AMD goes Asset Smart; splits into two

Tuesday 07th October 2008, 02:55:00 PM, written by Arun

In a move that has been worked on for more than a year, AMD has now divested its foundry operations as a separate company. It will be owned 55.6% by ATIC, a subsidiary of the Abu Dhabi government. In total, AMD's liquidity position will improve by $2.4B while massively reducing capital expenditures.

Despite the deal including the creation of new shares and warrants worth $314M, diluting shareholder value, the stock market is reacting very positively to the news: AMD is up nearly 20% as of 11AM ET. And rightfully so, it seems to us: this completely eliminates any risk of AMD running out of cash or eventually having to default on its debts. AMD's mid-term fortunes still depend on the quality of its 45nm CPU line-up and eventually on its upcoming Bulldozer architecture (in 2010?), but this should give them a solid footing to invest in the future.

A number of other data points and implications seem noteworthy:

  • AMD will exclusively manufacture its CPUs at this new entity, putting TSMC and Chartered off the table completely for microprocessors. However, it seems plausible that Abu Dhabi might eventually want to buy Chartered too and integrate it. This would be a big step in the creation of a true IBM Common Platform-based foundry giant, a clear alternative to TSMC for leading-edge foundry work.
  • The new company will "join the IBM joint development alliance for both silicon-on-insulator (SOI) and bulk silicon through the 22nm generation", indicating that AMD might be interested in not solely focusing on SOI-based solutions in the future. Alternatively, bulk silicon might be reserved for other foundry customers.
  • A further $3.6B to $6.0B will be invested in the next 5 years, in order to convert Fab30 into Fab38 in 2009 and start building the New York fab in the middle of 2009. Also, "The Foundry Company envisions expanding its global manufacturing footprint over time, if commercially justified, to also include new fabrication facilities in Abu Dhabi".
  • This might also have interesting implications for next-generation game console contracts. Not only could this new entity compete for contracts involving AMD, but also IBM and even NVIDIA.

IBM's disclosures about its upcoming 32nm High-K process have been received very well by the technical press, and so this is certainly a promising new venture. Right now, it remains overly dependent on AMD's success to be profitable, but in the long-term it should become a very credible alternative to other foundries for leading-edge work. In the current economic conditions, this move should reassure both investors and technology enthusiasts that AMD is there to stay.
 
AMD's long-term success depends on the performance and cost-efficiency of its upcoming Bulldozer architecture as well as the overall pricing trends in the microprocessor industry, but for now the company is back in a stronger position than they have been in for quite some time.


Discuss on the forums

Tagging

amd ± foundry


Latest Thread Comments (23 total)
Posted by 3dilettante on Tuesday, 14-Oct-08 14:36:18 UTC
Quoting WaltC
Late by whose yardstick? Intel was years "late" with Core 2, but then that only applies if you use AMD's product yardstick to measure lateness. I don't. Core 2 may have been very late to the party, but I'd conclude that had much more to do with Intel's own internal scheduling and priorities and capabilities than it did with what AMD was doing at the time.
AMD planned Bulldozer in 1H 2009. That's not the release date now.

Quote
The point is that "being late" is a purely subjective observation and as we've seen with both companies, sometimes tardiness doesn't matter much over the longer haul if what you're tardy with happens to be a compelling product.
If you say one date and that date is pushed back, you are late. Unless you want to render human language useless.

Quote
The markets don't seem to have a lot of problem in buying such products whether they are "late" or not.
Only true if a company is taken in isolation, to the exclusion of the execution undertaken by its competitors.

Quote
I think that practically speaking, we need to wait and see what Intel is going to be able to ship in quantity when it comes to Nehalem.
Nehalem early silicon has been live demonstrated and sampling for months now.
The design itself appears to be on firm standing.
The manufacturing is now being done on fabs that have been churning 45nm for quite some time.
The quantity will be somewhat limited, since Intel is delaying the release of the higher volume mainstream chips until 2009, and there will be some time before there is a crossover between Penryn-based and Nehalem-based cores.
Intel may be hindered by the current consumer market conditions, and the incremental single-threaded gains Nehalem's early SKUs will have over the highest-clocked Penryns in less optimized applications. That being said, Nehalem does cover a number of weaknesses in Core2 that should shine in a number of multimedia applications.

Quote
I recall that just a few months before Intel scrapped the P4 and moved on to Core that the top dogs inside Intel--Otellini, especially--made lots of public statements about the 4-5GHz P4's that *would be shipping* before the end of the year.
...
There's a good lesson there about the dangers of taking pre-release product publicity too seriously. Sometimes it proves worthless in the end.
Intel was called on the its P4 debacle.
My evaluation of Nehalem comes from the design information given out, not the hype.
It's a very sound design.


Quote
Here again, you are making all sorts of assumptions that Intel is going to be executing everything perfectly and perfectly in line with your chronological expectations--and you are also assuming that during that time period AMD won't be shipping much of anything and any kind of timetable...;)
The former has in the last few years done just that. The latter in the last few years has (or more properly, hasn't) done just that.
Call me conservative by going by recent history.

Quote
Bottom line is that we won't know how competitive Shanghai will be until Shanghai ships and can be contrasted with the shipping products from Intel at that time, whatever they may be.
The general outlines of the designs are pretty well known. We can get a very good idea of how it will do, just as we had a pretty good idea of Conroe vs. Barcelona 6 months ahead of time.
We had this argument then.
The amount of magic pixie dust in system design is rather low.

Quote
AMD has historically had nowhere near the fiscal latitude to be so reckless with its investments and its endeavors, so the fact that AMD doesn't burn up billions of $ trying to create market standards the market doesn't want should surprise nobody. Only Intel has the money to waste it like that.
Incorrect, AMD massively overpaid for ATI.
Or perhaps correct, AMD didn't have the money, so it dug itself massively into debt by spending money it didn't have to burn.

Quote
I wouldn't call it "quality"--I'd call it profligate wealth the great majority of it earned on the back of the x86 high-end monopoly Intel held for so long--a time before AMD's K7/K8 and beyond. Try the mental exercise of fiscally trading positions between AMD and Intel. Think Intel could have done as well as AMD if Intel had been constrained to AMD's budget? No way...;)
If the money and fabs don't care about the moral quality of the money flow, I'm not going to either.

Quote
Except for someone who's an Intel stockholder, how does this scenario benefit the market at large?
Irrelevant to the point being made. Intel's expenditures and efforts had an impact on the competition, even if Itanium struggled.

Quote
Isn't the idea behind the mythical Larabee essentially GPGPU, at its core?
It's many-core x86 with extended vector capability. It emphasizes a readily accessible memory model, a more predictable and well-understood processor pipeline, and massive amounts of Intel programming help.

