What to Expect When Advanced Micro Devices Inc. Reports Earnings

cageymaru

cageymaru

Fully [H]
Joined
Apr 10, 2003
Messages
22,132
What to Expect When Advanced Micro Devices Inc. Reports Earnings.
What to Expect When Advanced Micro Devices Inc. Reports Earnings -- The Motley Fool

I just posted the generalized opening paragraphs to give you an idea what's in the article. Go read the article for the juicy stuff! Very informative.

With new product launches still at least a few months away, signs of a turnaround are unlikely to show up in Advanced Micro Devices' (NASDAQ:AMD) first-quarter earnings report. Results will be announced Thursday following the market close, and if analyst estimates are correct, investors can expect a steep decline in revenue and earnings.


Demand for PCs remained weak during the first quarter, with IDC reporting that global unit shipments declined by 11.5% year over year. AMD's core business is heavily dependent on the PC, and without any indication that demand is stabilizing, returning to revenue growth will be difficult for the company. AMD plans to launch new GPUs and CPUs later this year, and both are expected to improve the company's competitive position, but a weak PC market will make executing a turnaround far more difficult.

 
That doesn't make for good reading. Hopefully Polaris and Zen can be a big hit for AMD. If these two don't, I'd hate to think what it could mean for AMD.
 
It's not all doom and gloom. Sony is doing the ps4k thing which probably involved giving AMD some cash up front. Also the new crypto coin thing may help on the gpu front. CPU front is probably dismal though. Should be interesting.
 
I wouldn't expect much this past quarter either, actually I'd expect red ink. I never did a FA for AMD, but after the stocks tanks with the expected bad quarter it could be a quick buy and sell opportunity.

At your own risk of coarse.
 
I think AMD is well positioned. They are already working on APU with HBM and that will probably be a killer product. That is totally different market and people are still buying ultrabooks and corporations are still using laptops. They will go for design wins and get as much market as they can. PC market may be shrinking but for AMD only place to go is up with new products coming out. They have a lot of marketshare to gain and if they have the right product set they will do better than expected.
 
In the words of Mr. T in Rocky III when asked what to expect in his upcoming fight: "Expect pain".
 
I see lots of upside for AMD in the near future. I think everyone is well aware of their current situation and is not expecting a good quarter. I bought a couple thousand shares @ $1.80 a little while back. I was tempted a week ago to sell because AMD is so heavily shorted and almost always sees a large drop in value around 10 days before their quarterly report. But I noticed that their sell off behavior is a little different at the moment, so I rolled the dice and did not sell betting that they will not crash hard after the report. But I did plop another couple thousand into my account just in case they do drop near $2/share so I can buy more.
 
jhatfie said:
I see lots of upside for AMD in the near future. I think everyone is well aware of their current situation and is not expecting a good quarter. I bought a couple thousand shares @ $1.80 a little while back. I was tempted a week ago to sell because AMD is so heavily shorted and almost always sees a large drop in value around 10 days before their quarterly report. But I noticed that their sell off behavior is a little different at the moment, so I rolled the dice and did not sell betting that they will not crash hard after the report. But I did plop another couple thousand into my account just in case they do drop near $2/share so I can buy more.
Click to expand...
Have to agree with your sentiment. Nothing shocking that could really come from their earnings report. That said, I see some solid future potential as they're one of the only IHVs with IP for CPU and graphics. I foresee discrete graphics going away in favor of APUs and small form factors. Polaris and Zen on a single die I see being a significant market opportunity. Intel thinking they need to diversify out of the segment indicates to me they think someone else may make some gains. The future in my mind means CPU and graphics on the same package. Graphics would also be more significant than CPU capability with the new APIs and ability to accelerate tasks with the GPU. Zen with Polaris 11(or perhaps larger) in a chromebox would be a really capable desktop replacement for a lot of people.
 
And what are the negatives that you see Anarchist and jhatfie. because there are many, many more than the positives. Positives are all attached together, even one negative can hinder that plan.

There are many things that have to align just right to make a risk into a possible positive.

Although its not all doom, but the way things change slowly, will take time before we see positives.

Look at Fury r3xx line, how many people thought it was going to equalize the share loss they already had, in this very forum. What did I say, it will take time, possible one or two quarters to see any potential gains, and that gain was much lower then I expected. I was thinking around 5% upswing, it ended up around 2%.

The market is not steady as the other players are doing their things too, they are also aware of what AMD is trying to do, so there are moves and counter moves, which ones will work, which ones are in a better position to capitalize so far nV and Intel has always been in a better position and have been capitalizing on the situation. Will that change? Yes but that means those companies have to falter and AMD has to take the upper hand at the same time.
 
Last edited:
AMD's first-quarter revenue fell 19.2 percent to $832 million, but beat average analysts' estimate of $818.2 billion, according to Thomson Reuters I/B/E/S.

Net loss narrowed to $109 million, or 14 cents per share, in the first quarter ended March 26, from $180 million, or 23 cents per share, a year earlier.

