The Private Parts Networking

One of the big topics in wireless today is Private Networks. This is the idea that someone other than a telecom operator sets up its own network, and either uses licensed spectrum (i.e. leased from the operators) or unlicensed spectrum, open to anyone (i.e. CBRS). Private networks are not a new idea, but they have received outsized attention recently and gotten conflated with 5G and the recent, wild US spectrum auctions. We have heard very few people actually ask the basic question about what are private networks good for? And does it make business sense to run one?

The pitch for private networks goes something like this. Private networks provide their owners with flexibility to run the network they way they need. This may mean the network is cheaper than signing up for service from the operators. As we will detail below, the savings are not entirely clear. But flexibility can also mean building a network tailored to the user’s specific needs. The shorthand for this is IoT, a network scaled for large amounts of low-power, low-data devices. Alternatively, flexibility can also mean the ability to integrate the network more closely with the owner’s own IT systems. The fact that so little of this is defined is a clue that no really knows what private networks are best suited for, there is a lot of experimentation ahead.

Who could really use a private network? Our view is that these are best suited for companies that have to cover large geographic ranges, with a lot of special purpose devices. This could be someone with a large real estate plot – like an oil field, a massive factory complex or a logistics center, or it could be a company with a large delivery fleet or a utility with thousands of meters to read. As far as we can tell, it makes little sense for companies that are largely concentrated in an office – here Wi-Fi or even wired Ethernet make more sense.

To build these out, the owner will have to build a network – buy and install base stations and antennas. This is probably the source of cost savings. The telecom operators have to cover a wide range of needs with their general purpose network. For instance, they still have to support legacy 3G and even 2G phones A company building a greenfield 4G or 5G network today could use radically less expensive gear. But that is where the cost savings end. They will also need to operate the network which can be expensive as it will require a dedicated work force trained in supporting this network. Depending on the importance of the network (i.e. what happens if it goes down for an hour or two?) this can become a fairly costly proposition.

In our view, the biggest gain likely will come from the integration of the private network into the owner’s IT systems. One of the biggest problems the telecom operators face is their legacy billing and support system. And when we say legacy, we mean some of these systems are 100 years old. One of the big benefits of 5G is the way in which it allows for core telecom networks to be run in a way that looks a lot like the way that Internet companies have been running their networks for years – containers, virtualization and all the rest. In theory, this means that a private network owner could take ‘off the shelf’ 5G gear and then run their private network as part of their IT system. This is useful not only in that it makes configuration and set-up much easier, but should also allow the owner to pull all the telemetry off the end devices, so that they could know, for instance, where all their drivers were during the day. These things are possible today, but private networks would likely make this work much simpler.

When you factor this all together, it is hard to see how many companies really need their own private network. The number is greater than zero, but only a small number of companies are large enough that they could manage this in their own right. A private network might make sense for Amazon, for example, as they grow their logistics fleet and eventually move into delivery by drones. (5G’s low latency makes drones one of the few use cases where 5G is truly meaningful.) Amazon is of course operating at a scale matched by few others. From what we have heard, the only customer in advanced stages of private network deployment anywhere today is the US military.

That being said, there is probably an opportunity for a company to serve as a Private Network Operator – setting up and maintaining private networks for other customers. We think the support costs have the potential to be significant, and by having multiple customers this kind of operator could amortize those support costs more effectively.

All in all, we like the technology behind private networks, but the business model leaves us with a lot of questions.

Photo by the blowup on Unsplash

3 responses to “The Private Parts Networking

  1. I believe that the ability to use network slices under 5G might address some of the rationale for private networks that you mention here. These can be configured for the ultra-reliable low latency communications (URLLC) needed to support the use case mentioned here (drone operation). I believe that The other key 5G capability, massive machine to machine communications (mM2M) covers the IoT needs.

  2. Correct me if I’m wrong, but I believe that according to this article only the biggest organisations (such as Amazon) could really justify having a private cellular network. However, there seems to be quite a bit of interest at smaller organisations too. Examples could include the company that runs airports in Paris (https://presse.groupeadp.fr/4g5g-network/) or a Mercedes factory in Germany (https://www.daimler.com/innovation/production/5g-network-production.html). Granted, these aren’t exactly tiny ventures, but nevertheless, an airport or a factory that is 20000 square meters isn’t really a U.S. Military or Amazon kind of operation in terms of size and complexity.
    What factors are needed to justify rolling out a private cellular network in such environments?

    • Yeah, this post has some issues, things that I should have written more clearly. I am thinking here mainly of the US where there are very few private networks. The picture is different in Europe and other parts of the world. But I agree that airports and factories are the kinds of places where these might make sense. It’s just that these oppportunities haven’t really taken off in the US yet. They might, but it is still very early.

Leave a Reply