As the global conversation on artificial intelligence shifts from experimentation to real-world deployment, Ericsson’s Chief Technology Officer Erik Ekudden believes the next disruption will not come from consumer-facing applications alone—but from how AI fundamentally reshapes digital infrastructure itself.

Speaking to CNBC-TV18 at the World Economic Forum 2026 in Davos, Ekudden said the rise of agentic AI, where multiple AI agents collaborate across complex workflows, will dramatically alter network traffic patterns and force a rethinking of how telecom networks are designed, built and operated.

“When agents start talking to each other and creating bigger solutions, that’s where the real change happens,” Ekudden said. “A single workflow could involve tens or even hundreds of agents interacting simultaneously. That completely changes the nature of network traffic.”
From apps to infrastructure

So far, much of the AI narrative has centred on automating tasks or improving productivity within enterprises. Ekudden argues that agentic AI marks a step-change—connecting systems end-to-end rather than optimising isolated functions.

This shift, he said, has direct implications for telecom networks. Instead of largely one-directional traffic—such as content flowing from the cloud to users—agentic AI creates constant, multi-directional data exchanges between devices, sensors, wearables and cloud-based AI systems.

“Will AI traffic only be downlink, from an AI agent to a device or AI glasses?” Ekudden asked. “Or will it be the other way around, where massive amounts of data are sent from sensors and devices up to the cloud? In reality, it will be both—and that changes how networks need to be built.”

The rise of the ‘intelligent fabric’

According to Ekudden, agentic AI will make networks far more “chatty”, with continuous communication between machines. This requires what he describes as an “intelligent fabric”—networks that can dynamically understand, prioritise and manage different types of traffic in real time.

“Not all traffic is created equal,” he said. “Networks need awareness and intelligence built into them so they can optimise performance, latency and reliability depending on the use case.”

Ericsson is already moving in this direction through AI-powered autonomous 5G networks. These systems use embedded AI to optimise performance, reduce energy consumption and accelerate service deployment—capabilities that will become even more critical as agentic AI scales.

Why 5G matters more than ever

While the industry often frames agentic AI as a future, Ekudden stressed that the foundation is being laid today through 5G standalone networks. AI-powered 5G, he said, enables the low latency, high reliability and strong uplink performance required for AI-driven workflows.

“You don’t need to wait for 6G to benefit from AI,” Ekudden said. “AI-powered 5G is already delivering real value, and it is the starting point for what comes next.”

As agentic AI use cases expand across sectors—from smart manufacturing and logistics to transportation and retail—the demands on networks will only increase, particularly on uplink capacity and responsiveness.

A clear path to AI-native 6G

Looking ahead, Ekudden described the transition to 6G as an evolution rather than a disruption. The work being done today on AI-powered 5G will flow naturally into AI-native 6G, where intelligence is embedded across every layer of the network.

“The journey towards 6G really starts with 5G standalone today,” he said. “From there, you have a smooth upgrade path as networks become more natively intelligent.”

This evolution will be essential to support large-scale agentic AI deployments, where networks must operate almost as active participants in digital workflows rather than passive data pipes.

Also Read | AI, telecom infra vital for each other, concerns remain on AI-based system’s impact: Trai Chair

Why this matters for India and beyond

Ekudden noted that countries with advanced, scalable 5G networks—such as India—are particularly well positioned to lead this transition. Large-scale deployments provide real-world learning environments where new AI-driven network capabilities can be tested, refined and commercialised.

For telecom operators globally, the message is clear: the value of networks in the AI era will increasingly depend on intelligence, programmability and adaptability—not just coverage or speed.

Below is the excerpt of the interview.

Q: AI continues to be one of the dominant themes at World Economic Forum 2026 in Davos. CEOs are talking about how they intend to use it and what kind of investments they intend to make. And at Ericsson, it has been a few years of using AI in an embedded way. Take me through what the experience has been like, but more importantly, what you see now in the future.

Ekudden: I think it’s a good framing that AI is already powering 5G networks today. But if I think back to the discussion here in Davos, it has very much been about the ROI of AI and the diffusion of AI. I think those are hugely important because the most advanced companies are now getting returns on their investments, and we are starting to see more AI applications and agentic flows permeating throughout enterprises. But I think the discussion that needs to be connected to that is how you actually get AI to the end users, or how you get data from sensors. Advanced connectivity is really what will power the future of AI, and that discussion is just starting right now.

To your point about AI in the network, we are already powering 5G with AI to create a better user experience, lower energy consumption, and also to be able to create more services faster. So AI in terms of autonomous networks, autonomous 5G networks, is actually already deployed in the most advanced networks today.

