On Aug. 27, Intel (NASDAQ:INTC) held its first "data center" day. During this event, the general manage0rs of the various divisions that comprise the company's Data Center Group gave roughly 30-minute presentations outlining the strategies for their respective segments.
During this event, Intel executive Sandra Rivera, the general manager of the company's Network Platforms Group, outlined the company's strategy to tackle the $18 billion market for networking silicon.
Rivera offered plenty of insight into the company's networking platform strategy. Here are three of the key points she made during her talk.
Why is Intel able to address this market today?
During the question-and-answer session with Rivera, one analyst pointed out that the networking-chip market has generally been serviced by chips built around the PowerPC and MIPS architectures. The analyst wanted to know what had kept Intel from being a major player in the networking market in the past (even today, the company has less than 10% revenue share) and what dynamics could help Intel capture more share in the future.
"Historically, the networking workloads required specialized functionalities, specialized I/O, specialized crypto compression capabilities, and specialized acceleration technology," Rivera said. "But over the last 10 years, we have been investing in the Intel architecture road map to integrate more of that capability into the processors and the chipsets."
Rivera also discussed the value proposition of moving away from what she described as a "very fragmented" market filled with a number of chipmakers serving different niches toward standard, Intel-based solutions.
Finally, she highlighted the "scalability" of Intel's offerings, which allows the company to offer potential solutions that range from very-high-performance Xeon processors all the way down to very low-power and dense Atom-based solutions.
A shift that benefits Intel
One trend in the networking market that Rivera discussed is the shift away from specialized, purpose-built chips toward general-purpose processors such as the ones that Intel builds.
Although a chip designed to perform a singular task is likely to be more efficient in running said task than a general-purpose processor will be, Rivera points out that such solutions are expensive, requiring significant volumes to justify the investments required.
"We see the industry shifting from [application specific integrated circuit] technology and purpose-built technology and network processors to CPUs, to general-purpose processing, that can run those same workloads in a much more cost effective manner," Rivera said.
Rivera also said Intel is seeing what she describes as a "convergence of workloads." She argued that by moving all of the key networking workloads to Intel architecture, customers would be able to benefit from lower solution development risk, time to market, and ultimately costs.
Intel has been investing heavily to capitalize on this opportunity
Rivera indicated that Intel has doubled its networking-specific research-and-development spending over the past two years, which is a clear signal that the company is serious about trying to continue to outgrow the overall networking market.
If we look at this slide that Rivera showed, we see that Intel plans to more than double its revenue from the networking market by 2018:
Given that Intel said it brought in about $1 billion from the networking market during 2014, it looks like the company plans to be at an annual run rate of well over $2 billion by the end of 2018. Assuming Intel's data-center group operating margins stay at around 50%, the growth in networking should also contribute quite nicely to the growth in the operating income of the highly lucrative-data center group.