Quote
Late might not be the proper word to use since Larabee isn't shipping yet and when it will nobody knows, not to mention that nobody knows just what precisely Larabee will be when it ships.
Larrabee is known to an increasing level of precision with each passing disclosure.
16-32 cores running 1.5-2.0 GHz, each with 4 threads, 32 64B vector register per thread, 32KiB L1, 256 KiB L2, 1024 bit ring bus, and at the top end well over 100 GiB/sec in memory bandwidth.
Each core's non-vector units are based off of the Pentium core.

The listed specs indicate Larrabee can muster 1/2 DP throughput per clock, and it will be more IEEE compliant than the GPUs. This is great news for HPC, not so much for consumer graphis.

Quote
What? The FABs will be manufacturing x86 cpus. What are you talking about? Do you think they are going to use the foundry to make go karts??? Abu Dhabi, of course, isn't interested in "designing x86 processors"--it's interested in *making them* in its jointly owned FABs with AMD. AMD will be the company doing the x86 designs that the FABs will be manufacturing.
Abu Dhabi in interested in owning a *foundry*, which means pulling in other customers. The more successful it is at that, the worse off AMD's x86 process will be.

Quote
Who's beating their way to the Intel FABs besides Intel? ?
Nobody, and with good reason.
The processes and economics the CPU fabs use are very hostile to the kinds of designs the bulk of the foundry market's customers want.

Quote
Yes, but that distinction applies equally to Intel as it does to anyone else.
Intel's not fabbing designs for foundry customers that will care.

Quote
You know, the thing about yields and FABs is this: the work at a FAB is an ongoing process of improvement and debugging. Every complex chip I've ever seen starts out initially with less-than-adequate yields, but then the process manufacturing is tweaked and improved to the point where the yields become satisfactory. This is true for Intel and AMD alike.
AMD and Intel accept mature yields that are significantly worse than those of a foundry because they sell comparatively high-margin chips that make up for the shortfall.
AMD's fab company cannot assume this if it wants to steal business from other foundries.

Quote
Says who?...;)
I guess just posts from professionals on the net and apparently those who shifted as fast as they could from IBM's foundry efforts, which is the same base process.

Quote
The point of the money is that not that you have so much of it you are constantly throwing it away down blind alleys just to be able to spend it--the point is rather that you have *enough* money and the judgment to use it wisely so that you throw away as little of it as you can possibly manage.
And AMD's done so well over the last year plus of hundreds of millions of dollar shortfalls per quarter?

Quote
Williamette was a disappointment compared to K7...;)
And to the P3. The call for those who wouldn't go AMD was wait for Northwood, which was a worthwhile wait.


Quote
Northwood was a temporary process advantage over Athlon. Just a soon as Athlon's process problems were solved (took AMD but 6-8 months IIRC), Northwood was left in the dust. Funny how people remember when Northwood ran ahead of Athlon, but seem to have forgotten that later Northwood fell behind Athlon.
Nonsense. Northwood was the reason for AMD's insipidly over-inflated performance ratings.
K8 was what brought things in line, and even then it was not a decisive win until Prescott came out.
The big win came with Opteron in the server market, where the base architecture was much better suited. At the desktop, it was a closer fight for a longer time, and on the laptop the Pentium M was a very strong hint of what was to come with Conroe.

Quote
Prescott was mainly a screwup because it didn't come close to living up to the 18-months of pre-release HYPE Intel lavished on it.
It was an overly large and hot chip that took the P4's penchant for hyperpipelining and extreme speculation to the furthest extent, and because of thermal problems couldn't even fully match Northwood.
The thermals were enough to counteract a number of enhancements Prescott used to lessen some of the numerous P4 glass jaws.

It would have been unimpressive compared to what came before it, even if AMD weren't there to capitalize on it.

Quote
Very late Intel figured out that the market wasn't going to go Itanium to get to 64-bits, and so Intel literally borrowed from AMD's x86-64 in a cross-license agreement, and Core 2 was born. A very good cpu. AMD has its work cut out for it--but then again this has always been the case for AMD, and I think it knows what needs to be done and will do it.
You must mean Prescott, since x86-64 came to Intel first with that architecture.
Core 2's heritage came from Merom, a core whose integer performance was such that people were bunging it into desktops.

Quote
But Intel currently sells 3d integrated graphics that also suck behind both AMD and nVidia's offerings, too.
And it outsells them by a massive margin. Who knew?

Quote
Where has that gotten AMD? Well, for starters I'd say its gotten the company billions in revenue it would not have had otherwise, and it greatly enhances the platform-centric competitiveness of AMD versus Intel.
Where are the profits? The graphics division might eke something out this quarter, but the overall picture is of mild financial underperformance. It would be survivable, if it weren't saddled with billions in debt.
Unfortunately, the graphics portion of the equation is bogged down by the CPU side.

Quote
Unlike Intel which will spend huge sums for the ignoble, express goals of trying to force AMD to take a bath, AMD doesn't have enough money to waste to respond in like fashion, and so AMD is forced to redeem itself on the basis of the quality of the products it sells.
Debate on points of design, architecture, economics, and long-term plans.
Resorting to moralistic terms is a waste of time. Neither Jesus, Allah, or Buddah is on AMD's board.
If you wanted to go for the x86 manufacturer with old-time values, go buy a Via Isaiah.

Quote
Ah, but this deal has stemmed to flow of red ink and leveraged AMD into a far better position to compete with Intel mano 'y mano...;) You'd best forget about the past and look to the future because that's where the action will be.
What's been promised. A flat-lined capex spend and the shoehorning of general foundry fab processes at the expense of the x86 line?
CPU designs that have not materialized, platform evolutions delayed?
I'm going off a list of AMD delays. Where's Sandtiger, Bobcat, Montreal, HT3, HiK and metal gates?
When's 32nm?
How long can they count on IBM's lead?
Whither SOI?

Quote
Or, it could be that AMD wouldn't sell much more of AMD to Abu Dhabi, and that Abu Dabhi took what AMD would sell it og the AMD FABs--and as Abhu Dabhi agreed to AMD management of the FABs and to sharing 50-50 voting rights with AMD inside the FABs--maybe it's Abhu Dabhi who is so keen on AMD's acumen and experience that Ahbu Dabhi is content to take a ground-floor seat and hang around for the ride that no doubt Abhu Dabhi expects to catapult it to the top of a far more interesting market.
If AMD gave up more than 50% voting rights on the fab, it would have been slammed by the cross-licensing agreement.


Quote
What's certain to me is that these investments into AMD were made in absolute confidence in the company and its leadership and its talent. And, too, Abhu Dabhi is prehaps optimistic about AMD's prosepct of putting a serious dent into Intel in the years ahead, and so considers AMD's game eminently workable and doable.
How about the fact that even AMD's implosion would not put much of a dent in a sovereign wealth fund estimated at near a trillion dollars.
They get the fabs, regardless.

