AMD Q3 Results

[MiddleOfTheRoad]

Doesnt this go back to the ATI purchase though? They had so much debt they had little choice but to sell them off.

I really wish AMD had purchased S3, 3dfx or XGI (instead of ATi) on the cheap. Had they had that GPU design team, kept their fabs and a lower debt load -- they'd be a much stronger company right now. I tinkered with S3 and XGI video cards -- they made decent hardware.... The drivers, meh... not so much.

 

[Idontcare]

I thought it was Hector that did the fab deal.

He was, the context of my list was about CEO's that oversaw large transitions within AMD.

I wasn't intending to attribute to Meyers the decision or leadership vision that went into initiating the spinning off of the fabs, that was Ruiz and the fabulous AMD BoD, but I was trying to attribute the ensuing changes in corporate culture and business direction to that Meyers.

Haven't read my own post since I posted it, but I'm gonna guess I totally botched the delivery of my intended context given that there are quite a few replies to it. oops!

 

[CHADBOGA]

I wasn't intending to attribute to Meyers the decision or leadership vision that went into initiating the spinning off of the fabs, that was Ruiz and the fabulous AMD BoD, but I was trying to attribute the ensuing changes in corporate culture and business direction to that Meyers.

Not sure that I would use the term "leadership vision" here. 😉

 

[DrMrLordX]

I really wish AMD had purchased S3, 3dfx or XGI (instead of ATi) on the cheap. Had they had that GPU design team, kept their fabs and a lower debt load -- they'd be a much stronger company right now. I tinkered with S3 and XGI video cards -- they made decent hardware.... The drivers, meh... not so much.

S3 sold out to VIA in 2001, 3dfx got snapped up by Nvidia in 2000 (during bankruptcy), and XGI was picked apart between 2006 and 2010 by ATI and SiS. Of the three companies mentioned, the only one that would have been available during any stage of AMD's Fusion initiative would have been XGI (unless AMD wanted to try buying out S3 from VIA, which is essentially what happened when HTC bought S3 from VIA in 2011).

 

[AtenRa]

Yes, I do.

They were second only to Intel at that point. Had they diversified and opened a foundry division much like Samsung then they could have kept their fabs, their process advantage, and grown revenue to continue paying for R&D.

AMD was leasing IBMs process, they didn't spend much on R&D to produce their own. It is why they went with 32nm Gate First SOI, exactly the same process IBM had. And they would transfer to IBMs 20nm Gate First SOI later on in 2014.
Also, im not sure how much more capacity they had to produce for more than AMD with a single fab at the time.
And im not familiar with the leasing terms about any exclusivities, could they use each IBM process and also produce for others or they could only use it internally for AMD products ?? There was a 32nm SOI AMD exclusivity with GloFo, something no one ever asked why.

 

[NostaSeronx]

AMD was leasing IBMs process, they didn't spend much on R&D to produce their own.

AMD's 130-nm SOI to 45-nm SOI was what IBM's nodes were based on. It isn't till the 32-nm PD-SOI node, can you accurately claim that AMD was leasing IBM's SOI. In fact, 32-nm PD-SOI was IBM's SOI in name only. All research and development of 32-nm PD-SOI's FEOL was at GlobalFoundries.

UTBB FD-SOI is a near direct copy of AMD's FD-SOI work from 2004 to 2007.

The 14-nm node UTBB FD-SOI node, is essentially the shrink of the 22-nm FD-SOI node, that AMD was going to detail in 2009. Since, AMD already detailed the 32-nm PD-SOI node in 2008.

The reason AMD dropped 22-nm PD-SOI was because ET-SOI was a year away.

GlobalFoundries gave AMD two choices in 2009.
- 2012 for PD-SOI
- 2013 for ET-SOI

ET-SOI unlike PD-SOI uses standardized Bulk EDA flows. This comes in handy with hybridization that ET-SOI can do. Essentially, AMD could design for Bulk or ET-SOI or both. All using the same tools, [Open]PDKs, etc.

 

[ShintaiDK]

AMD was leasing IBMs process, they didn't spend much on R&D to produce their own. It is why they went with 32nm Gate First SOI, exactly the same process IBM had. And they would transfer to IBMs 20nm Gate First SOI later on in 2014.
Also, im not sure how much more capacity they had to produce for more than AMD with a single fab at the time.
And im not familiar with the leasing terms about any exclusivities, could they use each IBM process and also produce for others or they could only use it internally for AMD products ?? There was a 32nm SOI AMD exclusivity with GloFo, something no one ever asked why.

AMD used 50-70% more on R&D before the fab spinoff than they do today. But again, they didnt starve the R&D pipeline back then.

The selling of the fabs simply made sure AMD would never get a comeback and is now settled in the big blob of fabless companies.

