Discussion AMD Reports Fourth Quarter and Full Year 2020 Financial Results

moinmoin

Diamond Member
Jun 1, 2017
4,926
7,608
136
Those percentage changes...

GAAP Quarterly Financial Results
Q4 2020Q4 2019Y/YQ3 2020Q/Q
Revenue ($M)$3,244$2,127Up 53%$2,801Up 16%
Gross profit ($M)$1,451$949Up 53%$1,230Up 18%
Gross margin45%45%Flat44%Up 1pp
Operating expenses ($M)$881$601Up 47%$781Up 13%
Operating income ($M)$570$348Up 64%$449Up 27%
Tax Valuation Allowance Release Benefit ($M)$1,301--------
Net income ($M)$1,781$170Up 948%$390Up 357%
Earnings per share$1.45$0.15Up 867%$0.32Up 353%
Non-GAAP(*) Quarterly Financial Results
Q4 2020Q4 2019Y/YQ3 2020Q/Q
Revenue ($M)$3,244$2,127Up 53%$2,801Up 16%
Gross profit ($M)$1,452$950Up 53%$1,231Up 18%
Gross margin45%45%Flat44%Up 1pp
Operating expenses ($M)$789$545Up 45%$706Up 12%
Operating income ($M)$663$405Up 64%$525Up 26%
Net income ($M)$636$383Up 66%$501Up 27%
Earnings per share$0.52$0.32Up 63%$0.41Up 27%
Annual Financial Results
GAAPNon-GAAP(*)
20202019Y/Y20202019Y/Y
Revenue ($M)$9,763$6,731Up 45%$9,763$6,731Up 45%
Gross profit ($M)$4,347$2,868Up 52%$4,353$2,874Up 51%
Gross margin45%43%Up 2pp45%43%Up 2pp
Operating expenses ($M)$2,978$2,297Up 30%$2,696$2,094Up 29%
Operating income ($M)$1,369$631Up 117%$1,657$840Up 97%
Tax Valuation Allowance Release Benefit ($M)$1,301----------
Net income ($M)$2,490$341Up 630%$1,575$756Up 108%
Earnings per share$2.06$0.30Up 587%$1.29$0.64Up 102%

Outlook:
For the first quarter of 2021, AMD expects revenue to be approximately $3.2 billion, plus or minus $100 million, an increase of approximately 79 percent year-over-year and down 1 percent sequentially. The year-over-year increase is expected to be driven by growth in all businesses. AMD expects non-GAAP gross margin to be approximately 46 percent in the first quarter of 2021.

For the full year 2021, AMD expects revenue growth of approximately 37 percent over 2020 driven by growth in all businesses. AMD expects non-GAAP gross margin to be approximately 47 percent for 2021.

I guess the project annual growth of 20-30% is a thing of the past now.
 
Last edited:

Saylick

Diamond Member
Sep 10, 2012
3,082
6,170
136
Kudos to Dr. Su and AMD! In typical AMD stock fashion, the stock price drops after a blow-out earnings report, but I expect analysts to give AMD price target upgrades in the next week with the stock price following suit over the next few weeks.
 

Gideon

Golden Member
Nov 27, 2007
1,598
3,520
136
Q1 outlook looks extermely optimistic as well:


Ian Cutress said:
Outlook Q1 2021
  • $3.2B Revenue (Flat QoQ, +79% YoY)
  • 46% GM - $830m OpEx FY 2021
  • 37% Revenue ($13.37B)
  • 26% OpEx ($3.5B full year)

That $3.2B revenue outlook for Q1 2021 vs a $1.786B revenue in Q1 2020 is a crazy +79% expected YoY jump.
Products AMD is generating revenue in Q1/H1 '21:
  • Ryzen 5000
  • Ryzen 5000 Mobile
  • Radeon RX 6000
  • Radeon RX 6000 Mobile
  • 2 x Consoles
  • EPYC Milan
  • More Discrete

Lisa said that 2021 is going to be a big year for AMD's Enterprise (EPYC) and Commercial (Ryzen Pro) business lines. These are regular revenue generators that AMD needs to develop sustainable contracts for. Sorry consumers, that wafer supply is going to customers with contracts.
 

JasonLD

Senior member
Aug 22, 2017
485
445
136
I think they can separate embedded and semi-custom from Enterprise instead of grouping them together, as well as CPUs and GPUs from computing section. I know they intentionally grouped them together during tough times, but I think the time has come to separate them for clearer picture.
 

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
Excellent growth. The issue comes when the Xilinx acquisition is complete (end of year?), which will dilute shares. Xilinx forecast revenue growth over next couple of years isn't as rosy as AMD's, though they run much larger gross margins (~70%). That being said, the combined company will not see the same growth as AMD alone. This may result in share price dropping. If that happens, I could see AMD leveraging profit to buy back stock over the 24 months after the acquisition is complete.
 