Posted by Florin on Tuesday, 14-Oct-08 15:48:52 UTC
Lots of reasoning based on wishful thinking in this topic. It's always entertaining to see the classic images of a noble smaller company vs the ruthless giant scenario being conjured up.

Let's face it, asset smart really is too nice a word for all of this. Clearly, it's the destruction of value that we've seen with the acquisition of ATI that was the catalyst for what is more like an act of desperation. We can only hope that this will be enough to guarantee AMD's continued survival as a budget alternative.

Posted by WaltC on Saturday, 18-Oct-08 00:14:43 UTC
Quoting 3dilettante

If you say one date and that date is pushed back, you are late. Unless you want to render human language useless.


Only true if a company is taken in isolation, to the exclusion of the execution undertaken by its competitors
It's sort of sad, really, to see you side-step the real reason behind most of AMD's financial problems of the last few years. I wish it was true that Intel's technology had been so much better--that, at least, would be understandable. But unfortunately it's a global matter of record that on numerous occasions with dozens of companies around the world over a span of years that Intel has used its money not to compete with AMD, but rather to pay OEMs either *not* to buy anything made by AMD or else to buy only a very small, non-competitive fraction of AMD's goods--in return for very lucrative Intel subsidies of one kind or another.

That's always sounded very desperate to me on Intel's part--which is what comes of Intel's being paranoid to a monomaniacal extreme I suppose. Yet there's no doubt that it happened consistently and with purpose and pattern, and over a protracted span of time. AMD, however, has had to compete solely on the basis of its technology because it lacks the funds to try and close off Intel's markets the way that Intel has paid to shut AMD out of AMD's markets.

That's the real survival story of AMD, that despite every moral and immoral roadblock Intel has placed in its path AMD is not only surviving, but actually growing. Intel's greatest fear, I imagine, is that AMD will find its way and simply become impossible to stop. I think that point has already been reached, although it isn't clear that Intel has realized it at this point. With every single x86 competitor Intel has faced historically its tactics have been to drive the smaller company out of business, and failing that, then and only then to belatedly and often half-heartedly compete. So far it looks to me as if AMD has survived the first stage--comparatively, then, the second stage should be a piece of cake...;)

Posted by Acert93 on Saturday, 18-Oct-08 10:07:57 UTC
That sounds like a rose-filtered image of AMD. e.g. The financial issues AMD has faced the last few years has been because Intel has continued to have a strong edge in the mobile market and C2D nailed a lot of the popular AMD bulletpoints (performance, performance/watt) for the consumer market. In the past AMD has also had difficulty delivering products due to a variety of reasons, including capacity issues. And AMD's current woes are also tied to investing a large sum of resources ($5B?) into ATI. AMD's execution, and targetting the right markets at the right time, has been spotty and surely needs to be considered when talking about their financial issues.

Posted by 3dilettante on Monday, 20-Oct-08 13:35:55 UTC
Quoting WaltC
It's sort of sad, really, to see you side-step the real reason behind most of AMD's financial problems of the last few years. I wish it was true that Intel's technology had been so much better--that, at least, would be understandable. But unfortunately it's a global matter of record that on numerous occasions with dozens of companies around the world over a span of years that Intel has used its money not to compete with AMD, but rather to pay OEMs either *not* to buy anything made by AMD or else to buy only a very small, non-competitive fraction of AMD's goods--in return for very lucrative Intel subsidies of one kind or another.
The confirmed and uncomfirmed pressures from Intel probably had a contributory role to AMD's predicament by constraining its ability to raise prices as much as it wanted in the days it was capacity constrained.

It is insufficient to explain AMD's full position, considering it survived for years as an x86 also-ran prior to K7, when it actually had a measure of restraint and some good fortune.

OEMs had a number of non-Intel reasons to be afraid of letting in AMD:
reluctance to have to design and qualify two separate CPU lines, AMD's slow rate of chipset design (when it made them at all), issues in quality and stability for third-party chipsets, AMD's lack of history in those markets, AMD's lack of capacity in meeting the volumes of the large OEMs, AMD's lack of reliability.

The alleged Intel abuses AMD is pursuing them is by AMD's own briefs actions that have curtailed years ago.
Let's note where AMD went when Intel's pressures eased and AMD's platform looked well-established and AMD's manufacturing seemed competitive.

AMD got some large OEM contracts, and then promptly struggled to meet them.
AMD, by its own admission, screwed its channel partners over in trying to woo low-margin OEM accounts.
AMD's design leadership floundered on the replacement for K8, and it rushed Barcelona through a razor-thin validation period that would miss an issue that brutalized AMD's reputation and hobbled its highest-margin sales for half a year. Recent converts to the AMD team got nailed. Sun, Cray, IBM, HP, Dell, and whoever else were left hanging.
AMD overextended itself in a massive purchase that in the end forced its market valuation at less than the purchase price.
AMD's process was lackluster, AMD's design was lackluster, and even by AMD's own admission in some of its court documents, its cost structure was inferior.

AMD, if it were in steadier hands, would have had a cash cushion to survive Barcelona without endangering the critical link between its fabs and CPU designers.

Quote
AMD, however, has had to compete solely on the basis of its technology because it lacks the funds to try and close off Intel's markets the way that Intel has paid to shut AMD out of AMD's markets.
And now AMD doesn't compete all that well on the basis of technology. Now what?

Counterexample: IBM has fallen back in several markets under Intel's product assault. IBM pours hundreds of millions of dollars into its microelectronics division and can afford to take a loss because of services.
Intel does not have that option.
Just which critical parties has Intel arm-twisted to shove IBM aside when IBM has a much longer history and far greater success in strongarm tactics?

Perhaps AMD's greatest hope for a fab customer is the hope that IBM gives up the ghost on its microelectronics business and tosses it to the Foundry Company, since IBM might be the only customer who will have any real interest in a process amenable to high-end x86, given that IBM made the process in the first place.

Quote
That's the real survival story of AMD, that despite every moral and immoral roadblock Intel has placed in its path AMD is not only surviving, but actually growing. Intel's greatest fear, I imagine, is that AMD will find its way and simply become impossible to stop. I think that point has already been reached, although it isn't clear that Intel has realized it at this point. With every single x86 competitor Intel has faced historically its tactics have been to drive the smaller company out of business, and failing that, then and only then to belatedly and often half-heartedly compete. So far it looks to me as if AMD has survived the first stage--comparatively, then, the second stage should be a piece of cake...;)
I will try to answer this with prose that is far less beautiful, though in my mind is far mor appropriate for the position AMD has found itself in.

AMD's business model is unsustainable. This was apparent years ago.
It's run a long race as a smaller lower-margin device manufacturer of an engineering-intensive and commoditized product manufactured in an in-house fabrication method that is capital-intensive and subject to rapid obsolescence.

