As part of this year’s MWC-related OpenRAN news this excellent piece by Morris Lore at Light Reading caught our eye. There are some interesting developments in telecom space here that we thought would be informative, but the real point of our post is to say “We told you so.”
First, some background. Wireless networks run over a bank of equipment – base stations, switches etc. The way these are designed and built is changing as telecom networks move to more modern software practices. One prominent change in all this is that much of this gear, especially base stations, are starting to look a lot like the kinds of computer servers we find in data centers. These base stations servers increasingly run on off the shelf processors as opposed to purpose-built chips designed by the equipment makers. Ericsson has long been a large Intel customer (and Huawei was in the Top Ten not too long ago).
Two years ago, we wrote a piece about Intel’s moves into this kind of telecom gear. As telco networks look a lot more like data centers, it makes sense that the leading provider of data center silicon would want to carve out a big share of the market. In that piece we noted that Intel was highly active within the data center working groups of the 3GPP, the body that creates the wireless standards. We noted that Qualcomm was entirely absent from those working groups, focusing entirely on other parts of the standard. Qualcomm’s rise to industry prominence was largely built on their management of the standards-setting process, so this absence seemed meaningful.
Which bring us to the Light Reading piece. This focuses on processors for the OpenRAN architecture, a new way of building the software running on all that networking equipment gear. We wrote about OpenRAN in our earlier piece about MWC, (and here is our background piece on OpenRAN). ORAN, in some form, is likely to become more prominent across telco space, and Light Reading holds that Intel is set to win the majority of processor demand for this gear.
However, there is an important distinction within the article. Base stations (or Distributed Units, DUs, as they are known in the ORAN acronym ocean) really need two workhorse chips – the CPU and some form of ‘accelerator’. The accelerator focuses on translating digital signals into radio waves, while the CPU does the heavy processing of general compute as in any server. This is directly analogous to the way the smartphones have two chips – the modem or baseband and the applications processor, respectively. Qualcomm just announced their line of ‘accelerators’ and has no plans to enter the CPU market. For their part, Intel is just focused on the CPU, having sold their modem business, they probably have no way to build an accelerator. So in theory, the two companies could split the market along those lines.
However, we believe Intel’s secret weapon is its ability to build software ecosystems. Their strength in the data center was built on their nurturing of the Linux ecosystem 20 years ago, followed by hundreds of venture investments in enterprise software tools (e.g. databases). Light Reading points out that Intel has invested in many ORAN software companies already. They are running the same playbook. We would argue that this could give them a big advantage and potentially even leverage over a wide swathe of future wireless standards, which could eventually push Qualcomm out of this part of the industry.
That being said, Intel has a lot going on right now. All this ORAN investment and product development could be seen as fairly far from Intel’s core markets of data centers and PCs. The company has been trying and failing to break into wireless for 20 years. So maybe this will all get re-org’ed away in some future cost cutting. Our guess is that the similarities between this business and data centers is sufficient to keep Intel in the mix.