Cloud Hosting

Advanced Micro Devices (AMD) is on the comeback trail, with a new generation of GPUs for graphics cards and CPUs for personal computers. The stock has come back strongly too, recently hitting a 52-week high of $15.55 after hitting a long-term low of $1.65 back in July of 2015.

A third leg of the AMD comeback could be CPUs for servers, which go mainly into datacenters and cloud farms. This is the least probable comeback. Once a significant second fiddle to Intel (INTC), by 2016 AMD's share of the server CPU market had dropped to less than 1%.

AMD has new Epyc server CPU processors in production that could possibly compete with Intel's. But AMD is essentially entering a new market, which will present difficulties in gaining market share. If you are not aware of these problems, you can find a short review here.

AMD EPYC server chip (source: AMD)

This article will assume that AMD overcomes these re-entry problems and gains significant server market share in the near future. I will examine the scenario where AMD gains 10% market share for three years starting in 2018. Then I will assess the current stock price in that light.

Intel's 100%

Intel releases quarterly figures on revenue from their data center group at each quarter's analyst conference. Here are the figures for 2016, from my notes:

Quarter of 2016

Datacenter segment revenue, $ billions

1

4.00

2

4.00

3

4.54

4

4.67

total

17.21

I'll round to $17.2 billion annual server revenue. I would note that Q4 revenue was up 8% from year earlier. It is not the safest assumption, but I am going to use 8% revenue growth to project the 2018, 2019, and 2020 numbers:

2018 $18.6 billion

2019 $20.1 billion

2020 $21.7 billion

We will also want to know how profitable the Data Center Group was in order to estimate AMD's potential profits under our hypothetical scenarios. From the Intel Q4 2016 transcript I found that...

"The Data Center Group had operating profit of $1.9 billion, down 14% year-over-year. Operating margin was impacted by the two one-time items I referred to earlier and the ramp of 14-nanometer on our server products, which we expect to generate continued cost improvements over time."

That gives a tremendous operating margin for the group of $1.9 / $4.67 = 40.7%. That is one reason why AMD wants to be in the space. In contrast, Intel's operating margin for its personal computers, aka the Client Computing Group, was not stated. But it must be lower, since it is the other major group and the overall operating margin was 30%. Gross margin was 63% overall.

Contrast that to AMD's Q4 2016 margins. AMD reported negative operating margins and a 32% gross margin.

AMD server scenarios

To compete with Intel in the server market I believe AMD will need to compete on pricing, and even at the same price would have lower gross and operating margins. I am going to pick a number out of the air, 20% for operating margin. If you know how to use a spreadsheet, you can examine what would happen at other margin points. Even margins of 18% or 22% would have significant impacts on how I would price AMD stock.

Here's what we get using Intel data center group revenue as the market size, an 8% annual growth rate, a 20% operating margin, and various AMD market shares. The years are just illustrative. I would be very impressed if AMD got 10% per year each year, but not surprised at the 5% per year rate:

year

TAM, millions

% of TAM

AMD revenue, millions

AMD operating profit, millions

2018

$18,600

5%

$ 930

$ 186

2019

20,100

10%

2,010

402

2020

21,700

15%

3,255

651

2018

18,600

10%

1,860

372

2019

20,100

20%

4,020

804

2020

21,700

30%

6,510

1,302

As the table shows, there is a huge difference between AMD taking 5% of the server market and then getting stuck there, and eventually taking 30% of the server market.

Note that it has been years since AMD has shown an operating profit on a full-year basis, even using non-GAAP accounting. Showing any operating profit at all would be a win, but would not justify the current stock price.

AMD stock price scenarios

Investors are enthusiastic about Ryzen CPUs and AMD's new graphics chip lineup, which are likely to add to revenue and profits. There could even be positive earnings for Q2. I am going to ignore that here and assume only the server CPU market will produce operating profits, earnings, and positive EPS. I am clear it is not the assumption most investors are making; it is to clarify the server segment potential contribution.

The 20% operating margin used in the calculations above is simply a guess, placed between AMDs past zero to negative margins and Intel's rather good margins.

I will use operating margins as earnings and EPS (earnings per share). In reality, at the very least, AMD's interest payments and any taxes would be subtracted from the earnings.

I am also using the latest number of shares published, 945 million. This would tend to be low since AMD will certainly issue stock-based compensation during the period covered.

Finally, I have to pick a P/E ratio. Right now that ratio is infinite, which is to say, there is no meaningful correspondence between the current stock price and 2016 earnings, which were negative. They were negative $0.60 per share on a GAAP basis, and negative $0.14 per share non-GAAP.

So you might say I am counting on Ryzen to at least get AMD to break-even, which is not an unreasonable assumption.

I am going to use a P/E of 20, which is a reasonable P/E for a company with a history of profitability but not very fast EPS growth, at least in a low-interest-rate environment. If AMD does turn out to grow revenue and EPS at a fast rate, a higher P/E would be reasonable.

So here are the calculated stock values for my table, with all the many assumptions I have made:

year

AMD server revenue, millions

AMD earnings, millions

EPS

AMD stock value, P/E = 20

2018

$ 930

$ 186

$0.197

$3.94

2019

2,010

402

0.425

8.51

2020

3,255

651

0.689

13.78

2018

1,860

372

Visit link:
Evaluating AMD's Server Market Potential - Seeking Alpha