and this would mean no additional capacity for more than a year.

The downside is from a risk mitigation perspective - if anything happens there (natural disaster, etc), AMD is sunk.

AMD has run with a single FAB for most of its existence. The risk is no greater now. The threat of a natural disaster won't be part of the decision. The actual decision will be based on cost and benefits from location. Remember that the NY offering of 1.5 Billion was substantial.

expanding it's relationship with Chartered

Chartered production is moving to 65nm and will increase somewhat in volume in 2008.

and or start working with UMC/TSMC.

This is not possible. UMC and TSMC use different process technologies than AMD/IBM/Chartered. This would mean that AMD would have to spend a great deal of time tweaking its design to fit what the foundry uses. Secondly, neither TSMC nor UMC is actually capable of producing AMD's 65nm chips. The best you could currently do would be to move 90nm production to an 80nm foundry process and this would be pointless since it would cost more than continuing production on FAB 30.

It's also not clear where the graphics production (both IPG and discrete) is happening - does anyone know if they are still outsourcing this?

ATI's graphics are still being produced at both TSMC (primary) and UMC (secondary).

I remember reading somewhere where AMD said they will start doing this inhouse in 2007

No. AMD has never said this. AMD won't manufacturer any GPU hardware any earlier than 2008. And, this would only be for MCM or GPU on the same die. The separately packaged chips will continue to be foundry produced.

BTW, why are you posting anonymously? Are you too embarassed to take credit for your post?

Scientia from AMDZone said...

anonymous
As a final example of a company that is perhaps the best (in my mind) at marketing look at Apple


Really? Then ask yourself why the average Mac owner is in their 40's. Apple's marketing is not nearly as successful as you think. In reality, the same people who used Apple II's in school bought Mac's. And, the same people who bought the first Mac's are still buying Mac's today. Apple's fortunes can change very quickly if it doesn't create true cross generation appeal. Presumably, this is what the current coffee house hippee commercials try to do.

Ho Ho said...

scientia
"Very true. This is why AMD is currently looking at MCM for fusion. AMD cannot do MCM with the processors because they don't use a shared bus. This also means that Intel will lose its ability to do MCM when it switches to an IMC/CSI architecture in early 2009."

...

"Again true, but Intel will likewise have to give up MCM when it drops the shared FSB."

So what will be. Does AMD move to MCM with fusion and it is possible to have MCM with IMC or not? If it is possible then why should Intel loose its ability to oroduce MCM cips with IMC?

Anonymous said...

Scientia - in YOUR OWN BLOG you have stated that you assume the 2 65nm processes are approximately the same cost:

"Now, some corrections. Intel does not have a lower cost product. On 65nm AMD's and Intel's costs are about the same." (Comments in "who won Q4'07 blog", 2/2/07, 4:57pm) Of course there was absolutely no data or links to support this statement. I'm also am not sure if you were comparing 65nm finished wafer cost, or factoring in yield, die size... (in virtually all cases, unless AMD's die size are significantly smaller, Intel will be marginally to significantly cheaper on both a finished wafer and good unit basis)

"AMD is selling the old 200mm equipment. Some is being sold to a Russian chip manufacturer."

200mm resale equipment will be lucky to get 20cents on the dollar, there is no warranty, no support and generally no performance guarantees. A typical equipment supplier will guarantee some performance spec, provide 1-2 year warranty on all (or sometimes just non-consumable) parts and will generally provide install/ qualification personnel, training, manuals, and sometimes up to a year of field service personnel - all of this an IC manufacturer who is reselling equipment cannot do (except maybe the manuals?).

I go through all of this as I think you will likely try to challenge the 20cents (again this is most likely BEST case) on the dollar estimate. This resale will not raise the amount of money everyone thinks it will - F30 has less than $2Bil of new equipment, which means at best they will get <400Mil for it. This is nothing to sneeze at, but it is not going to impact the bottom line a lot.