 

[Idontcare]

AMD was leasing IBMs process, they didn't spend much on R&D to produce their own. It is why they went with 32nm Gate First SOI, exactly the same process IBM had. And they would transfer to IBMs 20nm Gate First SOI later on in 2014.
Also, im not sure how much more capacity they had to produce for more than AMD with a single fab at the time.
And im not familiar with the leasing terms about any exclusivities, could they use each IBM process and also produce for others or they could only use it internally for AMD products ?? There was a 32nm SOI AMD exclusivity with GloFo, something no one ever asked why.

You are referring to the earlier decision to join the fab club and decimate the ranks of their internal process node R&D team, beginning with the 45nm node.

There were two transitions, the first was to go R&D-less for process node development. To outsource the R&D to the fab club, which was essentially an IBM led show in Fishkill.

That choice is what put them way behind in terms of node-release timeline because IBM was never one to worry about the lag between their nodes and that of the competition to their fab-club partners.

So long as the IBM node XYnm was out in time to compete with Intel whenever Intel finally released Itanium on node XYnm (usually about 2 yrs after x86 on node XYnm was released), IBM was fine. Not so for the other fab-club members.

This precipitated the inevitable situation in which AMD found itself fielding parts that could not command the ASPs necessary to command the margins and generate the revenue which were necessary to justify continuing to maintain the fabs.

(quite literally the exact same situation that IBM has found itself right now such that they find themselves compelled to sell their fabs to GF)

 

[mrmt]

You are referring to the earlier decision to join the fab club and decimate the ranks of their internal process node R&D team, beginning with the 45nm node.

But before joining the fab club (wasn't it on 65nm?), wasn't AMD licensing nodes from motorola? Afaik AMD never developed nodes on its own.

 

[TurtleCrusher]

The selling of the fabs simply made sure AMD would never get a comeback and is now settled in the big blob of fabless companies.

Either you're a paid poster or you have an unhealthy, negative obsession with anything AMD. You might want to step back from your computer and get some fresh air. You've been knocking AMD since god knows when and they're still here, competing. They have solid marketshare in video cards and budget OEM systems. Their products are in each of the major video game systems. While their CPU marketshare isn't the best, they're getting their products in more households than ever before.

The doom and gloom outlook you have is laughable.

 

[CHADBOGA]

Either you're a paid poster or you have an unhealthy, negative obsession with anything AMD. You might want to step back from your computer and get some fresh air. You've been knocking AMD since god knows when and they're still here, competing. They have solid marketshare in video cards and budget OEM systems. Their products are in each of the major video game systems. While their CPU marketshare isn't the best, they're getting their products in more households than ever before.

The doom and gloom outlook you have is laughable.

He is a technology enthusiast, calling it like he sees it.

Not every company has a rosy future to look forward to.

 

[krumme]

He is a technology enthusiast, calling it like he sees it.

Not every company has a rosy future to look forward to.

Enthusiasm is great - but i think its more than that - because there is an overload in this forum with crazy stuff as evident eg. in this thread also:
http://forums.anandtech.com/showthread.php?t=2404898

"[ASML] Intel to introduce EUV in 2016"

Sometimes its funny like the classic above, but the blind onesided approach can get boring.

The usual doom and gloom people also predict when there will be no AMD product:
"My guess is that this will happen once Intel 10nm products reach the market" (mrmt. SA, okt. 2014)

If thats the future i agree with you - its not rosy. Especially not as consumer - i guess you view yourself as consumer or what? - because sometimes reading these forums one have to doubt what identity people prefer.

 

[ShintaiDK]

Its always easier to attack posters rather than looking on realities and accept it. Its the same thing over and over again every single time.

AMD even said a 13% revenue drop is coming plus 7% layoffs in Q4. These are not happening because everything is rosy. And the consoles is only giving them so much for so long.

Q3 2012.
CPU+Graphics divisions. 1269M$ revenue. 328M$ R&D.

Q3 2014. (Without consoles revenue.)
CPU+Graphics division. 781M$ revenue. 278M$ R&D.

And with a projected revenuedrop company wide for Q4 on around 200M$ and having to fire 7% of the workforce.

Then people can scream and jump and whatever while claiming something else. But this is the reality.

AMD was only second to Intel when they still had their fabs, due to their fabs.

 

[Abwx]

Its always easier to attack posters rather than looking on realities and accept it. Its the same thing over and over again every single time.

AMD even said a 13% revenue drop is coming plus 7% layoffs in Q4. These are not happening because everything is rosy. And the consoles is only giving them so much for so long.

Reality is that their competitor supplied almost 10% of the X86 market with free chips and subsides to use thoses sub par CPUs in the very segment where AMD s offering is the most competitive, this has the consequence that they didnt,or wont, met their revenue target in quarterly revenue for Q3/Q4.

