Qualcomm has become the sixth company to announce plans to build ARM-based server SoCs, likely focusing at least in part on opportunities in wireless infrastructure.
At the company’s annual financial analyst event, Qualcomm announced its foray into the server market, a move widely anticipated after recent hires and organizational changes. While Qualcomm provided few details, we do know that it has created a dedicated group for this effort and that the products will be based on the company’s forthcoming custom 64-bit ARMv8-compatible CPU cores.
Qualcomm did not indicate what markets it will be targeting, but did receive an endorsement from Facebook, a key potential datacenter customer. The company did note, however, the growth of the cloud, the shift to virtualized environments, and the desire for custom solutions from large datacenter customers as key disrupters in the market. Given the company’s background, Tirias Research believes that it is likely to target communications infrastructure, particularly wireless communications.
The company is not providing any details on the products or their availability. If you consider the anticipated timeline for custom silicon products for the smartphone market, the announcements are likely in the near future.
This marks the sixth entrant into the 64-bit ARM server market after AMD, Applied Micro, Broadcom, Cavium, and Huawei. Many have doubted the potential for ARM in servers because the chips and servers have been slow in coming, especially after Samsung ended its efforts to enter the market. Entering any new market takes time, and success depends upon various market dynamics. Unlike the handset market, which is accustomed to new silicon and devices every twelve months, the server market moves in years. Just from the time a major customer receives a new server, deployment of the platform may take from six months, for an enterprise server, up to 18 months, for a communications platform.
The ARMv8 architecture was announced in late 2011. The first ARMv8-based products from ARM were announced in late 2012, and the first ARMv8 mobile products were announced in late 2013. The first ARM-based server chips hit general availability in late 2014, accompanied by the first ARMv8 server platforms from vendors like HP. We are likely to see first large ARM-based server deployments starting in mid-2015 to late 2015.
Over time, I believe Qualcomm will be successful for several reasons. The company has specialized knowledge and intellectual property in wireless communications. No single architecture can dominate all the diverse server workloads. The power-efficient ARM architecture will make it naturally suited for certain workloads.
Finally, there are several market inflection points occurring simultaneously. They include the growth of cloud solutions, the rise of software-abstracted and configurable networks based on SDN and NFV, and the trend toward custom silicon by large datacenter customers trying to optimize power and performance. These market inflection points will reward solutions that are power-efficient, flexible, scalable, and cost-effective, traits exhibited by the ARM architecture -- and particularly by Qualcomm.
Ultimately, Qualcomm’s entry into servers will be welcomed by more than just Facebook. Its greatest impact will be helping create a more competitive server market overall.