AMD'S stock price plummets amidst CFO resignation

[AnitaPeterson]

Actually it wouldnt. It would mean better and cheaper CPUs on the long run. I wouldnt be surprised if a 5nm fab costed 50billion$.

Hmmm... if history shows anything, it's that Intel prices were only forced down by competition. A virtual monopoly on the market would allow them to ask for as much money as they can. And since the money would be a sufficient incentive, progress would likely stagnate.

 

[Ventanni]

Hmmm... if history shows anything, it's that Intel prices were only forced down by competition. A virtual monopoly on the market would allow them to ask for as much money as they can. And since the money would be a sufficient incentive, progress would likely stagnate.

I agree with the post above, because we've experienced it before. If Intel gains a monopoly, expect to pay substantially more for your CPU, and experience far fewer architectural changes. It'd be a great thing for ARM though, as people would definitely look elsewhere for a CPU alternative.

 

[96Firebird]

I agree with the post above, because we've experienced it before. If Intel gains a monopoly, expect to pay substantially more for your CPU, and experience far fewer architectural changes. It'd be a great thing for ARM though, as people would definitely look elsewhere for a CPU alternative.

People always say this, but it is not always the case. Intel is also in competition with themselves. If they bring nothing new to the table, people don't need to upgrade. Therefore, no new chips are selling. Not good. Intel does not want the market to go stagnant, they need to make it worth upgrading so they can sell more.

 

[AnandThenMan]

People always say this, but it is not always the case. Intel is also in competition with themselves. If they bring nothing new to the table, people don't need to upgrade. Therefore, no new chips are selling.

Wrong on so many levels. There will always be new products, but the improvements will be smaller, and you will pay more for them. Plus people buy new products just for the sake of being "better" this is a very well known phenomenon in tech. Not to mention people buy new systems when the old one has a problem, it's often easier and sometimes cheaper to do so. The upgrade cycle exists irrespective of how good say a new generation of processors is.

Not good. Intel does not want the market to go stagnant, they need to make it worth upgrading so they can sell more.

Worth upgrading is a subjective thing, the average buyer knows little if anything about the actual performance and technical aspects of a machine. If they did, not a single person would have ever bought a Pentium 4 system by choice. Not a single person would buy an Intel GPU with the intention of gaming.

 

[RussianSensation]

I think the best thing AMD has going for it now, when considering avoiding a buyout, is they have too much debt to be worth absorbing. Hmmm, it is getting close, nVidia has enough liquid assets that they would just miss buying AMD outright and paying off all their debts. Don't think that would be a great thing for the discrete GPU sector, but it would probably help the CPU side of the x86 market out a great deal to have JHH helming Intel's main competition.

No, NV cannot at all acquire AMD with its liquid assets.

Market capitalization of a company is NOT the value of a company, especially not in an M&A transaction since you still have to account for the acquirer's "control and synergy" price premiums. At the very least, AMD's total value starts at the Enterprise Value (EV) level (or alternatively you have to derive the value using a combination of DCF, precent transaction, trading multiples comparable companies).

At minimum NV would have to pay this for AMD :

Enterprise value = market value of common equity + market value of debt + market value of preferred stock + minority interest (if any) + unfunded pension liabilities and other debt-deemed provisions – cash and cash equivalents.

However, in a real world market transaction, if the acquirer is buying the entire equity of the company, they generally have to pay an additional premium for "control." The current AMD shareholders will not just surrender AMD shares for $3.60. Furthermore, there is an additional premium associated with realizable synergistic value due to improved supply chain/ economies of scale/human capital that NV would realize due to acquiring another GPU maker (for example they would have higher purchasing bargaining power at TSMC since they would place 40-50% more chip orders reducing their cost per wafer). These synergies would demand a market premium on top of the TEV from AMD shareholders/Board of Directors. This so called control / market premium can be 20-30% alone on top of the current stock price.

 

[RussianSensation]

OT, why can't AMD catch a break. I sincerely hope they don't go under, as both the GPU and the x86 CPU markets rely on the competion AMD provides.

Starts with overpaying for the ATI acquisition at the peak of ATI's value (despite hardly any competing offers, but AMD paying a ~30% market premium), as a result burdening the company with $5.4B in debt on which AMD has had to pay interest all this time (the interest payments could have been used to drive R&D of the AMD CPU division, develop a strategy for mobile / smartphone/tablet CPU, etc.), resting on its laurels with Athlon 64 architecture, making a technical mistake of assuming that software would become multi-threaded much quicker than happened in reality (hence the 8-core Bulldozer modular design that sounded great in theory ended up inferior to strong IPC/less fast cores approach).

