gestern einen schönen Artikel auf dem amd stock subreddit gelesen.
Prior to Zen launch in 2016, AMD makes 4 billion/year and half of that is just PlayStation/Xbox. Out of the 2 billion, AMD probably makes around 1 billion in GPU and 1 Billion in CPU.
This year, AMD makes almost extra 1 billion due to the increase of CPU in the desktop. That accounts for 10 billion TAM and most OEM don't like Ryzen due to the missing iGPU. In any case, that's about 10% of the TAM. That generated almost 0.2 EPS over previous year.
If you use the same kind of adoption rate for EPYC and Ryzen Mobile, you can see that's there is at least 20 billion TAM in notebook and 20 billion of server. If we see the adoption is only half of the desktop due to OEM, That should translate to 2 billion additional revenue, which should be around 0.4 to 0.5 EPS. 30% of the growth over this year. That warrants the 10 to 15 ranges if we are using a P/E 23 of the standard S&P.
I'm guessing AMD signals 10% market share for EPYC by 2020, which is probably >0.75 EPS is coming from based on this model. 10% EPYC and 10% Ryzen mobile is about 4 billion additional revenue, 0.8 to 1.0 EPS. Takes the lower end of the profit and minus the 7nm higher R&D costs, that's probably how they get to 0.75EPS by 2020.
Notice that there is no significantly AI growth in the picture? Because there is not any. ROCm is at least 3 years behind CUDA and that's why Raja wants to go. There is no significant investment from AMD side to compete with NVDA on AI or even gaming. AMD is simply trying to use RTG for Semi-custom business and some gaming revenue. They won't and don't have the money for multiple mask for RTG. Navi has to be MCM because AMD probably only has enough money for 3 to 4 masks at most. 1 is Navi, 1 is Zen2, and the other 1 is APU.
If we are seeing significant adoption >5% this year, say 10%, the stock could double from here...reaching the range of 20+. The important question is how we look at Q1 guidance, if it's *30% like last year, we might get a good lift off. If it's the low to medium double digit, we are probably gonna get stuck in the 10 to 15 ranges until 2018. By earliest, i'm expecting Q2 a 35% YoY and Q3 40% YoY. That should break the stock passed 15 to 20+ territory.
If we are looking at 2018 Q3 report 25% YoY and 26% Q4 YoY in 2018. I think we will still get stuck where we are today.
The bond yield curve will probably converge around 2018h2 and early 2019 before 7nm comes out. If we don't see significant EPYC adoption by 2018 year end, I will have to exit all my position one way or the other.
My investment thesis is that Ryzen Mobile/Epyc will gain market share. To squeeze the shorts, they have to gain quickly next year before the market starts to drive downwards in 2019..
We will start to see couple of things in 2018