The battle between competing ARM and Intel architectures and ecosystems is reshaping the industry in many ways. Expectations are changing around form factors, programming models, costs, power budgets, and more.
For example, two ARM SoC vendors already offer their own server platforms -- AMD and Nvidia. As a result, some server OEMs likely will differentiate their products through vertical integration while others partner with select SoC vendors for customized solutions. I suspect Intel likely will follow one or both of these paths in the future.
The server segment is shaping up to be much more of a battle than anyone predicted, and that could have significant repercussions. I forecast ARM-based server solutions beginning in 2014 with the ARMv8 architecture. While the timing appears to correct, the battle for servers appears to be taking a turn I did not anticipate.
ARM is garnering a considerable amount of interest from the entire server value chain, from the SoC vendors to the large datacenter customers, but not just because of the potential power and cost savings ARM may offer. Much of the interest is in the potential flexibility of the architecture to develop custom solutions tailored to particular server requirements.
There have always been many differing server applications, ranging from small businesses to communications and datacenters. For example, in communications, the control plane that manages the routing of traffic and the data plane that carries the traffic typically use different CPU architectures.
Some things will not change. In client computing, it is unlikely ARM will conquer the traditional PC, just as it is unlikely the x86 will conquer smartphones and tablets.
I first wrote about the pending battle between ARM and Intel/ARM in an article published in the 2011 ARM Holdings annual report. I anticipated that the battle would be slow to develop and difficult for each party to enter into the incumbent's traditional market segments.
For the most part, the battle has played out as predicted. Despite a flurry of x86 tablets, they have failed to gain significant market share.
Microsoft has made strides to support ARM with Windows RT and the Surface tablet PC. Despite Microsoft's efforts and the growing dominance of the Android operating system in mobile devices, few ARM-based PCs have entered the market besides Surface.
In smartphones, Intel continues to scale down the Atom product family with the aid of advanced semiconductor process technology. But Intel's Atom design wins have not translated into significant market share to date.
Overall, I give the ARM architecture and ecosystem an edge in the current battle. It appears to be outflanking the industry heavyweight Intel, but there will be more battles to come.