  • Like
Reactions: lightmanek

CHADBOGA

Platinum Member
Mar 31, 2009
2,135
832
136
I wonder how much AMD is pressuring TSMC to build more capacity on both 7nm and 5nm?

It seems like it will be a while before Intel could possibly make AMD no longer capacity constrained, so one would imagine this is what AMD should be doing, but I have yet to hear anything along these lines? :confused:
 

itsmydamnation

Platinum Member
Feb 6, 2011
2,731
3,063
136
I wonder how much AMD is pressuring TSMC to build more capacity on both 7nm and 5nm?

It seems like it will be a while before Intel could possibly make AMD no longer capacity constrained, so one would imagine this is what AMD should be doing, but I have yet to hear anything along these lines? :confused:
i want to know what happened to all the Huawei and apple wafers?
if AMD is already consuming alot of those and its like it is right now.... then god damnit, my ideas of new rig in march are probably out the door!
 

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
i want to know what happened to all the Huawei and apple wafers?
if AMD is already consuming alot of those and its like it is right now.... then god damnit, my ideas of new rig in march are probably out the door!
We really only see little flashes of how much they're selling. One European retailer publishes figures monthly, and N7 sales (Zen2+Zen3) are way up (150%+) compared to this time last year. Combined with RDNA2, XSX, PS5 --- and let's not forget how many chiplets are going into Zen2 EPYC as well as how much Zen3 EPYC has been distributed ahead of the official release. Then add in the capacity they had to use in order to ensure OEMs for laptops had Cezanne, Lucienne shipped to them in time for integration...
 

gdansk

Golden Member
Feb 8, 2011
1,944
2,269
136
If that happens, I could see AMD leveraging profit to buy back stock over the 24 months after the acquisition is complete.
Wouldn't those profits be better spent guaranteeing capacity from TSMC for 6, 5 or even 3nm instead? Share buybacks have no chance of making AMD's dominance permanent, but being able to take full advantage of Intel's 7nm delay might.
 

positivedoppler

Golden Member
Apr 30, 2012
1,100
171
106
One has to wonder if Intel huge purchase of wafers from tsmc in 2021 is to muscle out amd. Even with 3.2 billions in sales, AMD is still small potato compare to Intel 20 billion. For AMD to capture 1/2 the market, they would need to triple their wafer supply. Intel buying it up for gpus that's going nowhere makes it unavailable for anyone else.
 

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
Wouldn't those profits be better spent guaranteeing capacity from TSMC for 6, 5 or even 3nm instead? Share buybacks have no chance of making AMD's dominance permanent, but being able to take full advantage of Intel's 7nm delay might.
AMD are already going to do that. There's limited wafer purchases AMD can make, and I don't think TSMC is interested in a bidding war between customers, which would only drive customers to other vendors, or make Intel invest more in their own production. As of now, all this hand-wringing about Intel swooping in to purchase N6 capacity that Huawei abandoned... it's not like Intel bought AMD's capacity. AMD's issue with N5 and N3 is going to end up being pure wafer cost. Density isn't going to be enough to offset the N5 wafer costing almost twice as much as N7, and N6 is just a pittance of an improvement over N7 that it's not even worth fighting over.

As for rationale for buyback, if AMD are confident about their growth prospects (and why wouldn't they be?), if there is deflation in stock price after the Xilinx merger, AMD could buyback the stock at that deflated price, await the market correction upward, and reissue at a higher price and increase equity capital to enable reinvestment in ongoing/new projects (or purchasing wafers!)... or perhaps they will use it to improve executive compensation to increase talent retention.

Intel are taking the opposite approach of buying back stock because they don't have a need for much equity funding. That could be because their profits are good enough -- and they are -- but it could also be that they aren't needing the capital for growth, which is also the case -- in fact, they're shedding their NAND division to SK Hynix, for instance.
 
Last edited:

eek2121

Platinum Member
Aug 2, 2005
2,883
3,859
136
“Sorry consumers, that wafer supply is going to customers with contracts.”

Yeah, about that...we just signed a new 4 year contract with a major OEM to deliver Intel systems because AMD can’t deliver volume...

I support AMD and all, but they won’t win like this. This contract was worth more than $100 million...
 

gdansk

Golden Member
Feb 8, 2011
1,944
2,269
136
AMD are already going to do that. There's limited wafer purchases AMD can make, and I don't think TSMC is interested in a bidding war between customers, which would only drive customers to other vendors, or make Intel invest more in their own production.

