The Business of Tech: AMD Q4 2009

by Ryan Smith on 1/22/2010 12:00 AM EST
POST A COMMENT

30 Comments

Back to Article

  • Morbid666 - Sunday, February 21, 2010 - link

    The Future is Fusion Reply
  • MJinZ - Saturday, January 23, 2010 - link

    "Nehalem regulating AMD to the sub-$200 market"..

    I think you meant to use "relegate".
    Reply
  • nilepez - Monday, January 25, 2010 - link

    You beat me by 2 days....too bad they don't read the comments. Most tech sites really need editors that catch this type of error, as well spelling and grammatical errors. I'd also welcome an end to the passive voice. Reply
  • Sahrin - Friday, January 22, 2010 - link

    While I haven't read the text of the settlement, I know that it doesn't include any admission of guilt on the part of Intel - making this sentence pretty misleading. I agree with it in principal, but using the word "official" lends the authors' opinion undue authority. Reply
  • blowfish - Friday, January 22, 2010 - link

    are you a lawyer or something?

    One thing that really gets my goat is when people, say Ken Lay, for example, come up for trial, they have a lawyer who says to the TV cameras something to the effect "my client is innocent, and this will be proved in the forthcoming trial". Of course, the trial goes badly for Ken Lay, he's found guilty, and banged up to rights. At least he then did the decent thing and topped himself, so there is some justice. But the effing scumbag lawyer never comes back to the TV people to say "oh I'm sorry, I was wrong, my client was guilty all along, I'm sorry I tried to mislead you".

    Well so it is with Intel. They are a dirty company, they used (and probably continue to use)dirty tricks to further their agenda, and just because they were able to settle without admitting liability doesn't mean they were innocent. I'm sure you know it yourself, why would an innocent party cough up $1.2B? It certainly wasn't because AMD had more money to spend on legal representation. At least the Justice Department might not have finished with Intel yet.
    Reply
  • Sahrin - Saturday, January 23, 2010 - link

    Not a lawyer, just someone with a healthy respect for the truth. The job of a journalist is not to report what the facts should be, it's what the facts are. It's fine to say, "guilty in everything but name" or "got off like OJ" or whatever. It's not OK to imply something is true that isn't - specifically, that Intel has officially admitted culpability.

    Why is it important, you ask?

    As you so rightly pointed, out, Intel hasn't had to say they were stealing yet - and while it may be a line in the sand for some, I believe it's very important for them to have to face their actions. By sticking our heads in the sand and ignoring their mistakes, we are allowing them to avoid confronting them.

    Your animosity towards lawyers is grossly misplaced. A lawyer is REQUIRED to represent his client as the client sees fit - that's why attorney-client privledge exists. When a lawyer stands up to say, "my client is innocent" - he's operating a legal strategy that's in the best interest of his client. To not do so would be even more dishonorable. I think everyone would agree society would be a pretty terrible place if lawyers were allowed to ignore their responsibility to their clients if they thought their client was guilty.

    And finally, it's the FTC that's investigating Intel at the moment, not Justice. Details matter, if only because if makes you sound like you don't know what you're talking about to those who do. (I DO NOT claim to be someone who does).
    Reply
  • drunkenmaster - Friday, January 22, 2010 - link

    Global foundries being fully bought out isn't actually fantastic for AMD's stock/long term value, but not bad either. Their losses are short term, they are spending HUGE right now to get up and running, when they are, frankly, they'll be a massive player, with higher quality than realistically any other foundry group out there. These are good things for AMD and their stock price, short term $100mil losses are nothing in a game this big, and the future stock sharing of profits for AMD specific shares would be far more beneficial than the short term losses. In the real world of owners though, the same guys who own AMD also part own Global, just under a separate name, so the split of the companies doesn't make a jot of difference to them, their pockets get filled when the companies do well, their cheque's just have a different name on them.

