News AMD earnings report 1Q 2020

Hitman928

Diamond Member
Apr 15, 2012
5,179
7,630
136
On current outlook for 2Q and full year 2020:

For the second quarter of 2020, AMD expects revenue to be approximately $1.85 billion, plus or minus $100 million, an increase of approximately 21 percent year-over-year and 4 percent sequentially. The year-over-year increase is expected to be primarily driven by growth of Ryzen and EPYC processor sales. The sequential increase is expected to be primarily driven by EPYC processor and semi-custom sales. AMD expects non-GAAP gross margin to be approximately 44 percent in the second quarter of 2020. The sequential decrease is expected to be primarily due to the initial ramp of next-generation semi-custom products.


Despite expectations of weaker COVID-19 related consumer demand in the second half of the year, AMD expects 2020 revenue to grow by approximately 25 percent, plus or minus 5 percentage points, compared to 2019 and non-GAAP gross margin to be approximately 45 percent.

AMD had previously predicted 28-30% revenue growth in 2020 so a 25% revenue growth given the global recession/depression would still be quite good I think if they can actually get that.
 
Last edited:

Hitman928

Diamond Member
Apr 15, 2012
5,179
7,630
136
Some segment breakdown highlights:

  • Computing and Graphics segment revenue was $1.44 billion, up 73 percent year-over-year. . . Revenue was higher year-over-year driven by strong Ryzen processor and Radeon™ product channel sales
  • Client processor average selling price (ASP) was up year-over-year driven by Ryzen processor sales. Client processor ASP was down slightly quarter-over-quarter due to higher notebook sales.
  • GPU ASP was lower year-over-year and quarter-over-quarter due to product mix.
  • Enterprise, Embedded and Semi-Custom segment revenue was $348 million, down 21 percent year-over-year and 25 percent sequentially primarily due to lower semi-custom sales, partially offset by higher EPYC processor sales.
 

tamz_msc

Diamond Member
Jan 5, 2017
3,710
3,554
136
Enterprise, Embedded and Semi-Custom segment revenue was $348 million, down 21 percent year-over-year and 25 percent sequentially primarily due to lower semi-custom sales, partially offset by higher EPYC processor sales.
Ok so how much contribution does EPYC actually have? Surely the console volume should be negligible by now.
 

Hitman928

Diamond Member
Apr 15, 2012
5,179
7,630
136
Ok so how much contribution does EPYC actually have? Surely the console volume should be negligible by now.

Since AMD doesn't do any breakdown between Enterprise and Semi-custom it's impossible to know for sure. Epyc is still a much smaller piece of the pie versus consumer though, that's for sure. Quick math which anyone is free to double check:

Edit: lol, too distracted trying to do multiple things at once, they already listed the revenue in the breakdown.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
@Hitman928

Still makes me wonder why the enterprise/embedded/semi-custom figures came in so low. There must be some other semi-custom contracts expiring.

edit: the other possibility is that AMD is dumping Rome at a discount to combat Intel's aggressive negotiated price strategies.
 
  • Like
Reactions: lightmanek
Mar 11, 2004
23,031
5,495
146
Its not surprising that EPYC hasn't made as much inroads as expected. There's a lot working against them there. I still don't think the big OEMs have adopted EPYC much, and the major players (Amazon, Facebook, Microsoft, etc) seem to be looking at a variety of things (Amazon's ARM chip, etc). I think those will come though. I also think that gaming will be big for AMD as they'll have both consoles and developing game streaming stuff. That stuff I think will really start to takeoff in like 2022.

I think the supercomputers are big wins for AMD, as, and I could be just inferring this and it be my own speculation and not how things will go, but that it was going to help develop sorta dev kits for HPC hardware, which I think would go a long way to boosting AMD in that segment.

Which, there's just general uncertainties in there. Not even talking about the pandemic. There's been talk about next steps. Stuff like fiber optic interconnects (internally), mixed chip solutions, etc. Both Intel and AMD are working on solutions. I think that was actually a big part of Facebook's self-developed platform. Same with Google.

