Image source: Intel.
Back in 2013, Intel (INTC 5.69%) released a server processor code-named Avoton and a networking-oriented counterpart by the name of Rangeley. These processors were based on the company's then-new Silvermont architecture, then the latest iteration of Intel's low-power, low-cost Atom processor family.
Avoton was intended to power a type of server known as a "micro-server," a type of server that uses many relatively weak processor cores rather than fewer high-performance processor cores for lightweight tasks. Rangeley was essentially Avoton with some specialized hardware on the chip intended to aid certain tasks popular in the networking world.
Intel is supposed to follow Avoton and Rangeley up with Denverton, a part using next-generation Goldmont CPU cores and built on the company's 14-nanometer chip manufacturing technology. However, the product continues to be delayed, which leads me to doubt whether it will ever actually be released.
The latest delay
Previous Intel public product road maps had suggested that Denverton would arrive sometime in 2015. Intel's public road maps later showed this part as coming sometime beyond 2015, presumably in 2016. However, the latest public road map now shows Denverton pushed out to sometime in the future, beyond 2016 .
If the product ever sees the light of day, it is now almost certainly a part that will show up in 2017.
Likely cancellation
The problem with delays in the semiconductor industry is that product cycles tend to be short. Intel typically refreshes its server products at an annual clip, although sometimes delays (lately due to manufacturing issues) extend those cycles to about a year and a half.
I suspect that by the time Denverton is actually released, it will not offer a particularly compelling value proposition in the marketplace. Competitors are likely to have superior products, and Intel's own higher-performance Xeon processors will probably also render this product mostly pointless. A product that was supposed to arrive in 2015 or 2016 just isn't going to be competitive if it is released in 2017.
I expect it to ultimately be cancelled.
What of Atom-based server chips in general?
Intel's data center group is extremely well funded and it has shown an enormous degree of competence for quite a long time. If Intel viewed Denverton (and whatever the networking-oriented variant of the chip is called) as a high-impact product, the company would have done whatever it took to get the product out in a much more timely fashion.
The fact that Intel keeps kicking this product down the road tells me that it doesn't care all that much. I wouldn't be surprised if this was due to Avoton and Rangeley seeing weak demand in the marketplace.
And if Intel doesn't care all that much about Denverton, it is unlikely that it particularly prioritizes Atom-based server chips in general.
Based on this, not only do I think Denverton will be cancelled, but I strongly suspect that Intel will not invest any further in Atom-based server chips. Instead, I expect these resources will be moved toward building highly integrated system-on-chip products for various market segments based on its higher performance Core architecture. In other words, its Xeon D product line.