It's not even clear that the model in the long-run is one even Intel can keep up, but it is currently significantly better positioned to survive the commoditization of x86, given its economies of scale and more secure financial status.

The split is not a massive win for AMD. The current plan at best will allow AMD to mostly keep the curtailed promises it made years ago. The fact that such a massive change was a wash means that AMD was in very serious trouble.

The interest of the investment group that owns most of the fabs is not aligned with the x86 design group. The more successful the foundry becomes, the less tied AMD will be to the fab process.
The size of the foundry is small. Dresden is going to have one fab kept on the 45nm SOI process, while the other is going to be trained on bulk.
That speaks volumes of what AMD's x86 group expects to sell, and speaks volumes on how badly the NY fab is needed for the foundry, if it wants any significant market share at all in the face of the larger number of wafer starts pure-play foundries have.

This is all in the face of an oncoming capacity glut, if the process race continues as it has and larger wafer sizes come around (possibly around 2015?). In the future, a few large mega fabs could supply the entire world's demand for VLSI wafer starts, and only those fabs will have the economies of scale and revenues to support the process.

AMD is not that wealthy or that high volume. It has yet to show plans on how to woo others to its process or even how it will use that process its for its own products to any significant effect.
Both tasks will be extremely difficult for AMD the x86 manufacturer.

To top it off, the ATI purchase and the split is a ready-made way for AMD to simply give up on x86 altogether.
The process was already outlined in the latest conference call. If AMD decides not to match future Abu Dhabi capital calls for the fabs to keep its ownership stake (the voting rights are not the only condition) high enough, AMD will decouple from the fabs sufficiently that they cease to be a subsidiary per the terms of its cross licencing.
AMD's x86 efforts could end with a simple bit of inaction. The fabs will have Abu Dhabi to fund them, and AMD can simply walk away from that chapter of its history.

Posted by plonk420 on Wednesday, 22-Oct-08 03:14:20 UTC
Quote
I think that practically speaking, we need to wait and see what Intel is going to be able to ship in quantity when it comes to Nehalem. I recall that just a few months before Intel scrapped the P4 and moved on to Core that the top dogs inside Intel--Otellini, especially--made lots of public statements about the 4-5GHz P4's that *would be shipping* before the end of the year. That didn't happen as the P4 was scrapped before the end of the year, instead. With Intel it is wise indeed to wait to see what Intel actually ships before getting all hot and bothered by pre-release publicity speculation. Prescott was hyped to the moon pre-release so that when it shipped it shipped not with a bang but with a whimper. There's a good lesson there about the dangers of taking pre-release product publicity too seriously. Sometimes it proves worthless in the end.
i'm pretty sure it's not hype, unless Anandtech's benchmarks aren't to be believed. x264 performance is all i'm looking at and care about (and slightly distributed computing in the winter months)... numbers can be faked, but it would take a lot of balls to fake something as solid as video encoding numbers...

Posted by Raqia on Friday, 31-Oct-08 07:45:10 UTC
AMD has had definite teething problems with its burgeoning market share, and it's learned from its mistake if its new, conservative roadmap says anything with its focus on the established STARS core and usage of MCMs. I think it will do just fine by making contact instead of swinging for the fences every time; executing to meet market demand has always been its biggest problem. Dirk Meyer, who is more familiar with the day to day engineering affairs of AMD, should also be more realistic CEO and know what can and can't be done.

As for business models, if AMD isn't sustainable, then nVidia is completely toast. Integration in the form of Fusion will pose a huge problem for nVidia once it gets off the ground. Intel will also be playing catchup in the next few years if AMD manages to sell customers both a CPU and a GPU at the same time with each sale.

I do agree that the sale of its foundries is an admission of defeat in a very real sense and that Intel is very clearly more sustainable than AMD. However, the extra billions Intel is capable of throwing at problem sees diminishing returns especially in cutting edge process enhancements which often require fundamental breakthroughs; AMD has always had very good engineering (despite the focus on its missteps by some) and as was suggested AMD just needs to keep pace with a steady hand to survive. It probably won't recapture the very high end unless Intel commits a major mistake or if Fusion really takes off giving AMD a lead on the field in on-die GPU integration; AMD doesn't need to do this though, and can survive and sustain its market share if it's within striking distance of Intel in terms of performance and offer better value through integration, low power consumption, or pricing.

Posted by WaltC on Friday, 14-Nov-08 23:01:34 UTC
Quoting Joshua Luna
That sounds like a rose-filtered image of AMD. e.g. The financial issues AMD has faced the last few years has been because Intel has continued to have a strong edge in the mobile market and C2D nailed a lot of the popular AMD bulletpoints (performance, performance/watt) for the consumer market. In the past AMD has also had difficulty delivering products due to a variety of reasons, including capacity issues. And AMD's current woes are also tied to investing a large sum of resources ($5B?) into ATI. AMD's execution, and targetting the right markets at the right time, has been spotty and surely needs to be considered when talking about their financial issues.
I'll go along with these sentiments completely. (Actually, when I came back to this topic I was flabbergasted to see how much time I'd already spent on it....;) Sheeeesh!...;))

The only thing I'd add, again, is that paying system OEMs not to buy AMD's products isn't something you do for competitive reasons--especially, when you know there are some areas in which your competitor is flagging anyway. Rather you'd compete directly on valuie and price and just wouldn't need to offer the kickbacks. I think that the only reason you'd pay people not to buy your competitor's products, especially if you think yours are better anyway, is to run them out of business before they can become competitive again. Historically it has worked for Intel every time so it isn't surprising to see Intel repeat those tactics. I think they may ultimately backfire in AMD's case, however, which would benefit all of us as consumers down the line.

Everyone wants to look at the situation as though AMD has had nothing but a level, competitive field in front of it the whole time. But that's just not the case--it would have been the case had Intel not engaged in this kind of thing so often and for so long. Except for Intel paying people not to buy AMD products, I have no other ax to grind where Intel is concerned.

Beginning with Core 2 Intel has certainly become more competitive, and I think that in itself is a very positive change. Should Intel be persuaded through some means to stop the kickback scheme based primarily on limiting the number of it's competitor's products that OEMs can be paid not to buy, then I think at long last we will have a level playing field, and may the best man win...;) (And as consumers all of us will win in the end.)

Posted by WaltC on Saturday, 15-Nov-08 01:43:58 UTC
3d, I've enjoyed the give and take, but I'm about spent on the subject so tonight will be my last foray into this subject. Enjoyed your comments, though--wouldn't want you to think otherwise...;) We'll have to agree to disagree, even though of course my point of view is unassailable and invulnerable. (Hah!...;))


Quoting 3dilettante
The confirmed and uncomfirmed pressures from Intel probably had a contributory role to AMD's predicament by constraining its ability to raise prices as much as it wanted in the days it was capacity constrained.
Yes, but the "supply problem" of course is what Intel alleges has been AMD's major past obstacle--yet it hasn't ever been AMD's stated observation all.

