In late 2013, Intel (INTC 1.61%) introduced its first truly viable Atom-based server and networking processors, code-named Avoton and Rangeley, respectively. These chips featured up to eight of the company's then-new Silvermont CPU cores and were built on the company's 22-nanometer manufacturing technology.
The move seemed to be a pre-emptive strike against a number of smaller chip companies looking to field low-power chips based on mobile-focused cores from ARM Holdings (ARMH).
Although Avoton and Rangeley seemed to be very strong parts for what they were, it turns out that server chips with a bunch of relatively weak CPU cores weren't as desirable to many of Intel's server customers as its higher-performance Xeon chips.
It makes sense that, roughly two years after the launch of the Atom-based Avoton and Rangeley chips, Intel seems to be dragging its feet in releasing Atom-based successors to these chips.
Wait, how do we know that these chips haven't been desirable?
At a recent investor conference, Kirk Skaugen -- who ran Intel's server group until early 2012 and now runs the company's Client Computing Group -- said that one of the last things he did before leaving that role was to "aggressively pursue Atom-based servers."
The rationale, he indicated, was that Intel would rather disrupt itself rather than let a competitor do it.
That being said, it doesn't seem that much "disruption" has taken place. Indeed, Skaugen claims that only a "very, very small percentage" of the server market is actually well-served with such products. In fact, he noted that when customers actually went to give these Atom-based server chips a shot, those customers realized that "Xeon is a pretty good purpose built product."
What does this mean for future Atom-based server chips?
According to Intel's most recently published public product road map, the successor to Avoton -- known as Denverton -- is scheduled to arrive at some point in the "future." It seems reasonable to expect these chips to arrive in early 2016, when Intel is expected to refresh its higher-end Xeon E5 processors.
At any rate, it would seem that these Atom-based server processors are relatively low on the company's list of priorities. We will probably see Intel release a few more generations of Atom-based server chips, but unless demand for these chips picks up, it wouldn't be surprising to see Intel ultimately killing this product line over the next five years or so to focus on other product lines.
What will Intel focus on going forward?
Intel's workhorses in the server processor market are its Xeon E5/E7 processors. Intel offers a very broad spectrum of processors within these families to address many different workload types, and I suspect that the majority of the company's efforts in the server CPU market will be spent in trying to make these processors and their accompanying platforms even better.
Another segment worth watching is the company's Xeon D family, which debuted earlier this year. These are low-power, highly integrated chips aimed at the same sorts of workloads that Intel's Atom-based server chips are. The key difference between Xeon D processors and the Atom-based server processors, though, is that the former use the company's "big" processor cores -- the same ones found in its PC processors as well as its high-end server processors -- rather than the relatively weak Atom cores found in the latter.