Which, there's also got to be lots of Intel hardware that is due to be swapped out. I'm not sure on the full life cycle of server equipment, but I'd expect there's ebbs and flows. AMD had to establish EPYC as reliable and not a flash in the pan. Then I think companies will start trialing it for potential as base platform, and then sales will go from there. I have to imagine that once a lot of companies start testing EPYC internally they'll be impressed. Some have likely been waiting to see if Intel would have a realistic response (which seems to me that even if Intel rights their ship, they're not going to eclipse AMD so much that it'd completely curtail EPYC viability - but companies had to see if Zen was Athlon 64 part 2 with a Core 2 stomping followed by Phenom issues, etc).

I think AMD knew and this seems to be right in line with what they expected, and why they focused a lot (at least to me it seems), on having a long term plan for Zen. To keep iterating. It actually seems to me they've been a bit aggressive (but in smart ways, meaning, I wasn't expecting 64 cores til Zen 3, I thought we'd get 48).

@Hitman928

Still makes me wonder why the enterprise/embedded/semi-custom figures came in so low. There must be some other semi-custom contracts expiring.

edit: the other possibility is that AMD is dumping Rome at a discount to combat Intel's aggressive negotiated price strategies.

I wouldn't be surprised there's been ones they were working on that got cancelled or are on hold or something. I think that Atari console is likely not going to see the light of day (think someone from Microsoft is even suing them over it). I believe there was talk about another one that wasn't Sony or Microsoft too, and I'd have a hunch that might have gotten dropped due to the current situation. Could also see a lot of the tinker board crowed dropping (there seemed to be some enthusiasm for embedded Ryzen, but those situations are already often skimming by on the skin of their teeth).

Didn't Sony cut their order for PS5, too? I seem to recall there being talk that Sony was gonna be pretty conservative for the PS5 launch and I think that was early in the pandemic, so it might not have even accounted for that.
 
  • Like
Reactions: randomhero

randomhero

Member
Apr 28, 2020
180
247
86
Its not surprising that EPYC hasn't made as much inroads as expected. There's a lot working against them there. I still don't think the big OEMs have adopted EPYC much, and the major players (Amazon, Facebook, Microsoft, etc) seem to be looking at a variety of things (Amazon's ARM chip, etc). I think those will come though. I also think that gaming will be big for AMD as they'll have both consoles and developing game streaming stuff. That stuff I think will really start to takeoff in like 2022.

I think the supercomputers are big wins for AMD, as, and I could be just inferring this and it be my own speculation and not how things will go, but that it was going to help develop sorta dev kits for HPC hardware, which I think would go a long way to boosting AMD in that segment.

Which, there's just general uncertainties in there. Not even talking about the pandemic. There's been talk about next steps. Stuff like fiber optic interconnects (internally), mixed chip solutions, etc. Both Intel and AMD are working on solutions. I think that was actually a big part of Facebook's self-developed platform. Same with Google.

Which, there's also got to be lots of Intel hardware that is due to be swapped out. I'm not sure on the full life cycle of server equipment, but I'd expect there's ebbs and flows. AMD had to establish EPYC as reliable and not a flash in the pan. Then I think companies will start trialing it for potential as base platform, and then sales will go from there. I have to imagine that once a lot of companies start testing EPYC internally they'll be impressed. Some have likely been waiting to see if Intel would have a realistic response (which seems to me that even if Intel rights their ship, they're not going to eclipse AMD so much that it'd completely curtail EPYC viability - but companies had to see if Zen was Athlon 64 part 2 with a Core 2 stomping followed by Phenom issues, etc).

I think AMD knew and this seems to be right in line with what they expected, and why they focused a lot (at least to me it seems), on having a long term plan for Zen. To keep iterating. It actually seems to me they've been a bit aggressive (but in smart ways, meaning, I wasn't expecting 64 cores til Zen 3, I thought we'd get 48).

