Advanced Micro Devices (NYSE:AMD) is making headlines across the tech blogosphere with the long-awaited release of its ARM-based Seattle chips that have been delayed by almost a year now. The tech community may be rejoicing the maturing ARM ecosystem in the datacenter segment with this release, but I believe there is little to cheer about for AMD shareholders with a long-term picture in mind. To put it bluntly, Seattle chips may be downright irrelevant and incompetent right at their release.
The Threat from Intel
Let me start by saying that in the cutthroat semiconductor industry, even a few months of delay can make or break the prospects of a product or a company. But the release of AMD Seattle chips is about a full year late than its originally planned schedule. The company and its management may get some points for finally delivering on its long-anticipated Seattle chips and for showing to the world that AMD still has some fight left, but the delayed chips won't necessarily bring in any meaningful business for the troubled chipmaker.
To put things in perspective, over the course of the past year, Intel (NASDAQ:INTC) has moved to 14nm lithography, while the just-released Seattle chips are being manufactured using a 28nm fabrication process. This huge delay has seemingly impaled AMD's efforts to enter the server business right from the start. Technology review site AnandTech reports that the newly released Seattle chips are about 10-20% slower and drain more power compared to Intel's two-year old Silvermont-based Atom chips, in spite of having similar price points.
Intel is scheduled to release Goldmont-based Atom chips next year, which would represent an architectural jump of two generations. If these Seattle chips can't compete with Intel's two-year old chips in terms of performance or pricing at the time of their release, how can we expect them to compete (and stay relevant) against their larger counterpart's upcoming offerings? It just doesn't make any sense pay the same price to purchase technologically inferior AMD product for building expensive hyperscale data centers.
As I explained in my previous article, the only real edge AMD has over Intel right now is its semi-custom offerings. It would theoretically allow the company to tailor its ARM-based chips as per the specific needs of its clients and give it an upper hand over the competition. But that is missing from Seattle chips. Dan Bounds, senior director of data center products at AMD, just noted that custom ARM chips will be available with its next CPU architectures termed as Zen and K12, which are still one to two years away from mass production.
Without the custom ARM chips, there isn't much reason to pick AMD over its other ARM-based competitors, or Intel for that matter. Purchasing decisions would be made on benchmarks run on standard chips and their pricing. Since AMD's Seattle doesn't offer any significant performance gains over existing Intel offerings, I believe AMD would have to lower its prices to become competitive in the segment. This could lead to lost revenue and lower profitability.
Enterprise clients may also purposely delay their projects and wait for Intel's upcoming Denverton chips - about 6-12 months away from mass release - and make their purchase only after conducting thorough comparisons and testing. In essence, AMD's 28nm Seattle chips will be competing with Intel's upcoming 14nm Goldmont-based chips. This could make Seattle chips an even tougher sell. Altogether, the aforementioned factors present a very bleak outlook for AMD's Seattle chips.
Competition in the industry
The competition within the ARM-based server segment is fierce as well. First off, the market for ARM microservers is still in its nascent stages, and there aren't many vendors that offer ARM-based systems.
Cavium Networks (NASDAQ:CAVM) and Applied Micro Circuits Corp. (NASDAQ:AMCC) dominate the segment and have forged ties with server makers and enterprise clients, so displacing them right away may not be entirely possible with a generic ARM architecture.
Secondly, new entrants are expected to come in going forward. For instance, Qualcomm (NASDAQ:QCOM) recently entered the ARM server race with its 24-core processor, which could reportedly house more cores by the time the company starts mass production. Qualcomm can potentially bring its mobile expertise and technology into the server segment and achieve higher economies of scale as a result of that.
AMD will have to form ties with server vendors and enterprise clients and prove to them that its offerings are better than its other ARM-based competitors in terms of price and performance. This could be a long shot, as the troubled semiconductor company will have to compete using its year-old Seattle architecture until its K12 comes into mass production.
I just don't see Seattle chips as credible contenders in the server segment due to the aforementioned reasons. At best, without any real performance and pricing advantages over its competitors, the Seattle chips may just end up being just a testing platform for enterprise clients.
But that doesn't necessarily mean AMD can't ever crack the ARM-based server market. The company may be able to fiercely compete in the ARM segment if its K12 release is: 1) at a lower and competitive node; 2) released in a timely manner; and 3) offers semi custom chips to offer flexibility for enterprise clients.
Until then, don't expect any miracles from AMD. The Seattle chips won't sell like hotcakes in the current market scenario.
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.