Sometimes big industry changes come with trumpets blaring and Barbarian Disruptors storming the Corporate Gates. Other times, major, important happenings occur behind closed doors at tedious conference sessions in bland hotel meeting rooms.
This story is one of those.
As John Whorfin said “History is made at night. Character is what you are in the Dark”.
Bear with us, we are going to dive into the esoterica of mobile telecom standards.
A big part of the upcoming 5G standard deals with the mobile operators’ Core Networks. That is to say, not the wireless part of the standard, but how the network handles signals once they reach a base station and then get routed around. The basic idea is that the new standard builds in greater flexibility into how those networks get built and operated. Here is a great summary of the basic ideas being implemented.
For those who read that summary and whose eyes glazed over at the first acronym (apologies for that, but this is telecom after all, you knew the job was difficult when you took it), the basic idea is to pull apart all the different tasks the network does and handle them in software. Previously, many of these functions were handled in dedicated hardware appliances. Prompting one senior network engineer we know to comment “My job title says Architect, but really I’m just a plumber”, having to cable together endless of racks of gear. The new standard defines each of these functions in a clear way and lays out ways for them to work together.
For those of you who regularly operate deep inside networks, none of this is new. The whole point of the standard is to allow the Telecom Operators to build networks that look a lot like the way the Internet operators build their networks. API calls, message handling, heterogeneous data stores and function abstraction are bread and butter issues for companies like Netflix (they cover much of this in their excellent engineering blog). In fact, our sense is that the 5G core network standard is already kind of old-fashioned, lagging the state of the art by many years. Which is to be expected, the operators needed to trade off cutting edge best practices for common capabilities and interoperability.
But dig right down, and the crucial aspect of this is that separation between hardware and software. It presages a big change in how equipment vendors design their gear. This is a big deal for companies like Ericsson and Nokia. Traditionally, much of their margin came from software captive inside their hardware, and customers had to pay full hardware prices. An equally large, but unsung change is that the hardware can now be built with general purpose processors – especially data center grade CPUs – a market for which Intel holds a near monopoly. The equipment vendors have long used a lot of Intel silicon, Huawei is one of Intel’s Top 10 customers, but 5G opens the doors to a vast expansion of that market. The equipment vendors will likely shed silicon designers who had been tasked with proprietary chips, and at the same time, the overall number of CPUs needed will grow dramatically. The equipment vendors will (probably) figure out new ways to monetize the bits they do well, and for Intel this could be a big opportunity.
This change has been coming a long time. And for people who participate in such endeavors (largely from beige, windowless airport hotel conference rooms) this is has been a multi-year effort. Nor has the opportunity escaped Intel’s notice. In fact, Intel has been steadily growing its participation in the standards bodies, notably 3GPP which drives the 5G standard. For years, Intel has been contributing to advancing the standard. You can see their past activity when you search for Intel + NFV. Their fingerprints are all of over NFV – let’s not worry about what that stands for, its just label for a big part of the architecture of what we are discussing. This has been a smart, forward-looking move by Intel, and they deserve credit for nurturing this development. Ecosystem building is clearly a core competency of Intel’s Data Center Group.
So now we come to the ‘disruptive’ part. For the better part of 20 years Qualcomm has dominated the standards bodies. People will interpret that word ‘dominate’ differently, but it is unarguable that the company has played a major role in the advancement of the wireless standards. And now Intel appears to be encroaching on their turf.
You could argue that maybe this does not matter that much. The standard are complicated and the ability of any one company being able to fully monetize the licensing of essential standards patents are much diluted. And Qualcomm remains the undisputed leader on the wireless side of the standard. That being said, there is a real risk that this could be Intel’s beachhead into broaching the wireless world. The may never be able to compete on the smartphone modem side , but the networking side of the standard is likely to grow in importance in the future.
It is also worth comparing Intel’s approach to ‘standards’ with Qualcomm and the operators’ Standards. Intel, as noted, knows how to build complex ecosystems of software that runs on top of its hardware. There are no formal standards for what Intel has built around the data center, just thousands of companies locked in by virtue of a decade of optimizations and tuning to run close to x86 metal. As software opens up in the operators’ networks, they can replicate the feat. More precisely, they would really just be expanding from an overlapping domain of expertise.
Further compounding the issue is that Qualcomm really has no way to counter Intel’s move. They have exited the data center CPU market. And they really have no networking assets. There appear to be no vestiges of Atheros’ nascent Ethernet products. Qualcomm apparently failed to even bid against Avago for Broadcom. And as we wrote about a few weeks back there are no scale players left to buy in the networking semis space.
It is hard to forecast how this will play out. To be more precise, it is easy to forecast a lot of possible scenarios, but the range of outcomes is too wide to forecast with much confidence. Nonetheless, the threat from Intel has the potential to shake up the standards bodies in many ways. The really pernicious part of this is that small changes in how the network passes message or routes calls can be pushed back to the air interface area where Qualcomm’s expertise rests. The parts of the standard affect each other and Intel’s entrance and many small changes could herald a big shift in this corner of the industry.