News AMD earnings report 1Q 2020

[Hitman928]

Data still coming in but first reports are:

EPS $0.18, in line with expectations
Revenue $1.79B, up 40% y/y and in line with expectations
Gross margin at 46%, up 5 y/y and up 1 Q/Q

Will post more as it comes in.

AMD Reports First Quarter 2020 Financial Results

seekingalpha.com

 

[Hitman928]

I dont believe embedded is only 20M in 1Q20

You may not believe it, but in the earnings call Devinder Kumar (AMD CFO) said that datacenter revenue was a high teens percentage of overall revenue in the quarter. So if you take $1.79B and multiply it by 17-19% you get $304.3M - $340.1M. That only leaves somewhere between $7.9M to $43.7M for both embedded and semi-custom.

Edit: If your argument is that you think ~$40M is more accurate, that's fine, these were all approximate numbers so the smallest contributor is obviously going to have the biggest error range. I mean, $20M is only about 1% of the overall revenue so yeah, there was obviously going to be some error range in there. The point was trying to figure out (roughly) the amount of Epyc revenue, not have a < 1% error rate on trying to line item out every segment based upon only overview guidance given by AMD.

If you are arguing that embedded/semi-custom should be more like $100M, then I would say the numbers don't support that.

 

[DisEnchantment]

So if you take $1.79B and multiply it by 17-19% you get $304.3B - $340.1B

Is that a $B (Billion) ...

 

[Hitman928]

Is that a $B (Billion) ...

Lol, fixed, thanks.

 

[AtenRa]

You may not believe it, but in the earnings call Devinder Kumar (AMD CFO) said that datacenter revenue was a high teens percentage of overall revenue in the quarter. So if you take $1.79B and multiply it by 17-19% you get $304.3M - $340.1M. That only leaves somewhere between $7.9M to $43.7M for both embedded and semi-custom.

Edit: If your argument is that you think ~$40M is more accurate, that's fine, these were all approximate numbers so the smallest contributor is obviously going to have the biggest error range. I mean, $20M is only about 1% of the overall revenue so yeah, there was obviously going to be some error range in there. The point was trying to figure out (roughly) the amount of Epyc revenue, not have a < 1% error rate on trying to line item out every segment based upon only overview guidance given by AMD.

If you are arguing that embedded/semi-custom should be more like $100M, then I would say the numbers don't support that.

Yea im more in line that datacenter is close to ~300M

 

[beginner99]

Rome is AMD's strongest product relative to its competition. That's a bit depressing.

Yeah dressing. On the other hand if you go to big OEMs you can see them still offering naples and broadwell based servers, yeah no joke. So server market always lags behind. I bet it will start to rise a lot in H2.

 

[DrMrLordX]

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

Ditto? Even if semi-custom/embedded revenue bottomed out at $0, that's still not a lot from EPYC.

I dont know about Epyc but the rest looking good.

That's the weird thing. AMD is making a lot more than I expected they would outside of the datacentre. Especially looking at what the plans were for the product stack in 2017 moving forward.

Is there are chance the "console volume" is actually negative right now?

It might be, from an accounting perspective, though I don't know if that would affect revenue reporting.

 

[piokos]

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.

Obviously, this is not even remotely close to the topic of this forum, so let's keep the accounting discussion as short as possible.

Earlier I thought you meant some creative accounting, i.e. booking the revenue whenever you want. This is illegal, obviously.
What you described now is called "instalment sale", i.e. when you recognize revenue at the moment of cash transfer. So: I sell you something for $10 and you pay me $1 every year for a decade. I report revenue of $1 every year.
Whether this is allowed or not is down to an accounting standard.
It's acceptable in US GAAP (but not mandatory!).

Publication 537 (2024), Installment Sales | Internal Revenue Service

www.irs.gov

The other approach is when you report the whole revenue on a day of sales - even if the buyer will give you the money in the future.
So in the above example, I'd have to book $10 in first year. To offset this, I'd also book $-9 as so called Accounts receivable, which is the money I expect to get later (it's an Asset).

It looks like AMD is not using the instalment sale method. Their GAAP and Non-GAAP revenue figures are identical. And they have very large receivables ($1.7B).
This means that the revenue we see is exactly how much CPUs they sold in a quarter and $1.7B is the cash still missing.
Precise info can be found in their financial statement, but I'm just too tired today. I'll edit this post tomorrow.

 

[tajoh111]

Ditto? Even if semi-custom/embedded revenue bottomed out at $0, that's still not a lot from EPYC.



That's the weird thing. AMD is making a lot more than I expected they would outside of the datacentre. Especially looking at what the plans were for the product stack in 2017 moving forward.



It might be, from an accounting perspective, though I don't know if that would affect revenue reporting.

This report along with the next quarterly forcast just shows just how much work AMD needs to get a foothold in the CPU market.

What should make people concerned it AMD is only expecting 21% year on year quarterly growth next quarter.

That isn't good considering how weak Intel's products are at this point and next year, Intel should definitely have their 10nm products available.

While we might be tempted to blame it all on Covid 19, Intel is expecting nearly 10% growth year on year for Q2. That is 1.5 billion dollars which should be AMD's if they were absorbing more of Intel's share. With Intel's revenue growing healthily still, AMD should be smashing it for next quarter but it isn't and that why the stock has been dropping today.

With the release of mobile ryzen, the partial ramp up of next gen consoles and likely no supply issues at this point for 7nm's, I would expect year on year growth to still be 40+%.

