Indian operators well placed as early adopters of AI in networks: Ekudden

India's rapid 5G rollout is powering early AI adoption in telecom networks, giving local operators a global edge in performance, efficiency and innovation

India is at the forefront of AI adoption in telecom networks with operators in a strong position compared to the rest of the world, says Erik Ekudden, senior vice president and chief technology officer, Ericsson. In an interview with Business Standard’s Gulveen Aulakh in New Delhi, he said India, which is continuing to invest in high performing 5G networks, will benefit by attracting more innovation on top of those networks. Edited excerpts: 

What’s the outlook of AI adoption and readiness, when comparing Indian telcos with global telcos? 

The fact that India is at the forefront of 5G, with 5G adoption reaching some 99.6% of all Indian districts already, is very significant. I think 5G is well established as a digital platform — a fabric for India — and that means India is now ready to take the next step to adopt AI widely. Not just AI in data centers, but AI for consumers, businesses, governments, and beyond. From that point of view, India is in a good place. Now, on AI adoption in networks, our customers here — Jio, Airtel, and Vodafone Idea — are already enjoying some of the benefits of AI in the network because we have implemented it in our products. We talk about it as a journey toward autonomous networks. These networks benefit in terms of better performance, improved energy efficiency, enhanced user experience, and better customer care through AI in the network. I think the Indian operators are in a strong position compared to the rest of the world, as early adopters of AI in networks. 

How do you see the adoption of 5G Advanced evolving among Indian telcos from this point onward? 

When we talk about 5G Advanced, we are referring to capabilities such as network slicing and the ability to offer dedicated, differentiated services. For example, financial institutions, healthcare providers, agricultural enterprises, and manufacturing facilities may require specific performance characteristics. In advanced manufacturing environments you need the latest 5G standalone architecture, combined with network slicing and programmable network APIs. These enable operators to tailor quality of service, support capabilities like precise indoor positioning, and enhance areas such as security and fraud management. Given the pace at which India has deployed 5G and its broader digital ecosystem maturity, the country is well positioned to move toward 5G Advanced at an early stage. 

What is your perspective on how AI-driven traffic will reshape telecom networks in the coming years? To what extent do you see this transformation altering network architecture and operations? 

 

Our expectation is that AI-driven use cases will significantly change traffic patterns. In particular, uplink traffic — data sent from devices such as robots, smartphones, AI glasses, or sensors to the network — is likely to grow faster than downlink traffic. Over the next five years, we estimate that globally, uplink traffic could grow by roughly three times on average, while downlink traffic may grow by around two times. This shift means networks must be designed with stronger uplink capabilities. Operators will need technologies optimised not only for downlink capacity, but also for uplink performance and low latency. 

 

At Ericsson, we have developed solutions specifically aimed at improving uplink efficiency and user experience. One example is our FDD Massive MIMO (often referred to as FD-MIMO) technology. This is an advanced beamforming solution for lower-frequency FDD bands, which enhances spectrum efficiency and improves both uplink coverage and capacity — effectively extracting more performance from existing spectrum assets.

 

Looking longer term, we believe AI-enabled wearables, robotics, and intelligent devices will not only proliferate over the next three to five years but also unlock new waves of innovation. To support this, Ericsson is helping open up networks through standardised network APIs. These APIs enable developers and enterprises to access capabilities such as enhanced uplink performance, quality of service controls, improved security features, and precise positioning.

 

From a network deployment point of view, and given that Vodafone wants to increase its capex into networks going forward, does the outlook for the Indian industry, with a better-placed third player, look positive?

 

I will not comment on individual customers. However, what we are seeing is that markets like India, which continue to invest in high performing 5G networks, will benefit by attracting more innovation on top of those networks. Not all countries are investing at the same pace. If you look at the overall market outlook for RAN and Core over the coming years, it remains relatively flat globally, with around 1 to 2 per cent growth. That is the broader industry context. Within that, however, certain markets will outperform. Use cases such as enterprise 5G and mission critical networks, for example railways in India and other industrial applications, are expected to grow at a healthy rate in the coming years. Markets that continue investing in network quality and coverage will be well positioned to capture that growth.

 

Given the significant R&D and your collaborations with IITs, to what extent can the AI research and products being developed in India be integrated into telecom networks or end-use solutions?