    But the difference between last year and this year in losses, is also manufacturing. Foundry costs dwarf chip R&D costs, production R&D is far more intensive than chip design R&D. Look at the New York State fab, its costing about $7billion to build, only $1.5-2bil is construction the rest is equipment. Actually I guess I stated it wrongly, the R&D isn't more expensive(well it probably is) but the massive cost difference is the equipment, outfitting the $2billion building will cost them in the region of 4-5billion, of which every new process will require a new 2-3billion of equipment, which means every single 2 years you're chucking out 3-4billion on new equipment.

    This is the difference between AMD of the past, whose profits are killed by spending 3-4billion every 18-24 months, and AMD of now/future, where Global Foundries takes all that cost over, and AMD also drop all the process R&D which is very costly, and they only have costs of their chips now. Meaning once they start selling chips, everything is pure profit for them, that money isn't being saved up to sink into buying new equipment. When you realise AMD has been spending the same on process R&D as intel for decades, and has been paying more for equipment per fab than Intel, that being only a few billion in debt, when they've been spending that every 2 years on new equipment is actually very well done.

    The reason this works and Global Foundries can make a profit, is instead of spreading the R&D and equipment costs against a single company making a single chip stuck at 20% market share, they can now make chips for more people in more fabs, the same R&D and equipment will be used for longer, on more products, on far more chips. Intel has a better margin because its always been able to spread its costs over 4 times the amount of chips AMD sells. Now AMD/Global combo can do the same, or even dare say it, better.

    Anyway, from this point on that sucking unavoidable spending of 3-4billion simply won't happen anymore. AMD will continue to have profitable quarters from now on, and with Global footing the bill for process's, and spending big to get them out on time we have several huge benefits for AMD.

    If AMD can get a killer architecture again Global can afford to push into Intel's market share, at AMD's best they were stuck at 20-22% market share because, quite simply their fabs could not produce more chips, and you can't sell more than you make. if they had 50% more fab capacity, they could have pushed to 30% market share quite easily at that point. Global opens this possibility in the future. LIkewise Global is now fighting for business and has the cash to get new processes going fast and on time without delaying new processes till they can afford the equipment, this means within 1-2 process nodes, Intel will never be noticeably ahead on process technology anymore.

    In other words, the future is very bright for AMD, and Intel finally has a competitor that has the financial/production capacity to hurt it if it gets behind on technology again.

    Will Global ever hurt AMD and not give them preferential treatment? No because while AMD might not own any of Global soon, Global(and their owners and several Abu Dhabi investment groups) own a good 15-20% of AMD already, with a commitment to buy whats set to be over a billions worth of more shares in a year or two, so they'll move up to anything up to owning 30% of AMD. Global and AMD are a single company in everything other than name, in so much as they both heavily rely on each other, and Global will always favour AMD and do whatever it can and offer them fantastic pricing.
    Reply
  • BSMonitor - Friday, January 22, 2010 - link

    I totally agree! Missed your post by 2 minutes, I am above! Reply
  • BernardP - Friday, January 22, 2010 - link

    This seems as good a place as another to post my Special Request to AMD:

    Please give us an Athlon II X4 640 BE - 3 GHz, unlocked multiplier, 95W TDP, no heatsink.

    Thanks in advance :-)
    Reply
  • ambientmf - Friday, January 22, 2010 - link

    It's nice to see AMD finally getting out of their financial hole. They basically have no competition from Nvidia on the GPU front and their Phenom II quad's are a much better value than Intel's Lynnfield/Bloomfield chips.
    Especially after the disaster of Phenom I...
    Reply
  • Spivonious - Friday, January 22, 2010 - link

    Getting out? They lost almost a billion dollars before the Intel settlement. Unless AMD comes up with a real competitor to Nehalem I don't think they'll be around in 5 years. Reply
  • bsoft16384 - Friday, January 22, 2010 - link

    Well, whether or not Bulldozer is a competitor to Sandy Bridge, AMD is going to be around in 5 years. Now that they don't have an unprofitable fabrication business dragging on the balance sheet, even the graphics side of the business is enough to keep AMD afloat.