With huge, fairly new FABs on line, newer than the great bulk of Intel's much more voluminous number of older FABs, it's difficult to understand Intel's talk about "supply problems" while at the same time Intel seems to think its policies of: "Get rich Mr. OEM by artificially constraining your AMD purchasing to 10% of the total cpus you buy (meaning you'll buy 90% from us); or get really, really rich by eschewing AMD altogether and buying 100% of your cpus from US!" would have had no effect whatsoever on AMD's demand and sales.

That argument has always struck me as very telling because in it Intel belatedly admits that demand for AMD cpus was very high--something only Intel would of course know based on the money it had to dole back to OEMs to artificially keep AMD's orders much lower than they would have been otherwise. Why would Intel have spent the money otherwise? Well, of course, it wouldn't have.

The whole line of AMD's "not having the capacity to meet its demand," while ignoring paying OEMs *not to order AMD cpus* is one of those beautiful and rare contradictions that come along when fingers get caught in the cookie jar. Of course, this sort of thing is the only thing that Intel might have said, even as illogical as it is, because at the time Intel was offering the anti-competitive subsidies, it was thinking about every thing else *except* having to explain this activity to a court at a later date...;) And why should Intel have been worried about that? The tactic has always worked for the company in the past. AMD is far more tenacious, however, as Intel has discovered.

Quote
It is insufficient to explain AMD's full position, considering it survived for years as an x86 also-ran prior to K7, when it actually had a measure of restraint and some good fortune.
On the contrary, AMD's statement, if not the charge itself, is that this kind of behavior is nothing new from Intel and has been going on for quite awhile. Again, the difference here is that unlike Intel's past would-be competitors, AMD did not fold up its tent and go home just because Intel was throwing its money around. I'm sure that AMD's damage declarations go back well beyond a decade ago.

Quote
OEMs had a number of non-Intel reasons to be afraid of letting in AMD:
reluctance to have to design and qualify two separate CPU lines, AMD's slow rate of chipset design (when it made them at all), issues in quality and stability for third-party chipsets, AMD's lack of history in those markets, AMD's lack of capacity in meeting the volumes of the large OEMs, AMD's lack of reliability.
OEMs have willingly testified, however, that Intel was livid when the K7 shipped and made all sorts of threats as to pricing and availability should any of those OEMs also wish to pick up AMD with supported products. The amazing thing about all of that is that the OEMs supported AMD anyway, Intel's ire be damned.

And why? That's the most compelling reason of all. By helping to make AMD a viable Intel competitor, as many OEMs in the early 90's surely did, with motherboards and lots of other things, the OEMs were liberating themselves from the inflexible Intel thumb. The result is plain to see: a thriving AMD is in the direct interests of OEMs as they can shop both AMD and Intel for the best deals and the days of being forced into line by a monolithic cpu company are long gone--for good.

In short, supporting AMD with products and services has always been in the direct interest of OEMs around the world--and *that's why* Intel had to start paying those who would take the money not to buy AMD's products. Even Dell eventually had to wise up and smell the coffee, as competitors embracing AMD were running Dell right off the charts. In short, even for Dell, the exclusive Kickbacks from Intel simply proved not enough, and when Dell realized that even Intel didn't have enough money to keep its Intel-exclusive OEM business afloat that it was time for a change.

Quote
AMD got some large OEM contracts, and then promptly struggled to meet them.
Not at all unusual. Before dumping its exclusivity with Intel, Dell often and verbosely complained about the fact that Intel could not keep up with Dell's ordering demand. There is nothing special or unique about AMD in that regard at all.

Quote
AMD, by its own admission, screwed its channel partners over in trying to woo low-margin OEM accounts.
AMD's design leadership floundered on the replacement for K8, and it rushed Barcelona through a razor-thin validation period that would miss an issue that brutalized AMD's reputation and hobbled its highest-margin sales for half a year. Recent converts to the AMD team got nailed. Sun, Cray, IBM, HP, Dell, and whoever else were left hanging.
No--actually "SUN, et al, and Cray and so on," didn't get screwed at all. They got their chips. But the little guys in the secondary channels didn't, and that was quite simply because for a brief period of time AMD's demand far outstripped it's ability to supply. This is *not* a problem unique to AMD--Intel's had it in the past, so has nVidia, ATi, and lots of other companies. Apple for years was perpetually unable to meet demand. That argument is not a good one, and really has no bearing on the discussion at at hand--which is "Why was Intel so invested in paying OEMs not to buy AMD chips which Intel now says AMD couldn't produce?" You see the logical fallacy there, I'm sure...;)

Second thought of the day in that regard, is why do you think AMD built its own FABs if not to be able to satisfy a burgeoning demand? Which, of course, yet again provides Intel with the obvious rationale of seeking to curb not AMD's supply, but AMD's demand through the vehicle of paying OEMs *not to buy AMD chips.*

Quote
AMD overextended itself in a massive purchase that in the end forced its market valuation at less than the purchase price.
What do you mean "in the end"?...;) There is no end in sight for AMD that I can see. Additionally, practically every stock out there has lost or is continuing to lose market valuation because of the general economy. Some of these companies indeed may be "close to the end," but AMD isn't to be numbered among them anymore than Intel is...;)

Quote
AMD's process was lackluster, AMD's design was lackluster, and even by AMD's own admission in some of its court documents, its cost structure was inferior.
Well, the now extinct Netburst was by all accounts grossly inferior to AMD 64, and Intel isn't yet six feet under as a result, is it? The biggest difference between Intel and AMD is surely not which of them has made the most mistakes and thrown away the most money--because of the two, Intel clearly and overwhelmingly wins that prize. The difference is that AMD has never had the money to burn and throw away that Intel has had. Reverse the bank accounts of the two companies over the past decade while leaving the rest of their histories unchanged, and my bet is that Intel would have joined Cyrix in the sky by now...;)

Quote
AMD, if it were in steadier hands, would have had a cash cushion to survive Barcelona without endangering the critical link between its fabs and CPU designers.
What I'm reading today about Shanghai is that OEMs are eagerly embracing it, and thanking their gods that they once again are liberated from under the heel of the Intel monopoly. (Sorry for the drama, but it's kind of fun...;)) Cash cushion? AMD recently acquired an investment of several billion dollars--along with an industry-applauded Shanghai, along with an industry-applauded gpu family, so I see nothing but positives there.

