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

[mrmt]

Sure the price may have been too high, just like when Intel bought McAfee for $7.7B! But the question is whether or not it's good business strategy to own the GPU part of an APU (compared to e.g. licensing it from an external partner).

McAfee was 15% of Intel annual revenue, ATI was 100% of AMD annual revenue. Intel paid the acquisition with shares and cash generated by its business, AMD had to make a huge leveraged buyout in order to close the deal. See why you can't compare both?

 

[mrmt]

I am curious how many commentators here work for a large to very large corporation.

Because if you are working for a company that does not have cold hard cash coming in every month to pay bills, you are screwed.

If a large company truly has problems, IE, no cash flow at all, it would be chopped up and sold in a heartbeat.

Maybe you should tell that to AMD CFO, every single quarter he has to show a couple of slides, one showing expenditures cuts, the other showing how they are managing liquidity. Cash is a problem for AMD, period.

 

[frozentundra123456]

I suspect what's actually going to happen is that the R9 390X is going to give Nvidia TITAN X cards a run for its money. I'll also run hotter, use more power, and be around 25-30% cheaper. Same damn thing we see from the rest of the R9 line almost across the board.

Zen will be lauded by tech people as a great product, but it has no home. You can't sell it to enthusiast gamers because we'll just buy a top-of-the-line Intel processor and pair it with a top-of-the-line NVIDIA GPU. You can't sell it to the corporate world, because they don't need the integrated graphics power it provides... an Intel Core processor is more than enough for business. The only people interested are budget gamers, and they're not enough to make up the difference. Console gamers are like the humans from Men in Black. For the most part, don't have a clue; they don't want one or need one, either, they're happy... they think they have a good... bead on things.

AMD will declare bankruptcy and/or sell to Samsung or some other large multinational technology corporation that can leverage their IP portfolio.

Steam boxes take over gaming; Microsoft, Sony, & Nintendo all become 3rd party developers.

May not work out quite like that, but I can dream...

I agree with you for the most part. The one place where I would have interest in Zen is if they can leverage HBM to improve igpu performance enough to make a decent midlevel gaming laptop without a discrete card. But they also have to make a considerable improvement in cpu performance and sell it at a considerably cheaper price than whatever intel/nvidia offers when Zen is finally available.

Zen is also a totally unknown product at this point except for some statement that AMD fans are hanging onto for dear life about a hulking 300 watt apu, which I have no interest in anyway. And considering the apparent difficulties in getting to the smaller nodes, there is absolutely no assurance that Zen will come out on time. In fact, I will be very surprised if it does, considering they are introducing a new process node, a new architecture, new memory technology, and also developing ARM. Seems like an extremely ambitious endeavor, especially for a company with the limited resources of AMD. I mean, how long have we been waiting for 390X, and that would seem a much simpler undertaking Zen.

 

[pablo87]

70%?

AMD today is worth 35-40% of what they payed for ATI.

What is ATI worth today though?

A) Market - AMD's EV = market cap + Long Term Debt = $4B. Arguably, the WSA is a $2B off balance sheet liability so EV is really $6B (just a lot of debt and liabilities!!!). What is x86 worth (w/o ATI, w/o WSA, w/o LTD), $1B? $2B? Not much more. Based on this, ATI would be worth anywhere from $3-$5B.

B) Comparative - based on NV's EV which is 12.3B+1.4B (debt) less $4B excess cash (my estimate) = $9.7B. @1/3 of NV, ATI would be worth $3.2B, @50% $4.85B.

So $3 to $5B.

 

[pablo87]

I think this entire thread can be summed up quite simply with three words:

Board of Directors

Fixed that for you.

 

[Phynaz]

It's because the customer does not have any option if they want to buy an Intel CPU. The iGPU is included on the die regardless.

If the customer could have bought an Intel CPU with an iGPU from AMD or nVidia, you can bet they would.

So back to the original question: Would Intel have been better better off licensing the iGPU from e.g. nVidia or AMD/ATI compared to designing their own from scratch?

What Steamroller CPUs are available without an iGPU?

Intel has bought a couple of graphics companies, and has graphics cross licensing deals in place with both Nvida and AMD.

 

[mrmt]

It's because the customer does not have any option if they want to buy an Intel CPU. The iGPU is included on the die regardless.

If the customer could have bought an Intel CPU with an iGPU from AMD or nVidia, you can bet they would.

So back to the original question: Would Intel have been better better off licensing the iGPU from e.g. nVidia or AMD/ATI compared to designing their own from scratch?

I don't know if you realized but you just validated Intel iGPU strategy. Give the consumer a low performance iGPU but top notch CPU and you'll end up with a good product and making tons of money. AMD strategy of sacrificing CPU performance for the sake of the iGPU didn't work. It doesn't matter whether AMD has a better iGPU, the consumers don't care at all.

