AMD Q1 2015 Earnings - 23 cents a share loss, to exit dense server (SeaMicro)

[DrMrLordX]

What about where cost is critical? People don't buy BT/CT just due to perf/w and heat, they do it mainly because they are cheap.
Perf/$ is important too.

Actually, OEMs snap them up because they are cheap. Then they saturate the shelves at big-box retailers with machines featuring such CPUs and mark up anything better by $80-$100 or more to improve margins.

At least Cherry Trail will be a little faster . . . maybe. And all signs point to AMD moving Carrizo-L into the AIO space as a replacement for Kabinis, which will be a big improvement.

People can argue 'til they're blue in the face about whether or not Beema (Mullins) would have been suitable for laptops or tablets. Fact is, so few OEMs even gave us samples to examine that it's hard to know for sure how it would have stacked up vs. Bay Trail. But if you think Beema had the wrong kind of thermals for anything mobile, then it would have been just peachy in those desktop AIOs that have featured bad Kabini processors for going on two years now. And as an added bonus, the Carrizo-L incarnation of the Puma core should have higher clockspeeds than Beema/Mullins ever did.

OEMs like HP, Gateway, Lenovo, etc. sell a LOT of those AIO machines, and many of them feature bad Jaguars. Puma with a clockspeed bump could actually give Intel some headaches in that arena. Unlike in the notebook realm, OEMs have been more than happy to shlep cat-based CPUs in AIOs. Let's just hope they stick to the quads.

Anyone buying BT/CT over a core Celeron or Pentium for a desktop machine is utterly clueless.

You're right. Lots of people ARE clueless. They buy what's immediately available and do no homework. The OEMs and big-boxes are banking on that, as are many e-tailers.

The price difference is minimal, but the performance difference is big.

The price difference between getting a Bay Trail or Kabini and something better is often $100-$150 in an OEM machine. I don't know if I'd call that minimal. Sure, if you look at the tray prices, it isn't like that, but . . .

Did AMD not have access to Pentium M and its immediate successors? They were on par with or bested K8 in performance. All it lacked was 64-bit. At this point AMD was wiping the floor with Netburst. Yet did they not realize that all Intel had to do was bring Pentium M to the desktop and it would be all over? Core 2 even manages higher IPC than K8 without an IMC. Mismanagement at the highest level: can't just rest on your laurels like AMD did when they were able to sell $1000 CPUs.

A lot of enthusiasts were overclocking Dothan on desktop motherboards and hitting 2.8 ghz with them. They were good, but they were single-core. Still, they stacked up extremely well in single-threaded software (of which there was quite a bit back in '05 and earlier) versus K8. AMD really should have seen that coming.

Pentium M was releases Q1 2003. I would presume Phenom was in development to counter Conroe. But they should have also been developing a counter to Conroe+IMC (ie Nehalem). I'm guessing that was Bulldozer, which took too long, and they were then surpassed by Sandy Bridge.

Don't forget that AMD had a massively-parallel design (the so-called "K9") that consumed an unknown amount of time and R&D expenditure before cancellation. Stars was their go-to as a replacement for K9. Phenom was probably not seriously on the table as far back as '03, though I can't say so for sure.

 

[mrmt]

Seriously? Did you even bother to look?

(...)

But sure, go ahead and ignore that.

The segment you are mentioning isn't Enterprise, but Enterprise, Embedded and Semi-Custom. The Enterprise segment, meaning servers, shouldn't be much further from 20-30 million per AMD revenue share according to IDC Research, and AMD is basically a nobody as far as the embedded market is concerned, to the point that they never reported meaningful sales and never showed up in any industry report... that leaves us with the Semi-Custom segment for the bulk of the revenue, and the Semi-Custom segment only have the console contracts so basically all the sales on this segment are from the console chips. So basically all CPU sales at AMD income statement happens on the Computing and Graphics segment.

The 70%+ drop in CPU sales is real, it started with CMT designs first in 2012 and just got worse in 2013 and 2014, but was partially offset by higher cat core sales, but now with the demise of the cat core family the CPU business basically imploded.

 

[Boze]

There is nothing wrong with the FX series.

Many people are examining it through the lens of the latest Intel release, but that is flawed.

Technologically, it is a competitor to the i7 920 and was improved a little bit over the years compared to that.

But Sandy Bridge and beyond is a class of its own. Full stop.

Our little microcosm of enthusiasts knows this and AMD knows this, no sense in beating the dead horse about ST performance over and over.

AMD moved towards APU and getting design wins on that, and it works for them.

I posted this once and it is 100% relevant in supporting the fact that APU is largely their future.