Quote
And now AMD doesn't compete all that well on the basis of technology. Now what?
Hmmm...apparently you may be the *only one* who thinks that Shanghai and the future of AMD cpus, along with the AMD 4000-series discrete gpus, along with AMD's Intel-smashing IGPs, along with AMD's energy-miser core-logic designs, among several other things--are not "technologically competitive"...;)

Quote
Counterexample: IBM has fallen back in several markets under Intel's product assault. IBM pours hundreds of millions of dollars into its microelectronics division and can afford to take a loss because of services.
Intel does not have that option.
Yes, IBM and Intel have been distinctly different types of companies for a long time now. IBM is, however, one of AMD's many dedicated and well-heeled partners.

Quote
Just which critical parties has Intel arm-twisted to shove IBM aside when IBM has a much longer history and far greater success in strongarm tactics?
Do you think that paying OEMs not to buy AMD cpus is a "strong-arm" tactic? I don't...;) Curious expression. However, there are no shortage of Taiwanese companies willing to write expletive-deleted accounts of their dealings with Intel in the past ("Willing to write" is a misnomer--they have already written about it) --especially when it concerned Intel's threats about supporting AMD's K7. I remember those days well, don't you?

Quote
Perhaps AMD's greatest hope for a fab customer is the hope that IBM gives up the ghost on its microelectronics business and tosses it to the Foundry Company, since IBM might be the only customer who will have any real interest in a process amenable to high-end x86, given that IBM made the process in the first place.
*choke*...;) Yes of course--it's well known that AMD has no hope of future FAB customers, and that's probably the reason that a company just bought > 50% in AMD's FABs for several billion dollars over the next few years--sure that makes sense...:D

TSMC should illustrate beyond a doubt that FABs can make money and lots and lots of money. The world of computer chips is growing and not shrinking and demand is rising and not falling--and so I have little doubt that the AMD FABs, currently state of the art and slated to remain so for the next several years--will have all of the business it can handle. Who knows? Perhaps even Intel will shut down some of its much older FABs and find it cheaper to use The Foundry Company's much more modern facilities...;) You really need to get your mind out of the "AMD is doomed" rut it's in--it seems counterproductive towards a balanced point of view, imo.

Quote
I will try to answer this with prose that is far less beautiful, though in my mind is far mor appropriate for the position AMD has found itself in.

AMD's business model is unsustainable. This was apparent years ago.
I agree that unless AMD had taken the concrete steps it has taken, in both the technology fields and the economic fields, and more recently the legal fields--that in the face of Intel's "We crush all competitors" business tactics--yes, AMD would not have been sustainable.

However, you need to go back and study Intel's extremely humble origins, long before Intel was ever a "big player," and temper your thoughts about AMD accordingly. Not to mention that you need to consider that during Intel's early days it had no entrenched monopolist to compete with in the sense of what AMD has had to endure for the past few decades. In all of the sentiment that you write I get the feeling that for you Intel has always been Intel as it is today instead of the pipsqueak it was when it got its very shaky start--and that start being nearly competitor free.

For me, AMD's "business model" in the face of a steamrolling, no-holds-barred, everything is legal Intel, with billions of dollars of money it was under no compunction to spend wisely--in the face of that environment I salute AMD and marvel at what the company has done. Quite literally, it's done and is continuing to do what nobody else has done before. Intel has not been able to derail AMD and that's because of AMD because Intel has surely left no stone unturned in its attempt to eliminate AMD from the competitive landscape. Indeed, sans AMD Intel would own the show lock, stock and barrel. Would we be paying $ 2k a pop these days for a 1.5GHz P4 in the absence of AMD? I can't say for sure, but I'd bet it would a very close thing.

Quote
It's run a long race as a smaller lower-margin device manufacturer of an engineering-intensive and commoditized product manufactured in an in-house fabrication method that is capital-intensive and subject to rapid obsolescence.
As AMD just divested its FABs, I can only gather you're talking about Intel?...;) Sure doesn't describe AMD. Seems to fit Intel to tee, though.

Quote
It's not even clear that the model in the long-run is one even Intel can keep up, but it is currently significantly better positioned to survive the commoditization of x86, given its economies of scale and more secure financial status.
Uh, just a few weeks ago there were several traditional financial institutions, some of them well over a century old with International reputations for stability and longevity, who were considered unassailable for the reasons you just laid out. More than one of them today is now extinct, however. Just goes to show that things can change very quickly not only for ill but also for good. In short, past performance is no guarantee of future profits...;)

Quote
The split is not a massive win for AMD. The current plan at best will allow AMD to mostly keep the curtailed promises it made years ago. The fact that such a massive change was a wash means that AMD was in very serious trouble.
A wash? Heh...;) By whose accounting? Remember that the AMD spin off was done with an eye to the future while saying good-bye to the past. AMD is growing in other words, as opposed to contracting. That's the part you keep missing somehow.

Quote
The interest of the investment group that owns most of the fabs is not aligned with the x86 design group. The more successful the foundry becomes, the less tied AMD will be to the fab process.
That's too bad for the "investment group" you cite, I guess, because when they made their investment and bought into the company, they deliberately left AMD in charge of the operations of the FAB in terms of management and they gave AMD 50% of the voting shares. You were saying something about "unaligned"?...;) The only "line" there will be AMD's.

Quote
The size of the foundry is small. Dresden is going to have one fab kept on the 45nm SOI process, while the other is going to be trained on bulk.
That speaks volumes of what AMD's x86 group expects to sell, and speaks volumes on how badly the NY fab is needed for the foundry, if it wants any significant market share at all in the face of the larger number of wafer starts pure-play foundries have.
Yes, you see, when your company is growing you need to expand, which is what AMD is doing. Apparently, I guess, you are unaware of the fact that Intel's FABs were also built one at a time over a period of time for the same reasons.

Quote
This is all in the face of an oncoming capacity glut, if the process race continues as it has and larger wafer sizes come around (possibly around 2015?). In the future, a few large mega fabs could supply the entire world's demand for VLSI wafer starts, and only those fabs will have the economies of scale and revenues to support the process.
You should write AMD's new company and inform them of the doom and gloom ahead, lest they proceed along a different path--probably because their data is different from yours and has lead them to reach much different conclusions.

But basically, you are pretty easy to understand in that you seem to have concluded that whatever Intel does or does not do will prove correct, while whatever AMD does or does not do will dismally fail...;) That's about the gist of your overall analysis, am I right?