 

[Lonyo]

Narrowminded crap. As long as there political interest in investing in gf tech - mubadala - there is cash for amd. Unless you beliewe gf can beat tsmc and you got to be insane to have that thought.

Never in amd history have it been more secure than now. Sharevalue will go near zero, ownership structure will change, but cash will be there.

So AMD is secure because you think that it will be basically bought by the same people who bought GF.

http://www.gartner.com/newsroom/id/3027717
Apparently GF has $4.4b in revenue.
We know for a fact that AMD is responsible for $1b of that revenue only, and has been for the last few years, since they are under the WSA and it's $1b as the floor they have to pay (in 2013, 2014 and probably 2015).

So AMD makes up around 23% of GF revenues (but doesn't use 23% of capacity).
Therefore, ATIC (GF owner) is going to buy AMD and invest billions just to protect 23% of their current revenue? Really?

GF is also continuing its capital investment in fab capacity and new fabs. We know that AMD isn't hitting the WSA and is not looking like it will increase its orders to GF, therefore any new investment in capacity is unrelated to AMD and is all non-AMD growth expectation. So they are investing in GF tech DESPITE AMD's inability to utilise that investment, so clearly AMD is in fact not that important to GF's overall plans because it operates in a global market for semiconductors which includes other parties who make up 77% of its revenues and is aiming to increase that revenue from exclusively non-AMD sources. The $1b hit would hurt, but it wouldn't make sense to put billions into AMD, which has shown it's incapable of competing while relying on GF just to secure a billion a year in revenue.

Also consider that if GF can't compete with TSMC, then AMD would need to continue using TSMC for their GPU products if they want those to remain competitive. Or do you think that GF would buy AMD, then use only GF foundries for all products, using non-competitive process tech to make their GPUs that would then also be non-competitive? (Plus the CPU tech which already can't compete with Intel).

 

[pablo87]

AMD's quarterly overhead including interest payments is ~ $410M - little change from Q4 to Q1. To break-even, they'd have to generate $1.2B revenue @ 35% gross margins. That seems to be the best, best case scenario for the rest of this year.

If the rest of the year stays the same as Q1, they'll probably burn through $300M (my WAG). Its a distinct possibility since Intel and NVidia aren't going to stand by and let AMD take share from them - as we all know, there's a tradition in silicon valley called meet comp. Moreover, the market is shrinking, if demand falters, the meet comp will become arm twisting aggressive.

Also, comparing a flat Revenue scenario for the rest of 2015, and the $839M remaining commit to GF under the WSA for 2015 versus the $1B actual in 2014, adjusted for the inventory reduction that transpired in late 2014/early 2015, it seems they're overcommitted by $200M. In other words, inventory could rise by $200M and cash reduce by $200M on that basis alone. Or the WSA will be re-negotiated for the umpteenth time.

Either way, that 2015 WSA commitment ain't going to be cash flow positive!!!

 

[pablo87]

Narrowminded crap. As long as there political interest in investing in gf tech - mubadala - there is cash for amd. Unless you beliewe gf can beat tsmc and you got to be insane to have that thought.

Never in amd history have it been more secure than now. Sharevalue will go near zero, ownership structure will change, but cash will be there.

Krumme, Mubadala is probably getting tired of the losses at Global Foundries, the oil business isn't generating cash like it did in the past. That would explain why Global Foundries is squeezing AMD for every pound of flesh - less money from Mubadala = more money from AMD. As the Chinese saying goes, better you cry than I cry.

 

[pablo87]

I don't know if you realized but you just validated Intel iGPU strategy. Give the consumer a low performance iGPU but top notch CPU and you'll end up with a good product and making tons of money. AMD strategy of sacrificing CPU performance for the sake of the iGPU didn't work. It doesn't matter whether AMD has a better iGPU, the consumers don't care at all.

I don't agree with this at all. Consumers that are still buying - kids - care. Most complain about having a GPU that is too slow, virtually no one asks about a faster processor. Tied for 1st might be an SSD. Battery life might be the 3rd vector in demand. Better screen might be 4th. Processor is perhaps 5th.

Just my take, obviously hard to survey customers as they don't know what they're missing out on.

It means Intel has done a great marketing job. But they've squeezed what they could out of the Consumer PC, and now demand is falling. Time for a new strategy.

 

[Fjodor2001]

McAfee was 15% of Intel annual revenue,

It was still a bad deal and bad business for Intel.

 

[Fjodor2001]

What Steamroller CPUs are available without an iGPU?

You're missing the point.

The point is that Intel selling lots of APUs with iGPU doesn't mean their iGPU is any good (which some here implied). In fact it's bad compared to the competition, but something the customers are forced to accept if they want to buy an Intel APU. I.e. they are willing to accept that, since the CPU part of the APU is good.

