Ericsson sees AI agents reshaping mobile network demand

Ericsson ConsumerLab has published research on how consumer use of AI could reshape demand on mobile networks. It found that 38% of smartphone users expect to use agentic AI by 2030.

The findings are based on a survey of more than 43,000 smartphone users aged 15 to 69 across 27 markets, representing about 1.7 billion smartphone users worldwide.

At present, about 4% of smartphone users use agentic AI tools, defined in the research as services that handle tasks normally carried out by the user. Respondents predicted a steep rise in adoption by the end of the decade, with common uses expected to include scheduling and coordination, comparing products, managing recurring purchases, and health monitoring.

The report also points to broader use of multimodal AI, as consumers combine text, voice, image, and other forms of interaction in a single service. In Australia, multimodal AI use is expected to rise from 6% to 23% by 2030.

Daily use is also expected to become more common. According to the research, two-thirds of smartphone users are expected to interact with AI every day by 2030. Users also said agentic AI could cut their screen time by as much as one hour a day.

Heavy users

One group stands out in the study: AI super users. About 10% of smartphone owners globally currently fall into that category, described as early adopters who engage heavily with multimodal AI.

That share is expected to grow to about 30% by 2030. The report said this group would account for 80% of all AI activity, underlining how a relatively small segment of users could drive a large share of network demand.

Current AI super users already spend about 2.5 times longer using AI apps and services than the average smartphone user, according to the research. By 2030, that figure is expected to rise to around four times longer.

For telecommunications providers, the study frames AI agents as a new category of network activity that may operate in the background rather than through direct, visible use on a screen. That could shift traffic patterns as software carries out searches, comparisons, bookings, and other tasks on behalf of users.

Rapid response times will be important to user satisfaction as consumers come to rely on AI in everyday settings at work and at home. The report also highlighted expectations around AI agent identity, mobile network security, and the role communications service providers should play.

Jasmeet Singh Sethi offered one view of how early usage patterns may develop. "We have seen many times that early adopter behavior and expectations - in this case the AI super users - are a strong indicator of much wider-scale adoption to follow. We believe this will be the same with the use of agentic AI," said Jasmeet Singh Sethi, Head of Ericsson ConsumerLab, Ericsson.

The study argues that growth in multimodal and agentic AI will affect not only traffic volumes but also the nature of network activity. Much of that activity may happen without a user actively navigating an app or website, as digital agents perform tasks continuously or at set intervals.

That matters because conventional mobile usage has tended to be shaped by visible consumer actions such as streaming, messaging, or browsing. Agentic AI could introduce a different pattern, with more automated interactions taking place in the background and placing fresh demands on latency, reliability, and the secure handling of transactions.

Research from Ericsson ConsumerLab also suggests that user perceptions of connectivity already influence the experience of AI super users. That gives service providers another reason to examine how networks cope with services that depend on quick data exchanges and a more constant flow of requests.

The 27 markets covered by the survey were Australia, Brazil, Canada, Chile, China, France, Germany, Greece, India, Indonesia, Ireland, Italy, Japan, Malaysia, Mexico, the Philippines, Qatar, Saudi Arabia, Singapore, South Korea, Spain, Sweden, Taiwan, Thailand, the United Arab Emirates, the United Kingdom, and the United States.

Sethi said the network effect of broader AI use would extend beyond visible consumer experience. "Multimodal and agentic AI growth will not only create network quality expectations for consumers in terms of their experience but will also result in significantly increased background actions in the networks that need to be handled securely and efficiently," said Sethi.