"K10 increases this to 11 metal layers."

For you and the rest of the folks - the number of metal layers is NOT PRODUCT DEPENDENT IT IS TECHNOLOGY NODE DEPENDENT! The # of metal layers is based on the 65nm (vs 90nm) node not K8 vs K10. This shows me what kind of background you have in Si technology and manufacturing and allows me to dismiss many of your process technology statements like:
- bulk Si fails more quickly than SOI (?)
-65nm costs between Intel and AMD are about the same (?)
- Intel only needs to find a high K material with a suitable bandgap (?)
- From what each company has said it looks like the tri-Gate will not be able to be shrunk while AMD's finFet will be (?)
....especially when you provide no supporting data or even some technical reasoning/opinion to support them.

"The best you could currently do would be to move 90nm production to an 80nm foundry process and this would be pointless since it would cost more than continuing production on FAB 30."

I was talking about ADDING capacity, not replacing existing fab capacity. I understand it takes time to port a process file over to a different manufacturer, which is why I think AMD should have explored this approach a long time ago (of course they may have done this and found it not to be cost effective). Of course now it is too late and they should just focus on ramping/converting existing fabs. Yes it means maintaining potentially 2 different designs for the same part, but this is the trade off vs the capex cost/risk of building a fab when you don't have the income. Frankly I don't know which is better financially (it is obviously easier just to do everything in house).

You say it would take a while - but AMD is forecasting added NY fab capacity would be fully up in 2012-2014 (analyst day foils), it's not like I'm trying to say they'd be able to put something out within a year - my point (taken out of context) was back when AMD realized they had a superior product (2004) they could have considered other means of increasing capacity like porting the design to a foundry realizing they would have limited capacity for the forseeable future. Risky? Sure. Economically viable? Don't know. Did AMD look into this? Maybe.

BTW chartered production is peanuts compared to overall AMD capacity. So saying this will somewhat increase is not saying much as it is so little to begin with.

"To anyone with common sense, the situation today is better than it was in late 2002."

Yeah I guess the amount of debt back in 2002 was WORSE than today? They also have to carry ATI operations which right now is at best breakeven (might break into the black in 2007). I'm not going to sit here and yell BK like an idiot, but unless K10 is significantly better than Core2, AMD will have to react to any changes in Intel pricing (again in under 4P space). If it is similar enough to Core2, then Intel can just turn the pricing screws whenever they want. At this point Intel's gross margin and cash position is far better than AMD's so they are in a better position to do this than AMD. If K10 is significantly enough better than AMD may not have to couple pricing and price cuts to Intel's...

Keep in mind while further 65nm conversion will help, AMD is still moving to greater dual core mix, and intro of quad core - the larger aggregate die size will offset a lot of the technology shrink gains.

The problem I have with your statement is you are looking at things mainly through a technical roadmap perspective and are ignoring the financial side - you are assuming AMD was in a similar financial situation in 2002 as they are today. While the product roadmap may be in a situation similar/better than 2002, the financial situation is clearly not ("to anyone with common sense").

And look at more than just earnings outlook... debt/equity, capex needs, interest payment on debt, free cash flow, declining bond rating, etc...

"Again true, but Intel will likewise have to give up MCM when it drops the shared FSB."

You continually state 2009 as if this is some sort of fact - this is an opinion and as of right now wrong, as reported Nehalem schedules are for H2'08 which will have CSI and IMC. Also Intel is not moving to CSI/IMC across the board - the speculation is that it will be server and maybe some enthusiast parts (I think this was Inq, so take it with a grain of salt). If so, the other parts can continue to use this approach (if needed). Regardless of all of this Intel can continue to milk the MCM cost and manufacturing advantage over AMD for the next 1.5-2 years.

On a side note - I am frustrated to keep hearing Intel fans say, wow when Intel introduces CSI/IMC they will blow AMD away as this is AMD's main technical advantage over Intel - unless these folks are running 2P+ servers I doubt they will see much of a difference (except maybe overall power). The Nehalem architecture will hopefully yield a significant benefit but I don't see the CSI/IMC part of the architecture doing much in "1P" space.