Intel has bought a couple of graphics companies, and has graphics cross licensing deals in place with both Nvida and AMD.

What we're discussing here is the option of licensing the complete iGPU as an IP-block, like ShintaiDK suggested AMD should do.

If that's such a great solution, then why didn't Intel go down that route? They could have licensed a much better iGPU from nVidia or AMD compared to what their own R&D department has been able to design.

 

[DrMrLordX]

What Steamroller CPUs are available without an iGPU?

The Athlon x4 860k and 840, FX 770k. Might be others.

 

[Fjodor2001]

I don't know if you realized but you just validated Intel iGPU strategy. Give the consumer a low performance iGPU but top notch CPU and you'll end up with a good product and making tons of money. AMD strategy of sacrificing CPU performance for the sake of the iGPU didn't work. It doesn't matter whether AMD has a better iGPU, the consumers don't care at all.

Sure, you can always argue what to prioritize, iGPU vs CPU performance. But that's another story.

The point is that Intel could have gotten a better iGPU if they had bought or licensed it from nVidia or ATI/AMD, instead of Intel trying to design the iGPU on their own. Intel is actually sacrificing a huge part of the die area to iGPU these days. That's because their design is worse than AMD's and nVidia's, so they have to compensate by adding more GPU cores and thus die area and to it. I.e. it is less cost efficient.

Sacrificing die area due to bad iGPU design is never good. That is regardless of to what extent you chose to prioritize iGPU over CPU performance, or vice versa.

 

[jpiniero]

Sacrificing die area due to bad iGPU design is never good. That is regardless of to what extent you chose to prioritize iGPU over CPU performance, or vice versa.

Intel has a die size it 'likes' to maintain for the mainstream line. Since they aren't going to put more CPU cores in, something has to. Hence the GPU. Plus it's (apparently) easier on yield and hedges against GPU performance suddenly becoming important.

A big part of Intel's GPU problem is that for games you need to spend the money on drivers to develop the "optimizations" (cheats) for every game. If you don't do it, performance suffers immensely. You can see this in how much Kepler performance has tanked for most recent games.

 

[AtenRa]

McAfee was 15% of Intel annual revenue, ATI was 100% of AMD annual revenue. Intel paid the acquisition with shares and cash generated by its business, AMD had to make a huge leveraged buyout in order to close the deal. See why you can't compare both?

Would you like to count how much money ATI generated for AMD until today and how much McAfee for Intel ???

Count the operating Income from GPUs, APUs and Consoles from 2006 to 2014, you will find its a nice big positive number. I dont think you can say the same for McAfee after the Intel acquisition 🙄

 

[AtenRa]

AMD's quarterly overhead including interest payments is ~ $410M - little change from Q4 to Q1. To break-even, they'd have to generate $1.2B revenue @ 35% gross margins. That seems to be the best, best case scenario for the rest of this year.

If the rest of the year stays the same as Q1, they'll probably burn through $300M (my WAG). Its a distinct possibility since Intel and NVidia aren't going to stand by and let AMD take share from them - as we all know, there's a tradition in silicon valley called meet comp. Moreover, the market is shrinking, if demand falters, the meet comp will become arm twisting aggressive.

Also, comparing a flat Revenue scenario for the rest of 2015, and the $839M remaining commit to GF under the WSA for 2015 versus the $1B actual in 2014, adjusted for the inventory reduction that transpired in late 2014/early 2015, it seems they're overcommitted by $200M. In other words, inventory could rise by $200M and cash reduce by $200M on that basis alone. Or the WSA will be re-negotiated for the umpteenth time.

Either way, that 2015 WSA commitment ain't going to be cash flow positive!!!

Read the following from the Q1 2015 earnings call,

http://seekingalpha.com/article/307...arnings-call-transcript?page=4&p=qanda&l=last

Devinder Kumar - SVP and CFO
So, good question. I think two parts to it. If you go back and look at our inventory profile even in 2014, in particular with the significant revenue we're getting in the semicustom space, the semicustom customers that we have, we have to supply the product for them essentially a quarter ahead of when they build the parts, and therefore the seasonally high sales happen in Q4. We build the product and supply it in Q3, but we have to get the wafer supply and inventory up in Q2. And fundamentally, the increase from Q1 to Q2 is to support the semicustom ramp, as well as what we said earlier about the second half being stronger than the first half. And obviously, we need to be prepared for that.
So I think from an overall standpoint, despite the challenges we talked about in the PC market, managing inventory flat in Q1 was good. And we followed the same profile like we did in 2014, which is our plan. You should see inventory go up during the year to support the ramp of the semicustom and then supplying the product, and at the end of the year being able to manage down to flat inventory year on year.