"Looking at All In One PC's, sorting by best selling:

http://www.bestbuy.com/site/searchp...abcat0501005&qp=soldby_facet=Sold By~Best Buy

Number 1 is the $1,269.99 iMac

Number 2 is the $679.99 A8 powered HP Pavilion

AMD *is* doing something right, but in a lot of hardcore enthusiast eyes they are a complete and utter failure."

On what planet is losing money every quarter 'doing something right'?

I can tell you what they didn't do right... they didn't hire the people needed to redesign an Intel-challenging chip. And hopefully soon they'll up and die for it.

I look forward to them going out of business. Someone will step up to fill the void, maybe they won't have their heads up the asses like AMD's BOD.

 

[AtenRa]

20nm console chips would give lower revenue, not higher.

Why ?? volume and ASP will degrease because for the 20nm ???

 

[erunion]

Why ?? volume and ASP will degrease because for the 20nm ???

Per unit price decrease, thus less revenue per console sold.

 

[AtenRa]

Per unit price decrease, thus less revenue per console sold.

That is only if volume remains the same. Q2 and Q3 Console SoC sales will increase.

 

[Erenhardt]

Per unit price decrease, thus less revenue per console sold.

Do you have any details on supplying contract between amd and sony/ms for console soc?

 

[erunion]

Do you have any details on supplying contract between amd and sony/ms for console soc?

Are you suggesting that AMD would get to die shrink and pocket the difference as margins, contrary to past console agreements?

 

[Udgnim]

On what planet is losing money every quarter 'doing something right'?

I can tell you what they didn't do right... they didn't hire the people needed to redesign an Intel-challenging chip. And hopefully soon they'll up and die for it.

I look forward to them going out of business. Someone will step up to fill the void, maybe they won't have their heads up the asses like AMD's BOD.

being super optimistic about that

Intel tried getting into the dGPU space and eventually decided it wasn't worth continuing to spend R&D on it

who's going to compete in x86 versus Intel once AMD is gone?

AMD x86 license can't be bought out

 

[jhu]

who's going to compete in x86 versus Intel once AMD is gone?

Intel. They need to give people a reason to upgrade and give them money, otherwise why upgrade?

 

[TurtleCrusher]

Intel. They need to give people a reason to upgrade and give them money, otherwise why upgrade?

That is so naive. Don't be a monopoly apologist. They'll kill old hardware off with software. That's it.

 

[Mk pt]

Are you suggesting that AMD would get to die shrink and pocket the difference as margins, contrary to past console agreements?

AMD can keep same profit in every unit sold and Sony/Ms pay less.

A die shrink will make production of the chip cheaper.
Read: Production.

Maybe you don't know, but the producer can keep same profit selling for less $ if the production is cheaper.

That's what margins are. And that's what a die shrink in console SoC's are.


About AMD losing money in desktop cpu's and Apu's.. They hardly beat my 6 year old i7 920 I had with oc, and they can't beat my westmere with oc.
They stopped in time.

But I hope new gpu's will be nice. My R9 290 were the best bang for the buck, even with the bitcoin farming crazy.

But is hard to believe they can do a 'come back' against current and future (skylake) Intel offering.

 

[TurtleCrusher]

I think with 16nm there's a huge window about to be opened for a comeback. Just like GPUs, all but Intel has been stuck on 32/28nm. Ridiculous.

 

[ShintaiDK]

Why ?? volume and ASP will degrease because for the 20nm ???

The revenue would drop due to reduced unit cost. Even with improved sales. But AMDs profit may improve.

This is purely a shrink and cost saving for MS/Sony. AMD is just the supplier.

 

[ShintaiDK]

I think with 16nm there's a huge window about to be opened for a comeback. Just like GPUs, all but Intel has been stuck on 32/28nm. Ridiculous.

They still be in line after all the mobile products willing to pay more. Nothing changes with 16nm. It may just be more needed than 20nm was fro GPU makers and the like.

 

[AtenRa]

The revenue would drop due to reduced unit cost. Even with improved sales. But AMDs profit may improve.

This is purely a shrink and cost saving for MS/Sony. AMD is just the supplier.

This is the first time you are implying 20nm will be cheaper than 28nm. Until recently you were tooting how new nodes are more expensive than 28nm and only Intel managed to lower the price of their 14nm node. :whiste:

In order to have a decrease in Revenue, your selling price decrease should be greater than your volume increase.

 

[ShintaiDK]

This is the first time you are implying 20nm will be cheaper than 28nm. Until recently you were tooting how new nodes are more expensive than 28nm and only Intel managed to lower the price of their 14nm node. :whiste:

In order to have a decrease in Revenue, your selling price decrease should be greater than your volume increase.

No I dont.

A direct shrinks gives other advantages than per transistor cost saving in manufactoring. It improves yield and binning and a higher amount of useable chips per wafer in the edge. But I am sure AMD is locked by contract and not by price in terms of shrink.