I must say, I am still shocked how slow uptake on Epyc is.
 

piokos

Senior member
Nov 2, 2018
554
206
86
I think AMD knew and this seems to be right in line with what they expected, and why they focused a lot (at least to me it seems), on having a long term plan for Zen.
But how "long term" is that plan going to be?
There must have been some internal targets on sales and profit. If AMD matches those, then the plans had to be really unambitious...

Objectively, there are still some massive issues with the platform. Like something that really shocked me just few weeks ago: no support for nested virtualization on Windows, which makes EPYC unusable for many Windows-based systems.
AMD is slowly working on these, but it means the whole competitive advantage that TSMC gave them (realistically: 3-4 years before Intel catches up) will be wasted on polishing the product.

Corporate clients have money. They can afford the prices Intel asks. They can't afford the risk that (still) comes with the AMD alternative.

It would be nice to see AMD just raising prices and spending more on cooperation with partners, fixing issues and building a proper support network.

Performance advantage is just not enough to attract someone whose "life" (in an enterprise meaning) depends on a product.
In the same way recommendations for outstanding taste are not enough for most people to try seafood cooked on the street.

AMD struck some deals with large cloud providers because of a very simple issue: they have money, time and know-how to fix some problems on their own. And it's cloud. It's meant to be flexible.
Whereas a server running an important on-premise system in a company either works or not. :)
 
Last edited:

itsmydamnation

Platinum Member
Feb 6, 2011
2,743
3,075
136
Objectively, there are still some massive issues with the platform. Like something that really shocked me just few weeks ago: no support for nested virtualization on Windows, which makes EPYC unusable for many Windows-based systems.
AMD is slowly working on these, but it means the whole competitive advantage that TSMC gave them (realistically: 3-4 years before Intel catches up) will be wasted on polishing the product.
errrr. AMD supported nested virtualisation well before intel ever did, like back in the K8 days. Dont confuse a very crappy hypervisor ( hyper-v) with bad hardware. honestly the only thing worse then hyper-v network stack is azures network stack... lol . Use anything else , kvm, esx, etc . I run guests on kvm on a centos guest in esxi on a 3600 everyday without issue.

You know how many companies i know running hyper-v in production? none, plenty of Vmware, kvm and nutanix. Microsoft just need to have/fix support on there OS, its a pain for the hom/powere user but not a issue for EPYC.
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
I dont know about Epyc but the rest looking good.

Im waiting to see what the reaction will be when the market open later today.
 

piokos

Senior member
Nov 2, 2018
554
206
86
errrr. AMD supported nested virtualisation well before intel ever did, like back in the K8 days. Dont confuse a very crappy hypervisor ( hyper-v) with bad hardware.
Use anything else , kvm, esx, etc . I run guests on kvm on a centos guest in esxi on a 3600 everyday without issue.
But we're talking about enterprise products and you can't tell a company to scrap their Windows systems and use KVM and Linux. That's the point.

I'm not saying there's something wrong with hardware either. But it isn't supported by the dominant OS. MS and AMD should fix this.
And it's AMD who should be the most interested in this matter. If they are working with MS on fixing this - great, but why so long?
You know how many companies i know running hyper-v in production? none, plenty of Vmware, kvm and nutanix. Microsoft just need to have/fix support on there OS, its a pain for the hom/powere user but not a issue for EPYC.
In production: probably not many. Most companies stick to VMware for the same reason they stick to Intel - because it's tried and tested.
But in dev/analytics environments? In private cloud providing VMs for employees? I bet: a lot.
Also, think about consequences this has on developers/analysts who use Docker or any other Hyper-V-based software.

It's not just Hyper-V either. Nested virtualization doesn't work in Virtualbox (on Windows host).
Ryzen 1000 was problematic for some emulators (at least the Android Studio issue was fixed).
It's an ongoing theme.

And it doesn't mean AMD did something wrong. But they're coming with a new architecture and they should cooperate with companies to make it more usable.
They don't offer support and they end up with 5% market share.
 