If AMD are confident about their growth prospects (and why wouldn't they be?), if there is deflation in stock price after the Xilinx merger, AMD could buyback the stock at that deflated price, await the market correction upward, and reissue at a higher price and increase equity capital to enable reinvestment in ongoing/new projects... or perhaps they will use it to improve executive compensation to increase talent retention.
There are more mythical unpurchased wafers if AMD was able to design their products on multiple nodes. As an example, consider that 6nm EUV is in production now with compatible design rules. AMD has many 7nm designs they could shrink. AMD simply missed the opportunity to have one of the earliest 6nm products ready. Now that AMD is 'becoming a company of scale' they really should exploit these opportunities again.

Secondly, what's the point of buying Xilinx anyway? FPGA and interconnects? Invest in exploiting newly acquired Xilinx IP before Nvidia-Mellanox-ARM consumes the entire data center. Stock compensation for employees for talent acquisition/retention doesn't sound bad either.
 
Last edited:

TheELF

Diamond Member
Dec 22, 2012
3,967
720
126
We really only see little flashes of how much they're selling. One European retailer publishes figures monthly, and N7 sales (Zen2+Zen3) are way up (150%+) compared to this time last year. Combined with RDNA2, XSX, PS5 --- and let's not forget how many chiplets are going into Zen2 EPYC as well as how much Zen3 EPYC has been distributed ahead of the official release. Then add in the capacity they had to use in order to ensure OEMs for laptops had Cezanne, Lucienne shipped to them in time for integration...
These are the official numbers straight from AMD, if they would be selling more than that they would be going straight to jail.
i dont think 6nm buys them anything really, a few switched layers onto EUV steppers that im sure TMSC would prioritise for 5nm anyway.

but if Q1 is really almost double rev on equal margin then you would assume almost double the wafers bought in Q4 2020 vs Q4 2019
This comes from tax benefits doubling the income, not from anything AMD actually did.
Tax Valuation Allowance Release Benefit ($M)$1,301
 

Kocicak

Senior member
Jan 17, 2019
982
973
136
Option 1

CASH $ 2 290 000 000
TOTAL DEBT $ 338 000 000

Option 2

CASH $ 1 952 000 000
TOTAL DEBT $ 0

Why not to do Option 2 ?
I mean - every reponsible person would do that, right?
 

moinmoin

Diamond Member
Jun 1, 2017
4,926
7,608
136
What happened to this thread? :oops:

That being said, the combined company will not see the same growth as AMD alone.
Depends on what kind of growth. AMD's 5 year outlook was talking about achieving ~25% annual growth which I think is perfectly feasible even including Xilinx.

It seems like it will be a while before Intel could possibly make AMD no longer capacity constrained, so one would imagine this is what AMD should be doing, but I have yet to hear anything along these lines? :confused:
Intel making AMD no longer capacity constrained? Am I missing something?
Anyway as long as demand outpaces supply for newer nodes there is no way around capacity constraints. Even GloFo appears to be capacity constrained at the moment.

Wouldn't those profits be better spent guaranteeing capacity from TSMC for 6, 5 or even 3nm instead?
That's what's essentially happening I think. AMD's outlooks have been pretty much spot on in the last years since it had to order capacity way in advance, and that sets the frame within which AMD can perform in the future well in advance.

One has to wonder if Intel huge purchase of wafers from tsmc in 2021 is to muscle out amd.
TSMC's CapEx shows TSMC is massively expanding as well. It's not so much about one customer outmuscling another but customers registering their demand and TSMC investing accordingly. So Intel's involvement just ensures more investment into increasing capacity, not some fantasy of one customer robbing another customer's capacity. And AMD, being a long time (graphics and semi custom chips at TSMC go way farther back than Zen 2) and now big time customer, shouldn't be in danger of losing anything.

“Sorry consumers, that wafer supply is going to customers with contracts.”

Yeah, about that...we just signed a new 4 year contract with a major OEM to deliver Intel systems because AMD can’t deliver volume...

I support AMD and all, but they won’t win like this. This contract was worth more than $100 million...
This is why Intel in my opinion is not in danger as long as it keeps its foundries. AMD covers the cutting edge technology. Intel is capable of offering the commodity quantity.

There are more mythical unpurchased wafers if AMD was able to design their products on multiple nodes. As an example, consider that 6nm EUV is in production now with compatible design rules.
Sorry to be blunt, but this is an inane suggestion. All newer nodes are capacity constrained already at the moment, even stuff at GloFo which nobody would call cutting edge anymore. And they are all constrained since demand outpaces planned supply in all areas. All those companies planned with less supply being sufficient and ordered and invested accordingly. For that to get fixed takes multiple years since it both involves investing in capacity increase as well as better capacity planning for the future. In my opinion it's only Apple that even gets close to predicting the actual demand for its products well, and Apple got a decade of knowledge to look back at for that.

