Mavenir launches telecom AI token billing infrastructure

Mavenir has launched a billing-grade infrastructure platform allowing telecom operators to monetise AI tokens directly through existing BSS (Business Support Systems).

Network operators face deep integration challenges when attempting to monetise large language model usage. Providers selling AI plans rely entirely on third-party cloud vendors to count tokens generated on external servers. This structural deficit prevents operators from enforcing usage quotas, exposing them to unbounded variable costs. Mavenir has released an infrastructure layer designed to close this metering gap, bringing AI token consumption directly into the operator Business Support Systems.

The Mavenir Digital Enablement platform counts input and output tokens applying the exact precision standards used for regulated data usage. Telecom companies integrate this capability without replacing their existing charging infrastructure. The billing path mirrors established data plan mechanisms, routing the network event through mediation, rating, and finally invoice generation. Network providers achieve billing-grade accuracy over token consumption, eliminating the financial risk associated with unmetered API calls to external language models.

Governing edge compute access and model routing

Controlling the cost of goods sold determines the viability of any network-delivered service. Routing all subscriber prompts to expensive frontier models guarantees commercial failure. Mavenir solves this economic barrier through an intelligent ‘Model Router’ positioned at the network edge. This router inspects incoming requests and evaluates them against the active subscriber profile and available processing resources.

Operators host efficient small language models within their own data centres to handle high-volume, low-complexity tasks. These models incur zero per-token licensing fees. When a user submits a complex reasoning task, the router evaluates the request, validates the subscriber tier, and forwards the prompt to external frontier models via secure APIs. Every transaction, regardless of the final processing location, registers within the operator metering system.

Telecom providers deploy the Mavenir platform to establish sovereign AI environments tailored to enterprise clients. The operator dedicates isolated compute resources and specific model instances to designated corporate accounts. Employees access the models using SIM-anchored identity authentication, providing a hardware-level security verification that public cloud applications cannot replicate.

Translating metered usage into consumer allowances

Consumer retail divisions struggle to market pure API access to mobile phone users. The Mavenir platform bridges this marketing divide by translating strict BSS metering into outcome-based consumer plans.

The billing engine tracks the exact byte size and processing cost of automated voicemail transcription or intelligent photograph edits. The operator packages these micro-transactions into recognisable consumer quotas, offering subscribers precise allowances based on specific applications. This translation layer ensures the network recovers the compute cost while presenting the customer with a highly tangible service benefit.

Enterprise billing models follow a separate, strictly quantitative path. CIOs typically allocate massive budgets toward external cloud providers using strict API metrics. Operators position their governed AI access directly against these existing corporate expenditures. Mavenir’s platform generates distinct invoices detailing exact usage across various model tiers. A corporate client receives a consolidated monthly bill demonstrating precisely how many queries routed through low-cost internal models versus external APIs.

Monetising physical network assets via distributed inference

Network operators maintain thousands of central offices and edge locations equipped with reliable power and cooling infrastructure. The Mavenir platform transforms these physical assets into distributed AI processing nodes.

Operators can install GPU clusters within these local facilities to create a vast, decentralised inference grid. Developers building latency-sensitive applications can then deploy their models directly onto this operator-owned hardware.

Third-party applications are able to execute code mere miles from the end user, driving response times down and keeping data in-country. The operator monetises this capability by packaging physical colocation, high-speed connectivity, and bare-metal compute resources into a unified commercial product. 

Operators build and deploy their own fine-tuned models using the Mavenir SLM Builder to run on this distributed hardware. Training smaller models on specific operator data sets yields high-performance tools capable of executing distinct network management or customer service functions. This deployment strategy converts unpredictable cloud service expenses into fixed and manageable hardware depreciation schedules.

See also: Verizon deploys Ericsson private 5G edge networks worldwide

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the IoT Tech Expo and Cyber Security & Cloud Expo. Click here for more information.

Telecoms is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.