Add, half a decade of uncompetitive desktop and mobile CPU vision that made less and less cash, poor marketing and execution along the way, lack of diversification strategy to capitalize on growing consumer trends in case the PC market was to slow down - which is exactly what we are seeing now with the tremendous growth of smartphones and tablets. Then you have the cannibalization of the sub-$100 discrete GPU with Intel moving into the APU space, general preference of discrete GPU consumers for NV products, i.e., NV loyalists refuse to buy AMD GPUs even when AMD is giving them for $269/299.

AMD's current product line is barely making enough $ to cover the interest on their existing debt and leaving room for marketing, hiring top engineers, top management and R&D and then you have desktop PCs and laptops as an entire industry segment becoming less popular overall with the average consumer who is now far more excited about smartphones and tablets.

It's the perfect storm of being squeezed from all sides at the same time.

However, it does appear like a combination of some talented AMD employees receiving lucrative positions with better opportunity elsewhere (Eric Demers to Qualcomm) and a case of Rory Read cleaning up the old team that got the company into this mess in the first place. If you look at the details of the new management, it now includes Lisa Su and Mark Papermaster, senior vice presidents, who both had lengthy tenures at IBM (same as Rory Read). That's not a coincidence. Then you have the case of AMD hiring John Gustafson as the new main product architect for AMD's graphics business unit. It just seems the new CEO is trying to restructure the company during the "adapt or die" phase.

The biggest risk factor is if AMD can survive this turbulent period and not run out of cash flow. At the same time, as we've seen from NV going into Tegra/smartphone/tablet markets, it seems inevitable that AMD would need to follow suit into this extremely competitive and low margin business simply because the market growth in the smartphone/tablet space is going to outpace desktop / laptop PC growth until at least 2015.

 

[RussianSensation]

Actually it wouldnt. It would mean better and cheaper CPUs on the long run. I wouldnt be surprised if a 5nm fab costed 50billion$.

That's the opposite of how a capitalist system works - you need real competition to have price pressures and similar pace of innovation. Lack of serious competitive price pressures generally results in terrible price competition - telecomm companies in US and Canada are the perfect example. If AMD is out, even if Intel is split into 2 "competing" companies, all it would do is create an illusion of competition. Even now Intel is already ripping us off. Instead of raising prices, they are selling us less die space (but that's directly related to CPU performance and core count).

Intel's first quad core came out in 2006 and by 2007, you were able to buy one for around $300-330 in the form of Q6600. Next year Intel is launching Haswell and it will still be $300 for a quad-core i7. If you don't see anything wrong with this picture, I don't know what to tell you. When AMD actually had competitive Athlon 64 CPUs, Intel wouldn't even dare charge extra for hyper-threading. Now they are tacking on an $80-100 price premium for this 1 feature in the i7 2600/2700/3770K. That's ridiculous and the only reason they are getting away with it is because Bulldozer has failed.

Also, when AMD CPUs had no competition from Intel, we had this:

Intel may lower prices if another competitor emerges, but your theory that in the long-run Intel will magically lower prices on their CPUs because they have less competition is an odd one.

 

[Blitzvogel]

What if VIA was to buy AMD? It would get VIA immediate access to higher end processors and GPUs. I wonder what the guys at Centaur working with AMD's Bobcat/Jaguar team could do in terms of low power envelope x86 design.

 

[Siberian]

Ruht Roh. AMD's Client Division chief is also leaving.
http://techreport.com/news/23594/amd-client-division-chief-is-also-leaving

Lets see what the stock does today after this news.

It says that he was "responsible for operations that made up about 80% of AMD'S revenue"

That sounds serious. These are not guys you replace over night.

 

[Lonyo]

Everyone talking about lack of competition seems to be ignoring ARM.
Yes, currently they aren't really a player in some markets, but you can bet if Intel started being greedy they would quickly gain a LOT more attention for certain uses.

Intel doesn't have free reign of the overall CPU market, encompassing desktops, servers and mobile. Even without AMD there will still be competition.

 

[BenSkywalker]

At minimum NV would have to pay this for AMD :

Enterprise value = market value of common equity + market value of debt + market value of preferred stock + minority interest (if any) + unfunded pension liabilities and other debt-deemed provisions – cash and cash equivalents.

