When it comes to new telecom services, we may have been missing the most important question to ask, one that could be phrased in a number of ways. Do we want evolution, or revolution? Improvement or transformation? Comfort or results? They all raise another question, of course. What’s the answer, or at least how might we approach getting an answer? Lets give it a try, because telco business models and the success of standards like 6G likely depend on it.
Evolution is what operators are comfortable; transformation or revolution in service technology would delight vendors but pose a big threat to operators. The problem with evolution is that it’s not clear what kind of revenue upside it can generate. The presumption in an evolution model is that growth in traffic, usage, or both will drive incremental revenue. But with the dominant service of the day, that’s rarely true.
Networking today, surely in the next five years, and likely even beyond that, is dominated by consumer Internet. Because the Internet reaches virtually every tech-literate person and every business, everything has adapted to use the Internet in some way as “dialtone”. Even business services have tended to migrate to MPLS VPNs, which use IP routers, or SD-WAN, which rides on the Internet itself. We’ve had broadband Internet long enough that there is very limited potential growth in the number of customers. We have smartphones and mobile broadband that’s approaching saturation in most markets. People are looking to pay less, not pay more, for their services, and anyone dreaming up new applications wants those applications to work over the Internet, with no incremental cost to users.
The reason this is important is simple; if you have a service, an application, a mission that you want to support, the best strategy is to base it on Internet connectivity because you’ll have that if you have any connectivity at all. If you can ride on connectivity your prospects already pay for, and they don’t need to pay for more connectivity, whatever they’re willing to pay for your service is yours to keep. Pick something more exotic, and you may have to wait for it to be available to your targets, and if they have to pay extra for it, you’ll see your piece of their willingness-to-pay pie shrink.
The corollary point to this is that if you want to sell a connectivity service other than consumer-grade Internet, you’ll face two significant problems; on-ramp and ROI.
The on-ramp problem is directly related to the consumer-Internet comments. How will an application for your new service develop? Either you have to deploy it in anticipation of applications that need it being developed and deployed, or developers will have to build the applications without suitable services being available. Since the latter is surely problematic, that means that operators are forced build out in anticipation of demand. That’s been the rule since 5G, and I think it’s pretty clear it hasn’t worked out for operators.
The reason it hasn’t is that you can’t start a revolution with supply-side changes unless there’s a huge pent-up need for the service, meaning unless the application(s) you expect to support exist in limited form with the current service set, and all that’s needed for them to expand is your new set of services. This, in effect, goes back to expecting the application developer to build something that will eventually demand new service features, but can hobble along with basic Internet. Do we have that? Maybe.
The trick is to find stuff that can “hobble along” with current Internet, but which would make a lot more money for the companies who invent the “stuff” if some service barriers associated with the current Internet were lifted. If hobbling along is good enough, then there’s no migration incentive because the presumed new service target, because it would make money for operators, would leave less money for our stuff-sources to keep.
This two-way balance (stuff-source versus operator) is bad enough, but if the “stuff” requires a series of technology evolutions or costs, then all the stakeholders have get enough of the retail pie to make their own investment and risk justified. The problem is that it appears that the “stuff” that could drive transformational opportunity is a multi-stakeholder proposition.
5G, essentially, promises two things. First, network slicing. Second, lower latency. Network slicing allows a mobile operator to separate traffic on their network for security or QoS reasons. Lower latency lets them support applications that are latency-sensitive, which tend to be “real-time” applications. Slices to separate traffic for security reasons hasn’t been compelling; why not use encrypted tunnels instead? Slices to separate traffic that’s already using low-latency mobile-RAN features? Maybe that would work. Thus, it’s real-time stuff we have to depend on, and if we want to do that, we have to assume that some application is interacting with our real world, our real lives. At the minimum, such an application has to know a lot about the real world, and a lot about us. The latter is complicated but possible, but the former relies on a fairly broad deployment of things that can tell the application about real-world conditions, and likely also things that can influence the behavior of some of those things.
If we look at the 5G promise in a cynical way (which is always my recommendation), we see that the presumption operators are making is that since we’re not generating humans with paying potential at a satisfactory rate, we need non-human subscribers, meaning M2M in some form. That means deploying “things” that need connectivity. This is why things like connected cars, smart cities, autonomous this and that, is so important to operators and their vendor community. All the “thing” applications/missions demand substantial buy-in from multiple classes of vendors, not to mention a substantial willingness to pay at the retail business/consumer level.
The real difference, I think, between telecommunications service evolution and revolution/transformation is that you could expect the former to be driven by growth in the customer base, which could come about by expanding the service area or lowering the price. With the latter, you need an application set we don’t yet have to be developed, implicating a host of ecosystem partners. The question, the big question, is whether there are any real-world, real-time applications that can deliver enough retail revenue to fill all the requisite pockets and still make a business case at the top of the value chain, where the retail buyer has to be the money pump. That’s the mission for any who want to promote a revenue-driven telco transformation, period.
But who wants it? Telcos? Vendors?
As I pointed out in my blog yesterday, network equipment vendors tend to think like telcos, meaning they fear competition more than they value opportunity. A revolutionary infrastructure change would put everyone’s incumbency up for grabs, so you can guess what that means.
Telcos should, but they’re the originators of the fear-competition mantra, and like enterprises they’re reluctant to embrace multi-vendor networks, fearing the usual complications. They’re also historically captive to vendor thinking, even in a standards forum.
It may be, as I’ve suggested, that chip vendors are the logical candidates, but Nvidia and Broadcom are focused on validating AI spending. Nvidia has been quietly doing more to support the “real” agent model, as their acquisition of SchedMD (arguably an example of the embedded agent model enterprises mention) shows. The have to be careful, though, because ramping up realistic agent technology will take time, and getting really aggressive about how agents are the future could hurt Nvidia’s chip sales to those cloud-AI giants. If nobody steps up, I think we’d eventually blunder into a solution path, but it would likely take beyond 2030 to do that, and we may have to figure out how to rescue the industry before then.