"BTW, why are you posting anonymously? Are you too embarassed to take credit for your post?"

Embarassed? It's not like I think the number of metal layers is product dependent, or say that I never stated something (except in your own blog!). Let's be real here - everyone who is posting on this blog is effectively anonymous.

So to answer your question - just lazy. It's fairly simple to hit publish your comment button....Also don't need any more junkmail and another password to remember... (I don't like the remember my password feature on firefox). If Sharikou180 thinks my comments are inappropriate, I assume he will not post them - I don't see the relevance of this question.

My only real significant error was thinking AMD would start doing IGP/DGP in house in 2007 - I looked back over the Ruiz interviews during the acquisition and he stated he would keep this based on foundry for forseeable future. Not sure where I thought I had read that this would be brough in house sometime in 2007 - I was wrong.

Anonymous said...

For those looking for some comic relief - you should head over to Sharikou's blog and pull up some of the Q1'06 blogs in the archives...

As if 40% market share(runrate) wasn't bad enough - the original prediction was 55%! And Shari-kook had the capacity calculations to support AMD being able to produce that by end of 2006!

Conroe was a "temporary solution"

AMD was winning the price war, based on the Top 5 NewEgg chip sales!

"AMD to rule 2006 full year unchallenged"

"I expect AMD to exit 2006 with 45% server market share, almost triple the current 16.4% market share."

"AMD stock is under valued"

Expected Barcelona to be launched in 7 months (this was from an August post) - I guess he's still got a few days left in the month on this one...

...and of course the Intel quarterly operating losses starting in Q3'06 leading up to the eventual BK in 2008.

Again if you are looking for some comic relief from whatever you are doing it is rather funny to look through his older predictions.... and as he continually reminds us his "predictions are almost always right"!

Anonymous said...

"Really? In what way? AMD will begin production of 45nm chips at midyear as well. This would put Intel at most 1/4 of a technology node ahead."

Man, you should go to work for politicians - you put Intel's 45nm product OUT into 2008 as Q3/Q4 would only have small amount of production. You then compare this timeline to the earliest possible AMD 45nm "begin production" timeline.

If Intel is trickling out product in late Q3/Q4'07 and AMD is trickling out product in Q3'08, this is about the same as 65nm... in my opinion this is the best AMD can hope for (actually the best case wold be Intel can fall down and slip 45nm out, but I think this is unlikely)

I love how the lead has now shrunk to "AT MOST" 6 months (as 6 months is a 1/4 of a generation). 45nm "mid 2008" seems to me should be interperted the same way Barcelona "mid 2007": paper announcements/ shipping in Q2, actual product in Q3.

By your statements, if Intel has product out at beg Q4'07, AMD will AT WORST have product at beginning of Q2'08 (if not earlier since this is worst case and you said AT MOST 6 months). You must either think Intel is slipping their 45nm schedule substantially or AMD is pulling theirs in even further.

Here's a bit of timeline for you from a manufacturing perspective: Assuming equipment is ordered and built and sitting on a dock outside the fab....
- 3-6 months for equipment installation and qualification,
- min 3 months for 45nm process qualification (I'll assume most of the work is done at IBM),
- 2 tapeouts (best case) would mean another 3-6 months
- actually product production ~4 months start to finish when the tapeout product has been validated...
- Also generally speaking when initial technology is released a more extensive product burn-in is done to validate the technology node (this is generally accelerated lifetime testing to validate the lifetime of a chip and operating range of the chip - not sure if this is on the order of weeks or months)

So we are talking ~13-19 months - they might shave 3 months or so if they can do the tapeout work at IBM and/or move the lots thru the fab quickly by hot boxing them which is typically done for initial steppings, they also could risk-start production wafers while the final stepping is validated but this is fairly risky (if there is a problem with the validation then you would end up scrapping all of the risk starts).