    Intel has shown that getting into the GPU business isn't as easy as it looks. There are basically two players: AMD and NVIDIA, and NVIDIA keeps shooting themselves in the foot with GPUs that are too late and too expensive. The bulk of the graphics business is in mainstream and mid-range parts ($50-$150) and ATI has a much stronger lineup for the foreseeable future.

    Then we have design wins with things like Congo, AMD's more-competitive low-end and mid-range parts (Athlon II and Phenom II), and the upcoming launch of the new 890GX chipset that should have no problem humiliating even the "new and improved" Intel GMA.

    We're probably not going to have another "Athlon 64 / Opteron" moment. Intel doesn't screw up as badly as they did with Prescott very often. The dominance Athlon 64 was a combination of Intel's inability to scale P4 to the clocks they wanted combined with AMD's best executed design since the Athlon. There's no "real competitor" to Nehalem coming for a long time.

    What (hopefully) will happen is that AMD will improve power consumption in mobile parts, get better chipsets with lower power consumption, better graphics and features like USB 3.0, and continue to execute well in their graphics division.
    Reply
  • BSMonitor - Friday, January 22, 2010 - link

    Funny. How can one talk about AMD financial success and Global Foundries financial woes seperately. It is all Accounting nonsense. So you take the cost of development and upgrading chip-making Fab's and stick on a seperate balance sheet called "Global Foundries". Global Foundries only real revenue comes from AMD.

    AMD makes a few dollars profit now.. Because the cost of chip-building is on a different balance sheet. Well guess what, the cost of chip-building is still too much for "Global Foundries" and they consistently lose money. Eventually Global Foundries goes bankrupt and AMD has no chip-builder...

    Ask GM and Delphi(or its many other suppliers) how well this works.
    Reply
  • subflava - Friday, January 22, 2010 - link

    You would be right if Global Foundries didn't plan on doing anything else except make chips for AMD. It would be just an accounting trick in that case. However, I assure you that the guys that bought Glo-Fo, (Abu Dhabi ATIC), did not spend $2.1B out of kindness to AMD. In addition they have since gone and spent another $3.9B for Chartered and merged the two companies. They plan on turning Glo-Fo into a legitimate foundry business to compete with Taiwan Semiconductor. This move was definitely about more than just accounting shenanigans. Reply
  • Penti - Saturday, January 23, 2010 - link

    Remember GF bought Chartered which already did have other customers on the AMD APM process. They already has all the tools for others to use them, both in Dresden and Singapore. And AMD was already fabbing CPUs at CSM. AMD plan to move some of the ATi-graphics stuff to GF. As well as chipsets. GF got plenty of other customers today which will scale up their business with them if GF offers a good deal to them. AMD only wants one fab to deal with that's pretty clear. (For CPUs, chipsets, APUs any way). They can streamline all the tools then. And it's not TSMC. There's plenty of companies ready to use and are using something other then TSMC. So all they need to do is get people designing for their plants. Something Chartered already has plenty of experience with. Reply
  • cocoviper - Saturday, January 23, 2010 - link

    Yep- I spoke with some ARM individuals at CES and they intend on doing a massive push for 28nm with GF.

    It won't be AMD only for long...
    Reply
  • Taft12 - Friday, January 22, 2010 - link

    Eventually Global Foundries goes bankrupt and AMD has no chip-builder...

    Why does Global Foundries need to be the chip-builder for AMD?
    Reply
  • ssj4Gogeta - Friday, January 22, 2010 - link

    I don't think any other foundry will be able to keep up with Intel's fabs. Reply
  • OCedHrt - Saturday, January 23, 2010 - link

    TSMC is ramping 28 and starting 22. In terms of keeping up in that area, they're doing fine..in terms of capacity...well that's another story. Reply
  • Hector1 - Monday, January 25, 2010 - link

    Re "TSMC is ramping 28"

    You've misread whatever you read. "Ramping" implies volume production has begun and that the volume is being increased week-to-week.