Excluding items, AMD posted a loss of 12 cents per share, compared with analysts' estimate of a 13 cents per share loss.
Click to expand...

AMD revenue beats, company signs $293 million licensing agreement

Its bad news just not as bad as expected.
 
razor1 said:
And what are the negatives that you see Anarchist and jhatfie. because there are many, many more than the positives. Positives are all attached together, even one negative can hinder that plan.

There are many things that have to align just right to make a risk into a possible positive.
Click to expand...
Intel cutting jobs to move away from the PC sector and coming back with hugely improved integrated graphics is a risk. Nvidia releasing a CPU capable of fighting with Intel and Zen. Complete shutdown of worldwide powergrids.

It appears the competition is moving away from the sector knowing they won't be able to compete for a while and I don't see a whole lot of other possibilities for SOCs. Apple appears moving towards small form factor, and away from Intel which is significant. They wouldn't even consider AMD if there were issues. Discrete graphics will transition to small form factor, requiring integrated GPUs, that means Nvidia is in trouble there without a CPU. And I can't help but think all these console refreshes will amount to some income. Another 300M from the Chinese server market means they are competing there as well.
 
Anarchist4000 said:
Intel cutting jobs to move away from the PC sector and coming back with hugely improved integrated graphics is a risk. Nvidia releasing a CPU capable of fighting with Intel and Zen. Complete shutdown of worldwide powergrids.

It appears the competition is moving away from the sector knowing they won't be able to compete for a while and I don't see a whole lot of other possibilities for SOCs. Apple appears moving towards small form factor, and away from Intel which is significant. They wouldn't even consider AMD if there were issues. Discrete graphics will transition to small form factor, requiring integrated GPUs, that means Nvidia is in trouble there without a CPU. And I can't help but think all these console refreshes will amount to some income. Another 300M from the Chinese server market means they are competing there as well.
Click to expand...

I don't think the lack of a cpu is an issue, I remember the same things being stated with the first APU introduction, it just hasn't happened because of other factors. Doesn't matter what they do with iGPU, since you will need more silicon in the CPU to match dGPU's, it will still increase costs and more so because to compete they have to make a die that is much larger since they have to integrate a similar performance GPU with the CPU and thus far this hasn't happened and possible won't.
 
Last edited:
I'm guessing the financial report will be something like this:

inferno14.jpg
 
razor1 said:
I don't think the lack of a cpu is an issue, I remember the same things being stated with the first APU introduction, it just hasn't happened because of other factors. Doesn't matter what they do with iGPU, since you will need more silicon in the CPU to match dGPU's, it will still increase costs and more so because to compete they have to make a die that is much larger since they have to integrate a similar performance GPU with the CPU and thus far this hasn't happened and possible won't.
Click to expand...
It won't even be "in" the CPU, it will be on the same package. Start transferring the FP portions of a CPU to the GPU, and I could see AMD getting a high end GPU, with HBM, onto the CPU package. Rather strong gaming performance in a small form factor box which seems to be the way the market is going. A discrete GPU, without HBM, is nearly the size of the motherboard in these things. Gaming on a chromebox sort of stuff.
 
Anarchist4000 said:
Rather strong gaming performance in a small form factor box which seems to be the way the market is going.
Click to expand...
Computers generally trend toward integration. It wasn't that log ago that discrete network and sound cards were the norm.
 
Darakian said:
Computers generally trend toward integration. It wasn't that log ago that discrete network and sound cards were the norm.
Click to expand...

In servers, they still are, and I think that is where 'desktop' parts are heading towards: Small servers.
 
Anarchist4000 said:
It won't even be "in" the CPU, it will be on the same package. Start transferring the FP portions of a CPU to the GPU, and I could see AMD getting a high end GPU, with HBM, onto the CPU package. Rather strong gaming performance in a small form factor box which seems to be the way the market is going. A discrete GPU, without HBM, is nearly the size of the motherboard in these things. Gaming on a chromebox sort of stuff.
Click to expand...


Yeah that could be true but it doesn't matter, look at Zen its supposed to be released with a GPU that will perform like a 7870. That is now low end by that time. Its not going to be doing anything not being done right now.
 
razor1 said:
Yeah that could be true but it doesn't matter, look at Zen its supposed to be released with a GPU that will perform like a 7870. That is now low end by that time. Its not going to be doing anything not being done right now.
Click to expand...
Like Polaris 11 before any performance was known? If they're doing what I think they're doing the GPU isn't integrated into the CPU, they will be separate dies. This was already hinted at with the giant HPC Zen being two chips connected together. No reason the smaller version won't be varying configurations of cores and GPUs. 8 Core Zen with Polaris 11, 4 Core Zen with Polaris 10, 16 Core Zen with Polaris 10 because something thought it was interesting? The limit seems like it would be how much shit fits on an interposer. A relatively weak Zen with Polaris 11 for targeting Chromboxes really wouldn't come as a surprise.
 
and what is the cost incurred by doing that? Its all about cost vs benefit. The market doesn't just change because a new product like that is released. Your optimism is great, but there are other things that you are not looking at.