Now if you like to talk reality you can only try to give us your estimation of how much revenue was lost due to thoses dubbious practices, i say 75-100m for Q3 and 100m+ for Q4.

 

[ShintaiDK]

I am quite aware that the notion is its always someone elses fault but AMDs.

I am sure you also have plenty of excuses for all the previous years and quarters as well.

Lets just recap with a few more years.

Q3 2008.
CPU+Graphics divisions. 1776M$ revenue. 422M$ R&D.

Q3 2010.
CPU+Graphics divisions. 1616M$ revenue. 359M$ R&D.

Q3 2012.
CPU+Graphics divisions. 1269M$ revenue. 328M$ R&D.

Q3 2014. (Without console/embedded revenue.)
CPU+Graphics division. 781M$ revenue. 278M$ R&D.

 

[Abwx]

I am sure you also have plenty of excuses for all the previous years and quarters.

Lets just recap with a few more years.

The trend is quite clear.

What matters is the current situation, you can come with as much years as you want this wont change the fact that they have currently the best product for the mainstream tablet/notebook, and that they are barred from selling it at a time where they are in dire need of revenue.

If they had no competitive product in this segment then you could say that they have no excuse but point is that the product is here, just that their competitor artificialy locked this market using its cashes reserves.

In this respect Lisa Su did the right move by agressively reducing prices in the very segment where Intel make the bread and butter that allow them to throw cash for free as a mean to be "competitive".

 

[dealcorn]

Reality is that their competitor supplied almost 10% of the X86 market with free chips and subsides to use thoses sub par CPUs in the very segment where AMD s offering is the most competitive, this has the consequence that they didnt,or wont, met their revenue target in quarterly revenue for Q3/Q4.

Intel does not use contra revenue for Bay Trail-D or Bay Trail-M (desktops and laptops respectively). These non subsidized, full margin products are what took major market share from AMD during Q2, Q3 and Q4(projected). Intel used contra revenue for tablets (Bay Trail-T) where it's market share exploded from ~nothing to ~20%. The reason AMD took no market share from ARM (the 80% piece) is that it's products are not competitive with ARM. Bay Trail-T took no market share from AMD because it had no tablet market share to yield.

Because in the tablet market Intel went from nothing to a bit player, from an anti trust perspective contra revenue is a non event. It is OK for a non player to lose money entering a new market. Consumers benefit.

 

[Abwx]

Intel does not use contra revenue for Bay Trail-D or Bay Trail-M (desktops and laptops respectively). These non subsidized, full margin products are what took major market share from AMD during Q2, Q3 and Q4(projected). Intel used contra revenue for tablets (Bay Trail-T) where it's market share exploded from ~nothing to ~20%. The reason AMD took no market share from ARM (the 80% piece) is that it's products are not competitive with ARM. Bay Trail-T took no market share from AMD because it had no tablet market share to yield.

That s completely wrong, almost all thoses subsided chips are used in W8.1 tablets and notebooks, and this is directly impacting AMD who did target this market with their Mullins chip that is much better than Btrail, and is even rivaling Core M if we are to believe the Yoga 3 recent benches.


If there was no subsides then you wouldnt see W8 tablets and notebooks at 200$ and less, numbers are numbers and they dont lie.

Because in the tablet market Intel went from nothing to a bit player, from an anti trust perspective contra revenue is a non event. It is OK for a non player to lose money entering a new market. Consumers benefit.

Consumer benefit.?.
So it s all benefit for the consumer if he cant buy a product that has a chip that is better than what is forced to the consumer by Intel s shenanigans.?.
You are surely joking i hope because Btrail is one gen behind Mullins, that is, a big benefit that being forced to buy obsolete gear, for Intel of course...

Edit : you are misleading the general public by stating false information about perfs, Mullins is better than either BT, Snapdragon 805 or any other such chips, do your homework next time...

 

[raghu78]

Its always easier to attack posters rather than looking on realities and accept it. Its the same thing over and over again every single time.

AMD even said a 13% revenue drop is coming plus 7% layoffs in Q4. These are not happening because everything is rosy. And the consoles is only giving them so much for so long.

Q3 2012.
CPU+Graphics divisions. 1269M$ revenue. 328M$ R&D.

Q3 2014. (Without consoles revenue.)
CPU+Graphics division. 781M$ revenue. 278M$ R&D.

And with a projected revenuedrop company wide for Q4 on around 200M$ and having to fire 7% of the workforce.

Then people can scream and jump and whatever while claiming something else. But this is the reality.

AMD was only second to Intel when they still had their fabs, due to their fabs.

