MegaFon Estimates That Up to 40% of Telecom Operators' Routine Tasks Will Go to AI Within Three Years

MegaFon forecasts that in three years, around 2029, up to 40% of telecom operators' routine service and operations tasks could be handled without human intervention. CNews reports this. The estimate applies to the industry as a whole, not to the internal plans of a single operator. It is a benchmark, not a commitment: the available portion of the article does not disclose the calculation methodology.

Which tasks are meant

The wording splits routine work into two buckets. Service tasks involve working with subscribers: questions about tariffs and balance, connecting and disconnecting services, and first-line diagnostics of connectivity problems. Operations tasks involve working with the network itself: triaging alarms, monitoring, standard configuration changes, and processing requests between departments.

The two share one trait: the tasks repeat, follow clear rules, and leave a digital trail. That kind of routine is the easiest to automate. The word "typical" in the forecast matters: it does not cover non-standard outages or complex, disputed requests.

How realistic is the number

The idea itself is not new. Operators have been automating network operations for decades: self-organizing networks, automated alarm handling, scripts for standard changes. The industry association TM Forum describes autonomous networks as a scale of levels, from manual operation to fully autonomous operation. Language models and agents add what was previously missing: the ability to understand unstructured requests and to work across several systems to see a task through to completion.

The "up to 40%" figure reads as a ceiling for the most advanced players, not as an industry-wide average. Three years is a short time in telecom, where an AI agent has to work on top of billing and network-support systems accumulated over decades.

Key facts and unknowns:

• Forecast horizon: three years, around 2029.
• Share of automation: up to 40% of typical tasks.
• Scope: service and operations processes of telecom operators.
• Source of the estimate: MegaFon, as reported by CNews.
• Not disclosed: the calculation base, the number of tasks in the sample, and the impact on headcount and costs.

Where automation hits a ceiling

The first obstacle is integration. An agent is useful only if it has access to billing, CRM, monitoring, and inventory systems. These systems were often built at different times and fit together poorly. The second is data quality: an agent will reproduce errors in reference data and equipment descriptions quickly and at scale. The third is accountability. If automation wrongly disconnects a service or changes a network configuration, there must be a clear rollback and audit mechanism. The fourth is regulation: handling subscribers' personal data imposes requirements on where and how models run.

What to do about it

Automating routine work does not, by itself, mean headcount cuts. The typical scenario looks different: people move away from mass requests toward exceptions, complex cases, and quality control of the agents' work. A new function emerges: analyzing the automation's errors.

For operators, the practical takeaway is to start with processes that have clear metrics, such as the share of requests resolved without handoff to a human agent and the time to resolve typical incidents, as well as the ability to roll back safely. For solution vendors, the key question is depth of integration with existing systems, not the quality of a model demo. For subscribers, the expected effect is simple: routine questions get resolved faster, while access to a live specialist for complex cases should remain. The forecast can be checked against operators' financial reporting: if automation is real, it will show up in the cost of serving a subscriber and in the speed of resolving outages.