  • Like
Reactions: spursindonesia

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,619
136
Enterprise, Embedded and Semi-Custom segment revenue was $348 million, down 21 percent year-over-year and 25 percent sequentially primarily due to lower semi-custom sales, partially offset by higher EPYC processor sales.
Ok so how much contribution does EPYC actually have? Surely the console volume should be negligible by now.
Is there are chance the "console volume" is actually negative right now? Last gen may not be produced anymore at this point, next gen is ramping up, so it's AMD role to order all those required dies for the next gen consoles and ramp up production of the chips. Is anybody familiar what the order and timing for this process may look like (e.g. AMD has to pay TSMC upfront for the dies, Sony and Microsoft only pay AMD for the finished chips, or something like that)?
 

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136
Last gen may not be produced anymore at this point

Both consoles will continue to be sold for at least 2-3 years after the next gen launch. The PS3 for instance was discontinued 3 years after the PS4's launch. The actual volume may not be much as you are sort of at the point that if you wanted the console you would have bought it by now.
 

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,619
136
Both consoles will continue to be sold for at least 2-3 years after the next gen launch. The PS3 for instance was discontinued 3 years after the PS4's launch. The actual volume may not be much as you are sort of at the point that if you wanted the console you would have bought it by now.
I didn't mean that production stopped altogether (I'm aware of the longevity of consoles, there were e.g. talks about "10 year plans" wrt to PS4 after all), more that stock of chips likely is still sufficient for Sony/Microsoft to assemble some more last gen consoles when demand warrants so there's no need/order for AMD to produce more chips at this time.
 

Hitman928

Diamond Member
Apr 15, 2012
5,179
7,630
136
AMD said that console revenue was negligible in this last quarter but that the very early ramp for next gen consoles will begin in 2Q. With that in mind and with the earlier numbers, I think the numbers are something like this:

  • 1Q19
    • Epyc ~$110M​
    • Licensing $60M​
    • Embedded/Semi-custom ~$270.5M​
  • 1Q20
    • Epyc ~$330M​
    • Embedded/Semi-custom <$20M​

That would give you more than $250M negative change in semi-custom, mostly offset by a $220M positive change in Epyc revenue but then some additional negative change due to no or negligible licensing revenue. These numbers could change significantly if AMD has much higher embedded/semi-custom revenue outside of consoles than we think.​
 
Last edited:

DisEnchantment

Golden Member
Mar 3, 2017
1,590
5,722
136
Somehow I get the feeling Lisa is being very cautious with announcing the EPYC results. I understood there were quite a bit of EPYC deals in the pipe which they are delivering but they are not reporting for this quarter, I listened to the Earnings call yesterday.
I think they met their guidance and want to keep it that way. Lisa might want to make sure the guidance for next Q are met by reporting the results for next Q instead, should there be unforeseen developments which could make AMD unable to meet their guidance which is understandable given current situations. Background, we are also doing it too in our company. We just specify a different payment date, it is just a date on when you want it paid. :)
 
  • Like
Reactions: lightmanek

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
AMD said that console revenue was negligible in this last quarter but that the very early ramp for next gen consoles will begin in 2Q. With that in mind and with the earlier numbers, I think the numbers are something like this:

  • 1Q19
    • Epyc ~$110M​
    • Licensing $60M​
    • Embedded/Semi-custom ~$270.5M​
  • 1Q20
    • Epyc ~$330M​
    • Embedded/Semi-custom <$20M​

That would give you more than $250M negative change in semi-custom, mostly offset by a $220M positive change in Epyc revenue but then some additional negative change due to no or negligible licensing revenue. These numbers could change significantly if AMD has much higher embedded/semi-custom revenue outside of consoles than we think.​

I dont believe embedded is only 20M in 1Q20
 
  • Like
Reactions: Gideon and gdansk

DisEnchantment

Golden Member
Mar 3, 2017
1,590
5,722
136
What you're describing here is generally illegal, so:
a) I wouldn't brag so much,
b) let's hope AMD isn't doing anything similar...
Sorry, it is not illegal. We have a legal, compliance, trade laws, finance and related departments that deal with such things.
The buyer and supplier agree on a payment date. This is normal business. The buyer will pay when that date specified has come to pass.