 

To a very large extent, the work done in India is integrated into our global portfolio. We have a strong R&D presence in Bengaluru, Chennai, and Delhi. Innovations developed here become part of our offerings both in India and internationally. We highly value our collaborations with institutions such as IIT Madras and IIT Kanpur, as well as other leading universities. These partnerships help us nurture talent who may later contribute to Ericsson’s R&D efforts. They also support thought leadership in areas such as responsible AI. For example, we are a partner of CERAI (Centre for Responsible AI) at IIT Madras in Chennai. Such initiatives are important given India’s strong ambitions and growing global role in AI, as reflected in forums such as this AI summit. The broader objective is the diffusion of AI, meaning moving beyond centralised data centers and extending AI capabilities to end users: consumers on smartphones, farmers in the field, doctors in hospitals, and enterprises across sectors.

 

How do you see upcoming products evolving to help streamline AI-driven traffic? Do you expect these innovations to be primarily integrated into end-user devices and applications, or will they largely remain embedded within the network infrastructure?

 

These innovations are fundamentally network capabilities. We integrate them into our products and solutions, which are deployed by leading operators in India and around the world. At the same time, we work very closely with device manufacturers — whether at the chipset level or with the device makers themselves. In fact, we will be showcasing several of these partnerships at the upcoming Mobile World Congress. Close collaboration between devices — which increasingly carry AI applications — and the network is essential to delivering a seamless end-to-end user experience. That said, many of the core enhancements happen on the network side. Improvements such as lower latency, stronger uplink performance, better reliability, and improved energy efficiency are primarily enabled through network infrastructure. There is, of course, a tight interaction between the network and devices. Some of these network capabilities also help reduce power consumption and battery drain in devices such as smartphones or AI-enabled wearables. 

 

By optimising signalling, traffic management, and radio performance, the network can contribute to longer battery life and a better overall user experience.

 

Going forward, would Ericsson consider pursuing multiple chip partnerships? We’ve seen several collaborations in the ecosystem, particularly with Nvidia, while Ericsson has largely worked with Intel and is also developing its own stack. 

 

Within the network domain, we have already taken significant steps to decouple hardware from software. This architectural approach allows us to work with multiple hardware and chipset partners. To your point, we have long-standing collaborations with Intel. We have also demonstrated — about a year and a half ago — the ability to run our solutions on Nvidia platforms. In addition, we have announced prototypes and collaborations involving AMD and ARM-based platforms, among others. Because this flexibility is built into our software code base, we can support different underlying hardware infrastructures depending on customer requirements. That gives us optionality and the ability to adapt to evolving ecosystem dynamics. When we announced our collaboration with US operator T-Mobile around a year and a half ago, that initiative was together with Nvidia. So, we have already demonstrated a multi-partner approach in practice.

 

How do you see large data centre investments coming to India impacting fiber or network opportunities? 

 

There is no doubt that increased data center investments will generate more traffic across networks. Large-scale data center build-outs will drive demand for both mobile technologies, such as 5G, and fixed infrastructure, including fiber and backbone networks. However, the most important trend to watch is how AI workloads evolve. In the initial phase, most AI capability is concentrated in centralised data centers, primarily focused on model training and early-stage inference. But as the industry matures, we expect more inference workloads to move closer to the edge — into telecom networks, enterprise environments, stadiums, campuses, and even onto end-user devices such as smartphones, laptops, and AI-enabled wearables. In other words, the shift will be from a training-optimised architecture to a more inference-optimised, distributed infrastructure. This transition will increase the need for what we describe as an ‘intelligent fabric’ — networks capable of efficiently connecting distributed compute resources, supporting low latency, high performance, and advanced connectivity across the ecosystem. For India, this presents a significant opportunity over the next three, five, and even 10 years. The country is currently seeing strong momentum in centralised data center build-outs, but the longer-term value creation will increasingly come from distributed, edge-enabled infrastructure built on advanced connectivity.

 

What do you think are the top two AI-led applications that are likely to grab most attention of consumers as well as enterprises?

 

The first is video-rich AI applications. As more devices, such as smart glasses, cameras, sensors, and industrial equipment, capture real-time visual data, that information will need to be processed through AI inference. This will generate substantial uplink traffic from devices to the network. These capabilities can enable services such as visual recognition in retail environments, enhanced safety and security in urban settings, industrial monitoring, and operational optimisation. Both enterprises and public-sector environments can benefit from these AI-enabled, vision-based services.

 

The second category relates to AI-augmented human interaction, particularly through voice and language. AI-powered glasses, wearables, and smartphones that support voice input, speech recognition, and real-time language translation could become highly valuable for consumers and enterprises alike. For a country like India — with its linguistic diversity — real-time translation and contextual voice services could be especially impactful. 

 

More broadly, AI-enabled communication services processed through edge inference can significantly enhance everyday experiences. Once users become accustomed to these AI-augmented services — whether through wearables, glasses, or smartphone applications — I believe they will see them as indispensable. These kinds of AI-enhanced human services are likely to remain with us for a long time.