Quote
AMD is not that wealthy or that high volume. It has yet to show plans on how to woo others to its process or even how it will use that process its for its own products to any significant effect.
Both tasks will be extremely difficult for AMD the x86 manufacturer.
I've been hearing essentially the same take on AMD since 1998, and it has yet to prove correct. Perhaps, though, what's been missing is a time frame that has thrown me off. Maybe that's it--maybe what they've always meant to say but have just assumed was understood by everyone is that AMD's failure is due to hit sometime between 2030 and 2040. Is that it? I hope so, because then I can understand the repetitive rationale a lot better...;)

Quote
To top it off, the ATI purchase and the split is a ready-made way for AMD to simply give up on x86 altogether.
Of course! Bingo! Yes, AMD bought ATi so that it could quit making cpus and make only gpus. Wow. That's insightful. I guess AMD will soon stop making cpus in about, oh, five years or so? After Shanghai and the rest of their recently released cpu roadmap sells in volume? How diabolically clever of AMD!--they are going to fool all of us until sometime in the future when they will suddenly quit making cpus and say, "Surprise! Fooled Ya!"

Quote
The process was already outlined in the latest conference call. If AMD decides not to match future Abu Dhabi capital calls for the fabs to keep its ownership stake (the voting rights are not the only condition) high enough, AMD will decouple from the fabs sufficiently that they cease to be a subsidiary per the terms of its cross licencing.
It's not just voting rights--it's also the day-to-day management of the FABs that AMD is tasked with. But I know that this is probably just a part of their clever ruse to make us *think* they are going to continue as a cpu company just to sell boatloads of cpus to heighten the element of surprise when they make their announcement about dropping cpu manufacturing and design completely!

That's breathtaking in its brilliance! Wish I'd have thought of that! Well, sigh...if only wishes were dollars....;/


Quote
AMD's x86 efforts could end with a simple bit of inaction. The fabs will have Abu Dhabi to fund them, and AMD can simply walk away from that chapter of its history.
Yes, what an incredible saga that will be! Clever AMD and its clever investors are faking all of this just to fool us into thinking they still want to make cpus! I see it now--yes, all of this nonsense about buying ATi so as to compete with Intel on a unified platform approach with AMD IGPs, core-logic, and cpus--why, it's all nonsense! By doing it this way Intel will be completely FOOLED on the day when AMD stands up, sticks out its tongue at Intel, and goes, "FLLLlll-l-l-l-l-l-l-lp!" (loud farting noise), "SURPRISE, Intel! We weren't serious, you schmucks! Had ya' goin', didn't we? So long, sucka's! We're off to compete with nVidia! We're following the yellow brick road, ha-ha, we're following the yellow brick road!..."!!!

Can you imagine? The deceit will be so unexpected that there won't be a dry pair of undies in all of Inteldom on that fateful day! And nVidia...OMG...JHH is going to defecate in his drawers when he hears the news--'cause clever old AMD and it's clever new investors will have fooled the world!

Posted by 3dilettante on Monday, 17-Nov-08 16:00:00 UTC
Quoting WaltC
On the contrary, AMD's statement, if not the charge itself, is that this kind of behavior is nothing new from Intel and has been going on for quite awhile.
Let's amend that to had been going on. AMD is not claiming any recent misbehavior.
Quote
Not at all unusual. Before dumping its exclusivity with Intel, Dell often and verbosely complained about the fact that Intel could not keep up with Dell's ordering demand. There is nothing special or unique about AMD in that regard at all.
Did Dell complain how Intel promised X number of chips and then delivered less than that number?

There is a difference between not making enough to meet demand, and failure to meet orders that were contracted, and possibly paid for.

Quoting WaltC
No--actually "SUN, et al, and Cray and so on," didn't get screwed at all. They got their chips.
Six months late.
HPC products could sell with Barcelona to select customers, with undisclosed promises for compensation or replacement.
Commercial server products could not be sold. A half year of product schedules were trashed by the TLB bug.
IBM had the PR black eye of having its Barcelona SPEC submissions disqualified in the time period, because the product was not available.


Quote
Second thought of the day in that regard, is why do you think AMD built its own FABs if not to be able to satisfy a burgeoning demand?
Perhaps because no PC manufacturer has amounted to much without a tightly coupled fab and chip design process?


Quote
What do you mean "in the end"?...
AMD took the writedown. It can't be taken back. Sounds pretty final to me.

Quote
There is no end in sight for AMD that I can see. Additionally, practically every stock out there has lost or is continuing to lose market valuation because of the general economy. Some of these companies indeed may be "close to the end," but AMD isn't to be numbered among them anymore than Intel is...
I should congratulate AMD's stock on being so visionary by losing most of its value half a year ahead of time.

Quote
Well, the now extinct Netburst was by all accounts grossly inferior to AMD 64, and Intel isn't yet six feet under as a result, is it?
It was not by all accounts inferior.
Northwood had the Athlon XP on the ropes. Those stupidly high PR numbers prior to K8 were symptomatic of that trend.

Even when Prescott was lagging, Intel's better marketing and broader manufacturing volumes meant that it could slash prices and still make money.
To top it off, the Pentium M with the Centrino branding push gave Intel a superior product to fall back on.
People were bunging Pentium Ms into desktop form factors. Discounting its weak FP resources, the chip was superior per clock to K8.

Quote
The difference is that AMD has never had the money to burn and throw away that Intel has had.
Tell that to Spansion.

Quote
Reverse the bank accounts of the two companies over the past decade while leaving the rest of their histories unchanged, and my bet is that Intel would have joined Cyrix in the sky by now...
So you're saying being secure financially covers a multitude of sins and is a very heavy point in a company's favor.
I've argued that too.

Quote
Cash cushion? AMD recently acquired an investment of several billion dollars--along with an industry-applauded Shanghai, along with an industry-applauded gpu family, so I see nothing but positives there.
AMD's overall debt position is better, but not massively so.
Shanghai is an improvement in one market segment.
These are positives, but there are serious background issues.

Quote
Hmmm...apparently you may be the *only one* who thinks that Shanghai and the future of AMD cpus, along with the
Shanghai will maintain competitiveness in 4 socket and above.
Nehalem and Dunnington will do very well below.
Nehalem 2-socket might eat out some of the bottom of the 4 socket space.

AMD's future CPUs will be Shanghai and its ilk for the next 2 years.

At the desktop, AMD's chips relative to Intels will be just about where they are right now.

Quote
AMD 4000-series discrete gpus, along with AMD's Intel-smashing IGPs, along with AMD's energy-miser core-logic designs, among several other things--are not "technologically competitive"...
AMD's GPUs have a fraction of the overall revenue slice. We'll have to see how things turn out once the fab company is totally divested.
Intel-smashing IGPs are a nice minority share of the IGP market.

Quote
Do you think that paying OEMs not to buy AMD cpus is a "strong-arm" tactic? I don't... Curious expression. However, there are no shortage of Taiwanese companies willing to write expletive-deleted accounts of their dealings with Intel in the past ("Willing to write" is a misnomer--they have already written about it) --especially when it concerned Intel's threats about supporting AMD's K7. I remember those days well, don't you?
You didn't address my point, and your dodge was not that artful.

