AMD’s EPYC Quarter

AMD posted better than expected earnings on the back of two significant points and milestones. The first was the increased share gains of their EPYC server chips. AMD, for the quarter, saw double-digit share gains in server thanks to the continued adoption of EPYC. Notable is the revenue increase for AMD in servers came almost exclusively from EPYC CPU’s as AMD’s server GPUs have not yet seen the adoption against Nvidia I’m sure management would like. The AMD GPU for datacenter story remains one of the potential upsides for AMD’s business. AMD also noted they hit their goal of double-digit market share in server and server revenue accounted for 20% of quarterly revenue.

The other area, also a significant milestone, was AMD’s record quarter share of notebook sales. The strength of the notebook market because of work from home and learn from home, combined with PC OEMs committing more design wins to AMD due to Intel’s supply constraints, led to the highest revenue from client processors in 12 years.

When you look at the business of Intel and Nvidia, the datacenter part of the story is one of the most important mostly because datacenter is extremely lucrative as I’ve written before. As important as client processors are, even when volume grows, it drags down ASP as Intel and AMD consistently see ASP fluctuation during periods of strong desktop/notebook sales. I make this point because the upside to AMD is mostly because they have so much headroom to grow in the datacenter. While the same is true for the client business, where AMD also has less than 20% share, growth there is not as lucrative.

In light of the competitive challenge and dynamic between AMD and Intel, it is worth highlighting AMD’s historical struggles and multiple fundamental architecture failures that nearly sunk the company several times. Knowing AMD’s history, the near-death experiences, the technical and architectural missteps along the way, only makes their position in the market today that much better of a comeback story. In large part of this turnaround is current CEO Lisa Su and the engineering team who overcame the architecture challenges AMD faced integrating ATI with the Bulldozer architecture. That team led them to the Ryzen core architecture, and EPYC server architectures that are as competitive of technical solutions to Intel AMD has ever had in the market.

At this point, it is hard not to stay bullish on AMD, given they are firing on all cylinders. Their guidance for the next quarter echoed the strong market for PCs we see due to COVID-19. Still, their server growth guidance was weaker, which is similar dynamics both Nvidia and Intel signaled, which does confirm the broader market slowdown of cloud and server/datacenter after the strong momentum from cloud infrastructure seen during the early stages of the pandemic.

Intel Needs to Pull and AMD
I’ve read several investor notes on AMD post-earnings this morning, and everyone says the same thing, which is a point I completely agree with, which is that Intel is not going to sit back and let AMD run away with the market. Intel still has a dominant market position, and their fate in the grave is far from sealed. But Intel must fix what is now obviously a broken architecture and process technology. In many ways, Intel is facing the same issue AMD has faced and overcome, which is why I say Intel must pull and AMD as ironic as that sounds.

Intel’s recent reorganization is their attempt to do so. Still, it will also require extremely difficult choices for Intel, which may very well include a new manufacturing strategy both in terms of outsourcing their X86 designs to someone like TSMC and possibly collaborating more with TSMC inside Intel foundries.

Intel making drastic changes, per a pivot, or bold decisions elsewhere are the biggest thing to bear in mind when considering AMD’s total upside. But ultimately, healthy competition in the semiconductor space between AMD and Intel is one of the best things to happen to the semiconductor space in a decade.

US and Semi Manufacturing
The last point I want to make, and one that I’ve been thinking about for a while is what cost to the West will come if Intel has to drastically change their US-based manufacturing. An argument can be made that from a US perspective, there is a matter of national security concern if the US is to lose domestic semiconductor manufacturing by a US company. Even if TSMC was to make more chips in Intel foundries, even if those are Intel chips, TSMC is still not an American company.

Given we know China is investing heavily in owning the full stack of semiconductor technology, there should be concern about certain government’s security future should the US no longer be a player in semiconductor manufacturing.

This should be a topic being discussed within the US government, and possible solutions need to be discussed due to the great risk of losing semiconductor manufacturing.

The next year will be extremely important for Intel, and the semiconductor space as a whole.

Published by

Ben Bajarin

Ben Bajarin is a Principal Analyst and the head of primary research at Creative Strategies, Inc - An industry analysis, market intelligence and research firm located in Silicon Valley. His primary focus is consumer technology and market trend research and he is responsible for studying over 30 countries. Full Bio

7 thoughts on “AMD’s EPYC Quarter”

Leave a Reply

Your email address will not be published. Required fields are marked *