Industry Odisha Bureau, Oct 11: Ericsson estimates AI could lift existing 5G performance by 25–30% through software. Operators may avoid immediate hardware spending. For Jio and Airtel, it offers a bridge towards 6G.
Artificial intelligence (AI) could enhance existing 5G network performance by 25–30%, according to Ericsson. The Swedish telecom equipment maker believes the gains can come through software. Compatible hardware already installed in networks could run these upgrades. That may spare operators an immediate round of fresh hardware spending. For Reliance Jio and Bharti Airtel, the opportunity is significant. It offers a way to extract more value from costly 5G rollouts. The figures, however, remain Ericsson’s estimate rather than independently measured results.
Upgrades through software
Erik Ekudden, Ericsson’s chief technology officer, outlined the approach to The Economic Times. He said a large share of equipment supplied in 2025 and 2026 supports AI features. Such equipment can gain new capabilities through software updates over time. Customers, he said, “do not necessarily incur an additional hardware cost.”
Results will vary, however. The age of installed equipment matters. So do the specific features each operator wants to deploy. The software itself is not free. Upgrades must still be procured and deployed across networks.
Making 5G investments pay
The economics matter for India’s telecom industry. Operators have invested billions of dollars in nationwide 5G networks since 2022. They now face pressure to monetise that infrastructure. At the same time, they must prepare for the next generation of mobile technology. Software-led upgrades could offer a cheaper path to better network utilisation. Instead of replacing equipment, operators could improve what they already own. Whether either operator adopts these upgrades remains to be seen.
AI before 6G
Ekudden argues operators need not wait for the next network generation. In his words, operators “do not need to wait for GPUs or 6G.” AI can already run within the Radio Access Network, or RAN. The RAN links mobile phones to the wider network through cell sites and antennas. New features and models can be added there progressively through software. This could move networks gradually towards more autonomous operations. Such autonomy is a direction of travel, not a finished reality.
Software also has limits. Newer hardware will eventually be needed for additional frequency bands and 6G capabilities.
Smaller models, tighter safeguards
Ekudden also set telecom AI apart from large general-purpose systems. He contrasted it with large models from firms such as OpenAI and Anthropic. Telecom networks, he said, need smaller, specialised models for specific tasks. Reliability is the priority, since these models help run critical infrastructure.
Deeper AI integration also brings cybersecurity concerns. Advanced models could behave unpredictably or be misused. Ericsson says it is stress-testing its systems against emerging threats. It is also deploying safeguards to limit unpredictable AI behaviour. Such measures reduce risk but cannot guarantee complete protection. Ekudden said the AI industry needs “a degree of self-control” over autonomous agents. Otherwise, he warned, “rogue agents” could be unleashed on society.
India’s 6G window
Looking further ahead, Ekudden sees India as a likely early 6G adopter. Adoption could come around 2029 or 2030, he said. That depends on the broader device ecosystem maturing in time. It is an industry view, not an official launch schedule. Until then, AI-powered 5G networks in India offer room for improvement. Today’s software upgrades could shape how operators approach tomorrow’s investment cycle.