Quote
Yes, IBM and Intel have been distinctly different types of companies for a long time now. IBM is, however, one of AMD's many dedicated and well-heeled partners.
IBM's microelectronics division has been losing money hand over fist. It has downsized again just recently, and it's offering a foundry service for its high-performance SOI process.
I guess that makes it both well-heeled and a dedicated friend.

Quote
Do you think that paying OEMs not to buy AMD cpus is a "strong-arm" tactic? I don't... Curious expression. However, there are no shortage of Taiwanese companies willing to write expletive-deleted accounts of their dealings with Intel in the past ("Willing to write" is a misnomer--they have already written about it) --especially when it concerned Intel's threats about supporting AMD's K7. I remember those days well, don't you?
If I recall correctly, AMD's finances imploded in this century.

Quote
*choke*... Yes of course--it's well known that AMD has no hope of future FAB customers, and that's probably the reason that a company just bought > 50% in AMD's FABs for several billion dollars over the next few years--sure that makes sense...:D
If you want to harp on Intel's being rich enough to throw billions of dollars down the drain, don't spare Abu Dhabi the same critique when its sovereign wealth fund was recently valued at a trillion dollars.

Quote
TSMC should illustrate beyond a doubt that FABs can make money and lots and lots of money. The world of computer chips is growing and not shrinking and demand is rising and not falling--and so I have little doubt that the AMD FABs, currently state of the art and slated to remain so for the next several years--will have all of the business it can handle.
Fab utilization has dropped below 85% industry-wide for foundries. Capital expenditures have been slashed across the board.
Foundry customer product transitions to the now years-old leading process nodes at the foundries have been slow.
IT spending has declined.

AMD picked one heck of a time to do this in a market that looks oh so receptive.

Quote
Quite literally, it's done and is continuing to do what nobody else has done before. Intel has not been able to derail AMD and that's because of AMD because Intel has surely left no stone unturned in its attempt to eliminate AMD from the competitive landscape. Indeed, sans AMD Intel would own the show lock, stock and barrel. Would we be paying $ 2k a pop these days for a 1.5GHz P4 in the absence of AMD? I can't say for sure, but I'd bet it would a very close thing.
Stop wasting my time with these tangents. What the world could have been if unicorns were allowed to become president is similarly irrelevant.
The question is what AMD needs to do to survive as an x86 manufacturer against a powerful competitor. I don't care what moral catastrophe we face if Intel finds the One Ring.

Quote
As AMD just divested its FABs, I can only gather you're talking about Intel?...
Sure doesn't describe AMD. Seems to fit Intel to tee, though.
Intel is a smaller also-ran in a high-capital industry which relies on high volumes?
What's the larger, vastly more financially successful competitor?

Quote
Uh, just a few weeks ago there were several traditional financial institutions, some of them well over a century old with International reputations for stability and longevity, who were considered unassailable for the reasons you just laid out. More than one of them today is now extinct, however. Just goes to show that things can change very quickly not only for ill but also for good. In short, past performance is no guarantee of future profits...
I guess the companies that go through extreme gymnastics with their financial and contractual schemes should be given a healthy dose of skepticism.

Quote
A wash? Heh... By whose accounting?
AMD's?

Quote
Remember that the AMD spin off was done with an eye to the future while saying good-bye to the past. AMD is growing in other words, as opposed to contracting. That's the part you keep missing somehow.
AMD just announced more workforce cuts. Just what is growing?
Don't forget, the foundry company isn't all AMD any more. Actually, if we go by ownership share, AMD has actually shrunk.

Quote
That's too bad for the "investment group" you cite, I guess, because when they made their investment and bought into the company, they deliberately left AMD in charge of the operations of the FAB in terms of management and they gave AMD 50% of the voting shares. You were saying something about "unaligned"?...;) The only "line" there will be AMD's.
Leading edge digital-only x86 manufacturing requires a level of process tuning and specialization that is at odds with the demands of mixed-signal low-margin and low-power circuitry that dominates the foundry market.
AMD wants one, the foundry company needs the other.
Reconcile these goals for me.

Quote
You should write AMD's new company and inform them of the doom and gloom ahead, lest they proceed along a different path--probably because their data is different from yours and has lead them to reach much different conclusions.
I think they are very aware of just how difficult this will be.
Going bankrupt without this less than ideal arrangment with Abu Dhabi was their other option.

Quote
But basically, you are pretty easy to understand in that you seem to have concluded that whatever Intel does or does not do will prove correct, while whatever AMD does or does not do will dismally fail... That's about the gist of your overall analysis, am I right?
If that's what you think, you are free to believe it. The politest response I can think of for you to do to yourself is strive to find a more accurate assessment of my argument.


Quote
Of course! Bingo! Yes, AMD bought ATi so that it could quit making cpus and make only gpus. Wow. That's insightful. I guess AMD will soon stop making cpus in about, oh, five years or so? After Shanghai and the rest of their recently released cpu roadmap sells in volume? How diabolically clever of AMD!--they are going to fool all of us until sometime in the future when they will suddenly quit making cpus and say, "Surprise! Fooled Ya!"
If they fail to maintain enough ownership by not matching Abu Dhabi's contribution for successive capital calls for the foundry company, they will have no choice.

Quote
It's not just voting rights--it's also the day-to-day management of the FABs that AMD is tasked with. But I know that this is probably just a part of their clever ruse to make us *think* they are going to continue as a cpu company just to sell boatloads of cpus to heighten the element of surprise when they make their announcement about dropping cpu manufacturing and design completely!
AMD's ownership and voting rights are what determines the foundry company's subsidiary status.
Both have lower bounds. Voting rights must stick at 50%. AMD cannot disburse more than 70% of profits to any co-owner, which means by some measure of non-voting ownership, it cannot fall below 30%.

Capital calls for the foundry that Abu Dhabi pays in will dilute AMD's share if it does not match the amount. This was stated in the conference call.

Quote
Yes, what an incredible saga that will be! Clever AMD and its clever investors are faking all of this just to fool us into thinking they still want to make cpus! I see it now--yes, all of this nonsense about buying ATi so as to compete with Intel on a unified platform approach with AMD IGPs, core-logic, and cpus--why, it's all nonsense!
Not on purpose, no.
But I can imagine that AMD the silicon design company could find it easier to slink into x86 obscurity than AMD the x86 fabrication company.


Add your comment in the forums

Related amd News

RWT Analyzes Bulldozer Benchmarks
AMD Bulldozer microarchitecture analysis
Say hello to GLOBALFOUNDRIES
AMD completes deal with ATIC to create The Foundry Company
AMD Propus to be released in Q2 & Q3
AMD launch 45nm Phenom II processor
Beyond Programmable Shading course notes available
AMD launches FireStream 9250 with 200Gflops DP via RV770
AMD GPGPU solutions get extra support from industry partners
AMD Phenom X3 released; reviewed