Q: You said that ROI, of course, is the big question, and that continues to be a question mark for many companies at this point in time. But is it working for you? Are you already starting to see returns?

Ekudden: We see that, and I think we have been fortunate to start very early in our own journey. So AI in R&D is already paying off. It is changing the way that we develop systems. Of course, developer experience is critically important for many talented engineers, many of them actually in India, as we talked about. This is really a journey that we started many years ago, and now, over the last year or so, we are starting to see significant productivity improvements, which is great. But more importantly, when you change the complete workflow, when you change the way you are working—not just coding and testing, but everything—it really changes how fast we can come to market with the latest products.

So yes, AI is already having an impact on us. And then, of course, in finance, in sales, in customer care, you see these kinds of use cases really starting to take off.

Q: The conversation has now moved towards agentic AI and what that could potentially mean. Many believe that this is a game changer for many sectors. I know you are a firm believer in that as well.

Ekudden: I am, because I think that is when we can start to go from very simple tasks that need to be automated—that is, of course, the first step—to connecting things in workflows. When agents start talking to each other and creating bigger solutions, that is really where the big change comes.

But I think it also has a big impact on the digital infrastructure itself. Ericsson builds mobile network infrastructure around the world. 50% of the world’s traffic outside of China passes through networks that we deliver. So of course, it matters a lot in terms of where that AI traffic will go. Will it only be downlink traffic from an AI agent down to a device, or AI glasses, or will it be more in the other direction, where you are sending data from sensors or from AI glasses up to the cloud? That changes traffic patterns. It changes how we need to dimension and build networks.

So AI has a huge impact on networks, especially agentic AI, because then you have chattiness between many agents. Think of a workflow that can consist of tens, if not hundreds, of different agents talking to each other. It becomes more like a fabric, a mesh of communication. I call it the intelligent fabric, because it needs to know about the traffic. Not all traffic is created equal. That is where we are going now with 5G standalone today, and then, of course, an upgrade with AI nativeness going forward.

Q: We are talking about 5G, but the move to 6G and India is at the forefront of that transition. Before I talk to you specifically about India, what does the move from 5G to 6G mean in an AI era?

Ekudden: Well, today, as I said, we are already building embedded AI into networks, creating a better user experience all the way up to customer care and so forth. That means AI-powered 5G is really the opportunity for our customers when it comes to the AI transition in society. You do not need to wait for 6G.

I think India has a fantastic starting point, with one of the absolutely best and biggest networks rolled out in the country. Once you have that better network—a better network for AI—you can start to create new services for society, AI-based services. And of course, you can create much better networks themselves.

So the journey towards the future 6G really starts with 5G standalone today, AI-powered 5G to get all these benefits, and then you have a relatively smooth upgrade path to AI-native 6G. It comes as a stepwise journey.

Q: Let’s talk about India. Ericsson has been a long-term believer in the India story and a partner to many telcos operating in India, including Bharti Airtel and others. How has the journey been so far for you? More importantly, as the pace of growth accelerates, what will it mean for Ericsson’s India plans?

Ekudden: You’re right. We have been supporting India for a very long time—more than 100 years—and this is, for us, the second-biggest market in the world. Being part of building out 5G at record speed a few years ago was a fantastic journey for us and, of course, for our customers. It is powering more or less all of India now with 5G.

It is also exciting to look at how it is growing over the coming five years. We are seeing that in perhaps less than five years, there will be over a billion 5G connections in India. So it is growing very fast, and that would mean that around 79–80% of connections will be on 5G in the relatively near future. Of course, it is a big and continuing upgrade of the networks, because you need coverage everywhere, better capacity, better uplink—all the things that I mentioned.

For us, it is really important to be working in India, but also to work within the Indian ecosystem with our customers, including the biggest customers in India and in the world. That gives us a lot of learnings. We can build better products and serve our customers even better. It also allows us to connect to the booming ecosystem, from startups to universities, that is leveraging better networks to create new AI applications—smart manufacturing, smart transportation, retail, and so on.

It also moves the value of what we are doing in India up the stack. As I mentioned, you need a better network for AI, but you also need to programme the network to actually get that better performance. So network programmability, network APIs, and creating a software layer on top of the network that enterprises can use—these are things we are really pioneering and driving in India. Working very closely with the Indian ecosystem is key, as is working with the Indian government.

Moving towards 6G, including the Bharat 6G initiative and the government’s efforts to pave the way, we are very much part of that, together with our partners and customers. India has set itself up for a very strong journey to 6G by having one of the strongest 5G networks today, upgrading with AI to AI-powered 5G and then to 6G. That vision and plan are very solid.

Watch accompanying video for entire conversation.