If you don’t know Latin, that means “Where are you going?” and it’s the real question that telcos, their vendors, and their customers all should be thinking about. The harsh realities telecom faces in 2026 are obviously generating both news stories and vendor/telco activity. Of course, this has been true in the past, too, and not much useful has been generated. How about now? Does anyone, really, know what to do?
On the story side, We have another Light Reading piece that cites the 5G slowdown. We have three stories (HERE, HERE, and HERE) on telco interest in AI, and (no surprise) another Light Reading piece touting 6G. Vendors, as I’ve pointed out in the past, are largely pinning their hopes on 6G, which makes them a part of the last of the stories. Ericsson says in the first story that it really hopes that humanoid robots running around will be as dependent on 5G as people. AI is, of course, a potential revolution for anyone who need to toss another idea into the mix. What’s really going on?
To start with, most of the telecom vendors’ people I know tell me that nobody is really waiting for humanoid robots to learn to use smartphones. Some say that was an example of a point I clearly think is valid (because I’ve made it myself); mobile real-world, real-time, stuff is the basis for any credible improvement in service revenues. Robots with smartphones would take too long, so other steps would be preferred (but they’d sure be welcome in a telco phone store if they ambled in).
The telcos I chat with agree with that, but to them any new application to drive new services is demand-side thinking that their supply-side mentality won’t deal with. That’s why they vacillate between yearning for 6G and eyeing a role in AI. The pragmatic (well, maybe, the disillusioned-by-5G) telco types think both these are destructive self-delusion, but their only alternative is to focus on cost management until (inevitably, they hope) the necessary applications will evolve.
This group points out that both 6G success and AI success would likely stem from an edge-hosting opportunity that arises from the orderly evolution of current real-time applications that I blogged about yesterday. Telcos have two assets in that space. First, they have local real estate where the hosting could be sited. Second, they have a lower internal rate of return and a familiarity with first cost that cloud providers lack. Most of these telco people don’t think that the current model of AI, the hosted chatbot model, is going to drive new monetizable services, for the simple reason that the great majority of the use of this kind of AI is for-free, not paid. Who’s going to pay to deliver a capability they don’t pay for?
Nobody who promotes either AI or 6G transformation of services is specific about just who would pay for these services, and why. The Light Reading piece admits that “6G will initially be a phased evolution of 5G, and the gains will ‘probably be incremental to start.’” They also say “The US is on record as saying they want to be a leader in 6G.” That makes it sound like 6G is a sort of low-PR-value space race that’s off to a slow start.
The main problem with 6G, then, is that any significant revenue value it generates would have to come from some sort of differentiated service that consumers and/or enterprises would be willing to pay for. Well, differentiated services are valuable only if you have differentiated applications, and the powerful force of economics has been working to mold applications into a connection model that fits the Internet. Why would you build an application that requires more expensive service if you can make the world’s data dialtone work?
The AI problem is partly what I’ve noted above; most use today is free and so is unlikely to drive a pay-for service. Businesses also say that AI doesn’t really change WAN traffic much if at all, and most don’t see it doing so in the next three years. They do believe that AI combines with real-time applications, and that this would become (at some point) a potential consumer of differentiated services, but they’re not convinced they know exactly when that could happen.
One point the Light Reading piece makes, quoting a former Nokia exec, is “The traditional ten-year IMT (International Mobile Telecommunications) cycle at the International Telecommunication Union (ITU) ‘although it has value … is no longer really aligned with the operator investment cycle, and … it’s completely unaligned with the current AI cycle.’” This, I agree, is something that’s usually overlooked, and yet is critically important. The demand side of any picture evolves at the state of the market. The supply side, in mobile networking, is constrained to a ten-year standards cycle. In the past, when mobile service use evolved to match what people did with wireline, applications were supply-constrained and so there was pent-up demand to be had. Now, telcos and vendors are proposing to deploy service capability to support applications that don’t really exist, meaning that they either hope the applications to come along before the standards and supply, or that the standards and supply would quickly pull through the applications. The former means a lot of application-side vendors and enterprises would be taking a big risk, and the latter that the telcos would be.
One possible solution would be to have telcos deploy new service technology/infrastructure for their own use, justified by cost savings or perhaps by seizing market share. The latter, many point out, is exactly what happened with 5G. There was no real justification, limited or no recognized user value, but the PR around 5G made upgrading to it competitively essential. The cynics (and as a cynic myself, I’d have to agree) say that the tech media will hype up 6G just like it did 5G, or perhaps even more, because they need clicks. Hype will work again, then.
Cost savings is another matter. The problem for telcos in using AI to cut costs is that for several decades they’ve focused relentlessly on “automation” to cut headcount. Most of the jobs that could be cut through IT mechanisms have already been cut, so most of the benefits are off the table. Can you do netops better with AI? Perhaps, but as one telco put it, “Let me get this straight. You have this thing that hallucinates maybe 20% of the time, and I’m supposed to let it run my network?” No, you say, it will simply tell a human what it thinks is wrong, and what would set it right, and the human will check. Then where does the savings to justify AI come from?
Enterprises think that the great majority of the stories about how AI has enabled companies to cut jobs are “cover” for a decision to cut jobs to boost profits. The ones who have offered comments on their own experience say that while AI agent deployments, and some hosted-AI-chatbot use, has boosted productivity for some workers, this has not translated into any significant job cuts.
Where does this leave us, and more directly leave telcos, in facing the future? You need to raise revenues, because you can’t cut costs continually. But there are no immediate prospects for significant new telco revenue, so in the near term you have to try to manage costs. In the long term, you have to face the revenue need, and you have to be realistic and not just hopeful. Fresh thinking is needed, because clearly what has been suggested and tried in the past has not worked. Don’t they say that “insanity is doing the same thing and expecting different results”? Think about that, ye in the telecom space.