Funnily AMD was GAAP profitable in Q3 2014 inspite of all the gloom talk. The Q4 projection is really bad but still AMD's fiscal discipline means they are not burning through cash as they did in 2012 and to some extent 2013. AMD has a difficult 18 months till they get their next gen ARMv8 and x86-64 architectures to launch. Their long term profitability and sustainability does depend on the 2016 architectures and their ability to diversify the revenue base further from the PC market where Intel has a stranglehold.

btw no way can AMD sustain the fab investments at 14nm, 10nm and 7nm even if they had the revenue which they had in 2006 (the peak of their Opteron success). In fact IBM sold their fab division because they did not have the wafer and product volumes required to sustain billions of dollars in fab spending at the leading edge.

Today other than Intel nobody has the product volume to sustain leading edge process R&D and fab investments. Even Samsung has to compete with TSMC and get sufficient foundry orders to sustain process R&D and leading edge fab investments.

 

[Abwx]

Their long term profitability and sustainability does depend on the 2016 architectures and their ability to diversify the revenue base further from the PC market where Intel has a stranglehold.

It depend on investment made now in RD and for this you must have the money.


Btw, Intel had and still has no stronghold in the mobile devices market and AMD did the necessary to be competitive in this market, but you know what followed, since Intel wasnt competitive technicaly they litteraly kept this market from being owned partialy by AMD by injecting massive amounts of cash, it must be a world record where the overall best performing chip did had a single design and still it s not launched yet, so it s not all to claim that they have to look for markets where Intel has no position, as you can see that dont work this way.

 

[Phynaz]

Q1 2013
Revenue = 1.088B
Income (Loss)= (98M)

Q1 2014
Revenue = 1.397B
income = 49M

---------

Q2 2013
Revenue = 1.161B
Income (Loss) = (29M)

Q2 2014
Revenue = 1.441B
income = 63M

-----------

Q3 2013
Revenue = 1.461B
Income = 95M

Q3 2014
Revenue = 1.429B
income = 63M

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

Q1+Q2+Q3 2013
Revenue = 3.71B
Income (Loss)= (32M)

Q1+Q2+Q3 2014
Revenue = 4.267B
Income = 175M

Your doom and gloom in everything AMD is pathetic.

Edit : Lets estimate Q4 2014 and entire year

Q4 2013
Revenue = 1.589B
Income = 135M

Q4 2014 (estimation with -13% Revenue from Q3)
Revenue = 1.243B
Income (Loss) = (10M) ?? (estimation)


-----------

2013
Revenue = 5.299B
Income = 103M

2014
Revenue = 5.51B
Income = 165M

Just a reminder that AMD reports in a few days. Are you still standing by this post?

 

[AtenRa]

Just a reminder that AMD reports in a few days. Are you still standing by this post?

With their guidance for -13% revenue from Q3 2014?? yeap, it should be close to that.

 

[Phynaz]

With their guidance for -13% revenue from Q3 2014?? yeap, it should be close to that.

No, not AMD's guidance (unless you are responsible for that), but the numbers you put together from their guidance - for example showing a $165M profit for 2014.

 

[AtenRa]

No, not AMD's guidance (unless you are responsible for that), but the numbers you put together from their guidance - for example showing a $165M profit for 2014.

The numbers are Operating Income.

 

[RussianSensation]

Considering how poorly GloFo has performed financially since AMD's spin off, I don't know how anyone here can even make a point that it would have been better to keep the fabs for AMD. The fab business is very difficult on its own as GloFo has struggled against TSMC. AMD would not have been able to have the money necessary to invest into those fabs to make them competitive enough because even as a stand-alone, GloFo is a horribly run business right now despite tens of billions having been sunk into it by Mubadala Group with $0 net return so far:
http://mobile.reuters.com/article/idUSBREA020EG20140103?irpc=932

Going into tablets and smartphones is also a failing strategy considering Intel flushed $4.2 billion in 2014 into the toilet trying to establish the Atom brand name while NV more or less brushed aside the smartphone and tablet market, instead focusing on the faster growing in-car tech business opportunities. Small companies like AMD can't afford to use their other profitable product lines to blow $4 billion as a possible gateway into new segments.

All the other anti-AMD comments are from the same posters, some of whom hate everything AMD, even the consoles. As I said only the blind f*n***s, Intel/NV employees/sharehoders, AMD stock put/shorter owners, ex-AMD buttburt employees would want for AMD to fail leaving us with Intel and NV only. I sincerely hope AMD recovers and we get a lot more needed competition in the CPU and GPU space. It also would be nice to have AMD bid for next gen console parts as that creates more choices and competition for other bidders. Also, can't wait for 980 owners to defend their precious GM204 purchase when 390 series mops the floor with it, when some of them have all but written off AMD's GPU division.

The next 24 months will be painful but this should have been expected as their CPU and GPU architectures have aged. I know in the GPU forum, a lot of people have already written off R9 300 series. I don't judge a company's future product based on its balance sheet position or popularity or badge or blind loyalty unlike some posters here from whom nothing positive ever comes out regarding AMD discussions.