AMD's Enterprise Value is $3.54 billion. nVidia has over $4 billion in liquid assets. You were saying?

http://ycharts.com/companies/AMD/enterprise_value

Wouldn't be a smart move IMO on nVidia's part, but it would almost assuredly save AMD.

ARM is dying.

W.T.F.......

2.2 Billion ARM processors sold per quarter, x86 sells about half of that per year(being generous, doesn't look like they are going to cross a billion units moved this year)-

http://www.eetimes.com/electronics-news/4235514/ARM-financial-results

The sad part is that ARM sales are exploding while x86 are struggling to maintain. Current estimates have ARM selling two processors for each person on Earth per year by ~'16.

Qualcomm probably doesn't know much about how to do dx11 or gpu compute but they will need to if they want to compete with nvidia who does. Buying Ati would give them that expertise.

AMD sent their people along with the IP when they sold the Snapdragon design to Qualcomm. They are already shipping a sub 1 watt dual core processor with DX9.3 feature set. Yes, moving to DX11 may take a bit of work, but already since Qualcomm acquired the engineers the Adreno line of GPU cores has made massive progress.

 

[Smoblikat]

Everyone talking about lack of competition seems to be ignoring ARM.
Yes, currently they aren't really a player in some markets, but you can bet if Intel started being greedy they would quickly gain a LOT more attention for certain uses.

Intel doesn't have free reign of the overall CPU market, encompassing desktops, servers and mobile. Even without AMD there will still be competition.

Intel is charging us 100$ for a feature that costs THEM money to disable, how arent they the opitomy of greed?

 

[Arkaign]

Intel is charging us 100$ for a feature that costs THEM money to disable, how arent they the opitomy of greed?

First, it's 'epitome', and second, that's the fundamental basis of corporatism, all public corporations are as greedy as possible because they have a fiduciary responsibility to the shareholders. Privately held corporations are also typically greedy because people like making money.

Second, if your example is Hyperthreading, then it's just like the rest of the CPU business. In the end, it takes the same amount of $ to make a CPU die from the top end of a particular models series that it takes to make the lowest end one, they just bin them and sort them to fill different markets.

The example above with the Athlon X2 pricing that RS so wisely posted shows this clearly. The X2's with 1MB Cache per core were sold at ~$900 (actually $1k+ for a little while) for the 4800+ model, while the 4400+ sold at $537, despite being the exact same CPU die (and virtually certain to hit the same stock speeds as the 4800+ anyway). ZOMG AMD was charging about $400+ for 200Mhz!? When it didn't cost them any more to make it? Alert the media!

Seriously. If anything, AMD's pricing when they've had highly competitive CPUs has been just as bad or worse than Intel. In the X2 days, they kept their lowest-end one at $300 for a long time, that 3800+ model. Today at least you can get a 3570K for less than that (considerably less factoring in inflation).

I'm not trying to beat up AMD here, but whenever I hear about how greedy Intel is, and how without AMD we're going to see sky-high prices, I just laugh.

What IS possible/probable, is that without competition from AMD, Intel may slow down release of new tech for longer periods, and I think we've perhaps already seen some examples of that. If BD had been an ass-kicking chip, who knows, we might have seen stock 4.4ghz and beyond i5/i7 models already (easily doable obviously), as well as a closer release date of Haswell.

 

[Ferzerp]

What IS possible/probable, is that without competition from AMD, Intel may slow down release of new tech for longer periods, and I think we've perhaps already seen some examples of that. If BD had been an ass-kicking chip, who knows, we might have seen stock 4.4ghz and beyond i5/i7 models already (easily doable obviously), as well as a closer release date of Haswell.

You're correct, but that state already exists really. There is absolutely no pressure to advance from AMD, and there hasn't been since the release on Conroe. Pricing hasn't increased, but who can say if they're holding back. No, we cannot assume that "moar cores!" not happening with Intel means they are holding back. We don't need consumer processors to go wider. The experience is no better with a wider processor. Scale up helps everything with the normal PC, scale out is of limited benefit. That's in no way proof of any sort of holding back. Scaling out is what you do when your other options are exhausted, because it's relatively easy, but is a heavy cost in die space.

 

[ShintaiDK]

Hmmm... if history shows anything, it's that Intel prices were only forced down by competition. A virtual monopoly on the market would allow them to ask for as much money as they can. And since the money would be a sufficient incentive, progress would likely stagnate.