So if AMD started installing 45nm equipment TODAY, the absolute best case they could do would be wafers out in ~Jun 08. This assumes very aggressive process qual times, minimal # of steppings, and of course the equipment is in transit as we speak, and no slips issues at ANY of the steps outline above.

What are your thoughts on the timeline, since you seem to have some specific insight into this area? Or are you just regurgitating things you have read on the web and creatively interperting them in the most favorable light to AMD?

Ho Ho said...

anonymous
"On a side note - I am frustrated to keep hearing Intel fans say, wow when Intel introduces CSI/IMC they will blow AMD away as this is AMD's main technical advantage over Intel"

For one thing it would give Intel much shorter memory latency. Having that it could also decrease its cache sizes which would bring either more cores or cheaper CPUs. IMC alone won't do anything but it will help in quite a few areas.

Anonymous said...

"Then ask yourself why the average Mac owner is in their 40's."

I love "facts" pulled out of magic hats. I guess iPod with 90 percent of the market is a result of marketing failure.

Scientia from AMDZone said...

anonymous

I'm not going to move the discussion wholesale from my blog to here; that would be silly. Why didn't you post these questions on my blog? And, why are you posting anonymously?

Anonymous said...

"Really? Then ask yourself why the average Mac owner is in their 40's. "

Does Apple only make Mac's? Why is the IPOD/itunes so prevalent? Is this a technologically superior product to a Samsung MP3 player, Zune or any other?

So I understand your argument - good marketing means selling to young people? (what the hell does age have to do with anything?).

The reason I say they have good marketing is that they are able to sell a product that is technically very similar to their competitors at a higher price than their competitors. There are basically two reasons you can do this - your competitors can't supply the entire market (which I doubt is the case for both PC and MP3 players) or you have some brand value where people are willing to pay more.

Ask yourself this - why do Mac's sell at all? What about iPODs? Do either of these offer better price/performance ratios?

Also exactly how is Apple only selling Mac's to people who previously owned their products, if their market share went UP? Are these "40 year olds" buying 2 or 3 computers each? Those must be some real good commercials... :)

You may not like Apple (I personally don't like them very much), but for you to say they don't have good marketing (especially with the reasons you gave in your post) shows your ignorance and/or lack of objectivity in this area.

Anonymous said...

Scientia from AMDZone said...
anonymous

I'm not going to move the discussion wholesale from my blog to here; that would be silly...


actually, counting reader comments is silly. you seem to be obsessed about it. I've seen you post anonymously at Sharikous blog promoting your new blog months ago. And i'm quite sure, it must be bothering you right now that Skarikou180 is getting more comments than your last blog post. Especially after you blasted his site as "only getting 4-6 comments per post".

Maybe your readers are starting to realized that your post have been completely off the mark.

Anonymous said...

"BTW, why are you posting anonymously? Are you too embarassed to take credit for your post?"

"So to answer your question - just lazy. It's fairly simple to hit publish your comment button....Also don't need any more junkmail and another password to remember... (I don't like the remember my password feature on firefox)."

"And, why are you posting anonymously?"

Someone stop the broken record... 180, you should block profile links. All these fanbois want to do is spew crapful of propaganda for ad clicks. Maybe they would have some sense if they didn't make no money.

Unknown said...

I'd just like to point out (because I don't feel like making a real argument) that most of the argument brought up against mine completely ignore the fact that AMD knew it wouldn't have a new process until 07, knew it needed to have an actual platform, couldn't invest in its own graphics segment and have one in time to compete, couldn't have a core to use for fusion (the only way it can realistically stay ahead of Intel in terms of innovation), wouldn't have a segment of the market place that isn't highly concerned with performance already secured, wouldn't have as much demand from the channel as they used to, and would still have tons of uncapitalized demand in the form of the previously mentioned companies. Is that enough of a problem?