What, you had some kind of idea that MS/Sony would pay more and raise the prices on consoles? No wonder you get AMDs financials so terrible wrong in your predictions.

 

[AtenRa]

No I dont.

A direct shrinks gives other advantages than per transistor cost saving in manufactoring. It improves yield and binning and a higher amount of useable chips per wafer in the edge. But I am sure AMD is locked by contract and not by price in terms of shrink.

What, you had some kind of idea that MS/Sony would pay more and raise the prices on consoles? No wonder you get AMDs financials so terrible wrong in your predictions.

Seams you dont understand basic economics,

We know that Console SOC sales(Volume) will increase in Q2 and peak in Q3 2015.

Unless AMD will degrease its margins for those SOCs going to 20nm, then there is no way Revenue from those sales will decrease.

I will let you think about it for a moment 😉

 

[Lonyo]

Seams you dont understand basic economics,

We know that Console SOC sales(Volume) will increase in Q2 and peak in Q3 2015.

Unless AMD will degrease its margins for those SOCs going to 20nm, then there is no way Revenue from those sales will decrease.

I will let you think about it for a moment 😉

Revenue and margins are unrelated.
If AMD sells 100 28nm chips at $100 with 40% margins you get:
100*100 = $10,000 revenue. $4,000 profit.

If AMD sells 120 20nm chips at $80 with 40% margins you get:
120*80= $9,600 revenue. $3,840 profit.
Same margins, more sales, less revenue. If margins increased, they would get more profit, if margins decreased they would get less profit. The revenue has nothing to do with margins though, it has to do with gross price.

 

[ShintaiDK]

Seams you dont understand basic economics,

We know that Console SOC sales(Volume) will increase in Q2 and peak in Q3 2015.

Unless AMD will degrease its margins for those SOCs going to 20nm, then there is no way Revenue from those sales will decrease.

I will let you think about it for a moment 😉

I wouldnt try to lecture anyone in economics after the blunders you make quarter after quarter.

So arcording to you, there is no cost saving for Sony/MS going 20nm. Because AMD will get all the benefits.

Lets see what happens.

 

[AtenRa]

Revenue and margins are unrelated.
If AMD sells 100 28nm chips at $100 with 40% margins you get:
100*100 = $10,000 revenue. $4,000 profit.

If AMD sells 120 20nm chips at $80 with 40% margins you get:
120*80= $9,600 revenue. $3,840 profit.
Same margins, more sales, less revenue. If margins increased, they would get more profit, if margins decreased they would get less profit. The revenue has nothing to do with margins though, it has to do with gross price.

Yes i forgot to say they will not decrease both revenue and their profits as well :whiste:

 

[AtenRa]

So arcording to you, there is no cost saving for Sony/MS going 20nm. Because AMD will get all the benefits.

Lets see what happens.

I havent said anything about savings on MS and Sony, did I ?? I have only said that revenue will not decrease going to 20nm if you keep your margins and profits (forgot to say that).

Now if you believe AMD will loose income going to 20nm then that is something else. 😉

 

[ShintaiDK]

Margins and profits are calculated out of revenue. If a chip is 25% cheaper and you increase margins and units 10%. You still end up with lower revenue.

Nobody talks about income.

 

[mrmt]

I think with 16nm there's a huge window about to be opened for a comeback. Just like GPUs, all but Intel has been stuck on 32/28nm. Ridiculous.

They have 0 node advantage over Nvidia today and yet their market share on the GPU business was never so low. In fact, if Nvidia is an example, their share might even go further down on the CPU market even after migrating for a new node.

 

[mrmt]

Margins and profits are calculated out of revenue. If a chip is 25% cheaper and you increase margins and units 10%. You still end up with lower revenue.

Nobody talks about income.

This revenue discussion is a red herring, as it doesn't matter AMD makes from each chip sold, but how much it makes after paying for manufacturing costs. This gross profits is what matters, and usually in an embedded contract this tends to go down to account for yields increase. Also if AMD moves to 20nm the gross revenue should be down but gross profits shouldn't, so while it may reduce AMD first line it a move neutral from a bottom line POV.

E.g:

(...)

Lisa Su - President and CEO
They do. But the peak, you know, the peek quarter for semicustom is always going to be the third quarter because that's when they will build ahead of holidays. What we saw in the first half of 2014 was more, I would say, demand that was not satisfied from the previous holiday.

Stacy Ragson - Bernstein
Got it. It's fair to say that the pricing will be down year to year, correct?

Lisa Su - President and CEO
It's fair to say there are some ASP decline, that's correct.

ASP on the console chips will go down YoY, whether this will impact gross profits is a different story. I think it will, but in a smaller scale than the full price.