Devinder Kumar - SVP and CFO
Yes. So first of all, I didn't give a cash guidance for Q2, but what I said specifically is we'll manage within the $600 million to $1 billion balance, the range that we have provided, the target minimum versus the optimal. On the WSA, the $1 billion that we talked about and signed in the WSA that we just confirmed with Global Foundries, that is a mix of revenue there. Just as a reminder, and I think you know this, but the large portion of the amount that we have is a good mix of product across PC, graphics and semicustom, as opposed to if you go back two years ago. So I would say it's very much in line with our expectations of how the market is going to evolve this year, in particular in the semicustom space.

John Pitzer - Credit Suisse
Yeah, good afternoon guys. Thanks for letting me ask the question. Devinder, just a follow-up on the WSA. Do you have a forecast for what payments to Global Foundries might look like in the June quarter? And just help me understand, to get from the run rate in Q1 to kind of the full-year guide of $1 billion, how should we think about kind of the linearity of those payments?


Devinder Kumar - SVP and CFO
Yeah, I think Q2 amount, I won't get in the quarterly profile, but if you listen to the comments we're making in terms of some of the actions we are taking in Q1 and Q2, the first half of the year, relative to the PC market, and then you listen about the semicustom profile first part of the second half, while the purchases of the first quarter are at the $161 million level, you can expect that it's going to follow the trajectory of the business, second half being stronger than the first half, the semicustom ramp-up in the second half, a significant portion of the billion dollars is semicustom products, and that's exactly how it goes throughout the year.

 

[mrmt]

It was still a bad deal and bad business for Intel.

Yes, but one that Intel could afford. ATI wasn't a bad business AMD could afford.

 

[mrmt]

Also, comparing a flat Revenue scenario for the rest of 2015, and the $839M remaining commit to GF under the WSA for 2015 versus the $1B actual in 2014, adjusted for the inventory reduction that transpired in late 2014/early 2015, it seems they're overcommitted by $200M. In other words, inventory could rise by $200M and cash reduce by $200M on that basis alone. Or the WSA will be re-negotiated for the umpteenth time.

Either way, that 2015 WSA commitment ain't going to be cash flow positive!!!

I think AMD is working hard to move up the console chips for GLF. If they are able to then the commitment will be easily met.

 

[frozentundra123456]

Or you could count the interest on the 5 billion that they have paid since the acquisition, and the opportunity cost of having 5 billion dollars down the drain to purchase ATI. Or you could even theorize that if they hadnt been so overwhelmed with debt they would have been in a better position to negotiate a less onerous WSA with GF.

At this point in time, I dont think it is possible to say if AMD would have been better or worse off if they had not purchased ATI. But all you have to due is look at the standing of AMD then vs now to question the purchase.

Now *if* HSA ever takes off or *if* HBM makes their APUs and/or dgpus compelling products (to consumers, not just amd fans on forums) then the ATI purchase will have proved its worth, although nothing will negate the fact that they paid too much.

And yes, I think the purchase of McAfee by Intel was incredibly stupid, and cant understand how a group of supposedly intelligent executives could ever justify the purchase. But I dont know what place that has in a thread about AMD financials, except to say ha, ha, Intel can make mistakes too.

 

[svenge]

But I dont know what place that has in a thread about AMD financials, except to say ha, ha, Intel can make mistakes too.

It's just about the only solace die-hard AMD partisans have these days.

 

[imported_ats]

It was still a bad deal and bad business for Intel.

Yeah but its pretty much like buying a spoiled carton of milk, sucks to buy it, but it isn't going to put you in an early grave. In contrast, AMD leveraged itself to buy the whole dairy only to find out the cows had mad cow disease.

 

[imported_ats]

You're missing the point.

The point is that Intel selling lots of APUs with iGPU doesn't mean their iGPU is any good (which some here implied). In fact it's bad compared to the competition, but something the customers are forced to accept if they want to buy an Intel APU. I.e. they are willing to accept that, since the CPU part of the APU is good.

Question for ya, if a 9.5 is enough to win all the prizes, does it really matter if you get a 10 instead? iGPUs are pretty much like that, where better isn't necessarily better in a world of just good enough. For 99.99% of the market, there is no difference between Intel's and AMD's iGPUs. OTOH, Intel has significantly better CPU performance across the whole range of products.

If that's such a great solution, then why didn't Intel go down that route? They could have licensed a much better iGPU from nVidia or AMD compared to what their own R&D department has been able to design.

Intel has gone down that route with multiple products.

 

[CHADBOGA]

Yeah but its pretty much like buying a spoiled carton of milk, sucks to buy it, but it isn't going to put you in an early grave. In contrast, AMD leveraged itself to buy the whole dairy only to find out the cows had mad cow disease.

LOL that is sig worthy. 😀