AMD really needs to get more executives talent for their marketing team to help establish better business to business connections. While their engineering team is executing exceptionally, their marketing still leaves something to be desired. Their horrific ray tracing demo, the fake jebaiting and their cheap shots at the competition might score them high points on forums but it makes them look unprofessional. AMD needs to look professional and like a leader in their tech space to be taken seriously.

Right now, AMD has such a positive presence for tech enthusiasts and forum goers because of their strong connections with the media(e.g lisa Su actually does interviews with a focus on Anandtech and PCWORLD) and because of guerrilla marketing. But this isn't translating well with decision makers who are making purchasing and design decisions because what may entertain and build up good reputation to the public, isn't translating well to business customers.

What I think AMD needs to show business customers is that they are on board 100% for business support and a low asking price doesn't mean this support is going to suffer. A low price along with childish marketing schemes disguised as viral and guerrilla marketing is not going to get AMD customers on the server side. They need to act like professionals and stop blaming the competition.

 

[DrMrLordX]

@tajoh111

To be fair, I haven't seen much of AMD blaming the competition lately.

 

[piokos]

AMD really needs to get more executives talent for their marketing team to help establish better business to business connections

Just a formal note: "marketing team" focuses on communicating products to consumers, i.e. that's the division that gives you all the nice slides for CES and things like that.

Unlike one may think, companies aren't built from engineers and marketing.
What's you're thinking about is a corporate sales team, i.e. people who negotiate and analyze large deals. And AMD is probably covered on that front, since they're actually doing pretty well (consoles, cloud DC clients, gov supercomputers).

IMO they simply aren't that interested at the moment. They have limited supply of silicon. They focus on the product groups that benefit the most: high-core desktop and server CPUs.
They will give more attention to other clients when the priority markets are saturated.

DIY desktop and big datacenter segments are probably cheaper and easier. You don't have to offer that much support, you don't have to launch very well tested products. Hobbyists are more than willing to flash a BIOS or spend the evening on configuration. Large DC clients can solve stuff on their own (they usually build the servers as well).
But general consumer and business chips have to be rock solid out of the box. So launching "PRO" and mobile SoCs a few months after desktop lineup gives AMD time to polish it (and, let's be honest, run a public beta...).

 

[Gideon]

Nextplatform also estimates that datacenter CPU + GPU is now the vast majority of AMDs Enterprise, Embedded and Semicustom revenue:

The Steady Patience Of AMD In The Datacenter

We have said more than once in recent weeks that recessions accelerate technology transitions but they do not cause them. We now have a textbook

www.nextplatform.com

 

[piokos]

Nextplatform also estimates that datacenter CPU + GPU is now the vast majority of AMDs Enterprise, Embedded and Semicustom revenue:

The Steady Patience Of AMD In The Datacenter

We have said more than once in recent weeks that recessions accelerate technology transitions but they do not cause them. We now have a textbook

www.nextplatform.com

I'm not sure what graph you're looking at. This one shows they're ~even.

 

[AtenRa]

I'm not sure what graph you're looking at. This one shows they're ~even.

yea, that means that Datacenter (EPYC cpus and Instinct GPUs) are the vast majority of the Enterprise, embedded and Semi-Custom group.

Consoles (semi-custom) are near zero in that graph.

 

[Gideon]

I'm not sure what graph you're looking at. This one shows they're ~even.

You missed the point. Data-center CPU + GPU is part of the "Enterprice, Embedded and Semi-Custom" division. Essentially this graph says that 95%+ of the revenue of that division came from EPYC and Radeon Instinct (with the former being the majority) Almost 0 came from consoles. I quote:

Based on thin data, past trends, and hunches, we reckon that the Epyc CPUs drove $252 million in sales in the March quarter, up by 2.2X compared to a year ago and up 9 percent sequentially in a quarter that just plain finished weird for every company in the world. We think that datacenter GPU sales (which includes both Radeon and Radeon Instinct cards made by the Compute and Graphics division) amounted to $78 million in Q1, up a smidgen year-on-year and down a smidgen sequentially from Q4 2019.

 

[randomhero]

And also reconfirms that @Hitman928 was right on the money(pun intended).

 

[Hitman928]

And also reconfirms that @Hitman928 was right on the money(pun intended).

To be honest, I forgot about the Radeon element in the datacenter numbers. With that said, I think they are overestimating the Radeon datacenter numbers for this quarter. If their Epyc numbers were true and with Lisa Su's comments, then 1Q19 they would have had only ~$80M in Epyc revenue a year ago but datacenter revenue a year ago was ~$190M. Radeon datacenter numbers have also been trending down over the last few quarters, not up. They had a couple of peak quarters where Microsoft and Google were buying GPUs for cloud gaming, but that has largely dried up for the time being. I think it's more likely that in the data center, Epyc revenue was more like $290M and Radeon is ~$40M.

 

[moinmoin]

Better late than never: a transcript of the earnings call

Advanced Micro Devices (AMD) Q1 2020 Earnings Call Transcript | The Motley Fool

AMD earnings call for the period ending March 31, 2020.

www.fool.com

Some quotes, AMD wants to reach 10% market share in the middle of this year and Su is positive they are reaching the target:
"we expect about 20 million units a year in terms of single-socket and dual-socket servers. That's about 5 million units a quarter. So 10% share is about 0.5 million units. (...) Q2 is actually our strongest backlog quarter that we've seen."

On impact of Covid-19:
"the activity level continues to be high on both the cloud as well as the enterprise side. The only place where perhaps we see a little bit of a slowdown is, as I said, on some of the transactional business which we had plan to grow as we go through this year, and that might grow more slowly just as people aren't focused on new infrastructure right now. But in terms of cloud and large enterprise, there continues to be good activity on both current already one design platforms as well as new pipeline engagements."