Has it Ryzen high enough?
Over the last couple of months, I've been torn. I know a time is coming when either Advanced Micro Devices (NYSE:AMD) or Intel (NASDAQ:INTC) will be a no-brainer short. The thinking here is pretty clear:
- If AMD's Zen-based CPUs are as good as AMD longs believe they are, Intel will be a short.
- If AMD's Zen-based CPUs don't live up to expectations, AMD will be a short.
It has been my thinking that AMD's Zen will not live up to expectations. I've written on why I believe this, and it has mostly to do with circumstantial, leaked, benchmarks. These benchmarks included Ashes of the Singularity, Naples leaks and even AMD's slides on how Zen compared to "Orochi."
However, standing in contrast to these benchmarks, we have AMD's presentations which were reinforced during AMD's recent "New Horizon" event. In this event AMD supposedly set Ryzen (the 8-core, 16-thread 3.4GHz Zen-based enthusiast chip) against Intel's i7 6900K under several scenarios, and AMD's chip supposedly held its ground well.
For AMD longs, AMD's "New Horizon" event would offer all the proof they needed that Zen can't miss. I myself was led closer to the edge of believing so. I still have doubts on why AMD puts so much emphasis on very specific, highly parallelized workloads like blender. New doubts were thrown into the ring when AMD's own set files to replicate the test used a sample rate which, if different between the AMD and Intel CPUs, would hand one a definite advantage. But, honestly speaking, I was driven closer to AMD longs' belief that the chip might perform. Not that I considered the chance of buying AMD stock. I wouldn't, because it's already nearly 4x more expensive than before previous AMD successes. But I'd surely sell short Intel if AMD longs were right.
Yet, only a few days later, another doubt showed up. This doubt, mind you, is consistent with the work I had done before and again seems to show Zen as falling short on single-threaded workloads - which also can sometimes lead to a disadvantage in multi-threaded workloads.
The New Doubt
There were, according to wccftech.com, new Ryzen benchmarks leaked in Chinese forums. This time, the leaks concerned Cinebench R15 and Fritz chess benchmarks, and they compared Ryzen to the Intel i7 7700K (4 cores) and i7 6900K (8 cores). This is how Ryzen compared in both benchmarks:
Now, consider the following:
- The R15 benchmark is the multi-threaded benchmark.
- Yet, Ryzen appears awfully close to the Intel i7 7700K (966 for the 7700K, 1188 for Ryzen). This implies Ryzen will be significantly behind the 7700K in single-threaded workloads.
- 1188 does represent an improvement in excess of what one would expect from the 2.8GHz to 3.4GHz frequency bump. An estimate using "Orochi" (FX-8100, in my view) would put this at ~1080. Yet these calculations confirm the previous estimates as a 10% difference isn't exceptional.
- The chess benchmark is even more unfavorable for Ryzen.
Broadly speaking, if these new leaks are confirmed then they'd confirm the possible short thesis on AMD. This is so because AMD would again be forced to use eight core chips to compete with Intel four core chips. It's one thing to use Ryzen to compete with the $999 Intel i7 6900K, and quite another to use it to compete with the $349 Intel i7 7700K. Moreover, when we got to Naples or lower-end Zen-based CPUs with fewer cores, the problem would be worse still.
As For Shorting Intel
If we believe what AMD bulls say and AMD shows, then AMD would necessarily take a lot of market share from Intel. This would have two effects on Intel:
- Intel would lose in quantities sold since the x86 market is stagnated.
- And to protect its remaining market share Intel would lose on prices and margins even on the market share it held.
So, if AMD's Zen performs as AMD bulls expect it to, by 2018 it wouldn't be surprising for Intel to lose 30%-40% of its profits even on just a 10%-15% broad drop in market share.
I say 2018 because AMD's Zen launch will be rather slow over time, with Zen APUs only expected by H2 2017. But as we saw, if it happens, the effect on Intel is not proportional to market share loss.
On top of this Intel also faces other threats, like the possibility of Apple migrating toward its home-grown AMR-based CPUs or competition from ARM alternatives in general.
When Will We Know?
So, given this contradictory data, when will one know whether AMD or INTC are shorts? Basically, we'll only know when independent testers get their mitts on Ryzen chips and test them. Up until now we have:
- Anonymous leaks in the internet.
- Marketing efforts by AMD.
We can't know for sure whether the leaks or AMD are closer to the truth. But when independent testers test AMD's new chip, the doubts will dissolve. At that point, we'll either know AMD will get lots of market share because of having a very competitive chip, and Intel will suffer, or we'll know Intel will be safe for a couple of years more. We should know this by the end of January 2017.
This might also help those who keep criticizing me understand why I although have my doubts regarding Zen's performance, I haven't acted on those doubts yet. I'm waiting, patiently. That said, for sure things aren't as clear as AMD longs think they are, unless they believe these constant leaks are all fakes.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.