AMD crushes Q4 earnings estimates

[realibrad]

People want AMD to have the margins of Intel. AMD margins 36%, Intel margins 63%, Nvidia about 60%.

 

[scannall]

Margins are terrible. They need to increase these to become more profitable.

I wouldn't call them terrible. But I would agree they need to be higher. 10 points would make investors happy.

 

[Jan Olšan]

People want AMD to have the margins of Intel. AMD margins 36%, Intel margins 63%, Nvidia about 60%.

I'm fine with Intel, Nvidia, Apple... having the margins of AMD.

Getting into the datacenter is a waaaaaaaaaaaaaaaaay longer process than you might imagine.
I've yet to see AMD-based PowerEdges tbh.

https://twitter.com/Daniel_Bowers/status/958555512714522624
"Dell has three 14G PowerEdge servers with EPYC processors in its VMware compatibility matrix"

 

[realibrad]

I'm fine with Intel, Nvidia, Apple... having the margins of AMD.



https://twitter.com/Daniel_Bowers/status/958555512714522624
"Dell has three 14G PowerEdge servers with EPYC processors in its VMware compatibility matrix"

I was just telling someone at work, that if AMD can get some market share back in servers, their margins should go up quite a bit.

 

[Topweasel]

I was just telling someone at work, that if AMD can get some market share back in servers, their margins should go up quite a bit.

Yeah the trick is to get ASP up. If they make headroom in the server business you are looking at great margins. But realistically AMD is making more than enough margin without forcing their pricing up (and possibly losing sales).

 

[realibrad]

Yeah the trick is to get ASP up. If they make headroom in the server business you are looking at great margins. But realistically AMD is making more than enough margin without forcing their pricing up (and possibly losing sales).

AMD is looking to take market share in a very competitive market. That means much of the increased revenue from larger margins will likely go into R&D.

 

[Atari2600]

Investors like margins, and AMD's was predicted to be 36% I think but actually ended up 35%, both numbers being very low vs their competitors.

F*** investors and the ship Wall street took them in on.

Running companies from one quarter to the next in terms of financials is a large part of why the Western business world is so f**ked up.

If Lisa Su and the execs have a long term plan at AMD, stick to that, and don't give two fiddlers fks about the next quarter!

 

[Topweasel]

F*** investors and the ship Wall street took them in on.

Running companies from one quarter to the next in terms of financials is a large part of why the Western business world is so f**ked up.

If Lisa Su and the execs have a long term plan at AMD, stick to that, and don't give two fiddlers fks about the next quarter!

That's my point in regards to margin. Who the hell cares if the margin is 10% or 100% if the result is 100 million in profit either way?

What shareholder responses that push better Margins do is is force companies to raise the price of their volume products to hit those goals. AMD doesn't need that. They have several product lines. Let the volume increase on their high value but high margin products like Threadripper and EPYC drive up their margin. Like I said the trick isn't to drive up margin, its to drive up value, so that they can get a higher ASP. That will make them the profits they need.

 

[realibrad]

That's my point in regards to margin. Who the hell cares if the margin is 10% or 100% if the result is 100 million in profit either way?

What shareholder responses that push better Margins do is is force companies to raise the price of their volume products to hit those goals. AMD doesn't need that. They have several product lines. Let the volume increase on their high value but high margin products like Threadripper and EPYC drive up their margin. Like I said the trick isn't to drive up margin, its to drive up value, so that they can get a higher ASP. That will make them the profits they need.

Its important because a company that is making 100 million with 100% margins can buffer itself against competition or market changes. A company with 10% in that same market will have a much tougher time when compared to the 100% company.

 

[Topweasel]

Its important because a company that is making 100 million with 100% margins can buffer itself against competition or market changes. A company with 10% in that same market will have a much tougher time when compared to the 100% company.

Not if raising Margin's lower their sales numbers. Of the two noted in this thread one is a monopoly and the other is a lifestyle brand. Not saying that AMD selling their stuff with improved value doesn't make it more difficult to compete with Intel. But they aren't able to compete with Intel, that is part of the problem. Realistic pricing with value is going to do AMD more help then droning on about not having enough margin to survive while they are making a profit.

