And this will be the case for the next two years, minimum. How can I be so sure? Especially with AMD up 157% year-t0-date and Intel up only 1.8% year-to-date, obviously AMD is overvalued, and Intel is selling at a discount, right? Wrong!
Even with AMD's 15% decline in the past few days, I am still bullish on AMD, and I can sum up exactly why in one world, yield.
If you don't know who Jim Keller is, then you probably don't pay close attention to CPU designers. Jim Keller is a microprocessor engineer, who designed the very popular AMD Athon Chip, then later worked at Apple and designed the A4 and A5 chips found in iPhones and iPads. He went back to a struggling AMD and helped them ryze from the ashes with the Zen architecture, found in AMD's Ryzen chips.
When I noticed that Keller was back at AMD and designing Zen, I decided to buy a small stake in AMD ($2.11 at the time). I was betting that Keller would do what he has done in past, make a really good CPU. I think it goes without saying, that he has delivered.
Keller, at AMD, helped design a technology called infinity fabric, this allowed AMD to make a modular chip, where they have multiple smaller dies that work together as one CPU. Infinity fabric is the magic that currently allows AMD to eat Intel for breakfast. AdoredTV put together an excellent video on youtube explaining exactly how AMD is very likely getting significantly better yields than Intel. The video is 20 min long, but worth watching in its entirety (if you start to get a bit lost, just stick with it, it will start to make more sense as it goes on, I promise).
These high yields are allowing AMD to have the best price-to-performance chips on the market.
When Epyc first came out Intel mockingly said the chips were "glued together", claiming their monolithic die was superior. Ironically Intel is certainly now working on a modular design themselves (more on that later).
Recent articles claim that Intel is advancing in it's 10nm process and that it is time to ditch AMD and buy INTC. I think anyone that listens to such advice will be woefully sorry. Even if INTC is making strides with their 10nm process (which is doubtful given the recent track record of Intel shenanigans) they are still using a monolithic die (low yield) in their CPU's, and they are not bound to have a modular design any time in the next few years, since it took Keller and AMD 5 years to develop Zen. Additionally, AMD is still on track to release 7 nm chips before Intel can release 10nm ones.
Earlier I said that Intel is certainly working on a modular chip. Now they probably won't admit this. I am sure if you ask someone at Intel they will say they can't comment. So how can I be so sure? Well Jim Keller, the guy that design Zen, left AMD, then did a brief stint at Tesla, and now works at...you guessed it Intel. On top of that Intel recently acquired NetSpeed, which has IP (intellectual property) around modular CPU design. This next statement is just speculation, but I wouldn't be surprised if this acquisition was nothing more than a play/move to deter any lawsuits from AMD when Intel comes out with their own modular CPU. With Keller now at Intel you know the lawyers at AMD will be watching Intel's products like a hawk for any patent infringement.
The articles calling to ditch AMD and buy Intel are bit premature, Intel has a ways to fall before they come back. When the chips Keller is working on at Intel are almost ready to hit the market (I estimate 2-4 years from now) that will be the time to buy Intel.
Disclosure: I currently hold a position with AMD and intent to hold this position for at least the next year or so.