Modern telecom networks have reached a point where human effort alone can no longer keep pace with the scale, variability, and complexity of today’s network infrastructure. Operators are under pressure to maintain reliability across countless devices, functions, and traffic patterns, making traditional, hands-on assurance nearly impossible to sustain. Automation is emerging as the only viable path forward. It enables early detection of issues, cross-domain reasoning, symptom correlation, and rapid remediation, often before customers experience any impact. As the volume of traffic and potential failure points grows, automation is shifting from a cost-efficiency initiative to an operational necessity.
But autonomation does not eliminate the need for people. The industry remains in the early stages of its maturity model, where human operators still guide and supervise automated systems. Over time, their role will shift toward higher-level oversight, with AI handling more of the routine detection and corrective actions. This evolution requires responsible technology development, transparency, and close coordination among vendors, hyperscalers, and AI partners. Ensuring that automated systems do not act as “black boxes” is central to maintaining trust and reducing operational risk.
As the ecosystem advances, automation and AI will enhance human intelligence. Watch this insightful interview with Kal De, Senior Vice President of Product & Engineering at Nokia, to discover how the next phase of telecom innovation will focus on building value on top of the network, not just maintaining it.
Steve Saunders:
Let's talk about modern telecom networks. They're getting really complicated, aren't they? What's driving that complexity, and where do you think your operator customers are feeling the pressure the most?
Kal De:
That's collectively complexity that is almost hard to comprehend, let alone manage. It's certainly hard to visualize, hard to keep track of. And so yeah, we are at a point in time where it's no longer manageable through human trial. I think that opens up some interesting opportunities for what we vendors and operators need to do to work together.
Steve Saunders:
Automation is a big part of the solution to that problem. Do you agree with that?
Kal De:
I think we're at a point where automation has become essential, if not almost existential, or at least it's heading there. For a large operator, a CSP, it's probably 70% of their TCO and human-driven assurance activity to keep the network up, running, stable, reliable.
With automation, it is an opportunity to detect the potential of something going wrong, to be able to reason across multiple domains, to correlate the various salient symptoms of whether it's an anomaly, an outage, something that looks like a cyber incident, and then be able to ideally take corrective measures, remediate before the subscriber's experience is impacted.
It sounds like common sense, but at the scale of the combinatorial explosion of functions, devices, like the complexity of traffic, not possible to do without automation at this time.
Steve Saunders:
What's the endpoint of the vision with automation though? Is it to get rid of the humans, or will there be a human in the loop?
Kal De:
There will always be a human in the loop. The question is where and in what capacity? If one considers AI and the maturity model of autonomous networks as depicted by the TM Forum, as an example, their model has got these five levels.
I think the industry, even sophisticated mature operators, are somewhere between a level two and a level three, where there is still hands-on keyboard involvement, and pretty significantly so. I think the first step is going to be the role of the operator, be it in the SOC, be it in the NOC, changing much more to being initially a guide, and then eventually to more of an observer.
Then of course, we're all heading towards hopefully that Nirvana state where the human intelligence augments AI intelligence, and not the other way around.
Steve Saunders:
I think that Nokia has a responsibility actually to slow things down a little bit because you act as an ombudsman between the carriers, enterprises, and industries that are being tasked with rolling out these technologies.
Kal De:
Very much so. I think that's a very fair way of describing it. I think you make a great point. Yes, if we are delegating this to a black box, that is a very dangerous thing to do. We are doing the best we can as a vendor. I mentioned the companies we're partnering with, we are also deeply partnering with AI vendors, not just the hyperscalers or the frontier model providers, but also with leading edge startups and trying to explain what we are doing, where the sensitivities lie, and what technologies can we use in AI ops and ML ops to mitigate and mediate that risk more effectively.
Steve Saunders:
I have to say Nokia's having an amazing year, isn't it? You've got a new CEO, you've got all these deals coming together. You're optimistic about next year?
Kal De:
It promises to be an enormously interesting time ahead. I think we feel like we are set up to work well together with our ecosystem and hopefully really deliver meaningful value to the segments in which we operate in the industry in telco, again, on multiple fronts. And really collectively, we all focus on building value on top of the network versus being focused on the network itself exclusively.