COST is a big factor in something like this. It won't be for consumer end products it will be for specialty needs.

Do you expect it to be lower than before? You still need to fab the interposer, you still need to have it assembled etc. So what is the cost compared to before?
 
Yakk said:
Wow, that IS quite the jump! I expected a drop then small bounce...

Nope, straight up apparently.
Click to expand...

Partially due to China, partially due to better than expected quarter, and partly due to shorts having to cover. I'd expect it to drop back down to more reasonable levels either later today or tomorrow.
 
Nytegard said:
Partially due to China, partially due to better than expected quarter, and partly due to shorts having to cover. I'd expect it to drop back down to more reasonable levels either later today or tomorrow.
Click to expand...

When there is bad news, companies always try to balance that with since good news, nothing new here.

So... By what you are saying the stock jumps almost 38% just on people covering their shorts? That is quite a statement to make :notworthy:
 
razor1 said:
and what is the cost incurred by doing that? Its all about cost vs benefit. The market doesn't just change because a new product like that is released. Your optimism is great, but there are other things that you are not looking at.

COST is a big factor in something like this. It won't be for consumer end products it will be for specialty needs.

Do you expect it to be lower than before? You still need to fab the interposer, you still need to have it assembled etc. So what is the cost compared to before?
Click to expand...
With a die less than half the size of a prior comparable product and good yields from using tiny ones I think costs will be significantly lower on the chips. Throw in less power circuitry, coolers, shorter boards, and scaling efficiency on HBM1 I think they could get there. Seems likely they're using HBM or double GDDR5 busses from combining chips with an interposer. Since journalists keep leaking that they're seeing short boards and Zauba shows them constantly receiving what look like short boards, I'm going with HBM. A Fury with 4GB seems to be doing rather well at even 4K, I don't see why everyone thinks HBM2, which will be more expensive, is required for acceptable performance. Every time someone asks for any sort of confirmation they deflect.
 
Die size doesn't matter (well the die size has to be less than half of what it was before) if the cost of the per transistor has increased does it?
 
razor1 said:
Die size doesn't matter (well the die size has to be less than half of what it was before) if the cost of the per transistor has increased does it?
Click to expand...
It matters because of yields and perf/mm2 likely more than doubling.
 
err no they are not, did you read what the CFO stated in yesterday's reports? Sounds like they will get double the perf.watt and thats it. On top of that I would expect yields of 14nm vs. 28nm to be the same if and only if, 14nm now mature, if not , I expect 14nm to be less. Although you get more chips per wafer, the wafer costs twice as much.....
 
razor1 said:
err no they are not, did you read what the CFO stated in yesterday's reports? Sounds like they will get double the perf.watt and thats it. On top of that I would expect yields of 14nm vs. 28nm to be the same if and only if, 14nm now mature, if not , I expect 14nm to be less. Although you get more chips per wafer, the wafer costs twice as much.....
Click to expand...
CFO didn't add a whole lot of new information. Just 2x perf/watt, which tells us nothing, and targeting mainsteam, which again tells very little. 14nm WILL have lower yields. That's why 2 dies will have better yields than 1 double sized die. Yields increase as the defects become more localized.
 
that still doesn't drop costs, if anything it increases costs. Just an example, One CPU, two Polaris 10's on the same interposer will that be more or less expensive that One CPU, and larger Vega on one card . Mind you both will use HBM and so that will not affect costs. The difference will be the two polarsis's. Not to mention the first system now will have to share total HBM on both CPU and Polaris, that might save costs but it will not be that much and will incur performance problems.

The point is, it will be a niche product targeted not at the general consumer as it will be expensive, there will be no "upgrade" paths well not as easy as before, and costs for such upgrade path will be much more expensive hence selective market.
 
razor1 said:
that still doesn't drop costs, if anything it increases costs. Just an example, One CPU, two Polaris 10's on the same interposer will that be more or less expensive that One CPU, and larger Vega on one card . Mind you both will use HBM and so that will not affect costs. The difference will be the two polarsis's. Not to mention the first system now will have to share total HBM on both CPU and Polaris, that might save costs but it will not be that much and will incur performance problems.

The point is, it will be a niche product targeted not at the general consumer as it will be expensive, there will be no "upgrade" paths well not as easy as before, and costs for such upgrade path will be much more expensive hence selective market.
Click to expand...
It would be cheaper than the alternative at that point. Two half sized dies WILL cost less than a single large die the same size unless yields are absolutely stellar. It's the difference between 15% wasted die and 40%, and it could be even greater than that. Regardless, putting the potential equivalent of a Fury Nano Crossfire on an APU probably won't be for the general consumer. Dual Polaris 10 on a discrete card is probably doable, but adding in Zen as well is likely too much.
 
You must log in or register to reply here.
Back
Top