 

[realibrad]

Not if raising Margin's lower their sales numbers. Of the two noted in this thread one is a monopoly and the other is a lifestyle brand. Not saying that AMD selling their stuff with improved value doesn't make it more difficult to compete with Intel. But they aren't able to compete with Intel, that is part of the problem. Realistic pricing with value is going to do AMD more help then droning on about not having enough margin to survive while they are making a profit.

Even then, because the hypothetical was 100 million revenue. That gives them the ability to lower prices to compete whereas the other company has far less ability to do so.

 

[StinkyPinky]

AMD will have smaller margins for two main reasons

1 - They're a smaller company, so R&D costs will be a higher percentage of their operating costs
2 - They have to sell for less profit margin to make their products more attractive against Intel. Intel has brand awareness that AMD lacks.

 

[moinmoin]

AMD's lower margins are actually an advantage for AMD. Intel's high margins are the very reason Intel tried and failed to enter numerous markets as their business there could never dream to reach the margins their core business achieves, by which every effort is measured. It's in AMD's own interest to not back themselves into that corner anytime soon.

 

[Kenmitch]

People want AMD to have the margins of Intel. AMD margins 36%, Intel margins 63%, Nvidia about 60%.

Death and taxes are the only thing that's guaranteed in life. Nothing saying both Intel and Nvidia are guaranteed their current margins forever. As AMD becomes more competitive the gaps will tighten up somewhat. It's just a matter of how it plays out in the end.

 

[realibrad]

Death and taxes are the only thing that's guaranteed in life. Nothing saying both Intel and Nvidia are guaranteed their current margins forever. As AMD becomes more competitive the gaps will tighten up somewhat. It's just a matter of how it plays out in the end.

Yes, as AMD becomes more competitive it will attack the competitors in their margins. But it also means that if the competition has an issue, they have a buffer. If AMD has the same issue it hurts far more.

 

[piesquared]

I recall AMD stating that overall margins are in the mid 30's mainly because the margins of consoles are low but the development of those APUs were paid for by the semi-custom customers, so should not be used as

AMD will have smaller margins for two main reasons

1 - They're a smaller company, so R&D costs will be a higher percentage of their operating costs
2 - They have to sell for less profit margin to make their products more attractive against Intel. Intel has brand awareness that AMD lacks.

And 3. Console margins are artificially lowering overall margins. But since there was no development costs, it is skewed lower than what it would typically be.

 

[itsmydamnation]

Yes, as AMD becomes more competitive it will attack the competitors in their margins. But it also means that if the competition has an issue, they have a buffer. If AMD has the same issue it hurts far more.

There biggest competitor has 11x the costs/employees meaning the hurt comes very quickly regardless. That competitor gets most of its revenue for 2 sectors based largely off one base development platform. Intel financials look very good but they are spending the technical capital they had accrued over the last decade and if 10nm sees anymore delays they will exhausted all that technical capital and then it will playout on the books during '19:
upto 64core 256mb l3 server processors vs 28 core 14nm cascade lake?
6core coffee lake s vs maybe up 16core Ryzen 3
~18 core cascade lake x vs 32core threadripper 3.

 

[realibrad]

There biggest competitor has 11x the costs/employees meaning the hurt comes very quickly regardless. That competitor gets most of its revenue for 2 sectors based largely off one base development platform. Intel financials look very good but they are spending the technical capital they had accrued over the last decade and if 10nm sees anymore delays they will exhausted all that technical capital and then it will playout on the books during '19:
upto 64core 256mb l3 server processors vs 28 core 14nm cascade lake?
6core coffee lake s vs maybe up 16core Ryzen 3
~18 core cascade lake x vs 32core threadripper 3.

What is important is that Intel has far more cash, and revenue per their expenses. They are a far more healthy and robust company and can withstand far more shocks.

This should probably end though, as this is not P&N and this is getting off topic.

 

[eek2121]

It'd be nice if the analysts manage to get AMD to say how much revenue from mining they are getting.

