By Abigail Mbah
Telecommunications operators could unlock new revenue streams by managing and monetising artificial intelligence (AI) consumption, moving beyond traditional voice and data bundles to offer services measured in AI tokens, according to the Global System for Mobile Communications Association (GSMA).
In a new industry analysis, Kenechi Okeleke of GSMA Intelligence argues that the growing adoption of AI applications, autonomous agents and automated business workflows is creating demand for services that allow organisations and consumers to access, monitor and control AI usage.
Unlike conventional internet consumption measured in gigabytes, AI usage is increasingly calculated in tokens, which represent units of text or other information processed by AI models. Consumption varies depending on the model, task complexity and whether the tokens cover input, output, cached information or reasoning processes.
Okeleke said telecom operators, which have largely deployed AI to improve internal efficiency and reduce operating costs, should now position themselves to sell AI-related services directly to customers.
Operators are already exploring opportunities across three areas: infrastructure, AI models and enterprise solutions.
In infrastructure, companies are investing in data centres, cloud capacity, fibre networks and edge computing facilities to support AI factories, sovereign AI platforms and GPU-as-a-service offerings.
At the model level, some operators are developing or partnering on foundation models tailored to local languages and markets. However, enterprise AI solutions currently represent the most active area, with more than 200 telecom companies worldwide offering AI copilots, conversational tools, productivity applications and industry-specific services.
Asia-Pacific accounts for 43 per cent of global telecom operator involvement in AI infrastructure and 38 per cent at the model level.
China is emerging as an early market for token-based AI services. Average daily token consumption in the country reportedly surged from about 100 billion at the beginning of 2024 to 140 trillion by March 2026, driven by the rapid adoption of AI agents and coding assistants.
The country’s major telecom operators are pursuing different monetisation strategies.
China Telecom bundles AI model access with connectivity and cybersecurity services, while China Mobile combines tokens with cloud PCs and AI agent capabilities. China Unicom is testing a credit-based system that allows customers to manage token spending across multiple AI models.
However, Okeleke cautioned that China’s model may not be easily replicated in other markets, given its extensive cloud infrastructure, substantial AI investments and availability of open-weight models.
He argued that operators elsewhere must develop strategies that reflect their infrastructure capabilities, customer needs and competitive environments.
Beyond selling AI tokens, the GSMA sees a bigger commercial opportunity in the control layer, where telecom operators can help businesses select AI models, route workloads, manage costs and meet governance requirements.
This opportunity rests on three capabilities: providing access to multiple AI models through a single interface and billing system; intelligently routing workloads based on cost, quality, speed, availability and regulatory requirements; and optimising AI workflows to reduce operating expenses.
The analysis cited China Mobile’s MoMA platform, which has achieved reported token cost reductions of 20 to 30 per cent through intelligent routing, caching and context reuse.
AT&T has also reported internal savings of up to 90 per cent through its AI gateway, demonstrating how operators can establish the value of AI management within their own businesses before commercialising similar services.
According to Okeleke, telecom operators have several advantages, including established customer relationships, billing and identity management systems, cybersecurity expertise, regulatory experience and distributed network infrastructure.
However, they face growing competition from hyperscalers, cloud providers and specialised AI companies developing their own model management platforms.
He urged operators to identify their positions in the AI value chain, develop the necessary technical capabilities and pursue strategic partnerships rather than limit themselves to low-margin AI service resale.
Operators must also prepare their networks for changing traffic patterns, including increased uplink demand and distributed processing, while leveraging network APIs and edge computing to differentiate their offerings.
The GSMA warned that operators risk losing a significant share of the emerging AI economy to technology companies that control the models and service platforms if they remain focused solely on providing connectivity.