Secondly, what's the point of buying Xilinx anyway?
I strongly suggest you to read the relevant articles and threads on that topic. In short, Xlinix inclusion expands AMD's TAM, most significantly datacenters with a focus on AI, ML and flexible specializations, automotive and so on. Xilinx and AMD already cooperate, and patents already show the possibility of including FPGA units in CPUs to offer custom acceleration in essentially real time. Xilinx is essentially the key for AMD to significantly expand beyond the confines of what's possible with just a combo of CPU and GPU.
 
Last edited:

LightningZ71

Golden Member
Mar 10, 2017
1,627
1,898
136
i dont think 6nm buys them anything really, a few switched layers onto EUV steppers that im sure TMSC would prioritise for 5nm anyway.

but if Q1 is really almost double rev on equal margin then you would assume almost double the wafers bought in Q4 2020 vs Q4 2019

The most significant area where N6 would have helped AMD where it was even possible that AMD could have targeted the node, is for Cezanne. The MT benchmarks that are coming our are hinting at a notable limitation with respect to power/thermals when all the cores are heavily taxed that is resulting in sublinear scaling and even minor regressions in some areas as compared to Renoir. The increased power efficiency of N6, especially in mobile, would have been welcome. The question is, would the extra cost of moving to N6, and the risk at the time of making the decision (2 years ago) that the node wouldn't pan out, have made it worth the investment? Would the processors sell at a high enough ASP to result in the same net profits per chip? I suspect that the answer would be that it would have broken even in the long run, but would have resulted in taking on additional risk, which is not a wise decision in most businesses.
 

Hitman928

Diamond Member
Apr 15, 2012
5,160
7,595
136
We really only see little flashes of how much they're selling. One European retailer publishes figures monthly, and N7 sales (Zen2+Zen3) are way up (150%+) compared to this time last year. Combined with RDNA2, XSX, PS5 --- and let's not forget how many chiplets are going into Zen2 EPYC as well as how much Zen3 EPYC has been distributed ahead of the official release. Then add in the capacity they had to use in order to ensure OEMs for laptops had Cezanne, Lucienne shipped to them in time for integration...

From the AMD earnings call:

Sell-through of our new Ryzen 5000 processors featuring our Zen 3 core was particularly strong, more than doubling the launch quarter sales of any prior generation Ryzen desktop processor.

They made a ton of them, demand was/is just crazy high.
 

Hitman928

Diamond Member
Apr 15, 2012
5,160
7,595
136
The most significant area where N6 would have helped AMD where it was even possible that AMD could have targeted the node, is for Cezanne. The MT benchmarks that are coming our are hinting at a notable limitation with respect to power/thermals when all the cores are heavily taxed that is resulting in sublinear scaling and even minor regressions in some areas as compared to Renoir. The increased power efficiency of N6, especially in mobile, would have been welcome. The question is, would the extra cost of moving to N6, and the risk at the time of making the decision (2 years ago) that the node wouldn't pan out, have made it worth the investment? Would the processors sell at a high enough ASP to result in the same net profits per chip? I suspect that the answer would be that it would have broken even in the long run, but would have resulted in taking on additional risk, which is not a wise decision in most businesses.

I agree, given the cost it would have taken to launch on N6, I don't think the risk outweighed the potential reward. Once AMD moves to 5 nm, it should help AMD increase overall volume as their 7 nm products will still be competitive and sell well. Obviously the new products on 5 nm are likely to have supply issues meeting demand if the current trends continue, but overall volume will increase significantly.

As far as increased leading edge volume, it's probably too late to have a very significant increase in volume for their coming 5 nm products, but I believe AMD and TSMC have made plans for a large increase in volume at the 3 nm node.
 
  • Like
Reactions: moinmoin

LightningZ71

Golden Member
Mar 10, 2017
1,627
1,898
136
Rembrandt, being later in the development pipeline, makes sense on N6 as when it was finalized for production, the risk on N6 was much lower. The speculated move to PCIe4, DDR5, and RDNA2 should make it something special.
 

Mopetar

Diamond Member
Jan 31, 2011
7,794
5,897
136
Option 1

CASH $ 2 290 000 000
TOTAL DEBT $ 338 000 000

Option 2

CASH $ 1 952 000 000
TOTAL DEBT $ 0

Why not to do Option 2 ?
I mean - every reponsible person would do that, right?

It's more complicated than this in the real world. Which countries are those cash reserves and that debt located. If they're all over the place, moving that cash around can get quite expensive.

Until recently the US corporate tax rates were higher than almost anywhere else on the planet, which meant companies would never bring overseas revenue back to the US and subject it to additional taxation. It's cheaper to have outstanding debt that gets paid off over time.