It would be nice if you told us redacted but AMD has no idea what percentage of sales go to mining vs gaming. Neither does Nvidia, though Nvidia actually has a plan to start monitoring card usage, it's invasive and pretty much all users are going to disable it when it launches.













No profanity allowed in tech forums.


esquared
Anandtech Forum Director

 

[eek2121]

Margins are fantastic where are your numbers?

This. IMO Topweasel doesn't even know what weasel carcasses sell for, much less the 'top weasel'. I'm no defender of AMD, but I do sell a product, and I could pitch it to most, they'd swear XYZ and it's actually X,XXX.

 

[eek2121]

There biggest competitor has 11x the costs/employees meaning the hurt comes very quickly regardless. That competitor gets most of its revenue for 2 sectors based largely off one base development platform. Intel financials look very good but they are spending the technical capital they had accrued over the last decade and if 10nm sees anymore delays they will exhausted all that technical capital and then it will playout on the books during '19:
upto 64core 256mb l3 server processors vs 28 core 14nm cascade lake?
6core coffee lake s vs maybe up 16core Ryzen 3
~18 core cascade lake x vs 32core threadripper 3.

Intel has hundreds of different product lines spread over a very large company. I try not to defend either company unless needed. However, 3 years ago, my Intel shares totaled 100% and my AMD shares totaled 0%. I got out before intel got slammed, and I got in with AMD before they went north. end result? No complaints. FYI there will never be a 32 core threadripper without a socket redesign anymore than there will be a 9-16 core Ryzen on AM4.

 

[HurleyBird]

FYI there will never be a 32 core threadripper without a socket redesign anymore than there will be a 9-16 core Ryzen on AM4.

It's possible that we could see both, but that would be due to backcompat where those processors would be designed around a new platform with DDR5 (eg. AM4+ or AM5), but perhaps also technically capable of running in the old socket with DDR4. It wouldn't be the first time something like that has happened, but if it did you can bet those products would be bandwidth starved on the old platform.

 

[itsmydamnation]

Intel has hundreds of different product lines spread over a very large company. I try not to defend either company unless needed. However, 3 years ago, my Intel shares totaled 100% and my AMD shares totaled 0%. I got out before intel got slammed, and I got in with AMD before they went north. end result? No complaints. FYI there will never be a 32 core threadripper without a socket redesign anymore than there will be a 9-16 core Ryzen on AM4.

yes they have lots of products, bust most of their revenue and profit comes from there primary micro architecture(even skylake-x is an extension of it, just look at the die shots of the cores themselves) . Everything else could hit home runs but if they had a catastrophic failure on delivery of a major uarch the rest of the companies performance would be irrelevant.

So we know minimum 7nm "zepplin" equivalent SOC is 12 cores, CPC say it's 16 with twice the L3 per core of 14nm Zepplin. You do the math, AMD can position whatever they want at that point.

For years people have overestimated CPU bandwidth requirements, do we need to go revisist the anandtech forum doom and gloom of dual channel with 8 cores? Hint there is a maximum amount of outstanding memory requests per Core ( they have to be tracked) outside of stream prefetching this is your limiting factor for amount of data a Core can get from the memory sub system.

 

[Yotsugi]

FYI there will never be a 32 core threadripper without a socket redesign anymore than there will be a 9-16 core Ryzen on AM4.

Slapping more cores on AM4/SP3 is piss easy.
They don't need to change anything, since both platforms were designed with longevity in mind.

CPC say it's 16 with twice the L3 per core of 14nm Zepplin.

That's probably the separate server die (CPC is usually right).

 

[itsmydamnation]

That's probably the separate server die (CPC is usually right).

It make 0 sense to make a separate server die, it wont help with max clock* and 7nm design and layout costs are massive. Look at it this way, they haven't hired 50-100% more engineers in the last year so what products aren't they working on to increase the amount of tape outs while effort per tape out increases massively?

*i think people who expect AMD to use HPC are wishfull thinkers, 1 SOC for enthusiast/HEDT/Sever just like now.