    TSMC has not started volume production of 28nm. Their press release of Jan 10 states "TSMC is currently conducting R&D for 28nm" and "will hand 28nm to the Phase 5 facility for volume production in the fourth quarter of 2010" They're a long way from shipping production units. They say they just fixed their yield problems on 40nm. It's unlikely that they'll be shipping real 28nm production volume until 2011 at the earliest
    Reply
  • The0ne - Monday, January 25, 2010 - link

    Intel does have good yields and this is in part due to their manufacturing processes. Once they have the processed "right" they locked it down preventing any further changes. This is arguable but it does work to a greater extent.

    Intel is the only company I know of that does this to maintain their yields. Typical manufacturing environments are constantly changing. Someday I'm going to have to visit these factories to truly observe how things are done :) Don't ask, I'm a Mfg Engineer myself.
    Reply
  • hyvonen - Sunday, January 24, 2010 - link

    TSMC is way behind Intel - if you recall, Intel was already showing a working 22nm wafer at the last IDF, while 32nm is mature enough to manufacture chips at yields high enough for sales... All this while TSMC has been struggling to get their 40nm to reasonable yields.
    Reply
  • subflava - Friday, January 22, 2010 - link

    Exactly...no one can keep up with Intel's fabs and that's why everyone has stopped trying. This is exactly why the foundries exist in the first place. Reply
  • MonkeyPaw - Tuesday, January 26, 2010 - link

    Intel's process tech it top notch because Intel is the first to buy the new manufacturing equipment needed for smaller nodes. They pay top dollar for that machinery, which is why Intel typically launches the highest-end CPUs on the newest process technology. The yields aren't great, but the prices are very high, so they get their investment back until the yields gets good enough for the rest of the product line. AMD on the other hand, has typically adopted new process tech much later, and they use the smallest nodes to sell their cheapest CPUs. As they improve the process, they ramp up speeds. It's basically the way a small volume company works versus a large volume company. Intel has to depend on tried-and-true processes to fill demand, while AMD depends on tried-and-true processes to compete at the high end. Both are valid approaches, and I really think that it works decently for both companies. Intel can react better to a bad node-shift versus AMD, so they can afford the risk. Reply
  • Shadowmaster625 - Friday, January 22, 2010 - link

    Looking at AMD's stock chart, it seems that players have been pricing in the intel settlement long before it happened.

    At 65% margin, intel is making 100 million for every 35 million they spend. That is highway robbery! If people had half a brain they would boycott...

    Want to see something funny? http://tinyurl.com/yda4ten">http://tinyurl.com/yda4ten This is a vietnamese thread talking about which is better, HD3300 or a 7300GT. Its funny as hell, but not too useful. I would go with the 7300GT, but I really wish you guys did more IGP vs discrete tests so I knew for sure.

    Reply
  • nofumble62 - Friday, January 22, 2010 - link

    Margin don't include all those payment. Reply
  • Hrel - Friday, January 22, 2010 - link

    who taught you math? If intel sold chips that in total cost 35Million and turned a 65% profit they'd bring in 57.75Million. Which is a profit of 22.75Million. Intel's profit margin would have to be 285% for 35Million in sales to bring in 100Million. Reply
  • je1117 - Sunday, January 24, 2010 - link

    Gross profit margin is gross proft (revenue - cogs) / total revenue. What he was saying is that AT 65% profit margin, Intel is making 100m for every 35m in selling costs. (100 - 35) / 100 = 65%.

    Profit and profit margin are two different things, who taught you business? ;)
    Reply
  • Hector1 - Friday, January 22, 2010 - link

    Do you think AMD wants Intel to lower prices so Intel's margin gets smaller ? What would happen ? AMD would be forced to lower prices as well, which would cause them to lose more money. Reply
  • fintel - Thursday, April 22, 2010 - link

    As far as blog is concerned here are so many interesting information Reply

Log in

Don't have an account? Sign up now