That's the opposite of how a capitalist system works - you need real competition to have price pressures and similar pace of innovation. Lack of serious competitive price pressures generally results in terrible price competition - telecomm companies in US and Canada are the perfect example. If AMD is out, even if Intel is split into 2 "competing" companies, all it would do is create an illusion of competition. Even now Intel is already ripping us off. Instead of raising prices, they are selling us less die space (but that's directly related to CPU performance and core count).

Intel's first quad core came out in 2006 and by 2007, you were able to buy one for around $300-330 in the form of Q6600. Next year Intel is launching Haswell and it will still be $300 for a quad-core i7. If you don't see anything wrong with this picture, I don't know what to tell you. When AMD actually had competitive Athlon 64 CPUs, Intel wouldn't even dare charge extra for hyper-threading. Now they are tacking on an $80-100 price premium for this 1 feature in the i7 2600/2700/3770K. That's ridiculous and the only reason they are getting away with it is because Bulldozer has failed.

Also, when AMD CPUs had no competition from Intel, we had this:

Intel may lower prices if another competitor emerges, but your theory that in the long-run Intel will magically lower prices on their CPUs because they have less competition is an odd one.

First of all, we never had cheaper CPUs when Core 2 hit and even today. Yet Intel have basicly been without competition since July 2006.

You both seems to be confused about a few basic things in capitalism and production. First of all CPUs aint something you constantly need new of like power, food and other consumption types without the ability to choose. You could perfectly avoid buying a CPU for 5-10 years for example. Or even avoid buying an x86 CPU.

Intel needs to sell 400million CPUs to keep fabs running. And they would rather sell 500million a year to make more money. Consumers aint willing to pay whatever Intel wants, because then they just dont upgrade or buy an alternative product. And if Intel charged 600$ today for an i7 3570K. I doubt Intel would make any profit at all due to the lower volume. The CPUs today aint priced in relation to AMD. AMD is nonexistant competition wise. They are priced for the perfect balance of margin/volume to get the maximum profit. Increase or lower price and your profit drops.

in short, Intel needs to give you a reason to buy a new CPU. That means innovation, low prices and so fourth.

And RussianSensation, comparing a Q6600 to Haswell as quadcore vs quadcore? One can only facepalm for the lack of insight. Also about the HT. You get a speedbin+2MB cache+HT. And all within the same TDP. Thats not something that comes for free.

Back to the reason why AMD is actually giving you higher prices. That again refers to both R&D and volume to keep fabs running. Remember fab cost is rapidly increasing and you need higher volume to pay for them. Without higher volume you either slow down node process or increase price. And you cant increase price since it will reduce the volume.

Same reason why there aint room for GloFo either very soon. And the final foundry battle will be between TSMC and Intel.

 

[Ferzerp]

Amusingly, the die size graph agrees in no way with the claim it was posted to present. Each of Intel's die shrinks gives approximately the same number of transistors in half the area. The growth you see over time was always across a single node (45nm). The later section of the graph only shows new data points with a node shrink (and of course the die size will shrink). The last data point represents 4 times the transistor density of the point two before it... The transistor count is what is important, and that has grown. Or are you trying to suggest that the value of a processor is how much silicon that it takes up?

Each node shrink costs more and more to manufacture a given unit of die area. The node shrinks keep happening because they allow more chips to be made per wafer though (which cancels out the additional cost per area). Suggesting that a shrinking die size means that you're being ripped off is preposterous.

edit: the original Pentium was a 294mm^2 die. That must have been twice the value of an IB!

 

[ShintaiDK]

Yep, he counterargues his own post with that graph. Hillarious.

But it returns back to the point of thinking static costs and a consumption demand. Both fails in this industry.

 

[Ferzerp]

The graph shows *me* that IB is twice the transistors in half the die area of lynnfield

Something tells me that's not what he was going for. But hey, it's a picture, so it must be supporting of whatever words get attached to it, right?

edit: It's also misleading because the 32nm tick is not included (which is understandable in a way I guess since Westmere ended up low end or high end, but not mid range in the consumer space, but had such a part existed, there would be a 32nm data point before SB that would be around 160-170 mm^2)

edit2: But then again, it has kentsfield in there too... And that was not a single die, so.... Yeah....

 

[Jaydip]

Guys relax, AMD is in safe hands again.I just bought it Send me your CV guys and we shall start anew

 

[ocre]

Guys relax, AMD is in safe hands again.I just bought it Send me your CV guys and we shall start anew

for real?

are you being serious?

 

[Jaydip]

I dare you for questioning me child