The telecom industry is entering a new phase where connectivity alone is no longer enough.

According to recent industry discussions, companies like Mavenir are encouraging operators to rethink their role—from traditional network providers to potential AI service platforms. The idea is bold: instead of only selling bandwidth, telcos could eventually offer AI computing power and “AI tokens”, powered by distributed data centers and GPU infrastructure.

This shift is driven by several forces:

Rapid progress of open-source AI models narrowing the gap with leading platforms like OpenAI

Growing investment in AI infrastructure and GPUs from companies such as Nvidia

Increasing demand for low-latency, localized AI services at the edge of telecom networks

Under this vision, telecom operators could evolve into “AI grids” that support:

AI-as-a-service for enterprise customers

Edge-based inference for real-time applications (translation, robotics, VR)

New monetization models beyond traditional connectivity

However, the industry is still divided. While the opportunity is significant, questions remain about execution capability, business model maturity, and whether telcos are ready to compete in the AI value chain.

Our perspective:

As telecom infrastructure continues to evolve toward high-density, high-bandwidth, and low-latency networks, the foundation for AI-enabled services is being quietly built today—especially in fiber, data center interconnect, and high-speed optical transport.

The question is no longer if telecom will intersect with AI, but how fast and at what level of the stack.

Source:https://www.lightreading.com/ai-machine-learning/mavenir-urges-telcos-to-sell-ai-tokens-and-not-rely-on-openai